{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.19853157538350275,
  "eval_steps": 1000,
  "global_step": 14000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.4180826813107339e-05,
      "grad_norm": 23.5,
      "learning_rate": 2.3584905660377358e-07,
      "loss": 2.2462,
      "step": 1
    },
    {
      "epoch": 2.8361653626214678e-05,
      "grad_norm": 5.40625,
      "learning_rate": 4.7169811320754717e-07,
      "loss": 2.0729,
      "step": 2
    },
    {
      "epoch": 5.6723307252429356e-05,
      "grad_norm": 83.5,
      "learning_rate": 9.433962264150943e-07,
      "loss": 2.1368,
      "step": 4
    },
    {
      "epoch": 8.508496087864403e-05,
      "grad_norm": 11.625,
      "learning_rate": 1.4150943396226415e-06,
      "loss": 2.0221,
      "step": 6
    },
    {
      "epoch": 0.00011344661450485871,
      "grad_norm": 13.375,
      "learning_rate": 1.8867924528301887e-06,
      "loss": 2.3639,
      "step": 8
    },
    {
      "epoch": 0.00014180826813107338,
      "grad_norm": 10.625,
      "learning_rate": 2.358490566037736e-06,
      "loss": 2.0759,
      "step": 10
    },
    {
      "epoch": 0.00017016992175728805,
      "grad_norm": 30.375,
      "learning_rate": 2.830188679245283e-06,
      "loss": 2.5909,
      "step": 12
    },
    {
      "epoch": 0.00019853157538350273,
      "grad_norm": 9.0625,
      "learning_rate": 3.30188679245283e-06,
      "loss": 2.3101,
      "step": 14
    },
    {
      "epoch": 0.00022689322900971742,
      "grad_norm": 9.875,
      "learning_rate": 3.7735849056603773e-06,
      "loss": 2.3223,
      "step": 16
    },
    {
      "epoch": 0.00025525488263593207,
      "grad_norm": 7.4375,
      "learning_rate": 4.245283018867925e-06,
      "loss": 2.3394,
      "step": 18
    },
    {
      "epoch": 0.00028361653626214677,
      "grad_norm": 17.125,
      "learning_rate": 4.716981132075472e-06,
      "loss": 2.3398,
      "step": 20
    },
    {
      "epoch": 0.00031197818988836146,
      "grad_norm": 10.1875,
      "learning_rate": 5.188679245283019e-06,
      "loss": 2.1757,
      "step": 22
    },
    {
      "epoch": 0.0003403398435145761,
      "grad_norm": 19.625,
      "learning_rate": 5.660377358490566e-06,
      "loss": 2.0609,
      "step": 24
    },
    {
      "epoch": 0.0003687014971407908,
      "grad_norm": 8.5625,
      "learning_rate": 6.132075471698113e-06,
      "loss": 2.0856,
      "step": 26
    },
    {
      "epoch": 0.00039706315076700545,
      "grad_norm": 8.125,
      "learning_rate": 6.60377358490566e-06,
      "loss": 1.9317,
      "step": 28
    },
    {
      "epoch": 0.00042542480439322015,
      "grad_norm": 8.625,
      "learning_rate": 7.0754716981132075e-06,
      "loss": 2.028,
      "step": 30
    },
    {
      "epoch": 0.00045378645801943485,
      "grad_norm": 10.75,
      "learning_rate": 7.547169811320755e-06,
      "loss": 2.0806,
      "step": 32
    },
    {
      "epoch": 0.0004821481116456495,
      "grad_norm": 8.375,
      "learning_rate": 8.018867924528302e-06,
      "loss": 2.1792,
      "step": 34
    },
    {
      "epoch": 0.0005105097652718641,
      "grad_norm": 7.46875,
      "learning_rate": 8.49056603773585e-06,
      "loss": 1.9408,
      "step": 36
    },
    {
      "epoch": 0.0005388714188980789,
      "grad_norm": 7.21875,
      "learning_rate": 8.962264150943396e-06,
      "loss": 1.8844,
      "step": 38
    },
    {
      "epoch": 0.0005672330725242935,
      "grad_norm": 10.875,
      "learning_rate": 9.433962264150944e-06,
      "loss": 2.0355,
      "step": 40
    },
    {
      "epoch": 0.0005955947261505082,
      "grad_norm": 8.5625,
      "learning_rate": 9.905660377358492e-06,
      "loss": 1.6167,
      "step": 42
    },
    {
      "epoch": 0.0006239563797767229,
      "grad_norm": 7.1875,
      "learning_rate": 1.0377358490566038e-05,
      "loss": 1.6281,
      "step": 44
    },
    {
      "epoch": 0.0006523180334029376,
      "grad_norm": 7.21875,
      "learning_rate": 1.0849056603773586e-05,
      "loss": 1.5859,
      "step": 46
    },
    {
      "epoch": 0.0006806796870291522,
      "grad_norm": 6.03125,
      "learning_rate": 1.1320754716981132e-05,
      "loss": 1.6459,
      "step": 48
    },
    {
      "epoch": 0.0007090413406553669,
      "grad_norm": 5.1875,
      "learning_rate": 1.179245283018868e-05,
      "loss": 1.4428,
      "step": 50
    },
    {
      "epoch": 0.0007374029942815816,
      "grad_norm": 7.0,
      "learning_rate": 1.2264150943396227e-05,
      "loss": 1.4423,
      "step": 52
    },
    {
      "epoch": 0.0007657646479077963,
      "grad_norm": 3.65625,
      "learning_rate": 1.2735849056603775e-05,
      "loss": 1.4288,
      "step": 54
    },
    {
      "epoch": 0.0007941263015340109,
      "grad_norm": 3.984375,
      "learning_rate": 1.320754716981132e-05,
      "loss": 1.3693,
      "step": 56
    },
    {
      "epoch": 0.0008224879551602257,
      "grad_norm": 3.8125,
      "learning_rate": 1.3679245283018869e-05,
      "loss": 1.4592,
      "step": 58
    },
    {
      "epoch": 0.0008508496087864403,
      "grad_norm": 2.90625,
      "learning_rate": 1.4150943396226415e-05,
      "loss": 1.4144,
      "step": 60
    },
    {
      "epoch": 0.0008792112624126549,
      "grad_norm": 2.59375,
      "learning_rate": 1.4622641509433963e-05,
      "loss": 1.4228,
      "step": 62
    },
    {
      "epoch": 0.0009075729160388697,
      "grad_norm": 2.9375,
      "learning_rate": 1.509433962264151e-05,
      "loss": 1.4243,
      "step": 64
    },
    {
      "epoch": 0.0009359345696650843,
      "grad_norm": 3.5625,
      "learning_rate": 1.5566037735849056e-05,
      "loss": 1.4232,
      "step": 66
    },
    {
      "epoch": 0.000964296223291299,
      "grad_norm": 4.0625,
      "learning_rate": 1.6037735849056604e-05,
      "loss": 1.5609,
      "step": 68
    },
    {
      "epoch": 0.0009926578769175137,
      "grad_norm": 2.796875,
      "learning_rate": 1.650943396226415e-05,
      "loss": 1.3382,
      "step": 70
    },
    {
      "epoch": 0.0010210195305437283,
      "grad_norm": 2.46875,
      "learning_rate": 1.69811320754717e-05,
      "loss": 1.2712,
      "step": 72
    },
    {
      "epoch": 0.001049381184169943,
      "grad_norm": 3.703125,
      "learning_rate": 1.7452830188679244e-05,
      "loss": 1.3288,
      "step": 74
    },
    {
      "epoch": 0.0010777428377961578,
      "grad_norm": 3.953125,
      "learning_rate": 1.7924528301886792e-05,
      "loss": 1.4028,
      "step": 76
    },
    {
      "epoch": 0.0011061044914223723,
      "grad_norm": 2.859375,
      "learning_rate": 1.839622641509434e-05,
      "loss": 1.3597,
      "step": 78
    },
    {
      "epoch": 0.001134466145048587,
      "grad_norm": 2.90625,
      "learning_rate": 1.8867924528301888e-05,
      "loss": 1.4342,
      "step": 80
    },
    {
      "epoch": 0.0011628277986748018,
      "grad_norm": 2.671875,
      "learning_rate": 1.9339622641509436e-05,
      "loss": 1.2457,
      "step": 82
    },
    {
      "epoch": 0.0011911894523010164,
      "grad_norm": 5.0625,
      "learning_rate": 1.9811320754716984e-05,
      "loss": 1.4336,
      "step": 84
    },
    {
      "epoch": 0.001219551105927231,
      "grad_norm": 2.875,
      "learning_rate": 2.0283018867924532e-05,
      "loss": 1.2609,
      "step": 86
    },
    {
      "epoch": 0.0012479127595534459,
      "grad_norm": 2.578125,
      "learning_rate": 2.0754716981132076e-05,
      "loss": 1.2695,
      "step": 88
    },
    {
      "epoch": 0.0012762744131796604,
      "grad_norm": 2.8125,
      "learning_rate": 2.1226415094339624e-05,
      "loss": 1.2783,
      "step": 90
    },
    {
      "epoch": 0.0013046360668058751,
      "grad_norm": 3.46875,
      "learning_rate": 2.1698113207547172e-05,
      "loss": 1.309,
      "step": 92
    },
    {
      "epoch": 0.00133299772043209,
      "grad_norm": 3.125,
      "learning_rate": 2.216981132075472e-05,
      "loss": 1.3988,
      "step": 94
    },
    {
      "epoch": 0.0013613593740583044,
      "grad_norm": 2.9375,
      "learning_rate": 2.2641509433962265e-05,
      "loss": 1.327,
      "step": 96
    },
    {
      "epoch": 0.0013897210276845192,
      "grad_norm": 4.25,
      "learning_rate": 2.3113207547169813e-05,
      "loss": 1.303,
      "step": 98
    },
    {
      "epoch": 0.0014180826813107337,
      "grad_norm": 3.828125,
      "learning_rate": 2.358490566037736e-05,
      "loss": 1.2823,
      "step": 100
    },
    {
      "epoch": 0.0014464443349369485,
      "grad_norm": 20.5,
      "learning_rate": 2.405660377358491e-05,
      "loss": 1.3277,
      "step": 102
    },
    {
      "epoch": 0.0014748059885631632,
      "grad_norm": 4.34375,
      "learning_rate": 2.4528301886792453e-05,
      "loss": 1.3433,
      "step": 104
    },
    {
      "epoch": 0.0015031676421893778,
      "grad_norm": 5.84375,
      "learning_rate": 2.5e-05,
      "loss": 1.2733,
      "step": 106
    },
    {
      "epoch": 0.0015315292958155925,
      "grad_norm": 3.15625,
      "learning_rate": 2.547169811320755e-05,
      "loss": 1.269,
      "step": 108
    },
    {
      "epoch": 0.0015598909494418073,
      "grad_norm": 5.15625,
      "learning_rate": 2.5943396226415094e-05,
      "loss": 1.2991,
      "step": 110
    },
    {
      "epoch": 0.0015882526030680218,
      "grad_norm": 3.296875,
      "learning_rate": 2.641509433962264e-05,
      "loss": 1.2476,
      "step": 112
    },
    {
      "epoch": 0.0016166142566942366,
      "grad_norm": 3.890625,
      "learning_rate": 2.688679245283019e-05,
      "loss": 1.283,
      "step": 114
    },
    {
      "epoch": 0.0016449759103204513,
      "grad_norm": 3.21875,
      "learning_rate": 2.7358490566037738e-05,
      "loss": 1.2708,
      "step": 116
    },
    {
      "epoch": 0.0016733375639466658,
      "grad_norm": 3.5,
      "learning_rate": 2.7830188679245282e-05,
      "loss": 1.2617,
      "step": 118
    },
    {
      "epoch": 0.0017016992175728806,
      "grad_norm": 3.234375,
      "learning_rate": 2.830188679245283e-05,
      "loss": 1.2324,
      "step": 120
    },
    {
      "epoch": 0.0017300608711990953,
      "grad_norm": 3.046875,
      "learning_rate": 2.8773584905660378e-05,
      "loss": 1.317,
      "step": 122
    },
    {
      "epoch": 0.0017584225248253099,
      "grad_norm": 4.875,
      "learning_rate": 2.9245283018867926e-05,
      "loss": 1.2038,
      "step": 124
    },
    {
      "epoch": 0.0017867841784515246,
      "grad_norm": 3.375,
      "learning_rate": 2.971698113207547e-05,
      "loss": 1.319,
      "step": 126
    },
    {
      "epoch": 0.0018151458320777394,
      "grad_norm": 3.8125,
      "learning_rate": 3.018867924528302e-05,
      "loss": 1.2748,
      "step": 128
    },
    {
      "epoch": 0.001843507485703954,
      "grad_norm": 3.59375,
      "learning_rate": 3.0660377358490567e-05,
      "loss": 1.3204,
      "step": 130
    },
    {
      "epoch": 0.0018718691393301687,
      "grad_norm": 3.125,
      "learning_rate": 3.113207547169811e-05,
      "loss": 1.3213,
      "step": 132
    },
    {
      "epoch": 0.0019002307929563834,
      "grad_norm": 3.09375,
      "learning_rate": 3.160377358490566e-05,
      "loss": 1.2092,
      "step": 134
    },
    {
      "epoch": 0.001928592446582598,
      "grad_norm": 3.09375,
      "learning_rate": 3.207547169811321e-05,
      "loss": 1.196,
      "step": 136
    },
    {
      "epoch": 0.0019569541002088125,
      "grad_norm": 3.625,
      "learning_rate": 3.254716981132075e-05,
      "loss": 1.1678,
      "step": 138
    },
    {
      "epoch": 0.0019853157538350275,
      "grad_norm": 3.390625,
      "learning_rate": 3.30188679245283e-05,
      "loss": 1.2782,
      "step": 140
    },
    {
      "epoch": 0.002013677407461242,
      "grad_norm": 4.3125,
      "learning_rate": 3.349056603773585e-05,
      "loss": 1.3041,
      "step": 142
    },
    {
      "epoch": 0.0020420390610874565,
      "grad_norm": 3.53125,
      "learning_rate": 3.39622641509434e-05,
      "loss": 1.1766,
      "step": 144
    },
    {
      "epoch": 0.0020704007147136715,
      "grad_norm": 4.34375,
      "learning_rate": 3.4433962264150943e-05,
      "loss": 1.2957,
      "step": 146
    },
    {
      "epoch": 0.002098762368339886,
      "grad_norm": 3.546875,
      "learning_rate": 3.490566037735849e-05,
      "loss": 1.2442,
      "step": 148
    },
    {
      "epoch": 0.0021271240219661006,
      "grad_norm": 3.765625,
      "learning_rate": 3.537735849056604e-05,
      "loss": 1.2219,
      "step": 150
    },
    {
      "epoch": 0.0021554856755923156,
      "grad_norm": 3.96875,
      "learning_rate": 3.5849056603773584e-05,
      "loss": 1.2118,
      "step": 152
    },
    {
      "epoch": 0.00218384732921853,
      "grad_norm": 3.265625,
      "learning_rate": 3.632075471698113e-05,
      "loss": 1.2072,
      "step": 154
    },
    {
      "epoch": 0.0022122089828447446,
      "grad_norm": 2.90625,
      "learning_rate": 3.679245283018868e-05,
      "loss": 1.2434,
      "step": 156
    },
    {
      "epoch": 0.0022405706364709596,
      "grad_norm": 3.90625,
      "learning_rate": 3.7264150943396224e-05,
      "loss": 1.242,
      "step": 158
    },
    {
      "epoch": 0.002268932290097174,
      "grad_norm": 3.9375,
      "learning_rate": 3.7735849056603776e-05,
      "loss": 1.2218,
      "step": 160
    },
    {
      "epoch": 0.0022972939437233887,
      "grad_norm": 4.875,
      "learning_rate": 3.820754716981133e-05,
      "loss": 1.2711,
      "step": 162
    },
    {
      "epoch": 0.0023256555973496036,
      "grad_norm": 3.453125,
      "learning_rate": 3.867924528301887e-05,
      "loss": 1.2997,
      "step": 164
    },
    {
      "epoch": 0.002354017250975818,
      "grad_norm": 4.0,
      "learning_rate": 3.9150943396226416e-05,
      "loss": 1.2355,
      "step": 166
    },
    {
      "epoch": 0.0023823789046020327,
      "grad_norm": 3.171875,
      "learning_rate": 3.962264150943397e-05,
      "loss": 1.2066,
      "step": 168
    },
    {
      "epoch": 0.0024107405582282477,
      "grad_norm": 4.03125,
      "learning_rate": 4.009433962264151e-05,
      "loss": 1.2166,
      "step": 170
    },
    {
      "epoch": 0.002439102211854462,
      "grad_norm": 3.640625,
      "learning_rate": 4.0566037735849064e-05,
      "loss": 1.2211,
      "step": 172
    },
    {
      "epoch": 0.0024674638654806767,
      "grad_norm": 3.921875,
      "learning_rate": 4.103773584905661e-05,
      "loss": 1.1586,
      "step": 174
    },
    {
      "epoch": 0.0024958255191068917,
      "grad_norm": 4.15625,
      "learning_rate": 4.150943396226415e-05,
      "loss": 1.1605,
      "step": 176
    },
    {
      "epoch": 0.0025241871727331062,
      "grad_norm": 3.46875,
      "learning_rate": 4.1981132075471704e-05,
      "loss": 1.1742,
      "step": 178
    },
    {
      "epoch": 0.002552548826359321,
      "grad_norm": 3.9375,
      "learning_rate": 4.245283018867925e-05,
      "loss": 1.2395,
      "step": 180
    },
    {
      "epoch": 0.0025809104799855358,
      "grad_norm": 4.53125,
      "learning_rate": 4.292452830188679e-05,
      "loss": 1.1823,
      "step": 182
    },
    {
      "epoch": 0.0026092721336117503,
      "grad_norm": 3.75,
      "learning_rate": 4.3396226415094345e-05,
      "loss": 1.1712,
      "step": 184
    },
    {
      "epoch": 0.002637633787237965,
      "grad_norm": 3.296875,
      "learning_rate": 4.386792452830189e-05,
      "loss": 1.2018,
      "step": 186
    },
    {
      "epoch": 0.00266599544086418,
      "grad_norm": 2.90625,
      "learning_rate": 4.433962264150944e-05,
      "loss": 1.1264,
      "step": 188
    },
    {
      "epoch": 0.0026943570944903943,
      "grad_norm": 3.578125,
      "learning_rate": 4.4811320754716985e-05,
      "loss": 1.2231,
      "step": 190
    },
    {
      "epoch": 0.002722718748116609,
      "grad_norm": 3.328125,
      "learning_rate": 4.528301886792453e-05,
      "loss": 1.1657,
      "step": 192
    },
    {
      "epoch": 0.0027510804017428234,
      "grad_norm": 3.890625,
      "learning_rate": 4.575471698113208e-05,
      "loss": 1.2448,
      "step": 194
    },
    {
      "epoch": 0.0027794420553690384,
      "grad_norm": 4.21875,
      "learning_rate": 4.6226415094339625e-05,
      "loss": 1.2995,
      "step": 196
    },
    {
      "epoch": 0.002807803708995253,
      "grad_norm": 3.90625,
      "learning_rate": 4.669811320754717e-05,
      "loss": 1.1806,
      "step": 198
    },
    {
      "epoch": 0.0028361653626214674,
      "grad_norm": 4.46875,
      "learning_rate": 4.716981132075472e-05,
      "loss": 1.1848,
      "step": 200
    },
    {
      "epoch": 0.0028645270162476824,
      "grad_norm": 4.78125,
      "learning_rate": 4.7641509433962266e-05,
      "loss": 1.179,
      "step": 202
    },
    {
      "epoch": 0.002892888669873897,
      "grad_norm": 4.21875,
      "learning_rate": 4.811320754716982e-05,
      "loss": 1.2387,
      "step": 204
    },
    {
      "epoch": 0.0029212503235001115,
      "grad_norm": 4.9375,
      "learning_rate": 4.858490566037736e-05,
      "loss": 1.1862,
      "step": 206
    },
    {
      "epoch": 0.0029496119771263265,
      "grad_norm": 4.15625,
      "learning_rate": 4.9056603773584906e-05,
      "loss": 1.1575,
      "step": 208
    },
    {
      "epoch": 0.002977973630752541,
      "grad_norm": 5.40625,
      "learning_rate": 4.952830188679246e-05,
      "loss": 1.2318,
      "step": 210
    },
    {
      "epoch": 0.0030063352843787555,
      "grad_norm": 4.25,
      "learning_rate": 5e-05,
      "loss": 1.2164,
      "step": 212
    },
    {
      "epoch": 0.0030346969380049705,
      "grad_norm": 3.90625,
      "learning_rate": 4.9999999900161665e-05,
      "loss": 1.1131,
      "step": 214
    },
    {
      "epoch": 0.003063058591631185,
      "grad_norm": 4.5625,
      "learning_rate": 4.999999960064667e-05,
      "loss": 1.1882,
      "step": 216
    },
    {
      "epoch": 0.0030914202452573996,
      "grad_norm": 3.703125,
      "learning_rate": 4.9999999101455004e-05,
      "loss": 1.2517,
      "step": 218
    },
    {
      "epoch": 0.0031197818988836145,
      "grad_norm": 3.71875,
      "learning_rate": 4.999999840258668e-05,
      "loss": 1.1909,
      "step": 220
    },
    {
      "epoch": 0.003148143552509829,
      "grad_norm": 3.484375,
      "learning_rate": 4.9999997504041694e-05,
      "loss": 1.1448,
      "step": 222
    },
    {
      "epoch": 0.0031765052061360436,
      "grad_norm": 4.0625,
      "learning_rate": 4.999999640582007e-05,
      "loss": 1.1793,
      "step": 224
    },
    {
      "epoch": 0.0032048668597622586,
      "grad_norm": 3.703125,
      "learning_rate": 4.9999995107921805e-05,
      "loss": 1.1572,
      "step": 226
    },
    {
      "epoch": 0.003233228513388473,
      "grad_norm": 3.203125,
      "learning_rate": 4.999999361034691e-05,
      "loss": 1.1737,
      "step": 228
    },
    {
      "epoch": 0.0032615901670146876,
      "grad_norm": 5.28125,
      "learning_rate": 4.99999919130954e-05,
      "loss": 1.1441,
      "step": 230
    },
    {
      "epoch": 0.0032899518206409026,
      "grad_norm": 3.703125,
      "learning_rate": 4.9999990016167286e-05,
      "loss": 1.1727,
      "step": 232
    },
    {
      "epoch": 0.003318313474267117,
      "grad_norm": 4.0,
      "learning_rate": 4.999998791956258e-05,
      "loss": 1.2494,
      "step": 234
    },
    {
      "epoch": 0.0033466751278933317,
      "grad_norm": 4.125,
      "learning_rate": 4.999998562328131e-05,
      "loss": 1.1364,
      "step": 236
    },
    {
      "epoch": 0.0033750367815195467,
      "grad_norm": 3.421875,
      "learning_rate": 4.999998312732349e-05,
      "loss": 1.1748,
      "step": 238
    },
    {
      "epoch": 0.003403398435145761,
      "grad_norm": 4.1875,
      "learning_rate": 4.999998043168912e-05,
      "loss": 1.1553,
      "step": 240
    },
    {
      "epoch": 0.0034317600887719757,
      "grad_norm": 3.703125,
      "learning_rate": 4.999997753637825e-05,
      "loss": 1.1333,
      "step": 242
    },
    {
      "epoch": 0.0034601217423981907,
      "grad_norm": 3.71875,
      "learning_rate": 4.999997444139089e-05,
      "loss": 1.067,
      "step": 244
    },
    {
      "epoch": 0.0034884833960244052,
      "grad_norm": 3.5,
      "learning_rate": 4.9999971146727075e-05,
      "loss": 1.1889,
      "step": 246
    },
    {
      "epoch": 0.0035168450496506198,
      "grad_norm": 3.609375,
      "learning_rate": 4.999996765238682e-05,
      "loss": 1.1171,
      "step": 248
    },
    {
      "epoch": 0.0035452067032768347,
      "grad_norm": 2.921875,
      "learning_rate": 4.999996395837016e-05,
      "loss": 1.0563,
      "step": 250
    },
    {
      "epoch": 0.0035735683569030493,
      "grad_norm": 3.765625,
      "learning_rate": 4.9999960064677104e-05,
      "loss": 1.1624,
      "step": 252
    },
    {
      "epoch": 0.003601930010529264,
      "grad_norm": 5.65625,
      "learning_rate": 4.999995597130771e-05,
      "loss": 1.1688,
      "step": 254
    },
    {
      "epoch": 0.0036302916641554788,
      "grad_norm": 3.28125,
      "learning_rate": 4.9999951678262e-05,
      "loss": 1.1031,
      "step": 256
    },
    {
      "epoch": 0.0036586533177816933,
      "grad_norm": 2.953125,
      "learning_rate": 4.999994718554001e-05,
      "loss": 1.0506,
      "step": 258
    },
    {
      "epoch": 0.003687014971407908,
      "grad_norm": 3.671875,
      "learning_rate": 4.9999942493141766e-05,
      "loss": 1.1787,
      "step": 260
    },
    {
      "epoch": 0.003715376625034123,
      "grad_norm": 3.046875,
      "learning_rate": 4.999993760106732e-05,
      "loss": 1.1705,
      "step": 262
    },
    {
      "epoch": 0.0037437382786603374,
      "grad_norm": 4.4375,
      "learning_rate": 4.9999932509316706e-05,
      "loss": 1.165,
      "step": 264
    },
    {
      "epoch": 0.003772099932286552,
      "grad_norm": 3.53125,
      "learning_rate": 4.999992721788996e-05,
      "loss": 1.198,
      "step": 266
    },
    {
      "epoch": 0.003800461585912767,
      "grad_norm": 3.484375,
      "learning_rate": 4.999992172678713e-05,
      "loss": 1.1073,
      "step": 268
    },
    {
      "epoch": 0.0038288232395389814,
      "grad_norm": 3.109375,
      "learning_rate": 4.999991603600826e-05,
      "loss": 1.1029,
      "step": 270
    },
    {
      "epoch": 0.003857184893165196,
      "grad_norm": 3.71875,
      "learning_rate": 4.9999910145553386e-05,
      "loss": 1.1733,
      "step": 272
    },
    {
      "epoch": 0.0038855465467914105,
      "grad_norm": 3.609375,
      "learning_rate": 4.9999904055422565e-05,
      "loss": 1.1625,
      "step": 274
    },
    {
      "epoch": 0.003913908200417625,
      "grad_norm": 4.65625,
      "learning_rate": 4.999989776561584e-05,
      "loss": 1.116,
      "step": 276
    },
    {
      "epoch": 0.00394226985404384,
      "grad_norm": 4.84375,
      "learning_rate": 4.9999891276133274e-05,
      "loss": 1.1644,
      "step": 278
    },
    {
      "epoch": 0.003970631507670055,
      "grad_norm": 3.21875,
      "learning_rate": 4.99998845869749e-05,
      "loss": 1.1055,
      "step": 280
    },
    {
      "epoch": 0.003998993161296269,
      "grad_norm": 4.125,
      "learning_rate": 4.9999877698140783e-05,
      "loss": 1.147,
      "step": 282
    },
    {
      "epoch": 0.004027354814922484,
      "grad_norm": 4.21875,
      "learning_rate": 4.999987060963098e-05,
      "loss": 1.1561,
      "step": 284
    },
    {
      "epoch": 0.004055716468548699,
      "grad_norm": 3.796875,
      "learning_rate": 4.999986332144554e-05,
      "loss": 1.104,
      "step": 286
    },
    {
      "epoch": 0.004084078122174913,
      "grad_norm": 3.421875,
      "learning_rate": 4.9999855833584516e-05,
      "loss": 1.0773,
      "step": 288
    },
    {
      "epoch": 0.004112439775801128,
      "grad_norm": 3.390625,
      "learning_rate": 4.999984814604799e-05,
      "loss": 1.0895,
      "step": 290
    },
    {
      "epoch": 0.004140801429427343,
      "grad_norm": 3.546875,
      "learning_rate": 4.9999840258835994e-05,
      "loss": 1.0812,
      "step": 292
    },
    {
      "epoch": 0.004169163083053557,
      "grad_norm": 3.328125,
      "learning_rate": 4.999983217194862e-05,
      "loss": 1.0921,
      "step": 294
    },
    {
      "epoch": 0.004197524736679772,
      "grad_norm": 3.578125,
      "learning_rate": 4.999982388538591e-05,
      "loss": 1.1024,
      "step": 296
    },
    {
      "epoch": 0.004225886390305987,
      "grad_norm": 3.421875,
      "learning_rate": 4.999981539914794e-05,
      "loss": 1.0863,
      "step": 298
    },
    {
      "epoch": 0.004254248043932201,
      "grad_norm": 4.6875,
      "learning_rate": 4.999980671323479e-05,
      "loss": 1.2086,
      "step": 300
    },
    {
      "epoch": 0.004282609697558416,
      "grad_norm": 3.8125,
      "learning_rate": 4.99997978276465e-05,
      "loss": 1.1149,
      "step": 302
    },
    {
      "epoch": 0.004310971351184631,
      "grad_norm": 3.90625,
      "learning_rate": 4.999978874238316e-05,
      "loss": 1.0967,
      "step": 304
    },
    {
      "epoch": 0.004339333004810845,
      "grad_norm": 3.9375,
      "learning_rate": 4.9999779457444844e-05,
      "loss": 1.0695,
      "step": 306
    },
    {
      "epoch": 0.00436769465843706,
      "grad_norm": 3.109375,
      "learning_rate": 4.999976997283162e-05,
      "loss": 1.1057,
      "step": 308
    },
    {
      "epoch": 0.004396056312063275,
      "grad_norm": 3.265625,
      "learning_rate": 4.999976028854357e-05,
      "loss": 1.1259,
      "step": 310
    },
    {
      "epoch": 0.004424417965689489,
      "grad_norm": 3.0,
      "learning_rate": 4.999975040458076e-05,
      "loss": 1.0606,
      "step": 312
    },
    {
      "epoch": 0.004452779619315704,
      "grad_norm": 3.359375,
      "learning_rate": 4.9999740320943275e-05,
      "loss": 1.1007,
      "step": 314
    },
    {
      "epoch": 0.004481141272941919,
      "grad_norm": 3.34375,
      "learning_rate": 4.999973003763121e-05,
      "loss": 1.1371,
      "step": 316
    },
    {
      "epoch": 0.004509502926568133,
      "grad_norm": 3.828125,
      "learning_rate": 4.999971955464462e-05,
      "loss": 1.145,
      "step": 318
    },
    {
      "epoch": 0.004537864580194348,
      "grad_norm": 3.4375,
      "learning_rate": 4.999970887198361e-05,
      "loss": 1.1202,
      "step": 320
    },
    {
      "epoch": 0.004566226233820563,
      "grad_norm": 3.46875,
      "learning_rate": 4.999969798964825e-05,
      "loss": 1.1516,
      "step": 322
    },
    {
      "epoch": 0.004594587887446777,
      "grad_norm": 3.6875,
      "learning_rate": 4.9999686907638646e-05,
      "loss": 1.1018,
      "step": 324
    },
    {
      "epoch": 0.004622949541072992,
      "grad_norm": 4.34375,
      "learning_rate": 4.999967562595487e-05,
      "loss": 1.1226,
      "step": 326
    },
    {
      "epoch": 0.004651311194699207,
      "grad_norm": 4.375,
      "learning_rate": 4.999966414459701e-05,
      "loss": 1.0955,
      "step": 328
    },
    {
      "epoch": 0.004679672848325421,
      "grad_norm": 2.96875,
      "learning_rate": 4.999965246356517e-05,
      "loss": 1.047,
      "step": 330
    },
    {
      "epoch": 0.004708034501951636,
      "grad_norm": 3.734375,
      "learning_rate": 4.999964058285944e-05,
      "loss": 1.1238,
      "step": 332
    },
    {
      "epoch": 0.004736396155577851,
      "grad_norm": 3.65625,
      "learning_rate": 4.999962850247991e-05,
      "loss": 1.0768,
      "step": 334
    },
    {
      "epoch": 0.004764757809204065,
      "grad_norm": 4.375,
      "learning_rate": 4.999961622242669e-05,
      "loss": 1.1649,
      "step": 336
    },
    {
      "epoch": 0.00479311946283028,
      "grad_norm": 3.90625,
      "learning_rate": 4.999960374269986e-05,
      "loss": 1.1531,
      "step": 338
    },
    {
      "epoch": 0.004821481116456495,
      "grad_norm": 3.828125,
      "learning_rate": 4.999959106329952e-05,
      "loss": 1.1492,
      "step": 340
    },
    {
      "epoch": 0.0048498427700827094,
      "grad_norm": 3.984375,
      "learning_rate": 4.999957818422579e-05,
      "loss": 1.0841,
      "step": 342
    },
    {
      "epoch": 0.004878204423708924,
      "grad_norm": 3.609375,
      "learning_rate": 4.9999565105478755e-05,
      "loss": 1.0165,
      "step": 344
    },
    {
      "epoch": 0.004906566077335139,
      "grad_norm": 4.40625,
      "learning_rate": 4.9999551827058536e-05,
      "loss": 1.1411,
      "step": 346
    },
    {
      "epoch": 0.0049349277309613535,
      "grad_norm": 3.421875,
      "learning_rate": 4.999953834896521e-05,
      "loss": 1.174,
      "step": 348
    },
    {
      "epoch": 0.0049632893845875685,
      "grad_norm": 4.75,
      "learning_rate": 4.999952467119892e-05,
      "loss": 1.1168,
      "step": 350
    },
    {
      "epoch": 0.004991651038213783,
      "grad_norm": 3.953125,
      "learning_rate": 4.999951079375976e-05,
      "loss": 1.1348,
      "step": 352
    },
    {
      "epoch": 0.0050200126918399975,
      "grad_norm": 4.3125,
      "learning_rate": 4.9999496716647834e-05,
      "loss": 1.1131,
      "step": 354
    },
    {
      "epoch": 0.0050483743454662125,
      "grad_norm": 4.0625,
      "learning_rate": 4.999948243986325e-05,
      "loss": 1.1325,
      "step": 356
    },
    {
      "epoch": 0.0050767359990924275,
      "grad_norm": 3.578125,
      "learning_rate": 4.999946796340615e-05,
      "loss": 1.1337,
      "step": 358
    },
    {
      "epoch": 0.005105097652718642,
      "grad_norm": 3.453125,
      "learning_rate": 4.9999453287276624e-05,
      "loss": 1.151,
      "step": 360
    },
    {
      "epoch": 0.0051334593063448565,
      "grad_norm": 3.875,
      "learning_rate": 4.9999438411474794e-05,
      "loss": 1.1385,
      "step": 362
    },
    {
      "epoch": 0.0051618209599710715,
      "grad_norm": 3.71875,
      "learning_rate": 4.999942333600079e-05,
      "loss": 1.1325,
      "step": 364
    },
    {
      "epoch": 0.005190182613597286,
      "grad_norm": 2.984375,
      "learning_rate": 4.999940806085472e-05,
      "loss": 1.0621,
      "step": 366
    },
    {
      "epoch": 0.005218544267223501,
      "grad_norm": 3.203125,
      "learning_rate": 4.999939258603671e-05,
      "loss": 1.06,
      "step": 368
    },
    {
      "epoch": 0.0052469059208497155,
      "grad_norm": 3.59375,
      "learning_rate": 4.999937691154688e-05,
      "loss": 1.1121,
      "step": 370
    },
    {
      "epoch": 0.00527526757447593,
      "grad_norm": 3.3125,
      "learning_rate": 4.9999361037385366e-05,
      "loss": 1.0988,
      "step": 372
    },
    {
      "epoch": 0.005303629228102145,
      "grad_norm": 4.21875,
      "learning_rate": 4.9999344963552285e-05,
      "loss": 1.0685,
      "step": 374
    },
    {
      "epoch": 0.00533199088172836,
      "grad_norm": 3.765625,
      "learning_rate": 4.999932869004777e-05,
      "loss": 1.1825,
      "step": 376
    },
    {
      "epoch": 0.005360352535354574,
      "grad_norm": 3.171875,
      "learning_rate": 4.9999312216871944e-05,
      "loss": 1.1556,
      "step": 378
    },
    {
      "epoch": 0.005388714188980789,
      "grad_norm": 3.71875,
      "learning_rate": 4.999929554402495e-05,
      "loss": 1.07,
      "step": 380
    },
    {
      "epoch": 0.005417075842607004,
      "grad_norm": 3.828125,
      "learning_rate": 4.999927867150691e-05,
      "loss": 1.0872,
      "step": 382
    },
    {
      "epoch": 0.005445437496233218,
      "grad_norm": 3.40625,
      "learning_rate": 4.999926159931797e-05,
      "loss": 1.1259,
      "step": 384
    },
    {
      "epoch": 0.005473799149859433,
      "grad_norm": 3.296875,
      "learning_rate": 4.9999244327458256e-05,
      "loss": 1.063,
      "step": 386
    },
    {
      "epoch": 0.005502160803485647,
      "grad_norm": 4.375,
      "learning_rate": 4.999922685592791e-05,
      "loss": 1.138,
      "step": 388
    },
    {
      "epoch": 0.005530522457111862,
      "grad_norm": 3.953125,
      "learning_rate": 4.9999209184727076e-05,
      "loss": 1.09,
      "step": 390
    },
    {
      "epoch": 0.005558884110738077,
      "grad_norm": 4.03125,
      "learning_rate": 4.9999191313855884e-05,
      "loss": 1.0996,
      "step": 392
    },
    {
      "epoch": 0.005587245764364291,
      "grad_norm": 3.796875,
      "learning_rate": 4.999917324331449e-05,
      "loss": 1.0604,
      "step": 394
    },
    {
      "epoch": 0.005615607417990506,
      "grad_norm": 4.46875,
      "learning_rate": 4.999915497310302e-05,
      "loss": 1.1628,
      "step": 396
    },
    {
      "epoch": 0.005643969071616721,
      "grad_norm": 3.75,
      "learning_rate": 4.999913650322164e-05,
      "loss": 1.0717,
      "step": 398
    },
    {
      "epoch": 0.005672330725242935,
      "grad_norm": 3.0,
      "learning_rate": 4.9999117833670495e-05,
      "loss": 1.0827,
      "step": 400
    },
    {
      "epoch": 0.00570069237886915,
      "grad_norm": 3.65625,
      "learning_rate": 4.999909896444972e-05,
      "loss": 1.0837,
      "step": 402
    },
    {
      "epoch": 0.005729054032495365,
      "grad_norm": 3.09375,
      "learning_rate": 4.999907989555948e-05,
      "loss": 1.1118,
      "step": 404
    },
    {
      "epoch": 0.005757415686121579,
      "grad_norm": 4.40625,
      "learning_rate": 4.9999060626999914e-05,
      "loss": 1.0355,
      "step": 406
    },
    {
      "epoch": 0.005785777339747794,
      "grad_norm": 3.75,
      "learning_rate": 4.9999041158771195e-05,
      "loss": 1.1074,
      "step": 408
    },
    {
      "epoch": 0.005814138993374009,
      "grad_norm": 3.65625,
      "learning_rate": 4.999902149087345e-05,
      "loss": 1.1869,
      "step": 410
    },
    {
      "epoch": 0.005842500647000223,
      "grad_norm": 3.40625,
      "learning_rate": 4.9999001623306876e-05,
      "loss": 1.1101,
      "step": 412
    },
    {
      "epoch": 0.005870862300626438,
      "grad_norm": 3.53125,
      "learning_rate": 4.999898155607159e-05,
      "loss": 1.0199,
      "step": 414
    },
    {
      "epoch": 0.005899223954252653,
      "grad_norm": 3.1875,
      "learning_rate": 4.999896128916778e-05,
      "loss": 1.0981,
      "step": 416
    },
    {
      "epoch": 0.005927585607878867,
      "grad_norm": 4.25,
      "learning_rate": 4.999894082259559e-05,
      "loss": 1.0801,
      "step": 418
    },
    {
      "epoch": 0.005955947261505082,
      "grad_norm": 3.375,
      "learning_rate": 4.99989201563552e-05,
      "loss": 1.0877,
      "step": 420
    },
    {
      "epoch": 0.005984308915131297,
      "grad_norm": 3.515625,
      "learning_rate": 4.999889929044676e-05,
      "loss": 1.1017,
      "step": 422
    },
    {
      "epoch": 0.006012670568757511,
      "grad_norm": 4.3125,
      "learning_rate": 4.999887822487045e-05,
      "loss": 1.107,
      "step": 424
    },
    {
      "epoch": 0.006041032222383726,
      "grad_norm": 3.625,
      "learning_rate": 4.999885695962643e-05,
      "loss": 1.1297,
      "step": 426
    },
    {
      "epoch": 0.006069393876009941,
      "grad_norm": 4.0625,
      "learning_rate": 4.999883549471487e-05,
      "loss": 1.0946,
      "step": 428
    },
    {
      "epoch": 0.006097755529636155,
      "grad_norm": 4.0625,
      "learning_rate": 4.999881383013595e-05,
      "loss": 1.0647,
      "step": 430
    },
    {
      "epoch": 0.00612611718326237,
      "grad_norm": 4.0,
      "learning_rate": 4.999879196588983e-05,
      "loss": 1.1169,
      "step": 432
    },
    {
      "epoch": 0.006154478836888585,
      "grad_norm": 3.265625,
      "learning_rate": 4.9998769901976696e-05,
      "loss": 1.055,
      "step": 434
    },
    {
      "epoch": 0.006182840490514799,
      "grad_norm": 3.734375,
      "learning_rate": 4.999874763839671e-05,
      "loss": 1.0391,
      "step": 436
    },
    {
      "epoch": 0.006211202144141014,
      "grad_norm": 3.515625,
      "learning_rate": 4.999872517515007e-05,
      "loss": 1.0616,
      "step": 438
    },
    {
      "epoch": 0.006239563797767229,
      "grad_norm": 3.84375,
      "learning_rate": 4.999870251223694e-05,
      "loss": 1.1106,
      "step": 440
    },
    {
      "epoch": 0.006267925451393443,
      "grad_norm": 3.25,
      "learning_rate": 4.999867964965751e-05,
      "loss": 1.0544,
      "step": 442
    },
    {
      "epoch": 0.006296287105019658,
      "grad_norm": 3.515625,
      "learning_rate": 4.9998656587411954e-05,
      "loss": 1.1045,
      "step": 444
    },
    {
      "epoch": 0.006324648758645873,
      "grad_norm": 2.859375,
      "learning_rate": 4.999863332550045e-05,
      "loss": 1.1112,
      "step": 446
    },
    {
      "epoch": 0.006353010412272087,
      "grad_norm": 3.09375,
      "learning_rate": 4.9998609863923215e-05,
      "loss": 1.0418,
      "step": 448
    },
    {
      "epoch": 0.006381372065898302,
      "grad_norm": 3.828125,
      "learning_rate": 4.99985862026804e-05,
      "loss": 1.0795,
      "step": 450
    },
    {
      "epoch": 0.006409733719524517,
      "grad_norm": 3.3125,
      "learning_rate": 4.999856234177221e-05,
      "loss": 1.0808,
      "step": 452
    },
    {
      "epoch": 0.006438095373150731,
      "grad_norm": 4.09375,
      "learning_rate": 4.999853828119885e-05,
      "loss": 1.0493,
      "step": 454
    },
    {
      "epoch": 0.006466457026776946,
      "grad_norm": 3.46875,
      "learning_rate": 4.9998514020960485e-05,
      "loss": 1.0342,
      "step": 456
    },
    {
      "epoch": 0.006494818680403161,
      "grad_norm": 3.546875,
      "learning_rate": 4.999848956105733e-05,
      "loss": 1.1309,
      "step": 458
    },
    {
      "epoch": 0.006523180334029375,
      "grad_norm": 3.90625,
      "learning_rate": 4.9998464901489563e-05,
      "loss": 1.0944,
      "step": 460
    },
    {
      "epoch": 0.00655154198765559,
      "grad_norm": 3.46875,
      "learning_rate": 4.999844004225739e-05,
      "loss": 1.0529,
      "step": 462
    },
    {
      "epoch": 0.006579903641281805,
      "grad_norm": 3.484375,
      "learning_rate": 4.999841498336101e-05,
      "loss": 1.0789,
      "step": 464
    },
    {
      "epoch": 0.006608265294908019,
      "grad_norm": 3.640625,
      "learning_rate": 4.999838972480062e-05,
      "loss": 1.0436,
      "step": 466
    },
    {
      "epoch": 0.006636626948534234,
      "grad_norm": 3.328125,
      "learning_rate": 4.999836426657643e-05,
      "loss": 1.0754,
      "step": 468
    },
    {
      "epoch": 0.006664988602160449,
      "grad_norm": 2.984375,
      "learning_rate": 4.999833860868863e-05,
      "loss": 1.1059,
      "step": 470
    },
    {
      "epoch": 0.006693350255786663,
      "grad_norm": 3.78125,
      "learning_rate": 4.999831275113744e-05,
      "loss": 1.093,
      "step": 472
    },
    {
      "epoch": 0.006721711909412878,
      "grad_norm": 4.0625,
      "learning_rate": 4.9998286693923055e-05,
      "loss": 1.1176,
      "step": 474
    },
    {
      "epoch": 0.006750073563039093,
      "grad_norm": 4.0625,
      "learning_rate": 4.9998260437045686e-05,
      "loss": 1.0963,
      "step": 476
    },
    {
      "epoch": 0.006778435216665307,
      "grad_norm": 3.515625,
      "learning_rate": 4.9998233980505535e-05,
      "loss": 1.0579,
      "step": 478
    },
    {
      "epoch": 0.006806796870291522,
      "grad_norm": 3.0,
      "learning_rate": 4.999820732430284e-05,
      "loss": 1.0708,
      "step": 480
    },
    {
      "epoch": 0.006835158523917737,
      "grad_norm": 3.203125,
      "learning_rate": 4.9998180468437786e-05,
      "loss": 1.0267,
      "step": 482
    },
    {
      "epoch": 0.0068635201775439515,
      "grad_norm": 3.71875,
      "learning_rate": 4.9998153412910594e-05,
      "loss": 1.1024,
      "step": 484
    },
    {
      "epoch": 0.006891881831170166,
      "grad_norm": 3.375,
      "learning_rate": 4.999812615772149e-05,
      "loss": 1.1203,
      "step": 486
    },
    {
      "epoch": 0.006920243484796381,
      "grad_norm": 3.046875,
      "learning_rate": 4.999809870287068e-05,
      "loss": 0.9936,
      "step": 488
    },
    {
      "epoch": 0.0069486051384225955,
      "grad_norm": 3.609375,
      "learning_rate": 4.9998071048358384e-05,
      "loss": 1.0788,
      "step": 490
    },
    {
      "epoch": 0.0069769667920488105,
      "grad_norm": 3.03125,
      "learning_rate": 4.999804319418484e-05,
      "loss": 1.1216,
      "step": 492
    },
    {
      "epoch": 0.007005328445675025,
      "grad_norm": 3.640625,
      "learning_rate": 4.9998015140350255e-05,
      "loss": 1.037,
      "step": 494
    },
    {
      "epoch": 0.0070336900993012395,
      "grad_norm": 3.59375,
      "learning_rate": 4.999798688685484e-05,
      "loss": 1.1288,
      "step": 496
    },
    {
      "epoch": 0.0070620517529274545,
      "grad_norm": 2.984375,
      "learning_rate": 4.9997958433698856e-05,
      "loss": 1.0899,
      "step": 498
    },
    {
      "epoch": 0.0070904134065536695,
      "grad_norm": 3.265625,
      "learning_rate": 4.9997929780882504e-05,
      "loss": 1.106,
      "step": 500
    },
    {
      "epoch": 0.007118775060179884,
      "grad_norm": 3.296875,
      "learning_rate": 4.999790092840602e-05,
      "loss": 1.0528,
      "step": 502
    },
    {
      "epoch": 0.0071471367138060985,
      "grad_norm": 3.5,
      "learning_rate": 4.999787187626962e-05,
      "loss": 1.01,
      "step": 504
    },
    {
      "epoch": 0.0071754983674323135,
      "grad_norm": 3.484375,
      "learning_rate": 4.999784262447357e-05,
      "loss": 1.0104,
      "step": 506
    },
    {
      "epoch": 0.007203860021058528,
      "grad_norm": 3.203125,
      "learning_rate": 4.9997813173018074e-05,
      "loss": 1.0916,
      "step": 508
    },
    {
      "epoch": 0.007232221674684743,
      "grad_norm": 4.03125,
      "learning_rate": 4.999778352190337e-05,
      "loss": 1.0464,
      "step": 510
    },
    {
      "epoch": 0.0072605833283109576,
      "grad_norm": 3.015625,
      "learning_rate": 4.999775367112972e-05,
      "loss": 1.1138,
      "step": 512
    },
    {
      "epoch": 0.007288944981937172,
      "grad_norm": 3.578125,
      "learning_rate": 4.9997723620697335e-05,
      "loss": 1.0889,
      "step": 514
    },
    {
      "epoch": 0.007317306635563387,
      "grad_norm": 3.265625,
      "learning_rate": 4.999769337060646e-05,
      "loss": 1.0505,
      "step": 516
    },
    {
      "epoch": 0.007345668289189602,
      "grad_norm": 3.5,
      "learning_rate": 4.9997662920857344e-05,
      "loss": 1.0892,
      "step": 518
    },
    {
      "epoch": 0.007374029942815816,
      "grad_norm": 3.3125,
      "learning_rate": 4.9997632271450225e-05,
      "loss": 1.0791,
      "step": 520
    },
    {
      "epoch": 0.007402391596442031,
      "grad_norm": 3.421875,
      "learning_rate": 4.999760142238535e-05,
      "loss": 1.1102,
      "step": 522
    },
    {
      "epoch": 0.007430753250068246,
      "grad_norm": 3.546875,
      "learning_rate": 4.999757037366297e-05,
      "loss": 1.087,
      "step": 524
    },
    {
      "epoch": 0.00745911490369446,
      "grad_norm": 3.296875,
      "learning_rate": 4.999753912528332e-05,
      "loss": 1.0717,
      "step": 526
    },
    {
      "epoch": 0.007487476557320675,
      "grad_norm": 3.078125,
      "learning_rate": 4.9997507677246666e-05,
      "loss": 1.0555,
      "step": 528
    },
    {
      "epoch": 0.00751583821094689,
      "grad_norm": 3.671875,
      "learning_rate": 4.999747602955325e-05,
      "loss": 1.0958,
      "step": 530
    },
    {
      "epoch": 0.007544199864573104,
      "grad_norm": 3.859375,
      "learning_rate": 4.9997444182203316e-05,
      "loss": 1.022,
      "step": 532
    },
    {
      "epoch": 0.007572561518199319,
      "grad_norm": 4.40625,
      "learning_rate": 4.9997412135197134e-05,
      "loss": 1.0834,
      "step": 534
    },
    {
      "epoch": 0.007600923171825534,
      "grad_norm": 4.25,
      "learning_rate": 4.999737988853496e-05,
      "loss": 1.0382,
      "step": 536
    },
    {
      "epoch": 0.007629284825451748,
      "grad_norm": 3.328125,
      "learning_rate": 4.9997347442217035e-05,
      "loss": 1.0283,
      "step": 538
    },
    {
      "epoch": 0.007657646479077963,
      "grad_norm": 3.84375,
      "learning_rate": 4.9997314796243634e-05,
      "loss": 1.1022,
      "step": 540
    },
    {
      "epoch": 0.007686008132704177,
      "grad_norm": 3.296875,
      "learning_rate": 4.999728195061502e-05,
      "loss": 1.0579,
      "step": 542
    },
    {
      "epoch": 0.007714369786330392,
      "grad_norm": 3.4375,
      "learning_rate": 4.999724890533143e-05,
      "loss": 1.0533,
      "step": 544
    },
    {
      "epoch": 0.007742731439956607,
      "grad_norm": 3.5,
      "learning_rate": 4.999721566039316e-05,
      "loss": 1.083,
      "step": 546
    },
    {
      "epoch": 0.007771093093582821,
      "grad_norm": 3.21875,
      "learning_rate": 4.9997182215800455e-05,
      "loss": 1.0488,
      "step": 548
    },
    {
      "epoch": 0.007799454747209036,
      "grad_norm": 2.78125,
      "learning_rate": 4.999714857155359e-05,
      "loss": 1.0939,
      "step": 550
    },
    {
      "epoch": 0.00782781640083525,
      "grad_norm": 3.5625,
      "learning_rate": 4.999711472765283e-05,
      "loss": 1.0696,
      "step": 552
    },
    {
      "epoch": 0.007856178054461465,
      "grad_norm": 3.9375,
      "learning_rate": 4.999708068409845e-05,
      "loss": 1.0529,
      "step": 554
    },
    {
      "epoch": 0.00788453970808768,
      "grad_norm": 3.140625,
      "learning_rate": 4.9997046440890714e-05,
      "loss": 1.0261,
      "step": 556
    },
    {
      "epoch": 0.007912901361713895,
      "grad_norm": 3.5625,
      "learning_rate": 4.9997011998029905e-05,
      "loss": 1.016,
      "step": 558
    },
    {
      "epoch": 0.00794126301534011,
      "grad_norm": 3.25,
      "learning_rate": 4.99969773555163e-05,
      "loss": 1.0712,
      "step": 560
    },
    {
      "epoch": 0.007969624668966325,
      "grad_norm": 3.515625,
      "learning_rate": 4.999694251335016e-05,
      "loss": 1.0652,
      "step": 562
    },
    {
      "epoch": 0.007997986322592538,
      "grad_norm": 3.09375,
      "learning_rate": 4.999690747153178e-05,
      "loss": 1.0161,
      "step": 564
    },
    {
      "epoch": 0.008026347976218753,
      "grad_norm": 3.625,
      "learning_rate": 4.999687223006143e-05,
      "loss": 1.0946,
      "step": 566
    },
    {
      "epoch": 0.008054709629844968,
      "grad_norm": 3.421875,
      "learning_rate": 4.9996836788939396e-05,
      "loss": 1.0748,
      "step": 568
    },
    {
      "epoch": 0.008083071283471183,
      "grad_norm": 4.375,
      "learning_rate": 4.999680114816595e-05,
      "loss": 1.044,
      "step": 570
    },
    {
      "epoch": 0.008111432937097398,
      "grad_norm": 3.484375,
      "learning_rate": 4.9996765307741394e-05,
      "loss": 0.9668,
      "step": 572
    },
    {
      "epoch": 0.008139794590723613,
      "grad_norm": 3.21875,
      "learning_rate": 4.999672926766601e-05,
      "loss": 1.0584,
      "step": 574
    },
    {
      "epoch": 0.008168156244349826,
      "grad_norm": 3.484375,
      "learning_rate": 4.9996693027940076e-05,
      "loss": 1.09,
      "step": 576
    },
    {
      "epoch": 0.008196517897976041,
      "grad_norm": 3.203125,
      "learning_rate": 4.999665658856388e-05,
      "loss": 1.0735,
      "step": 578
    },
    {
      "epoch": 0.008224879551602256,
      "grad_norm": 3.140625,
      "learning_rate": 4.999661994953773e-05,
      "loss": 1.0357,
      "step": 580
    },
    {
      "epoch": 0.008253241205228471,
      "grad_norm": 3.5625,
      "learning_rate": 4.999658311086191e-05,
      "loss": 1.1033,
      "step": 582
    },
    {
      "epoch": 0.008281602858854686,
      "grad_norm": 3.265625,
      "learning_rate": 4.999654607253671e-05,
      "loss": 1.0494,
      "step": 584
    },
    {
      "epoch": 0.008309964512480901,
      "grad_norm": 2.859375,
      "learning_rate": 4.999650883456243e-05,
      "loss": 1.0304,
      "step": 586
    },
    {
      "epoch": 0.008338326166107114,
      "grad_norm": 3.0,
      "learning_rate": 4.9996471396939374e-05,
      "loss": 1.0845,
      "step": 588
    },
    {
      "epoch": 0.00836668781973333,
      "grad_norm": 3.265625,
      "learning_rate": 4.9996433759667826e-05,
      "loss": 1.0447,
      "step": 590
    },
    {
      "epoch": 0.008395049473359544,
      "grad_norm": 3.40625,
      "learning_rate": 4.99963959227481e-05,
      "loss": 1.0451,
      "step": 592
    },
    {
      "epoch": 0.00842341112698576,
      "grad_norm": 3.4375,
      "learning_rate": 4.9996357886180476e-05,
      "loss": 1.0794,
      "step": 594
    },
    {
      "epoch": 0.008451772780611974,
      "grad_norm": 4.0625,
      "learning_rate": 4.999631964996529e-05,
      "loss": 1.0491,
      "step": 596
    },
    {
      "epoch": 0.008480134434238189,
      "grad_norm": 3.265625,
      "learning_rate": 4.9996281214102826e-05,
      "loss": 1.0311,
      "step": 598
    },
    {
      "epoch": 0.008508496087864402,
      "grad_norm": 3.515625,
      "learning_rate": 4.99962425785934e-05,
      "loss": 1.0697,
      "step": 600
    },
    {
      "epoch": 0.008536857741490617,
      "grad_norm": 3.265625,
      "learning_rate": 4.999620374343732e-05,
      "loss": 1.0606,
      "step": 602
    },
    {
      "epoch": 0.008565219395116832,
      "grad_norm": 3.453125,
      "learning_rate": 4.999616470863489e-05,
      "loss": 1.0309,
      "step": 604
    },
    {
      "epoch": 0.008593581048743047,
      "grad_norm": 4.15625,
      "learning_rate": 4.9996125474186414e-05,
      "loss": 1.0405,
      "step": 606
    },
    {
      "epoch": 0.008621942702369262,
      "grad_norm": 3.1875,
      "learning_rate": 4.999608604009223e-05,
      "loss": 1.0147,
      "step": 608
    },
    {
      "epoch": 0.008650304355995477,
      "grad_norm": 3.5,
      "learning_rate": 4.999604640635264e-05,
      "loss": 1.0767,
      "step": 610
    },
    {
      "epoch": 0.00867866600962169,
      "grad_norm": 3.328125,
      "learning_rate": 4.999600657296796e-05,
      "loss": 1.0442,
      "step": 612
    },
    {
      "epoch": 0.008707027663247905,
      "grad_norm": 3.40625,
      "learning_rate": 4.99959665399385e-05,
      "loss": 1.0137,
      "step": 614
    },
    {
      "epoch": 0.00873538931687412,
      "grad_norm": 4.375,
      "learning_rate": 4.9995926307264594e-05,
      "loss": 1.0142,
      "step": 616
    },
    {
      "epoch": 0.008763750970500335,
      "grad_norm": 3.546875,
      "learning_rate": 4.999588587494656e-05,
      "loss": 1.0484,
      "step": 618
    },
    {
      "epoch": 0.00879211262412655,
      "grad_norm": 3.578125,
      "learning_rate": 4.9995845242984706e-05,
      "loss": 1.0453,
      "step": 620
    },
    {
      "epoch": 0.008820474277752764,
      "grad_norm": 3.6875,
      "learning_rate": 4.999580441137938e-05,
      "loss": 1.027,
      "step": 622
    },
    {
      "epoch": 0.008848835931378978,
      "grad_norm": 3.890625,
      "learning_rate": 4.9995763380130896e-05,
      "loss": 1.0741,
      "step": 624
    },
    {
      "epoch": 0.008877197585005193,
      "grad_norm": 3.703125,
      "learning_rate": 4.9995722149239575e-05,
      "loss": 1.0291,
      "step": 626
    },
    {
      "epoch": 0.008905559238631408,
      "grad_norm": 3.0625,
      "learning_rate": 4.999568071870576e-05,
      "loss": 1.0586,
      "step": 628
    },
    {
      "epoch": 0.008933920892257623,
      "grad_norm": 3.375,
      "learning_rate": 4.9995639088529776e-05,
      "loss": 1.0845,
      "step": 630
    },
    {
      "epoch": 0.008962282545883838,
      "grad_norm": 3.78125,
      "learning_rate": 4.9995597258711954e-05,
      "loss": 1.0618,
      "step": 632
    },
    {
      "epoch": 0.008990644199510052,
      "grad_norm": 3.140625,
      "learning_rate": 4.9995555229252635e-05,
      "loss": 1.0145,
      "step": 634
    },
    {
      "epoch": 0.009019005853136267,
      "grad_norm": 3.515625,
      "learning_rate": 4.999551300015214e-05,
      "loss": 1.1262,
      "step": 636
    },
    {
      "epoch": 0.009047367506762482,
      "grad_norm": 3.453125,
      "learning_rate": 4.999547057141082e-05,
      "loss": 0.9884,
      "step": 638
    },
    {
      "epoch": 0.009075729160388697,
      "grad_norm": 3.484375,
      "learning_rate": 4.9995427943029014e-05,
      "loss": 1.0423,
      "step": 640
    },
    {
      "epoch": 0.009104090814014911,
      "grad_norm": 3.390625,
      "learning_rate": 4.9995385115007055e-05,
      "loss": 1.0374,
      "step": 642
    },
    {
      "epoch": 0.009132452467641126,
      "grad_norm": 3.578125,
      "learning_rate": 4.9995342087345286e-05,
      "loss": 1.0917,
      "step": 644
    },
    {
      "epoch": 0.00916081412126734,
      "grad_norm": 3.21875,
      "learning_rate": 4.999529886004405e-05,
      "loss": 1.07,
      "step": 646
    },
    {
      "epoch": 0.009189175774893555,
      "grad_norm": 3.875,
      "learning_rate": 4.9995255433103696e-05,
      "loss": 1.1243,
      "step": 648
    },
    {
      "epoch": 0.00921753742851977,
      "grad_norm": 3.46875,
      "learning_rate": 4.9995211806524574e-05,
      "loss": 1.0334,
      "step": 650
    },
    {
      "epoch": 0.009245899082145985,
      "grad_norm": 3.359375,
      "learning_rate": 4.9995167980307024e-05,
      "loss": 1.0902,
      "step": 652
    },
    {
      "epoch": 0.0092742607357722,
      "grad_norm": 3.484375,
      "learning_rate": 4.9995123954451404e-05,
      "loss": 1.1037,
      "step": 654
    },
    {
      "epoch": 0.009302622389398415,
      "grad_norm": 3.0625,
      "learning_rate": 4.999507972895805e-05,
      "loss": 1.003,
      "step": 656
    },
    {
      "epoch": 0.009330984043024628,
      "grad_norm": 3.21875,
      "learning_rate": 4.9995035303827344e-05,
      "loss": 0.9697,
      "step": 658
    },
    {
      "epoch": 0.009359345696650843,
      "grad_norm": 3.453125,
      "learning_rate": 4.9994990679059616e-05,
      "loss": 1.0206,
      "step": 660
    },
    {
      "epoch": 0.009387707350277058,
      "grad_norm": 3.75,
      "learning_rate": 4.999494585465523e-05,
      "loss": 1.0407,
      "step": 662
    },
    {
      "epoch": 0.009416069003903273,
      "grad_norm": 3.71875,
      "learning_rate": 4.9994900830614546e-05,
      "loss": 1.0539,
      "step": 664
    },
    {
      "epoch": 0.009444430657529488,
      "grad_norm": 2.875,
      "learning_rate": 4.9994855606937916e-05,
      "loss": 1.0032,
      "step": 666
    },
    {
      "epoch": 0.009472792311155703,
      "grad_norm": 3.4375,
      "learning_rate": 4.999481018362571e-05,
      "loss": 1.0443,
      "step": 668
    },
    {
      "epoch": 0.009501153964781916,
      "grad_norm": 3.171875,
      "learning_rate": 4.9994764560678293e-05,
      "loss": 1.0054,
      "step": 670
    },
    {
      "epoch": 0.00952951561840813,
      "grad_norm": 3.265625,
      "learning_rate": 4.999471873809602e-05,
      "loss": 1.0936,
      "step": 672
    },
    {
      "epoch": 0.009557877272034346,
      "grad_norm": 4.09375,
      "learning_rate": 4.999467271587927e-05,
      "loss": 1.1194,
      "step": 674
    },
    {
      "epoch": 0.00958623892566056,
      "grad_norm": 3.140625,
      "learning_rate": 4.9994626494028384e-05,
      "loss": 1.0168,
      "step": 676
    },
    {
      "epoch": 0.009614600579286776,
      "grad_norm": 3.953125,
      "learning_rate": 4.999458007254376e-05,
      "loss": 1.0976,
      "step": 678
    },
    {
      "epoch": 0.00964296223291299,
      "grad_norm": 3.71875,
      "learning_rate": 4.9994533451425756e-05,
      "loss": 1.1002,
      "step": 680
    },
    {
      "epoch": 0.009671323886539204,
      "grad_norm": 3.0625,
      "learning_rate": 4.9994486630674744e-05,
      "loss": 1.0512,
      "step": 682
    },
    {
      "epoch": 0.009699685540165419,
      "grad_norm": 3.1875,
      "learning_rate": 4.999443961029111e-05,
      "loss": 1.0842,
      "step": 684
    },
    {
      "epoch": 0.009728047193791634,
      "grad_norm": 3.75,
      "learning_rate": 4.9994392390275216e-05,
      "loss": 0.9851,
      "step": 686
    },
    {
      "epoch": 0.009756408847417849,
      "grad_norm": 3.109375,
      "learning_rate": 4.999434497062743e-05,
      "loss": 0.9994,
      "step": 688
    },
    {
      "epoch": 0.009784770501044064,
      "grad_norm": 4.15625,
      "learning_rate": 4.999429735134816e-05,
      "loss": 1.0518,
      "step": 690
    },
    {
      "epoch": 0.009813132154670279,
      "grad_norm": 3.5,
      "learning_rate": 4.999424953243776e-05,
      "loss": 1.0893,
      "step": 692
    },
    {
      "epoch": 0.009841493808296492,
      "grad_norm": 2.921875,
      "learning_rate": 4.999420151389662e-05,
      "loss": 1.0451,
      "step": 694
    },
    {
      "epoch": 0.009869855461922707,
      "grad_norm": 3.71875,
      "learning_rate": 4.999415329572513e-05,
      "loss": 1.0503,
      "step": 696
    },
    {
      "epoch": 0.009898217115548922,
      "grad_norm": 3.09375,
      "learning_rate": 4.999410487792368e-05,
      "loss": 1.0118,
      "step": 698
    },
    {
      "epoch": 0.009926578769175137,
      "grad_norm": 3.359375,
      "learning_rate": 4.999405626049263e-05,
      "loss": 1.0035,
      "step": 700
    },
    {
      "epoch": 0.009954940422801352,
      "grad_norm": 3.125,
      "learning_rate": 4.99940074434324e-05,
      "loss": 1.0273,
      "step": 702
    },
    {
      "epoch": 0.009983302076427567,
      "grad_norm": 3.515625,
      "learning_rate": 4.999395842674336e-05,
      "loss": 1.0375,
      "step": 704
    },
    {
      "epoch": 0.01001166373005378,
      "grad_norm": 3.515625,
      "learning_rate": 4.99939092104259e-05,
      "loss": 1.1057,
      "step": 706
    },
    {
      "epoch": 0.010040025383679995,
      "grad_norm": 3.3125,
      "learning_rate": 4.9993859794480425e-05,
      "loss": 1.0436,
      "step": 708
    },
    {
      "epoch": 0.01006838703730621,
      "grad_norm": 3.3125,
      "learning_rate": 4.999381017890733e-05,
      "loss": 1.0869,
      "step": 710
    },
    {
      "epoch": 0.010096748690932425,
      "grad_norm": 2.859375,
      "learning_rate": 4.9993760363706996e-05,
      "loss": 1.0425,
      "step": 712
    },
    {
      "epoch": 0.01012511034455864,
      "grad_norm": 3.125,
      "learning_rate": 4.999371034887984e-05,
      "loss": 1.0183,
      "step": 714
    },
    {
      "epoch": 0.010153471998184855,
      "grad_norm": 3.1875,
      "learning_rate": 4.999366013442624e-05,
      "loss": 1.0579,
      "step": 716
    },
    {
      "epoch": 0.010181833651811068,
      "grad_norm": 3.375,
      "learning_rate": 4.999360972034663e-05,
      "loss": 1.1013,
      "step": 718
    },
    {
      "epoch": 0.010210195305437283,
      "grad_norm": 3.078125,
      "learning_rate": 4.999355910664138e-05,
      "loss": 1.0237,
      "step": 720
    },
    {
      "epoch": 0.010238556959063498,
      "grad_norm": 3.265625,
      "learning_rate": 4.9993508293310905e-05,
      "loss": 1.0493,
      "step": 722
    },
    {
      "epoch": 0.010266918612689713,
      "grad_norm": 3.390625,
      "learning_rate": 4.999345728035562e-05,
      "loss": 1.0047,
      "step": 724
    },
    {
      "epoch": 0.010295280266315928,
      "grad_norm": 3.859375,
      "learning_rate": 4.999340606777591e-05,
      "loss": 1.0385,
      "step": 726
    },
    {
      "epoch": 0.010323641919942143,
      "grad_norm": 5.03125,
      "learning_rate": 4.999335465557221e-05,
      "loss": 1.0464,
      "step": 728
    },
    {
      "epoch": 0.010352003573568356,
      "grad_norm": 3.5625,
      "learning_rate": 4.999330304374491e-05,
      "loss": 0.9834,
      "step": 730
    },
    {
      "epoch": 0.010380365227194571,
      "grad_norm": 3.421875,
      "learning_rate": 4.999325123229444e-05,
      "loss": 1.0757,
      "step": 732
    },
    {
      "epoch": 0.010408726880820786,
      "grad_norm": 3.609375,
      "learning_rate": 4.9993199221221206e-05,
      "loss": 1.0112,
      "step": 734
    },
    {
      "epoch": 0.010437088534447001,
      "grad_norm": 3.953125,
      "learning_rate": 4.999314701052562e-05,
      "loss": 1.1014,
      "step": 736
    },
    {
      "epoch": 0.010465450188073216,
      "grad_norm": 3.203125,
      "learning_rate": 4.99930946002081e-05,
      "loss": 1.0492,
      "step": 738
    },
    {
      "epoch": 0.010493811841699431,
      "grad_norm": 3.0625,
      "learning_rate": 4.999304199026907e-05,
      "loss": 1.0031,
      "step": 740
    },
    {
      "epoch": 0.010522173495325644,
      "grad_norm": 2.953125,
      "learning_rate": 4.999298918070894e-05,
      "loss": 1.0141,
      "step": 742
    },
    {
      "epoch": 0.01055053514895186,
      "grad_norm": 3.09375,
      "learning_rate": 4.999293617152815e-05,
      "loss": 1.0233,
      "step": 744
    },
    {
      "epoch": 0.010578896802578074,
      "grad_norm": 3.625,
      "learning_rate": 4.999288296272711e-05,
      "loss": 1.0374,
      "step": 746
    },
    {
      "epoch": 0.01060725845620429,
      "grad_norm": 3.46875,
      "learning_rate": 4.9992829554306234e-05,
      "loss": 1.0846,
      "step": 748
    },
    {
      "epoch": 0.010635620109830504,
      "grad_norm": 2.984375,
      "learning_rate": 4.9992775946265966e-05,
      "loss": 1.043,
      "step": 750
    },
    {
      "epoch": 0.01066398176345672,
      "grad_norm": 3.34375,
      "learning_rate": 4.999272213860674e-05,
      "loss": 0.9907,
      "step": 752
    },
    {
      "epoch": 0.010692343417082932,
      "grad_norm": 4.3125,
      "learning_rate": 4.999266813132896e-05,
      "loss": 1.088,
      "step": 754
    },
    {
      "epoch": 0.010720705070709147,
      "grad_norm": 3.203125,
      "learning_rate": 4.9992613924433085e-05,
      "loss": 1.0518,
      "step": 756
    },
    {
      "epoch": 0.010749066724335362,
      "grad_norm": 3.28125,
      "learning_rate": 4.999255951791952e-05,
      "loss": 1.0805,
      "step": 758
    },
    {
      "epoch": 0.010777428377961577,
      "grad_norm": 3.171875,
      "learning_rate": 4.9992504911788735e-05,
      "loss": 1.049,
      "step": 760
    },
    {
      "epoch": 0.010805790031587792,
      "grad_norm": 3.046875,
      "learning_rate": 4.9992450106041135e-05,
      "loss": 1.0692,
      "step": 762
    },
    {
      "epoch": 0.010834151685214007,
      "grad_norm": 2.984375,
      "learning_rate": 4.999239510067717e-05,
      "loss": 1.0051,
      "step": 764
    },
    {
      "epoch": 0.01086251333884022,
      "grad_norm": 3.578125,
      "learning_rate": 4.999233989569728e-05,
      "loss": 1.0144,
      "step": 766
    },
    {
      "epoch": 0.010890874992466435,
      "grad_norm": 3.125,
      "learning_rate": 4.9992284491101905e-05,
      "loss": 0.962,
      "step": 768
    },
    {
      "epoch": 0.01091923664609265,
      "grad_norm": 3.6875,
      "learning_rate": 4.999222888689149e-05,
      "loss": 1.0756,
      "step": 770
    },
    {
      "epoch": 0.010947598299718865,
      "grad_norm": 3.390625,
      "learning_rate": 4.9992173083066466e-05,
      "loss": 1.0974,
      "step": 772
    },
    {
      "epoch": 0.01097595995334508,
      "grad_norm": 3.6875,
      "learning_rate": 4.9992117079627296e-05,
      "loss": 1.0391,
      "step": 774
    },
    {
      "epoch": 0.011004321606971294,
      "grad_norm": 3.4375,
      "learning_rate": 4.999206087657442e-05,
      "loss": 1.013,
      "step": 776
    },
    {
      "epoch": 0.011032683260597509,
      "grad_norm": 3.375,
      "learning_rate": 4.999200447390828e-05,
      "loss": 0.9818,
      "step": 778
    },
    {
      "epoch": 0.011061044914223724,
      "grad_norm": 3.34375,
      "learning_rate": 4.999194787162934e-05,
      "loss": 0.9791,
      "step": 780
    },
    {
      "epoch": 0.011089406567849939,
      "grad_norm": 3.390625,
      "learning_rate": 4.999189106973804e-05,
      "loss": 1.0711,
      "step": 782
    },
    {
      "epoch": 0.011117768221476153,
      "grad_norm": 3.265625,
      "learning_rate": 4.9991834068234844e-05,
      "loss": 1.0469,
      "step": 784
    },
    {
      "epoch": 0.011146129875102368,
      "grad_norm": 3.1875,
      "learning_rate": 4.99917768671202e-05,
      "loss": 0.9811,
      "step": 786
    },
    {
      "epoch": 0.011174491528728582,
      "grad_norm": 3.0625,
      "learning_rate": 4.999171946639457e-05,
      "loss": 0.9695,
      "step": 788
    },
    {
      "epoch": 0.011202853182354797,
      "grad_norm": 2.875,
      "learning_rate": 4.99916618660584e-05,
      "loss": 1.0348,
      "step": 790
    },
    {
      "epoch": 0.011231214835981012,
      "grad_norm": 2.859375,
      "learning_rate": 4.999160406611218e-05,
      "loss": 1.0236,
      "step": 792
    },
    {
      "epoch": 0.011259576489607227,
      "grad_norm": 2.890625,
      "learning_rate": 4.9991546066556335e-05,
      "loss": 1.039,
      "step": 794
    },
    {
      "epoch": 0.011287938143233442,
      "grad_norm": 3.140625,
      "learning_rate": 4.9991487867391345e-05,
      "loss": 1.0567,
      "step": 796
    },
    {
      "epoch": 0.011316299796859657,
      "grad_norm": 3.265625,
      "learning_rate": 4.999142946861768e-05,
      "loss": 1.0297,
      "step": 798
    },
    {
      "epoch": 0.01134466145048587,
      "grad_norm": 3.765625,
      "learning_rate": 4.99913708702358e-05,
      "loss": 0.9283,
      "step": 800
    },
    {
      "epoch": 0.011373023104112085,
      "grad_norm": 3.359375,
      "learning_rate": 4.999131207224617e-05,
      "loss": 1.0652,
      "step": 802
    },
    {
      "epoch": 0.0114013847577383,
      "grad_norm": 3.203125,
      "learning_rate": 4.9991253074649266e-05,
      "loss": 1.0936,
      "step": 804
    },
    {
      "epoch": 0.011429746411364515,
      "grad_norm": 3.109375,
      "learning_rate": 4.9991193877445554e-05,
      "loss": 1.0151,
      "step": 806
    },
    {
      "epoch": 0.01145810806499073,
      "grad_norm": 3.34375,
      "learning_rate": 4.999113448063551e-05,
      "loss": 0.9899,
      "step": 808
    },
    {
      "epoch": 0.011486469718616945,
      "grad_norm": 3.703125,
      "learning_rate": 4.999107488421961e-05,
      "loss": 0.9936,
      "step": 810
    },
    {
      "epoch": 0.011514831372243158,
      "grad_norm": 2.953125,
      "learning_rate": 4.999101508819833e-05,
      "loss": 0.9536,
      "step": 812
    },
    {
      "epoch": 0.011543193025869373,
      "grad_norm": 3.28125,
      "learning_rate": 4.999095509257214e-05,
      "loss": 1.0291,
      "step": 814
    },
    {
      "epoch": 0.011571554679495588,
      "grad_norm": 3.140625,
      "learning_rate": 4.9990894897341534e-05,
      "loss": 1.0869,
      "step": 816
    },
    {
      "epoch": 0.011599916333121803,
      "grad_norm": 3.40625,
      "learning_rate": 4.999083450250698e-05,
      "loss": 1.0776,
      "step": 818
    },
    {
      "epoch": 0.011628277986748018,
      "grad_norm": 3.0,
      "learning_rate": 4.999077390806896e-05,
      "loss": 1.0554,
      "step": 820
    },
    {
      "epoch": 0.011656639640374233,
      "grad_norm": 2.984375,
      "learning_rate": 4.9990713114027966e-05,
      "loss": 1.0387,
      "step": 822
    },
    {
      "epoch": 0.011685001294000446,
      "grad_norm": 3.3125,
      "learning_rate": 4.9990652120384474e-05,
      "loss": 1.1099,
      "step": 824
    },
    {
      "epoch": 0.011713362947626661,
      "grad_norm": 3.59375,
      "learning_rate": 4.999059092713898e-05,
      "loss": 1.0429,
      "step": 826
    },
    {
      "epoch": 0.011741724601252876,
      "grad_norm": 2.953125,
      "learning_rate": 4.999052953429197e-05,
      "loss": 1.0017,
      "step": 828
    },
    {
      "epoch": 0.01177008625487909,
      "grad_norm": 3.234375,
      "learning_rate": 4.999046794184393e-05,
      "loss": 1.0694,
      "step": 830
    },
    {
      "epoch": 0.011798447908505306,
      "grad_norm": 3.265625,
      "learning_rate": 4.999040614979536e-05,
      "loss": 1.0168,
      "step": 832
    },
    {
      "epoch": 0.01182680956213152,
      "grad_norm": 3.796875,
      "learning_rate": 4.999034415814675e-05,
      "loss": 1.0606,
      "step": 834
    },
    {
      "epoch": 0.011855171215757734,
      "grad_norm": 2.953125,
      "learning_rate": 4.9990281966898595e-05,
      "loss": 1.0336,
      "step": 836
    },
    {
      "epoch": 0.011883532869383949,
      "grad_norm": 3.046875,
      "learning_rate": 4.9990219576051386e-05,
      "loss": 0.9751,
      "step": 838
    },
    {
      "epoch": 0.011911894523010164,
      "grad_norm": 3.203125,
      "learning_rate": 4.999015698560563e-05,
      "loss": 0.9751,
      "step": 840
    },
    {
      "epoch": 0.011940256176636379,
      "grad_norm": 3.375,
      "learning_rate": 4.999009419556182e-05,
      "loss": 1.0226,
      "step": 842
    },
    {
      "epoch": 0.011968617830262594,
      "grad_norm": 3.1875,
      "learning_rate": 4.999003120592047e-05,
      "loss": 1.0217,
      "step": 844
    },
    {
      "epoch": 0.011996979483888809,
      "grad_norm": 3.140625,
      "learning_rate": 4.998996801668206e-05,
      "loss": 1.061,
      "step": 846
    },
    {
      "epoch": 0.012025341137515022,
      "grad_norm": 3.109375,
      "learning_rate": 4.998990462784712e-05,
      "loss": 1.0456,
      "step": 848
    },
    {
      "epoch": 0.012053702791141237,
      "grad_norm": 4.28125,
      "learning_rate": 4.998984103941614e-05,
      "loss": 1.0514,
      "step": 850
    },
    {
      "epoch": 0.012082064444767452,
      "grad_norm": 3.1875,
      "learning_rate": 4.998977725138964e-05,
      "loss": 1.0695,
      "step": 852
    },
    {
      "epoch": 0.012110426098393667,
      "grad_norm": 3.328125,
      "learning_rate": 4.9989713263768115e-05,
      "loss": 1.0497,
      "step": 854
    },
    {
      "epoch": 0.012138787752019882,
      "grad_norm": 2.796875,
      "learning_rate": 4.998964907655208e-05,
      "loss": 1.0239,
      "step": 856
    },
    {
      "epoch": 0.012167149405646097,
      "grad_norm": 3.0,
      "learning_rate": 4.9989584689742056e-05,
      "loss": 0.9841,
      "step": 858
    },
    {
      "epoch": 0.01219551105927231,
      "grad_norm": 3.25,
      "learning_rate": 4.998952010333855e-05,
      "loss": 0.9966,
      "step": 860
    },
    {
      "epoch": 0.012223872712898525,
      "grad_norm": 3.921875,
      "learning_rate": 4.9989455317342076e-05,
      "loss": 0.9948,
      "step": 862
    },
    {
      "epoch": 0.01225223436652474,
      "grad_norm": 3.65625,
      "learning_rate": 4.998939033175316e-05,
      "loss": 1.0605,
      "step": 864
    },
    {
      "epoch": 0.012280596020150955,
      "grad_norm": 3.703125,
      "learning_rate": 4.998932514657232e-05,
      "loss": 1.0642,
      "step": 866
    },
    {
      "epoch": 0.01230895767377717,
      "grad_norm": 3.15625,
      "learning_rate": 4.998925976180007e-05,
      "loss": 1.038,
      "step": 868
    },
    {
      "epoch": 0.012337319327403385,
      "grad_norm": 3.015625,
      "learning_rate": 4.998919417743694e-05,
      "loss": 1.0077,
      "step": 870
    },
    {
      "epoch": 0.012365680981029598,
      "grad_norm": 3.546875,
      "learning_rate": 4.9989128393483445e-05,
      "loss": 1.0422,
      "step": 872
    },
    {
      "epoch": 0.012394042634655813,
      "grad_norm": 2.96875,
      "learning_rate": 4.9989062409940116e-05,
      "loss": 0.9849,
      "step": 874
    },
    {
      "epoch": 0.012422404288282028,
      "grad_norm": 3.5,
      "learning_rate": 4.998899622680747e-05,
      "loss": 1.0115,
      "step": 876
    },
    {
      "epoch": 0.012450765941908243,
      "grad_norm": 3.109375,
      "learning_rate": 4.998892984408605e-05,
      "loss": 0.9811,
      "step": 878
    },
    {
      "epoch": 0.012479127595534458,
      "grad_norm": 3.3125,
      "learning_rate": 4.998886326177638e-05,
      "loss": 1.0521,
      "step": 880
    },
    {
      "epoch": 0.012507489249160673,
      "grad_norm": 3.265625,
      "learning_rate": 4.998879647987901e-05,
      "loss": 1.0624,
      "step": 882
    },
    {
      "epoch": 0.012535850902786886,
      "grad_norm": 3.4375,
      "learning_rate": 4.998872949839444e-05,
      "loss": 1.0684,
      "step": 884
    },
    {
      "epoch": 0.012564212556413101,
      "grad_norm": 3.5625,
      "learning_rate": 4.9988662317323215e-05,
      "loss": 1.0897,
      "step": 886
    },
    {
      "epoch": 0.012592574210039316,
      "grad_norm": 2.953125,
      "learning_rate": 4.998859493666588e-05,
      "loss": 0.9441,
      "step": 888
    },
    {
      "epoch": 0.012620935863665531,
      "grad_norm": 3.015625,
      "learning_rate": 4.998852735642298e-05,
      "loss": 1.0306,
      "step": 890
    },
    {
      "epoch": 0.012649297517291746,
      "grad_norm": 3.171875,
      "learning_rate": 4.998845957659504e-05,
      "loss": 1.0119,
      "step": 892
    },
    {
      "epoch": 0.012677659170917961,
      "grad_norm": 2.96875,
      "learning_rate": 4.99883915971826e-05,
      "loss": 1.0314,
      "step": 894
    },
    {
      "epoch": 0.012706020824544174,
      "grad_norm": 3.203125,
      "learning_rate": 4.9988323418186224e-05,
      "loss": 1.0402,
      "step": 896
    },
    {
      "epoch": 0.01273438247817039,
      "grad_norm": 2.921875,
      "learning_rate": 4.998825503960643e-05,
      "loss": 0.9315,
      "step": 898
    },
    {
      "epoch": 0.012762744131796604,
      "grad_norm": 3.125,
      "learning_rate": 4.998818646144378e-05,
      "loss": 1.0016,
      "step": 900
    },
    {
      "epoch": 0.01279110578542282,
      "grad_norm": 3.25,
      "learning_rate": 4.998811768369882e-05,
      "loss": 1.0555,
      "step": 902
    },
    {
      "epoch": 0.012819467439049034,
      "grad_norm": 4.6875,
      "learning_rate": 4.9988048706372096e-05,
      "loss": 0.997,
      "step": 904
    },
    {
      "epoch": 0.01284782909267525,
      "grad_norm": 3.4375,
      "learning_rate": 4.998797952946416e-05,
      "loss": 1.0924,
      "step": 906
    },
    {
      "epoch": 0.012876190746301463,
      "grad_norm": 3.484375,
      "learning_rate": 4.998791015297556e-05,
      "loss": 0.9572,
      "step": 908
    },
    {
      "epoch": 0.012904552399927677,
      "grad_norm": 2.859375,
      "learning_rate": 4.998784057690686e-05,
      "loss": 1.0163,
      "step": 910
    },
    {
      "epoch": 0.012932914053553892,
      "grad_norm": 3.046875,
      "learning_rate": 4.9987770801258617e-05,
      "loss": 1.0219,
      "step": 912
    },
    {
      "epoch": 0.012961275707180107,
      "grad_norm": 3.0,
      "learning_rate": 4.998770082603137e-05,
      "loss": 1.0196,
      "step": 914
    },
    {
      "epoch": 0.012989637360806322,
      "grad_norm": 3.328125,
      "learning_rate": 4.9987630651225694e-05,
      "loss": 1.0136,
      "step": 916
    },
    {
      "epoch": 0.013017999014432537,
      "grad_norm": 3.0,
      "learning_rate": 4.998756027684215e-05,
      "loss": 1.0315,
      "step": 918
    },
    {
      "epoch": 0.01304636066805875,
      "grad_norm": 3.0625,
      "learning_rate": 4.998748970288129e-05,
      "loss": 1.0534,
      "step": 920
    },
    {
      "epoch": 0.013074722321684966,
      "grad_norm": 3.6875,
      "learning_rate": 4.998741892934368e-05,
      "loss": 1.0377,
      "step": 922
    },
    {
      "epoch": 0.01310308397531118,
      "grad_norm": 3.40625,
      "learning_rate": 4.99873479562299e-05,
      "loss": 0.9951,
      "step": 924
    },
    {
      "epoch": 0.013131445628937395,
      "grad_norm": 3.28125,
      "learning_rate": 4.998727678354049e-05,
      "loss": 0.9822,
      "step": 926
    },
    {
      "epoch": 0.01315980728256361,
      "grad_norm": 3.484375,
      "learning_rate": 4.998720541127605e-05,
      "loss": 0.9618,
      "step": 928
    },
    {
      "epoch": 0.013188168936189824,
      "grad_norm": 2.734375,
      "learning_rate": 4.998713383943712e-05,
      "loss": 0.9923,
      "step": 930
    },
    {
      "epoch": 0.013216530589816039,
      "grad_norm": 3.5625,
      "learning_rate": 4.998706206802429e-05,
      "loss": 1.0026,
      "step": 932
    },
    {
      "epoch": 0.013244892243442254,
      "grad_norm": 3.625,
      "learning_rate": 4.9986990097038134e-05,
      "loss": 1.0192,
      "step": 934
    },
    {
      "epoch": 0.013273253897068469,
      "grad_norm": 3.25,
      "learning_rate": 4.998691792647922e-05,
      "loss": 1.0003,
      "step": 936
    },
    {
      "epoch": 0.013301615550694684,
      "grad_norm": 3.21875,
      "learning_rate": 4.9986845556348116e-05,
      "loss": 0.9926,
      "step": 938
    },
    {
      "epoch": 0.013329977204320899,
      "grad_norm": 3.375,
      "learning_rate": 4.998677298664542e-05,
      "loss": 0.987,
      "step": 940
    },
    {
      "epoch": 0.013358338857947112,
      "grad_norm": 2.84375,
      "learning_rate": 4.9986700217371694e-05,
      "loss": 1.0087,
      "step": 942
    },
    {
      "epoch": 0.013386700511573327,
      "grad_norm": 3.15625,
      "learning_rate": 4.998662724852753e-05,
      "loss": 1.018,
      "step": 944
    },
    {
      "epoch": 0.013415062165199542,
      "grad_norm": 3.234375,
      "learning_rate": 4.9986554080113516e-05,
      "loss": 1.0081,
      "step": 946
    },
    {
      "epoch": 0.013443423818825757,
      "grad_norm": 3.015625,
      "learning_rate": 4.998648071213021e-05,
      "loss": 0.9899,
      "step": 948
    },
    {
      "epoch": 0.013471785472451972,
      "grad_norm": 3.21875,
      "learning_rate": 4.998640714457823e-05,
      "loss": 1.0116,
      "step": 950
    },
    {
      "epoch": 0.013500147126078187,
      "grad_norm": 3.34375,
      "learning_rate": 4.998633337745815e-05,
      "loss": 0.9632,
      "step": 952
    },
    {
      "epoch": 0.0135285087797044,
      "grad_norm": 3.859375,
      "learning_rate": 4.998625941077055e-05,
      "loss": 1.0225,
      "step": 954
    },
    {
      "epoch": 0.013556870433330615,
      "grad_norm": 2.96875,
      "learning_rate": 4.9986185244516035e-05,
      "loss": 0.9978,
      "step": 956
    },
    {
      "epoch": 0.01358523208695683,
      "grad_norm": 3.03125,
      "learning_rate": 4.9986110878695194e-05,
      "loss": 1.009,
      "step": 958
    },
    {
      "epoch": 0.013613593740583045,
      "grad_norm": 3.109375,
      "learning_rate": 4.9986036313308614e-05,
      "loss": 0.9876,
      "step": 960
    },
    {
      "epoch": 0.01364195539420926,
      "grad_norm": 2.78125,
      "learning_rate": 4.99859615483569e-05,
      "loss": 1.0383,
      "step": 962
    },
    {
      "epoch": 0.013670317047835475,
      "grad_norm": 3.78125,
      "learning_rate": 4.9985886583840634e-05,
      "loss": 1.0225,
      "step": 964
    },
    {
      "epoch": 0.013698678701461688,
      "grad_norm": 3.375,
      "learning_rate": 4.998581141976044e-05,
      "loss": 0.9649,
      "step": 966
    },
    {
      "epoch": 0.013727040355087903,
      "grad_norm": 3.109375,
      "learning_rate": 4.998573605611689e-05,
      "loss": 1.0711,
      "step": 968
    },
    {
      "epoch": 0.013755402008714118,
      "grad_norm": 3.75,
      "learning_rate": 4.998566049291061e-05,
      "loss": 1.0343,
      "step": 970
    },
    {
      "epoch": 0.013783763662340333,
      "grad_norm": 3.265625,
      "learning_rate": 4.9985584730142185e-05,
      "loss": 1.0372,
      "step": 972
    },
    {
      "epoch": 0.013812125315966548,
      "grad_norm": 4.125,
      "learning_rate": 4.998550876781224e-05,
      "loss": 0.9944,
      "step": 974
    },
    {
      "epoch": 0.013840486969592763,
      "grad_norm": 3.984375,
      "learning_rate": 4.998543260592136e-05,
      "loss": 1.0489,
      "step": 976
    },
    {
      "epoch": 0.013868848623218976,
      "grad_norm": 3.625,
      "learning_rate": 4.9985356244470165e-05,
      "loss": 1.0604,
      "step": 978
    },
    {
      "epoch": 0.013897210276845191,
      "grad_norm": 3.265625,
      "learning_rate": 4.9985279683459264e-05,
      "loss": 1.0474,
      "step": 980
    },
    {
      "epoch": 0.013925571930471406,
      "grad_norm": 3.25,
      "learning_rate": 4.998520292288927e-05,
      "loss": 0.9992,
      "step": 982
    },
    {
      "epoch": 0.013953933584097621,
      "grad_norm": 5.125,
      "learning_rate": 4.9985125962760794e-05,
      "loss": 1.0326,
      "step": 984
    },
    {
      "epoch": 0.013982295237723836,
      "grad_norm": 3.1875,
      "learning_rate": 4.998504880307444e-05,
      "loss": 0.9677,
      "step": 986
    },
    {
      "epoch": 0.01401065689135005,
      "grad_norm": 3.5625,
      "learning_rate": 4.998497144383084e-05,
      "loss": 1.0452,
      "step": 988
    },
    {
      "epoch": 0.014039018544976264,
      "grad_norm": 3.375,
      "learning_rate": 4.998489388503061e-05,
      "loss": 1.0298,
      "step": 990
    },
    {
      "epoch": 0.014067380198602479,
      "grad_norm": 3.53125,
      "learning_rate": 4.998481612667437e-05,
      "loss": 1.0814,
      "step": 992
    },
    {
      "epoch": 0.014095741852228694,
      "grad_norm": 3.15625,
      "learning_rate": 4.998473816876273e-05,
      "loss": 1.0021,
      "step": 994
    },
    {
      "epoch": 0.014124103505854909,
      "grad_norm": 3.671875,
      "learning_rate": 4.998466001129632e-05,
      "loss": 1.0291,
      "step": 996
    },
    {
      "epoch": 0.014152465159481124,
      "grad_norm": 3.21875,
      "learning_rate": 4.9984581654275764e-05,
      "loss": 1.038,
      "step": 998
    },
    {
      "epoch": 0.014180826813107339,
      "grad_norm": 3.46875,
      "learning_rate": 4.99845030977017e-05,
      "loss": 1.0501,
      "step": 1000
    },
    {
      "epoch": 0.014209188466733552,
      "grad_norm": 3.25,
      "learning_rate": 4.9984424341574724e-05,
      "loss": 0.9588,
      "step": 1002
    },
    {
      "epoch": 0.014237550120359767,
      "grad_norm": 3.40625,
      "learning_rate": 4.998434538589549e-05,
      "loss": 1.0341,
      "step": 1004
    },
    {
      "epoch": 0.014265911773985982,
      "grad_norm": 3.84375,
      "learning_rate": 4.998426623066463e-05,
      "loss": 1.0583,
      "step": 1006
    },
    {
      "epoch": 0.014294273427612197,
      "grad_norm": 2.9375,
      "learning_rate": 4.9984186875882764e-05,
      "loss": 0.9681,
      "step": 1008
    },
    {
      "epoch": 0.014322635081238412,
      "grad_norm": 3.203125,
      "learning_rate": 4.998410732155053e-05,
      "loss": 1.0716,
      "step": 1010
    },
    {
      "epoch": 0.014350996734864627,
      "grad_norm": 3.078125,
      "learning_rate": 4.998402756766857e-05,
      "loss": 1.0494,
      "step": 1012
    },
    {
      "epoch": 0.01437935838849084,
      "grad_norm": 3.390625,
      "learning_rate": 4.998394761423751e-05,
      "loss": 0.9572,
      "step": 1014
    },
    {
      "epoch": 0.014407720042117055,
      "grad_norm": 3.734375,
      "learning_rate": 4.9983867461257996e-05,
      "loss": 1.0078,
      "step": 1016
    },
    {
      "epoch": 0.01443608169574327,
      "grad_norm": 3.46875,
      "learning_rate": 4.998378710873067e-05,
      "loss": 1.0536,
      "step": 1018
    },
    {
      "epoch": 0.014464443349369485,
      "grad_norm": 3.21875,
      "learning_rate": 4.9983706556656165e-05,
      "loss": 0.9553,
      "step": 1020
    },
    {
      "epoch": 0.0144928050029957,
      "grad_norm": 3.078125,
      "learning_rate": 4.998362580503513e-05,
      "loss": 1.0132,
      "step": 1022
    },
    {
      "epoch": 0.014521166656621915,
      "grad_norm": 3.765625,
      "learning_rate": 4.9983544853868205e-05,
      "loss": 1.0496,
      "step": 1024
    },
    {
      "epoch": 0.014549528310248128,
      "grad_norm": 3.125,
      "learning_rate": 4.9983463703156045e-05,
      "loss": 1.0338,
      "step": 1026
    },
    {
      "epoch": 0.014577889963874343,
      "grad_norm": 3.265625,
      "learning_rate": 4.99833823528993e-05,
      "loss": 1.0567,
      "step": 1028
    },
    {
      "epoch": 0.014606251617500558,
      "grad_norm": 3.3125,
      "learning_rate": 4.99833008030986e-05,
      "loss": 1.0348,
      "step": 1030
    },
    {
      "epoch": 0.014634613271126773,
      "grad_norm": 3.0625,
      "learning_rate": 4.9983219053754627e-05,
      "loss": 1.0154,
      "step": 1032
    },
    {
      "epoch": 0.014662974924752988,
      "grad_norm": 3.53125,
      "learning_rate": 4.9983137104868005e-05,
      "loss": 0.9694,
      "step": 1034
    },
    {
      "epoch": 0.014691336578379203,
      "grad_norm": 3.84375,
      "learning_rate": 4.99830549564394e-05,
      "loss": 1.0494,
      "step": 1036
    },
    {
      "epoch": 0.014719698232005416,
      "grad_norm": 3.375,
      "learning_rate": 4.9982972608469474e-05,
      "loss": 0.997,
      "step": 1038
    },
    {
      "epoch": 0.014748059885631631,
      "grad_norm": 3.484375,
      "learning_rate": 4.998289006095888e-05,
      "loss": 1.0016,
      "step": 1040
    },
    {
      "epoch": 0.014776421539257846,
      "grad_norm": 2.984375,
      "learning_rate": 4.9982807313908273e-05,
      "loss": 1.0062,
      "step": 1042
    },
    {
      "epoch": 0.014804783192884061,
      "grad_norm": 3.25,
      "learning_rate": 4.9982724367318315e-05,
      "loss": 1.0157,
      "step": 1044
    },
    {
      "epoch": 0.014833144846510276,
      "grad_norm": 3.28125,
      "learning_rate": 4.998264122118968e-05,
      "loss": 1.0084,
      "step": 1046
    },
    {
      "epoch": 0.014861506500136491,
      "grad_norm": 2.921875,
      "learning_rate": 4.9982557875523016e-05,
      "loss": 1.0217,
      "step": 1048
    },
    {
      "epoch": 0.014889868153762705,
      "grad_norm": 3.515625,
      "learning_rate": 4.9982474330319e-05,
      "loss": 0.9879,
      "step": 1050
    },
    {
      "epoch": 0.01491822980738892,
      "grad_norm": 3.796875,
      "learning_rate": 4.9982390585578295e-05,
      "loss": 1.0983,
      "step": 1052
    },
    {
      "epoch": 0.014946591461015134,
      "grad_norm": 2.6875,
      "learning_rate": 4.998230664130157e-05,
      "loss": 0.9426,
      "step": 1054
    },
    {
      "epoch": 0.01497495311464135,
      "grad_norm": 3.21875,
      "learning_rate": 4.99822224974895e-05,
      "loss": 1.0467,
      "step": 1056
    },
    {
      "epoch": 0.015003314768267564,
      "grad_norm": 3.21875,
      "learning_rate": 4.998213815414274e-05,
      "loss": 1.0172,
      "step": 1058
    },
    {
      "epoch": 0.01503167642189378,
      "grad_norm": 3.515625,
      "learning_rate": 4.998205361126198e-05,
      "loss": 1.003,
      "step": 1060
    },
    {
      "epoch": 0.015060038075519993,
      "grad_norm": 3.765625,
      "learning_rate": 4.99819688688479e-05,
      "loss": 1.0466,
      "step": 1062
    },
    {
      "epoch": 0.015088399729146208,
      "grad_norm": 3.359375,
      "learning_rate": 4.998188392690116e-05,
      "loss": 0.9869,
      "step": 1064
    },
    {
      "epoch": 0.015116761382772423,
      "grad_norm": 3.0,
      "learning_rate": 4.9981798785422445e-05,
      "loss": 0.9381,
      "step": 1066
    },
    {
      "epoch": 0.015145123036398637,
      "grad_norm": 3.390625,
      "learning_rate": 4.998171344441244e-05,
      "loss": 1.0238,
      "step": 1068
    },
    {
      "epoch": 0.015173484690024852,
      "grad_norm": 3.125,
      "learning_rate": 4.998162790387183e-05,
      "loss": 0.9599,
      "step": 1070
    },
    {
      "epoch": 0.015201846343651067,
      "grad_norm": 3.40625,
      "learning_rate": 4.998154216380129e-05,
      "loss": 1.0138,
      "step": 1072
    },
    {
      "epoch": 0.01523020799727728,
      "grad_norm": 3.28125,
      "learning_rate": 4.998145622420149e-05,
      "loss": 0.9546,
      "step": 1074
    },
    {
      "epoch": 0.015258569650903496,
      "grad_norm": 3.46875,
      "learning_rate": 4.998137008507316e-05,
      "loss": 1.0203,
      "step": 1076
    },
    {
      "epoch": 0.01528693130452971,
      "grad_norm": 3.09375,
      "learning_rate": 4.998128374641694e-05,
      "loss": 0.9999,
      "step": 1078
    },
    {
      "epoch": 0.015315292958155926,
      "grad_norm": 3.03125,
      "learning_rate": 4.998119720823354e-05,
      "loss": 0.9725,
      "step": 1080
    },
    {
      "epoch": 0.01534365461178214,
      "grad_norm": 3.25,
      "learning_rate": 4.998111047052366e-05,
      "loss": 0.9585,
      "step": 1082
    },
    {
      "epoch": 0.015372016265408354,
      "grad_norm": 3.625,
      "learning_rate": 4.9981023533287984e-05,
      "loss": 1.0276,
      "step": 1084
    },
    {
      "epoch": 0.015400377919034569,
      "grad_norm": 3.265625,
      "learning_rate": 4.998093639652721e-05,
      "loss": 1.0176,
      "step": 1086
    },
    {
      "epoch": 0.015428739572660784,
      "grad_norm": 3.140625,
      "learning_rate": 4.998084906024202e-05,
      "loss": 0.9741,
      "step": 1088
    },
    {
      "epoch": 0.015457101226286999,
      "grad_norm": 3.71875,
      "learning_rate": 4.998076152443313e-05,
      "loss": 0.9763,
      "step": 1090
    },
    {
      "epoch": 0.015485462879913214,
      "grad_norm": 2.921875,
      "learning_rate": 4.9980673789101234e-05,
      "loss": 0.9724,
      "step": 1092
    },
    {
      "epoch": 0.015513824533539429,
      "grad_norm": 3.796875,
      "learning_rate": 4.9980585854247025e-05,
      "loss": 1.0816,
      "step": 1094
    },
    {
      "epoch": 0.015542186187165642,
      "grad_norm": 3.5625,
      "learning_rate": 4.998049771987121e-05,
      "loss": 1.0342,
      "step": 1096
    },
    {
      "epoch": 0.015570547840791857,
      "grad_norm": 3.28125,
      "learning_rate": 4.99804093859745e-05,
      "loss": 1.0689,
      "step": 1098
    },
    {
      "epoch": 0.015598909494418072,
      "grad_norm": 3.453125,
      "learning_rate": 4.998032085255759e-05,
      "loss": 0.9524,
      "step": 1100
    },
    {
      "epoch": 0.015627271148044287,
      "grad_norm": 3.203125,
      "learning_rate": 4.998023211962119e-05,
      "loss": 1.0013,
      "step": 1102
    },
    {
      "epoch": 0.0156556328016705,
      "grad_norm": 2.875,
      "learning_rate": 4.998014318716601e-05,
      "loss": 0.9643,
      "step": 1104
    },
    {
      "epoch": 0.015683994455296717,
      "grad_norm": 3.8125,
      "learning_rate": 4.998005405519276e-05,
      "loss": 0.9961,
      "step": 1106
    },
    {
      "epoch": 0.01571235610892293,
      "grad_norm": 3.234375,
      "learning_rate": 4.997996472370215e-05,
      "loss": 0.9969,
      "step": 1108
    },
    {
      "epoch": 0.015740717762549147,
      "grad_norm": 3.171875,
      "learning_rate": 4.9979875192694904e-05,
      "loss": 1.0208,
      "step": 1110
    },
    {
      "epoch": 0.01576907941617536,
      "grad_norm": 2.984375,
      "learning_rate": 4.997978546217172e-05,
      "loss": 0.9547,
      "step": 1112
    },
    {
      "epoch": 0.015797441069801573,
      "grad_norm": 3.40625,
      "learning_rate": 4.997969553213333e-05,
      "loss": 1.0023,
      "step": 1114
    },
    {
      "epoch": 0.01582580272342779,
      "grad_norm": 2.75,
      "learning_rate": 4.997960540258044e-05,
      "loss": 1.0083,
      "step": 1116
    },
    {
      "epoch": 0.015854164377054003,
      "grad_norm": 3.15625,
      "learning_rate": 4.997951507351378e-05,
      "loss": 0.9814,
      "step": 1118
    },
    {
      "epoch": 0.01588252603068022,
      "grad_norm": 3.28125,
      "learning_rate": 4.997942454493406e-05,
      "loss": 1.0643,
      "step": 1120
    },
    {
      "epoch": 0.015910887684306433,
      "grad_norm": 3.5,
      "learning_rate": 4.997933381684202e-05,
      "loss": 0.9972,
      "step": 1122
    },
    {
      "epoch": 0.01593924933793265,
      "grad_norm": 3.609375,
      "learning_rate": 4.997924288923836e-05,
      "loss": 1.0059,
      "step": 1124
    },
    {
      "epoch": 0.015967610991558863,
      "grad_norm": 3.515625,
      "learning_rate": 4.9979151762123834e-05,
      "loss": 1.0025,
      "step": 1126
    },
    {
      "epoch": 0.015995972645185076,
      "grad_norm": 3.203125,
      "learning_rate": 4.997906043549915e-05,
      "loss": 1.0496,
      "step": 1128
    },
    {
      "epoch": 0.016024334298811293,
      "grad_norm": 3.140625,
      "learning_rate": 4.9978968909365054e-05,
      "loss": 1.0303,
      "step": 1130
    },
    {
      "epoch": 0.016052695952437506,
      "grad_norm": 3.234375,
      "learning_rate": 4.997887718372226e-05,
      "loss": 1.0127,
      "step": 1132
    },
    {
      "epoch": 0.016081057606063723,
      "grad_norm": 3.34375,
      "learning_rate": 4.99787852585715e-05,
      "loss": 0.9693,
      "step": 1134
    },
    {
      "epoch": 0.016109419259689936,
      "grad_norm": 3.3125,
      "learning_rate": 4.997869313391353e-05,
      "loss": 1.0744,
      "step": 1136
    },
    {
      "epoch": 0.01613778091331615,
      "grad_norm": 3.28125,
      "learning_rate": 4.997860080974907e-05,
      "loss": 1.008,
      "step": 1138
    },
    {
      "epoch": 0.016166142566942366,
      "grad_norm": 3.15625,
      "learning_rate": 4.997850828607885e-05,
      "loss": 0.9879,
      "step": 1140
    },
    {
      "epoch": 0.01619450422056858,
      "grad_norm": 3.484375,
      "learning_rate": 4.997841556290362e-05,
      "loss": 0.9935,
      "step": 1142
    },
    {
      "epoch": 0.016222865874194796,
      "grad_norm": 3.140625,
      "learning_rate": 4.997832264022413e-05,
      "loss": 0.9493,
      "step": 1144
    },
    {
      "epoch": 0.01625122752782101,
      "grad_norm": 3.828125,
      "learning_rate": 4.9978229518041096e-05,
      "loss": 1.0122,
      "step": 1146
    },
    {
      "epoch": 0.016279589181447226,
      "grad_norm": 3.1875,
      "learning_rate": 4.997813619635529e-05,
      "loss": 0.9721,
      "step": 1148
    },
    {
      "epoch": 0.01630795083507344,
      "grad_norm": 3.046875,
      "learning_rate": 4.997804267516744e-05,
      "loss": 0.9836,
      "step": 1150
    },
    {
      "epoch": 0.016336312488699652,
      "grad_norm": 3.546875,
      "learning_rate": 4.997794895447829e-05,
      "loss": 1.0038,
      "step": 1152
    },
    {
      "epoch": 0.01636467414232587,
      "grad_norm": 3.359375,
      "learning_rate": 4.9977855034288603e-05,
      "loss": 1.0617,
      "step": 1154
    },
    {
      "epoch": 0.016393035795952082,
      "grad_norm": 3.234375,
      "learning_rate": 4.997776091459912e-05,
      "loss": 1.0078,
      "step": 1156
    },
    {
      "epoch": 0.0164213974495783,
      "grad_norm": 3.265625,
      "learning_rate": 4.99776665954106e-05,
      "loss": 1.0223,
      "step": 1158
    },
    {
      "epoch": 0.016449759103204512,
      "grad_norm": 3.4375,
      "learning_rate": 4.997757207672378e-05,
      "loss": 0.9568,
      "step": 1160
    },
    {
      "epoch": 0.016478120756830725,
      "grad_norm": 3.4375,
      "learning_rate": 4.997747735853943e-05,
      "loss": 0.9982,
      "step": 1162
    },
    {
      "epoch": 0.016506482410456942,
      "grad_norm": 3.5,
      "learning_rate": 4.99773824408583e-05,
      "loss": 1.0043,
      "step": 1164
    },
    {
      "epoch": 0.016534844064083155,
      "grad_norm": 4.03125,
      "learning_rate": 4.997728732368116e-05,
      "loss": 1.0374,
      "step": 1166
    },
    {
      "epoch": 0.016563205717709372,
      "grad_norm": 2.84375,
      "learning_rate": 4.997719200700875e-05,
      "loss": 0.9627,
      "step": 1168
    },
    {
      "epoch": 0.016591567371335585,
      "grad_norm": 3.140625,
      "learning_rate": 4.9977096490841846e-05,
      "loss": 1.0374,
      "step": 1170
    },
    {
      "epoch": 0.016619929024961802,
      "grad_norm": 2.859375,
      "learning_rate": 4.99770007751812e-05,
      "loss": 0.9399,
      "step": 1172
    },
    {
      "epoch": 0.016648290678588015,
      "grad_norm": 4.09375,
      "learning_rate": 4.9976904860027595e-05,
      "loss": 1.0486,
      "step": 1174
    },
    {
      "epoch": 0.01667665233221423,
      "grad_norm": 3.53125,
      "learning_rate": 4.997680874538178e-05,
      "loss": 1.0026,
      "step": 1176
    },
    {
      "epoch": 0.016705013985840445,
      "grad_norm": 3.328125,
      "learning_rate": 4.997671243124452e-05,
      "loss": 1.0324,
      "step": 1178
    },
    {
      "epoch": 0.01673337563946666,
      "grad_norm": 3.21875,
      "learning_rate": 4.9976615917616595e-05,
      "loss": 1.0327,
      "step": 1180
    },
    {
      "epoch": 0.016761737293092875,
      "grad_norm": 2.9375,
      "learning_rate": 4.997651920449878e-05,
      "loss": 0.9776,
      "step": 1182
    },
    {
      "epoch": 0.01679009894671909,
      "grad_norm": 2.75,
      "learning_rate": 4.9976422291891834e-05,
      "loss": 0.929,
      "step": 1184
    },
    {
      "epoch": 0.0168184606003453,
      "grad_norm": 3.265625,
      "learning_rate": 4.997632517979654e-05,
      "loss": 0.9847,
      "step": 1186
    },
    {
      "epoch": 0.01684682225397152,
      "grad_norm": 3.21875,
      "learning_rate": 4.997622786821367e-05,
      "loss": 1.0331,
      "step": 1188
    },
    {
      "epoch": 0.01687518390759773,
      "grad_norm": 3.390625,
      "learning_rate": 4.9976130357144e-05,
      "loss": 1.0048,
      "step": 1190
    },
    {
      "epoch": 0.016903545561223948,
      "grad_norm": 3.1875,
      "learning_rate": 4.997603264658832e-05,
      "loss": 0.9351,
      "step": 1192
    },
    {
      "epoch": 0.01693190721485016,
      "grad_norm": 3.484375,
      "learning_rate": 4.997593473654739e-05,
      "loss": 0.987,
      "step": 1194
    },
    {
      "epoch": 0.016960268868476378,
      "grad_norm": 3.328125,
      "learning_rate": 4.997583662702201e-05,
      "loss": 1.0626,
      "step": 1196
    },
    {
      "epoch": 0.01698863052210259,
      "grad_norm": 3.21875,
      "learning_rate": 4.997573831801296e-05,
      "loss": 1.0148,
      "step": 1198
    },
    {
      "epoch": 0.017016992175728805,
      "grad_norm": 3.09375,
      "learning_rate": 4.9975639809521005e-05,
      "loss": 0.9608,
      "step": 1200
    },
    {
      "epoch": 0.01704535382935502,
      "grad_norm": 2.90625,
      "learning_rate": 4.997554110154696e-05,
      "loss": 0.9711,
      "step": 1202
    },
    {
      "epoch": 0.017073715482981235,
      "grad_norm": 3.265625,
      "learning_rate": 4.997544219409161e-05,
      "loss": 1.0109,
      "step": 1204
    },
    {
      "epoch": 0.01710207713660745,
      "grad_norm": 3.453125,
      "learning_rate": 4.9975343087155726e-05,
      "loss": 1.0121,
      "step": 1206
    },
    {
      "epoch": 0.017130438790233665,
      "grad_norm": 3.6875,
      "learning_rate": 4.997524378074012e-05,
      "loss": 0.9851,
      "step": 1208
    },
    {
      "epoch": 0.017158800443859878,
      "grad_norm": 3.515625,
      "learning_rate": 4.9975144274845565e-05,
      "loss": 0.9395,
      "step": 1210
    },
    {
      "epoch": 0.017187162097486094,
      "grad_norm": 3.171875,
      "learning_rate": 4.997504456947287e-05,
      "loss": 0.9037,
      "step": 1212
    },
    {
      "epoch": 0.017215523751112308,
      "grad_norm": 3.203125,
      "learning_rate": 4.997494466462284e-05,
      "loss": 1.0069,
      "step": 1214
    },
    {
      "epoch": 0.017243885404738524,
      "grad_norm": 4.0,
      "learning_rate": 4.997484456029625e-05,
      "loss": 0.9966,
      "step": 1216
    },
    {
      "epoch": 0.017272247058364738,
      "grad_norm": 3.109375,
      "learning_rate": 4.997474425649391e-05,
      "loss": 0.9974,
      "step": 1218
    },
    {
      "epoch": 0.017300608711990954,
      "grad_norm": 2.96875,
      "learning_rate": 4.997464375321663e-05,
      "loss": 1.0135,
      "step": 1220
    },
    {
      "epoch": 0.017328970365617168,
      "grad_norm": 3.140625,
      "learning_rate": 4.997454305046519e-05,
      "loss": 0.9954,
      "step": 1222
    },
    {
      "epoch": 0.01735733201924338,
      "grad_norm": 3.0625,
      "learning_rate": 4.997444214824042e-05,
      "loss": 0.978,
      "step": 1224
    },
    {
      "epoch": 0.017385693672869598,
      "grad_norm": 3.15625,
      "learning_rate": 4.997434104654312e-05,
      "loss": 1.0173,
      "step": 1226
    },
    {
      "epoch": 0.01741405532649581,
      "grad_norm": 3.546875,
      "learning_rate": 4.997423974537408e-05,
      "loss": 1.0004,
      "step": 1228
    },
    {
      "epoch": 0.017442416980122027,
      "grad_norm": 3.25,
      "learning_rate": 4.997413824473413e-05,
      "loss": 1.0244,
      "step": 1230
    },
    {
      "epoch": 0.01747077863374824,
      "grad_norm": 3.65625,
      "learning_rate": 4.9974036544624063e-05,
      "loss": 1.0044,
      "step": 1232
    },
    {
      "epoch": 0.017499140287374454,
      "grad_norm": 3.25,
      "learning_rate": 4.9973934645044706e-05,
      "loss": 0.9482,
      "step": 1234
    },
    {
      "epoch": 0.01752750194100067,
      "grad_norm": 3.296875,
      "learning_rate": 4.997383254599687e-05,
      "loss": 0.995,
      "step": 1236
    },
    {
      "epoch": 0.017555863594626884,
      "grad_norm": 3.09375,
      "learning_rate": 4.9973730247481365e-05,
      "loss": 0.9939,
      "step": 1238
    },
    {
      "epoch": 0.0175842252482531,
      "grad_norm": 3.140625,
      "learning_rate": 4.997362774949901e-05,
      "loss": 1.0545,
      "step": 1240
    },
    {
      "epoch": 0.017612586901879314,
      "grad_norm": 3.53125,
      "learning_rate": 4.997352505205062e-05,
      "loss": 0.9429,
      "step": 1242
    },
    {
      "epoch": 0.017640948555505527,
      "grad_norm": 3.015625,
      "learning_rate": 4.997342215513703e-05,
      "loss": 1.006,
      "step": 1244
    },
    {
      "epoch": 0.017669310209131744,
      "grad_norm": 3.015625,
      "learning_rate": 4.9973319058759045e-05,
      "loss": 1.0089,
      "step": 1246
    },
    {
      "epoch": 0.017697671862757957,
      "grad_norm": 2.75,
      "learning_rate": 4.99732157629175e-05,
      "loss": 0.9778,
      "step": 1248
    },
    {
      "epoch": 0.017726033516384174,
      "grad_norm": 3.28125,
      "learning_rate": 4.9973112267613206e-05,
      "loss": 0.9763,
      "step": 1250
    },
    {
      "epoch": 0.017754395170010387,
      "grad_norm": 3.515625,
      "learning_rate": 4.9973008572847e-05,
      "loss": 0.9798,
      "step": 1252
    },
    {
      "epoch": 0.017782756823636604,
      "grad_norm": 3.171875,
      "learning_rate": 4.997290467861971e-05,
      "loss": 0.955,
      "step": 1254
    },
    {
      "epoch": 0.017811118477262817,
      "grad_norm": 3.078125,
      "learning_rate": 4.997280058493217e-05,
      "loss": 0.9724,
      "step": 1256
    },
    {
      "epoch": 0.01783948013088903,
      "grad_norm": 3.203125,
      "learning_rate": 4.9972696291785204e-05,
      "loss": 0.9792,
      "step": 1258
    },
    {
      "epoch": 0.017867841784515247,
      "grad_norm": 2.859375,
      "learning_rate": 4.997259179917964e-05,
      "loss": 0.9505,
      "step": 1260
    },
    {
      "epoch": 0.01789620343814146,
      "grad_norm": 3.234375,
      "learning_rate": 4.9972487107116336e-05,
      "loss": 0.9388,
      "step": 1262
    },
    {
      "epoch": 0.017924565091767677,
      "grad_norm": 3.609375,
      "learning_rate": 4.99723822155961e-05,
      "loss": 1.0486,
      "step": 1264
    },
    {
      "epoch": 0.01795292674539389,
      "grad_norm": 3.34375,
      "learning_rate": 4.997227712461978e-05,
      "loss": 1.0313,
      "step": 1266
    },
    {
      "epoch": 0.017981288399020103,
      "grad_norm": 3.578125,
      "learning_rate": 4.997217183418822e-05,
      "loss": 1.0311,
      "step": 1268
    },
    {
      "epoch": 0.01800965005264632,
      "grad_norm": 3.828125,
      "learning_rate": 4.997206634430226e-05,
      "loss": 1.0077,
      "step": 1270
    },
    {
      "epoch": 0.018038011706272533,
      "grad_norm": 4.40625,
      "learning_rate": 4.997196065496274e-05,
      "loss": 1.0111,
      "step": 1272
    },
    {
      "epoch": 0.01806637335989875,
      "grad_norm": 3.15625,
      "learning_rate": 4.99718547661705e-05,
      "loss": 0.9822,
      "step": 1274
    },
    {
      "epoch": 0.018094735013524963,
      "grad_norm": 3.65625,
      "learning_rate": 4.9971748677926396e-05,
      "loss": 0.998,
      "step": 1276
    },
    {
      "epoch": 0.01812309666715118,
      "grad_norm": 2.8125,
      "learning_rate": 4.9971642390231266e-05,
      "loss": 0.9695,
      "step": 1278
    },
    {
      "epoch": 0.018151458320777393,
      "grad_norm": 3.078125,
      "learning_rate": 4.9971535903085966e-05,
      "loss": 1.0122,
      "step": 1280
    },
    {
      "epoch": 0.018179819974403606,
      "grad_norm": 3.390625,
      "learning_rate": 4.997142921649134e-05,
      "loss": 0.9692,
      "step": 1282
    },
    {
      "epoch": 0.018208181628029823,
      "grad_norm": 3.0625,
      "learning_rate": 4.997132233044825e-05,
      "loss": 0.9654,
      "step": 1284
    },
    {
      "epoch": 0.018236543281656036,
      "grad_norm": 3.53125,
      "learning_rate": 4.9971215244957525e-05,
      "loss": 1.0124,
      "step": 1286
    },
    {
      "epoch": 0.018264904935282253,
      "grad_norm": 3.5,
      "learning_rate": 4.9971107960020054e-05,
      "loss": 1.0268,
      "step": 1288
    },
    {
      "epoch": 0.018293266588908466,
      "grad_norm": 3.390625,
      "learning_rate": 4.9971000475636674e-05,
      "loss": 0.964,
      "step": 1290
    },
    {
      "epoch": 0.01832162824253468,
      "grad_norm": 3.0,
      "learning_rate": 4.997089279180825e-05,
      "loss": 0.9454,
      "step": 1292
    },
    {
      "epoch": 0.018349989896160896,
      "grad_norm": 3.546875,
      "learning_rate": 4.997078490853563e-05,
      "loss": 1.0075,
      "step": 1294
    },
    {
      "epoch": 0.01837835154978711,
      "grad_norm": 2.890625,
      "learning_rate": 4.99706768258197e-05,
      "loss": 0.9953,
      "step": 1296
    },
    {
      "epoch": 0.018406713203413326,
      "grad_norm": 3.140625,
      "learning_rate": 4.99705685436613e-05,
      "loss": 0.9705,
      "step": 1298
    },
    {
      "epoch": 0.01843507485703954,
      "grad_norm": 3.34375,
      "learning_rate": 4.99704600620613e-05,
      "loss": 0.977,
      "step": 1300
    },
    {
      "epoch": 0.018463436510665756,
      "grad_norm": 3.25,
      "learning_rate": 4.997035138102058e-05,
      "loss": 1.0046,
      "step": 1302
    },
    {
      "epoch": 0.01849179816429197,
      "grad_norm": 3.015625,
      "learning_rate": 4.9970242500539985e-05,
      "loss": 0.9558,
      "step": 1304
    },
    {
      "epoch": 0.018520159817918182,
      "grad_norm": 3.421875,
      "learning_rate": 4.9970133420620415e-05,
      "loss": 0.9802,
      "step": 1306
    },
    {
      "epoch": 0.0185485214715444,
      "grad_norm": 2.984375,
      "learning_rate": 4.997002414126271e-05,
      "loss": 1.0092,
      "step": 1308
    },
    {
      "epoch": 0.018576883125170612,
      "grad_norm": 3.671875,
      "learning_rate": 4.9969914662467764e-05,
      "loss": 1.0219,
      "step": 1310
    },
    {
      "epoch": 0.01860524477879683,
      "grad_norm": 3.328125,
      "learning_rate": 4.996980498423644e-05,
      "loss": 1.0479,
      "step": 1312
    },
    {
      "epoch": 0.018633606432423042,
      "grad_norm": 3.25,
      "learning_rate": 4.9969695106569616e-05,
      "loss": 1.0206,
      "step": 1314
    },
    {
      "epoch": 0.018661968086049256,
      "grad_norm": 3.328125,
      "learning_rate": 4.996958502946818e-05,
      "loss": 0.986,
      "step": 1316
    },
    {
      "epoch": 0.018690329739675472,
      "grad_norm": 3.03125,
      "learning_rate": 4.9969474752933e-05,
      "loss": 0.9434,
      "step": 1318
    },
    {
      "epoch": 0.018718691393301685,
      "grad_norm": 2.90625,
      "learning_rate": 4.996936427696495e-05,
      "loss": 1.0413,
      "step": 1320
    },
    {
      "epoch": 0.018747053046927902,
      "grad_norm": 3.109375,
      "learning_rate": 4.9969253601564935e-05,
      "loss": 1.0068,
      "step": 1322
    },
    {
      "epoch": 0.018775414700554115,
      "grad_norm": 3.5,
      "learning_rate": 4.996914272673382e-05,
      "loss": 0.9977,
      "step": 1324
    },
    {
      "epoch": 0.018803776354180332,
      "grad_norm": 3.359375,
      "learning_rate": 4.9969031652472495e-05,
      "loss": 1.0753,
      "step": 1326
    },
    {
      "epoch": 0.018832138007806545,
      "grad_norm": 2.96875,
      "learning_rate": 4.996892037878185e-05,
      "loss": 0.975,
      "step": 1328
    },
    {
      "epoch": 0.01886049966143276,
      "grad_norm": 3.59375,
      "learning_rate": 4.9968808905662776e-05,
      "loss": 1.0303,
      "step": 1330
    },
    {
      "epoch": 0.018888861315058975,
      "grad_norm": 3.546875,
      "learning_rate": 4.996869723311616e-05,
      "loss": 1.0193,
      "step": 1332
    },
    {
      "epoch": 0.01891722296868519,
      "grad_norm": 3.359375,
      "learning_rate": 4.9968585361142885e-05,
      "loss": 1.0402,
      "step": 1334
    },
    {
      "epoch": 0.018945584622311405,
      "grad_norm": 3.265625,
      "learning_rate": 4.996847328974386e-05,
      "loss": 0.9332,
      "step": 1336
    },
    {
      "epoch": 0.01897394627593762,
      "grad_norm": 3.4375,
      "learning_rate": 4.996836101891997e-05,
      "loss": 1.0211,
      "step": 1338
    },
    {
      "epoch": 0.01900230792956383,
      "grad_norm": 3.625,
      "learning_rate": 4.996824854867212e-05,
      "loss": 1.0175,
      "step": 1340
    },
    {
      "epoch": 0.01903066958319005,
      "grad_norm": 3.8125,
      "learning_rate": 4.99681358790012e-05,
      "loss": 0.9963,
      "step": 1342
    },
    {
      "epoch": 0.01905903123681626,
      "grad_norm": 3.140625,
      "learning_rate": 4.996802300990811e-05,
      "loss": 1.0322,
      "step": 1344
    },
    {
      "epoch": 0.01908739289044248,
      "grad_norm": 3.953125,
      "learning_rate": 4.9967909941393764e-05,
      "loss": 0.9775,
      "step": 1346
    },
    {
      "epoch": 0.01911575454406869,
      "grad_norm": 3.109375,
      "learning_rate": 4.996779667345905e-05,
      "loss": 0.9432,
      "step": 1348
    },
    {
      "epoch": 0.01914411619769491,
      "grad_norm": 3.203125,
      "learning_rate": 4.996768320610489e-05,
      "loss": 0.9846,
      "step": 1350
    },
    {
      "epoch": 0.01917247785132112,
      "grad_norm": 3.171875,
      "learning_rate": 4.996756953933216e-05,
      "loss": 1.0315,
      "step": 1352
    },
    {
      "epoch": 0.019200839504947335,
      "grad_norm": 2.9375,
      "learning_rate": 4.9967455673141795e-05,
      "loss": 0.9563,
      "step": 1354
    },
    {
      "epoch": 0.01922920115857355,
      "grad_norm": 3.046875,
      "learning_rate": 4.9967341607534704e-05,
      "loss": 0.9794,
      "step": 1356
    },
    {
      "epoch": 0.019257562812199765,
      "grad_norm": 3.46875,
      "learning_rate": 4.996722734251178e-05,
      "loss": 1.0363,
      "step": 1358
    },
    {
      "epoch": 0.01928592446582598,
      "grad_norm": 3.1875,
      "learning_rate": 4.9967112878073944e-05,
      "loss": 1.0626,
      "step": 1360
    },
    {
      "epoch": 0.019314286119452195,
      "grad_norm": 3.078125,
      "learning_rate": 4.996699821422212e-05,
      "loss": 0.97,
      "step": 1362
    },
    {
      "epoch": 0.019342647773078408,
      "grad_norm": 3.546875,
      "learning_rate": 4.996688335095722e-05,
      "loss": 0.9886,
      "step": 1364
    },
    {
      "epoch": 0.019371009426704625,
      "grad_norm": 3.890625,
      "learning_rate": 4.996676828828014e-05,
      "loss": 0.95,
      "step": 1366
    },
    {
      "epoch": 0.019399371080330838,
      "grad_norm": 3.390625,
      "learning_rate": 4.996665302619183e-05,
      "loss": 0.9793,
      "step": 1368
    },
    {
      "epoch": 0.019427732733957054,
      "grad_norm": 3.125,
      "learning_rate": 4.9966537564693194e-05,
      "loss": 1.0197,
      "step": 1370
    },
    {
      "epoch": 0.019456094387583268,
      "grad_norm": 3.03125,
      "learning_rate": 4.996642190378515e-05,
      "loss": 1.0323,
      "step": 1372
    },
    {
      "epoch": 0.019484456041209484,
      "grad_norm": 3.078125,
      "learning_rate": 4.996630604346864e-05,
      "loss": 0.9732,
      "step": 1374
    },
    {
      "epoch": 0.019512817694835698,
      "grad_norm": 3.265625,
      "learning_rate": 4.996618998374458e-05,
      "loss": 1.0177,
      "step": 1376
    },
    {
      "epoch": 0.01954117934846191,
      "grad_norm": 2.921875,
      "learning_rate": 4.996607372461388e-05,
      "loss": 1.0088,
      "step": 1378
    },
    {
      "epoch": 0.019569541002088128,
      "grad_norm": 3.265625,
      "learning_rate": 4.99659572660775e-05,
      "loss": 0.9963,
      "step": 1380
    },
    {
      "epoch": 0.01959790265571434,
      "grad_norm": 3.328125,
      "learning_rate": 4.996584060813635e-05,
      "loss": 1.0344,
      "step": 1382
    },
    {
      "epoch": 0.019626264309340558,
      "grad_norm": 3.359375,
      "learning_rate": 4.996572375079136e-05,
      "loss": 0.9972,
      "step": 1384
    },
    {
      "epoch": 0.01965462596296677,
      "grad_norm": 4.15625,
      "learning_rate": 4.9965606694043474e-05,
      "loss": 1.049,
      "step": 1386
    },
    {
      "epoch": 0.019682987616592984,
      "grad_norm": 3.390625,
      "learning_rate": 4.996548943789362e-05,
      "loss": 0.9603,
      "step": 1388
    },
    {
      "epoch": 0.0197113492702192,
      "grad_norm": 3.03125,
      "learning_rate": 4.9965371982342735e-05,
      "loss": 0.9737,
      "step": 1390
    },
    {
      "epoch": 0.019739710923845414,
      "grad_norm": 3.65625,
      "learning_rate": 4.996525432739176e-05,
      "loss": 1.0291,
      "step": 1392
    },
    {
      "epoch": 0.01976807257747163,
      "grad_norm": 3.453125,
      "learning_rate": 4.996513647304164e-05,
      "loss": 1.0009,
      "step": 1394
    },
    {
      "epoch": 0.019796434231097844,
      "grad_norm": 3.65625,
      "learning_rate": 4.9965018419293306e-05,
      "loss": 0.998,
      "step": 1396
    },
    {
      "epoch": 0.019824795884724057,
      "grad_norm": 3.3125,
      "learning_rate": 4.99649001661477e-05,
      "loss": 0.9731,
      "step": 1398
    },
    {
      "epoch": 0.019853157538350274,
      "grad_norm": 2.984375,
      "learning_rate": 4.996478171360577e-05,
      "loss": 0.9657,
      "step": 1400
    },
    {
      "epoch": 0.019881519191976487,
      "grad_norm": 3.03125,
      "learning_rate": 4.996466306166847e-05,
      "loss": 0.9758,
      "step": 1402
    },
    {
      "epoch": 0.019909880845602704,
      "grad_norm": 3.125,
      "learning_rate": 4.996454421033674e-05,
      "loss": 1.0365,
      "step": 1404
    },
    {
      "epoch": 0.019938242499228917,
      "grad_norm": 3.234375,
      "learning_rate": 4.996442515961153e-05,
      "loss": 1.0159,
      "step": 1406
    },
    {
      "epoch": 0.019966604152855134,
      "grad_norm": 2.90625,
      "learning_rate": 4.996430590949379e-05,
      "loss": 0.9615,
      "step": 1408
    },
    {
      "epoch": 0.019994965806481347,
      "grad_norm": 3.15625,
      "learning_rate": 4.9964186459984474e-05,
      "loss": 1.0178,
      "step": 1410
    },
    {
      "epoch": 0.02002332746010756,
      "grad_norm": 3.25,
      "learning_rate": 4.996406681108453e-05,
      "loss": 0.9805,
      "step": 1412
    },
    {
      "epoch": 0.020051689113733777,
      "grad_norm": 3.234375,
      "learning_rate": 4.9963946962794925e-05,
      "loss": 1.0389,
      "step": 1414
    },
    {
      "epoch": 0.02008005076735999,
      "grad_norm": 3.046875,
      "learning_rate": 4.996382691511661e-05,
      "loss": 0.9989,
      "step": 1416
    },
    {
      "epoch": 0.020108412420986207,
      "grad_norm": 3.265625,
      "learning_rate": 4.996370666805055e-05,
      "loss": 1.0183,
      "step": 1418
    },
    {
      "epoch": 0.02013677407461242,
      "grad_norm": 3.71875,
      "learning_rate": 4.996358622159769e-05,
      "loss": 1.0716,
      "step": 1420
    },
    {
      "epoch": 0.020165135728238633,
      "grad_norm": 3.609375,
      "learning_rate": 4.9963465575759006e-05,
      "loss": 1.0465,
      "step": 1422
    },
    {
      "epoch": 0.02019349738186485,
      "grad_norm": 3.25,
      "learning_rate": 4.9963344730535455e-05,
      "loss": 0.9909,
      "step": 1424
    },
    {
      "epoch": 0.020221859035491063,
      "grad_norm": 3.0,
      "learning_rate": 4.9963223685928005e-05,
      "loss": 0.9544,
      "step": 1426
    },
    {
      "epoch": 0.02025022068911728,
      "grad_norm": 3.515625,
      "learning_rate": 4.9963102441937626e-05,
      "loss": 0.9657,
      "step": 1428
    },
    {
      "epoch": 0.020278582342743493,
      "grad_norm": 3.203125,
      "learning_rate": 4.996298099856528e-05,
      "loss": 0.9926,
      "step": 1430
    },
    {
      "epoch": 0.02030694399636971,
      "grad_norm": 3.3125,
      "learning_rate": 4.996285935581194e-05,
      "loss": 1.0087,
      "step": 1432
    },
    {
      "epoch": 0.020335305649995923,
      "grad_norm": 3.109375,
      "learning_rate": 4.996273751367858e-05,
      "loss": 0.9596,
      "step": 1434
    },
    {
      "epoch": 0.020363667303622136,
      "grad_norm": 3.328125,
      "learning_rate": 4.9962615472166166e-05,
      "loss": 1.0273,
      "step": 1436
    },
    {
      "epoch": 0.020392028957248353,
      "grad_norm": 2.859375,
      "learning_rate": 4.996249323127568e-05,
      "loss": 0.9797,
      "step": 1438
    },
    {
      "epoch": 0.020420390610874566,
      "grad_norm": 3.3125,
      "learning_rate": 4.9962370791008093e-05,
      "loss": 0.9683,
      "step": 1440
    },
    {
      "epoch": 0.020448752264500783,
      "grad_norm": 3.203125,
      "learning_rate": 4.996224815136439e-05,
      "loss": 1.0476,
      "step": 1442
    },
    {
      "epoch": 0.020477113918126996,
      "grad_norm": 3.359375,
      "learning_rate": 4.996212531234554e-05,
      "loss": 1.0006,
      "step": 1444
    },
    {
      "epoch": 0.02050547557175321,
      "grad_norm": 3.171875,
      "learning_rate": 4.996200227395254e-05,
      "loss": 1.0025,
      "step": 1446
    },
    {
      "epoch": 0.020533837225379426,
      "grad_norm": 3.265625,
      "learning_rate": 4.9961879036186355e-05,
      "loss": 0.9417,
      "step": 1448
    },
    {
      "epoch": 0.02056219887900564,
      "grad_norm": 3.4375,
      "learning_rate": 4.9961755599047975e-05,
      "loss": 0.9288,
      "step": 1450
    },
    {
      "epoch": 0.020590560532631856,
      "grad_norm": 3.390625,
      "learning_rate": 4.996163196253839e-05,
      "loss": 1.0549,
      "step": 1452
    },
    {
      "epoch": 0.02061892218625807,
      "grad_norm": 3.375,
      "learning_rate": 4.996150812665859e-05,
      "loss": 0.9749,
      "step": 1454
    },
    {
      "epoch": 0.020647283839884286,
      "grad_norm": 3.078125,
      "learning_rate": 4.9961384091409555e-05,
      "loss": 1.011,
      "step": 1456
    },
    {
      "epoch": 0.0206756454935105,
      "grad_norm": 2.96875,
      "learning_rate": 4.9961259856792285e-05,
      "loss": 0.9912,
      "step": 1458
    },
    {
      "epoch": 0.020704007147136713,
      "grad_norm": 2.953125,
      "learning_rate": 4.9961135422807765e-05,
      "loss": 0.9577,
      "step": 1460
    },
    {
      "epoch": 0.02073236880076293,
      "grad_norm": 3.046875,
      "learning_rate": 4.996101078945699e-05,
      "loss": 0.9743,
      "step": 1462
    },
    {
      "epoch": 0.020760730454389142,
      "grad_norm": 3.25,
      "learning_rate": 4.9960885956740956e-05,
      "loss": 1.011,
      "step": 1464
    },
    {
      "epoch": 0.02078909210801536,
      "grad_norm": 3.328125,
      "learning_rate": 4.9960760924660666e-05,
      "loss": 0.9702,
      "step": 1466
    },
    {
      "epoch": 0.020817453761641572,
      "grad_norm": 3.375,
      "learning_rate": 4.9960635693217103e-05,
      "loss": 0.99,
      "step": 1468
    },
    {
      "epoch": 0.020845815415267786,
      "grad_norm": 3.359375,
      "learning_rate": 4.996051026241129e-05,
      "loss": 0.9512,
      "step": 1470
    },
    {
      "epoch": 0.020874177068894002,
      "grad_norm": 2.9375,
      "learning_rate": 4.9960384632244216e-05,
      "loss": 1.0469,
      "step": 1472
    },
    {
      "epoch": 0.020902538722520216,
      "grad_norm": 2.640625,
      "learning_rate": 4.996025880271688e-05,
      "loss": 0.9317,
      "step": 1474
    },
    {
      "epoch": 0.020930900376146432,
      "grad_norm": 3.8125,
      "learning_rate": 4.996013277383029e-05,
      "loss": 0.9458,
      "step": 1476
    },
    {
      "epoch": 0.020959262029772645,
      "grad_norm": 3.046875,
      "learning_rate": 4.996000654558546e-05,
      "loss": 1.0269,
      "step": 1478
    },
    {
      "epoch": 0.020987623683398862,
      "grad_norm": 3.125,
      "learning_rate": 4.995988011798339e-05,
      "loss": 1.0523,
      "step": 1480
    },
    {
      "epoch": 0.021015985337025075,
      "grad_norm": 3.4375,
      "learning_rate": 4.9959753491025095e-05,
      "loss": 0.9595,
      "step": 1482
    },
    {
      "epoch": 0.02104434699065129,
      "grad_norm": 3.234375,
      "learning_rate": 4.995962666471158e-05,
      "loss": 1.0159,
      "step": 1484
    },
    {
      "epoch": 0.021072708644277505,
      "grad_norm": 2.75,
      "learning_rate": 4.9959499639043864e-05,
      "loss": 1.0325,
      "step": 1486
    },
    {
      "epoch": 0.02110107029790372,
      "grad_norm": 3.15625,
      "learning_rate": 4.995937241402297e-05,
      "loss": 0.9696,
      "step": 1488
    },
    {
      "epoch": 0.021129431951529935,
      "grad_norm": 3.015625,
      "learning_rate": 4.995924498964989e-05,
      "loss": 0.9668,
      "step": 1490
    },
    {
      "epoch": 0.02115779360515615,
      "grad_norm": 3.453125,
      "learning_rate": 4.9959117365925654e-05,
      "loss": 0.9939,
      "step": 1492
    },
    {
      "epoch": 0.021186155258782362,
      "grad_norm": 3.25,
      "learning_rate": 4.9958989542851283e-05,
      "loss": 0.9625,
      "step": 1494
    },
    {
      "epoch": 0.02121451691240858,
      "grad_norm": 2.78125,
      "learning_rate": 4.995886152042781e-05,
      "loss": 0.9667,
      "step": 1496
    },
    {
      "epoch": 0.02124287856603479,
      "grad_norm": 3.03125,
      "learning_rate": 4.995873329865624e-05,
      "loss": 0.9705,
      "step": 1498
    },
    {
      "epoch": 0.02127124021966101,
      "grad_norm": 4.125,
      "learning_rate": 4.9958604877537596e-05,
      "loss": 1.0069,
      "step": 1500
    },
    {
      "epoch": 0.02129960187328722,
      "grad_norm": 4.0625,
      "learning_rate": 4.9958476257072914e-05,
      "loss": 1.0081,
      "step": 1502
    },
    {
      "epoch": 0.02132796352691344,
      "grad_norm": 3.265625,
      "learning_rate": 4.995834743726322e-05,
      "loss": 0.9986,
      "step": 1504
    },
    {
      "epoch": 0.02135632518053965,
      "grad_norm": 3.875,
      "learning_rate": 4.995821841810954e-05,
      "loss": 0.9558,
      "step": 1506
    },
    {
      "epoch": 0.021384686834165865,
      "grad_norm": 2.96875,
      "learning_rate": 4.99580891996129e-05,
      "loss": 0.9662,
      "step": 1508
    },
    {
      "epoch": 0.02141304848779208,
      "grad_norm": 3.125,
      "learning_rate": 4.9957959781774344e-05,
      "loss": 0.9746,
      "step": 1510
    },
    {
      "epoch": 0.021441410141418295,
      "grad_norm": 3.375,
      "learning_rate": 4.9957830164594893e-05,
      "loss": 1.0243,
      "step": 1512
    },
    {
      "epoch": 0.02146977179504451,
      "grad_norm": 2.921875,
      "learning_rate": 4.9957700348075584e-05,
      "loss": 1.0064,
      "step": 1514
    },
    {
      "epoch": 0.021498133448670725,
      "grad_norm": 3.015625,
      "learning_rate": 4.995757033221747e-05,
      "loss": 0.9553,
      "step": 1516
    },
    {
      "epoch": 0.021526495102296938,
      "grad_norm": 3.09375,
      "learning_rate": 4.995744011702157e-05,
      "loss": 0.9645,
      "step": 1518
    },
    {
      "epoch": 0.021554856755923155,
      "grad_norm": 3.125,
      "learning_rate": 4.995730970248893e-05,
      "loss": 0.9829,
      "step": 1520
    },
    {
      "epoch": 0.021583218409549368,
      "grad_norm": 2.84375,
      "learning_rate": 4.995717908862059e-05,
      "loss": 0.9936,
      "step": 1522
    },
    {
      "epoch": 0.021611580063175585,
      "grad_norm": 3.3125,
      "learning_rate": 4.99570482754176e-05,
      "loss": 1.02,
      "step": 1524
    },
    {
      "epoch": 0.021639941716801798,
      "grad_norm": 2.96875,
      "learning_rate": 4.9956917262881e-05,
      "loss": 0.9703,
      "step": 1526
    },
    {
      "epoch": 0.021668303370428015,
      "grad_norm": 3.40625,
      "learning_rate": 4.995678605101184e-05,
      "loss": 0.9509,
      "step": 1528
    },
    {
      "epoch": 0.021696665024054228,
      "grad_norm": 4.34375,
      "learning_rate": 4.9956654639811165e-05,
      "loss": 0.9992,
      "step": 1530
    },
    {
      "epoch": 0.02172502667768044,
      "grad_norm": 3.203125,
      "learning_rate": 4.995652302928002e-05,
      "loss": 1.0186,
      "step": 1532
    },
    {
      "epoch": 0.021753388331306658,
      "grad_norm": 3.359375,
      "learning_rate": 4.9956391219419464e-05,
      "loss": 0.9736,
      "step": 1534
    },
    {
      "epoch": 0.02178174998493287,
      "grad_norm": 3.296875,
      "learning_rate": 4.9956259210230544e-05,
      "loss": 0.9685,
      "step": 1536
    },
    {
      "epoch": 0.021810111638559088,
      "grad_norm": 3.34375,
      "learning_rate": 4.995612700171432e-05,
      "loss": 1.0086,
      "step": 1538
    },
    {
      "epoch": 0.0218384732921853,
      "grad_norm": 3.0,
      "learning_rate": 4.9955994593871846e-05,
      "loss": 0.9383,
      "step": 1540
    },
    {
      "epoch": 0.021866834945811514,
      "grad_norm": 2.96875,
      "learning_rate": 4.9955861986704175e-05,
      "loss": 1.0382,
      "step": 1542
    },
    {
      "epoch": 0.02189519659943773,
      "grad_norm": 3.59375,
      "learning_rate": 4.995572918021237e-05,
      "loss": 0.9803,
      "step": 1544
    },
    {
      "epoch": 0.021923558253063944,
      "grad_norm": 3.5625,
      "learning_rate": 4.995559617439749e-05,
      "loss": 0.9705,
      "step": 1546
    },
    {
      "epoch": 0.02195191990669016,
      "grad_norm": 3.15625,
      "learning_rate": 4.995546296926061e-05,
      "loss": 1.0152,
      "step": 1548
    },
    {
      "epoch": 0.021980281560316374,
      "grad_norm": 3.1875,
      "learning_rate": 4.9955329564802774e-05,
      "loss": 0.9243,
      "step": 1550
    },
    {
      "epoch": 0.022008643213942587,
      "grad_norm": 3.1875,
      "learning_rate": 4.995519596102506e-05,
      "loss": 0.9974,
      "step": 1552
    },
    {
      "epoch": 0.022037004867568804,
      "grad_norm": 3.203125,
      "learning_rate": 4.995506215792852e-05,
      "loss": 0.9629,
      "step": 1554
    },
    {
      "epoch": 0.022065366521195017,
      "grad_norm": 3.390625,
      "learning_rate": 4.995492815551425e-05,
      "loss": 1.0142,
      "step": 1556
    },
    {
      "epoch": 0.022093728174821234,
      "grad_norm": 3.28125,
      "learning_rate": 4.995479395378329e-05,
      "loss": 0.9821,
      "step": 1558
    },
    {
      "epoch": 0.022122089828447447,
      "grad_norm": 3.140625,
      "learning_rate": 4.9954659552736736e-05,
      "loss": 0.9923,
      "step": 1560
    },
    {
      "epoch": 0.022150451482073664,
      "grad_norm": 2.78125,
      "learning_rate": 4.9954524952375646e-05,
      "loss": 0.9458,
      "step": 1562
    },
    {
      "epoch": 0.022178813135699877,
      "grad_norm": 3.1875,
      "learning_rate": 4.9954390152701104e-05,
      "loss": 1.0019,
      "step": 1564
    },
    {
      "epoch": 0.02220717478932609,
      "grad_norm": 3.1875,
      "learning_rate": 4.995425515371418e-05,
      "loss": 1.0018,
      "step": 1566
    },
    {
      "epoch": 0.022235536442952307,
      "grad_norm": 3.578125,
      "learning_rate": 4.995411995541596e-05,
      "loss": 0.9671,
      "step": 1568
    },
    {
      "epoch": 0.02226389809657852,
      "grad_norm": 2.953125,
      "learning_rate": 4.995398455780752e-05,
      "loss": 0.9419,
      "step": 1570
    },
    {
      "epoch": 0.022292259750204737,
      "grad_norm": 3.65625,
      "learning_rate": 4.995384896088994e-05,
      "loss": 0.9976,
      "step": 1572
    },
    {
      "epoch": 0.02232062140383095,
      "grad_norm": 3.1875,
      "learning_rate": 4.9953713164664294e-05,
      "loss": 1.0145,
      "step": 1574
    },
    {
      "epoch": 0.022348983057457163,
      "grad_norm": 3.71875,
      "learning_rate": 4.9953577169131696e-05,
      "loss": 1.0217,
      "step": 1576
    },
    {
      "epoch": 0.02237734471108338,
      "grad_norm": 2.96875,
      "learning_rate": 4.9953440974293194e-05,
      "loss": 0.9768,
      "step": 1578
    },
    {
      "epoch": 0.022405706364709593,
      "grad_norm": 3.234375,
      "learning_rate": 4.995330458014991e-05,
      "loss": 0.9724,
      "step": 1580
    },
    {
      "epoch": 0.02243406801833581,
      "grad_norm": 2.90625,
      "learning_rate": 4.9953167986702905e-05,
      "loss": 1.0012,
      "step": 1582
    },
    {
      "epoch": 0.022462429671962023,
      "grad_norm": 3.375,
      "learning_rate": 4.9953031193953285e-05,
      "loss": 0.9853,
      "step": 1584
    },
    {
      "epoch": 0.02249079132558824,
      "grad_norm": 3.0,
      "learning_rate": 4.9952894201902146e-05,
      "loss": 0.9182,
      "step": 1586
    },
    {
      "epoch": 0.022519152979214453,
      "grad_norm": 3.234375,
      "learning_rate": 4.995275701055058e-05,
      "loss": 0.9459,
      "step": 1588
    },
    {
      "epoch": 0.022547514632840666,
      "grad_norm": 2.921875,
      "learning_rate": 4.9952619619899665e-05,
      "loss": 0.9647,
      "step": 1590
    },
    {
      "epoch": 0.022575876286466883,
      "grad_norm": 3.109375,
      "learning_rate": 4.995248202995052e-05,
      "loss": 0.9983,
      "step": 1592
    },
    {
      "epoch": 0.022604237940093096,
      "grad_norm": 3.375,
      "learning_rate": 4.995234424070424e-05,
      "loss": 1.0178,
      "step": 1594
    },
    {
      "epoch": 0.022632599593719313,
      "grad_norm": 3.46875,
      "learning_rate": 4.995220625216192e-05,
      "loss": 0.9892,
      "step": 1596
    },
    {
      "epoch": 0.022660961247345526,
      "grad_norm": 3.21875,
      "learning_rate": 4.995206806432466e-05,
      "loss": 1.0128,
      "step": 1598
    },
    {
      "epoch": 0.02268932290097174,
      "grad_norm": 3.296875,
      "learning_rate": 4.9951929677193576e-05,
      "loss": 0.9676,
      "step": 1600
    },
    {
      "epoch": 0.022717684554597956,
      "grad_norm": 2.828125,
      "learning_rate": 4.995179109076976e-05,
      "loss": 0.8941,
      "step": 1602
    },
    {
      "epoch": 0.02274604620822417,
      "grad_norm": 3.171875,
      "learning_rate": 4.995165230505433e-05,
      "loss": 0.9697,
      "step": 1604
    },
    {
      "epoch": 0.022774407861850386,
      "grad_norm": 3.484375,
      "learning_rate": 4.9951513320048384e-05,
      "loss": 0.9746,
      "step": 1606
    },
    {
      "epoch": 0.0228027695154766,
      "grad_norm": 3.40625,
      "learning_rate": 4.9951374135753035e-05,
      "loss": 0.9942,
      "step": 1608
    },
    {
      "epoch": 0.022831131169102816,
      "grad_norm": 3.234375,
      "learning_rate": 4.995123475216941e-05,
      "loss": 1.0235,
      "step": 1610
    },
    {
      "epoch": 0.02285949282272903,
      "grad_norm": 3.0,
      "learning_rate": 4.995109516929859e-05,
      "loss": 0.9384,
      "step": 1612
    },
    {
      "epoch": 0.022887854476355243,
      "grad_norm": 3.09375,
      "learning_rate": 4.9950955387141715e-05,
      "loss": 0.9656,
      "step": 1614
    },
    {
      "epoch": 0.02291621612998146,
      "grad_norm": 3.390625,
      "learning_rate": 4.9950815405699905e-05,
      "loss": 1.0278,
      "step": 1616
    },
    {
      "epoch": 0.022944577783607673,
      "grad_norm": 3.3125,
      "learning_rate": 4.9950675224974256e-05,
      "loss": 1.0221,
      "step": 1618
    },
    {
      "epoch": 0.02297293943723389,
      "grad_norm": 3.0625,
      "learning_rate": 4.9950534844965904e-05,
      "loss": 1.017,
      "step": 1620
    },
    {
      "epoch": 0.023001301090860102,
      "grad_norm": 2.953125,
      "learning_rate": 4.995039426567596e-05,
      "loss": 0.9876,
      "step": 1622
    },
    {
      "epoch": 0.023029662744486316,
      "grad_norm": 3.6875,
      "learning_rate": 4.995025348710556e-05,
      "loss": 0.9416,
      "step": 1624
    },
    {
      "epoch": 0.023058024398112532,
      "grad_norm": 3.609375,
      "learning_rate": 4.995011250925582e-05,
      "loss": 1.0165,
      "step": 1626
    },
    {
      "epoch": 0.023086386051738746,
      "grad_norm": 4.0625,
      "learning_rate": 4.994997133212787e-05,
      "loss": 0.9674,
      "step": 1628
    },
    {
      "epoch": 0.023114747705364962,
      "grad_norm": 3.015625,
      "learning_rate": 4.9949829955722826e-05,
      "loss": 0.9787,
      "step": 1630
    },
    {
      "epoch": 0.023143109358991176,
      "grad_norm": 2.984375,
      "learning_rate": 4.994968838004184e-05,
      "loss": 0.9699,
      "step": 1632
    },
    {
      "epoch": 0.023171471012617392,
      "grad_norm": 3.15625,
      "learning_rate": 4.9949546605086015e-05,
      "loss": 0.9735,
      "step": 1634
    },
    {
      "epoch": 0.023199832666243606,
      "grad_norm": 3.59375,
      "learning_rate": 4.99494046308565e-05,
      "loss": 1.0388,
      "step": 1636
    },
    {
      "epoch": 0.02322819431986982,
      "grad_norm": 2.9375,
      "learning_rate": 4.9949262457354426e-05,
      "loss": 1.0158,
      "step": 1638
    },
    {
      "epoch": 0.023256555973496035,
      "grad_norm": 2.828125,
      "learning_rate": 4.9949120084580936e-05,
      "loss": 0.9289,
      "step": 1640
    },
    {
      "epoch": 0.02328491762712225,
      "grad_norm": 2.765625,
      "learning_rate": 4.994897751253715e-05,
      "loss": 0.9808,
      "step": 1642
    },
    {
      "epoch": 0.023313279280748465,
      "grad_norm": 3.171875,
      "learning_rate": 4.9948834741224226e-05,
      "loss": 0.9562,
      "step": 1644
    },
    {
      "epoch": 0.02334164093437468,
      "grad_norm": 3.09375,
      "learning_rate": 4.994869177064329e-05,
      "loss": 1.04,
      "step": 1646
    },
    {
      "epoch": 0.023370002588000892,
      "grad_norm": 3.015625,
      "learning_rate": 4.9948548600795485e-05,
      "loss": 0.9566,
      "step": 1648
    },
    {
      "epoch": 0.02339836424162711,
      "grad_norm": 3.140625,
      "learning_rate": 4.994840523168197e-05,
      "loss": 0.9769,
      "step": 1650
    },
    {
      "epoch": 0.023426725895253322,
      "grad_norm": 3.328125,
      "learning_rate": 4.994826166330386e-05,
      "loss": 0.9202,
      "step": 1652
    },
    {
      "epoch": 0.02345508754887954,
      "grad_norm": 3.5,
      "learning_rate": 4.994811789566234e-05,
      "loss": 0.9775,
      "step": 1654
    },
    {
      "epoch": 0.023483449202505752,
      "grad_norm": 3.296875,
      "learning_rate": 4.994797392875853e-05,
      "loss": 0.9515,
      "step": 1656
    },
    {
      "epoch": 0.02351181085613197,
      "grad_norm": 3.625,
      "learning_rate": 4.994782976259358e-05,
      "loss": 1.013,
      "step": 1658
    },
    {
      "epoch": 0.02354017250975818,
      "grad_norm": 3.515625,
      "learning_rate": 4.994768539716866e-05,
      "loss": 0.9947,
      "step": 1660
    },
    {
      "epoch": 0.023568534163384395,
      "grad_norm": 3.078125,
      "learning_rate": 4.9947540832484904e-05,
      "loss": 0.9654,
      "step": 1662
    },
    {
      "epoch": 0.02359689581701061,
      "grad_norm": 3.296875,
      "learning_rate": 4.994739606854348e-05,
      "loss": 0.9394,
      "step": 1664
    },
    {
      "epoch": 0.023625257470636825,
      "grad_norm": 2.96875,
      "learning_rate": 4.994725110534555e-05,
      "loss": 1.0203,
      "step": 1666
    },
    {
      "epoch": 0.02365361912426304,
      "grad_norm": 3.515625,
      "learning_rate": 4.9947105942892244e-05,
      "loss": 0.9882,
      "step": 1668
    },
    {
      "epoch": 0.023681980777889255,
      "grad_norm": 3.1875,
      "learning_rate": 4.994696058118476e-05,
      "loss": 0.9478,
      "step": 1670
    },
    {
      "epoch": 0.023710342431515468,
      "grad_norm": 2.9375,
      "learning_rate": 4.9946815020224215e-05,
      "loss": 0.9161,
      "step": 1672
    },
    {
      "epoch": 0.023738704085141685,
      "grad_norm": 3.1875,
      "learning_rate": 4.99466692600118e-05,
      "loss": 0.9529,
      "step": 1674
    },
    {
      "epoch": 0.023767065738767898,
      "grad_norm": 3.046875,
      "learning_rate": 4.994652330054868e-05,
      "loss": 0.9861,
      "step": 1676
    },
    {
      "epoch": 0.023795427392394115,
      "grad_norm": 2.953125,
      "learning_rate": 4.994637714183601e-05,
      "loss": 0.9934,
      "step": 1678
    },
    {
      "epoch": 0.023823789046020328,
      "grad_norm": 3.234375,
      "learning_rate": 4.994623078387497e-05,
      "loss": 0.9735,
      "step": 1680
    },
    {
      "epoch": 0.023852150699646545,
      "grad_norm": 3.3125,
      "learning_rate": 4.994608422666671e-05,
      "loss": 1.0267,
      "step": 1682
    },
    {
      "epoch": 0.023880512353272758,
      "grad_norm": 3.796875,
      "learning_rate": 4.994593747021242e-05,
      "loss": 0.9541,
      "step": 1684
    },
    {
      "epoch": 0.02390887400689897,
      "grad_norm": 3.046875,
      "learning_rate": 4.9945790514513255e-05,
      "loss": 1.0137,
      "step": 1686
    },
    {
      "epoch": 0.023937235660525188,
      "grad_norm": 3.09375,
      "learning_rate": 4.99456433595704e-05,
      "loss": 0.9788,
      "step": 1688
    },
    {
      "epoch": 0.0239655973141514,
      "grad_norm": 3.0,
      "learning_rate": 4.9945496005385036e-05,
      "loss": 0.9362,
      "step": 1690
    },
    {
      "epoch": 0.023993958967777618,
      "grad_norm": 3.25,
      "learning_rate": 4.994534845195832e-05,
      "loss": 0.9626,
      "step": 1692
    },
    {
      "epoch": 0.02402232062140383,
      "grad_norm": 3.578125,
      "learning_rate": 4.994520069929145e-05,
      "loss": 0.9866,
      "step": 1694
    },
    {
      "epoch": 0.024050682275030044,
      "grad_norm": 3.21875,
      "learning_rate": 4.994505274738559e-05,
      "loss": 0.9931,
      "step": 1696
    },
    {
      "epoch": 0.02407904392865626,
      "grad_norm": 3.078125,
      "learning_rate": 4.9944904596241936e-05,
      "loss": 1.0025,
      "step": 1698
    },
    {
      "epoch": 0.024107405582282474,
      "grad_norm": 3.125,
      "learning_rate": 4.994475624586167e-05,
      "loss": 0.9802,
      "step": 1700
    },
    {
      "epoch": 0.02413576723590869,
      "grad_norm": 3.171875,
      "learning_rate": 4.994460769624596e-05,
      "loss": 1.0029,
      "step": 1702
    },
    {
      "epoch": 0.024164128889534904,
      "grad_norm": 3.3125,
      "learning_rate": 4.994445894739601e-05,
      "loss": 0.9358,
      "step": 1704
    },
    {
      "epoch": 0.024192490543161117,
      "grad_norm": 3.34375,
      "learning_rate": 4.994430999931301e-05,
      "loss": 0.9599,
      "step": 1706
    },
    {
      "epoch": 0.024220852196787334,
      "grad_norm": 3.25,
      "learning_rate": 4.994416085199813e-05,
      "loss": 1.0063,
      "step": 1708
    },
    {
      "epoch": 0.024249213850413547,
      "grad_norm": 3.03125,
      "learning_rate": 4.9944011505452585e-05,
      "loss": 0.982,
      "step": 1710
    },
    {
      "epoch": 0.024277575504039764,
      "grad_norm": 3.484375,
      "learning_rate": 4.994386195967754e-05,
      "loss": 0.9737,
      "step": 1712
    },
    {
      "epoch": 0.024305937157665977,
      "grad_norm": 3.59375,
      "learning_rate": 4.994371221467422e-05,
      "loss": 0.9453,
      "step": 1714
    },
    {
      "epoch": 0.024334298811292194,
      "grad_norm": 3.140625,
      "learning_rate": 4.994356227044381e-05,
      "loss": 0.9314,
      "step": 1716
    },
    {
      "epoch": 0.024362660464918407,
      "grad_norm": 3.6875,
      "learning_rate": 4.994341212698749e-05,
      "loss": 1.0388,
      "step": 1718
    },
    {
      "epoch": 0.02439102211854462,
      "grad_norm": 3.265625,
      "learning_rate": 4.9943261784306485e-05,
      "loss": 0.998,
      "step": 1720
    },
    {
      "epoch": 0.024419383772170837,
      "grad_norm": 3.390625,
      "learning_rate": 4.9943111242401974e-05,
      "loss": 1.0044,
      "step": 1722
    },
    {
      "epoch": 0.02444774542579705,
      "grad_norm": 3.421875,
      "learning_rate": 4.994296050127518e-05,
      "loss": 0.9648,
      "step": 1724
    },
    {
      "epoch": 0.024476107079423267,
      "grad_norm": 3.34375,
      "learning_rate": 4.99428095609273e-05,
      "loss": 0.9665,
      "step": 1726
    },
    {
      "epoch": 0.02450446873304948,
      "grad_norm": 3.03125,
      "learning_rate": 4.9942658421359525e-05,
      "loss": 0.9839,
      "step": 1728
    },
    {
      "epoch": 0.024532830386675693,
      "grad_norm": 2.90625,
      "learning_rate": 4.994250708257308e-05,
      "loss": 0.9904,
      "step": 1730
    },
    {
      "epoch": 0.02456119204030191,
      "grad_norm": 3.515625,
      "learning_rate": 4.994235554456916e-05,
      "loss": 0.9641,
      "step": 1732
    },
    {
      "epoch": 0.024589553693928123,
      "grad_norm": 3.3125,
      "learning_rate": 4.9942203807348985e-05,
      "loss": 0.9384,
      "step": 1734
    },
    {
      "epoch": 0.02461791534755434,
      "grad_norm": 3.40625,
      "learning_rate": 4.994205187091377e-05,
      "loss": 1.0115,
      "step": 1736
    },
    {
      "epoch": 0.024646277001180553,
      "grad_norm": 3.625,
      "learning_rate": 4.994189973526472e-05,
      "loss": 0.9677,
      "step": 1738
    },
    {
      "epoch": 0.02467463865480677,
      "grad_norm": 2.890625,
      "learning_rate": 4.994174740040305e-05,
      "loss": 0.9269,
      "step": 1740
    },
    {
      "epoch": 0.024703000308432983,
      "grad_norm": 2.859375,
      "learning_rate": 4.994159486632999e-05,
      "loss": 0.9646,
      "step": 1742
    },
    {
      "epoch": 0.024731361962059197,
      "grad_norm": 3.359375,
      "learning_rate": 4.9941442133046735e-05,
      "loss": 0.9975,
      "step": 1744
    },
    {
      "epoch": 0.024759723615685413,
      "grad_norm": 3.09375,
      "learning_rate": 4.9941289200554524e-05,
      "loss": 0.9819,
      "step": 1746
    },
    {
      "epoch": 0.024788085269311626,
      "grad_norm": 3.671875,
      "learning_rate": 4.994113606885458e-05,
      "loss": 1.0289,
      "step": 1748
    },
    {
      "epoch": 0.024816446922937843,
      "grad_norm": 3.71875,
      "learning_rate": 4.9940982737948106e-05,
      "loss": 0.994,
      "step": 1750
    },
    {
      "epoch": 0.024844808576564056,
      "grad_norm": 2.9375,
      "learning_rate": 4.9940829207836346e-05,
      "loss": 0.8944,
      "step": 1752
    },
    {
      "epoch": 0.02487317023019027,
      "grad_norm": 3.1875,
      "learning_rate": 4.9940675478520516e-05,
      "loss": 0.9585,
      "step": 1754
    },
    {
      "epoch": 0.024901531883816486,
      "grad_norm": 2.75,
      "learning_rate": 4.994052155000185e-05,
      "loss": 0.9787,
      "step": 1756
    },
    {
      "epoch": 0.0249298935374427,
      "grad_norm": 3.140625,
      "learning_rate": 4.994036742228158e-05,
      "loss": 0.9633,
      "step": 1758
    },
    {
      "epoch": 0.024958255191068916,
      "grad_norm": 3.390625,
      "learning_rate": 4.994021309536092e-05,
      "loss": 1.0009,
      "step": 1760
    },
    {
      "epoch": 0.02498661684469513,
      "grad_norm": 3.0,
      "learning_rate": 4.9940058569241125e-05,
      "loss": 0.9903,
      "step": 1762
    },
    {
      "epoch": 0.025014978498321346,
      "grad_norm": 3.234375,
      "learning_rate": 4.993990384392341e-05,
      "loss": 0.9569,
      "step": 1764
    },
    {
      "epoch": 0.02504334015194756,
      "grad_norm": 3.15625,
      "learning_rate": 4.993974891940902e-05,
      "loss": 0.9532,
      "step": 1766
    },
    {
      "epoch": 0.025071701805573773,
      "grad_norm": 3.4375,
      "learning_rate": 4.99395937956992e-05,
      "loss": 0.9846,
      "step": 1768
    },
    {
      "epoch": 0.02510006345919999,
      "grad_norm": 3.03125,
      "learning_rate": 4.993943847279517e-05,
      "loss": 0.9399,
      "step": 1770
    },
    {
      "epoch": 0.025128425112826203,
      "grad_norm": 3.546875,
      "learning_rate": 4.9939282950698195e-05,
      "loss": 0.98,
      "step": 1772
    },
    {
      "epoch": 0.02515678676645242,
      "grad_norm": 3.078125,
      "learning_rate": 4.99391272294095e-05,
      "loss": 0.9429,
      "step": 1774
    },
    {
      "epoch": 0.025185148420078633,
      "grad_norm": 3.15625,
      "learning_rate": 4.993897130893033e-05,
      "loss": 1.0073,
      "step": 1776
    },
    {
      "epoch": 0.025213510073704846,
      "grad_norm": 3.25,
      "learning_rate": 4.9938815189261934e-05,
      "loss": 0.9578,
      "step": 1778
    },
    {
      "epoch": 0.025241871727331062,
      "grad_norm": 2.921875,
      "learning_rate": 4.993865887040555e-05,
      "loss": 0.926,
      "step": 1780
    },
    {
      "epoch": 0.025270233380957276,
      "grad_norm": 3.328125,
      "learning_rate": 4.9938502352362435e-05,
      "loss": 0.9475,
      "step": 1782
    },
    {
      "epoch": 0.025298595034583492,
      "grad_norm": 3.125,
      "learning_rate": 4.993834563513385e-05,
      "loss": 0.9597,
      "step": 1784
    },
    {
      "epoch": 0.025326956688209706,
      "grad_norm": 2.890625,
      "learning_rate": 4.9938188718721025e-05,
      "loss": 0.9446,
      "step": 1786
    },
    {
      "epoch": 0.025355318341835922,
      "grad_norm": 3.25,
      "learning_rate": 4.993803160312523e-05,
      "loss": 0.9787,
      "step": 1788
    },
    {
      "epoch": 0.025383679995462136,
      "grad_norm": 3.171875,
      "learning_rate": 4.9937874288347705e-05,
      "loss": 0.9527,
      "step": 1790
    },
    {
      "epoch": 0.02541204164908835,
      "grad_norm": 3.125,
      "learning_rate": 4.993771677438972e-05,
      "loss": 0.9587,
      "step": 1792
    },
    {
      "epoch": 0.025440403302714566,
      "grad_norm": 3.078125,
      "learning_rate": 4.993755906125253e-05,
      "loss": 0.9844,
      "step": 1794
    },
    {
      "epoch": 0.02546876495634078,
      "grad_norm": 3.40625,
      "learning_rate": 4.993740114893739e-05,
      "loss": 1.0143,
      "step": 1796
    },
    {
      "epoch": 0.025497126609966995,
      "grad_norm": 2.8125,
      "learning_rate": 4.993724303744556e-05,
      "loss": 0.9416,
      "step": 1798
    },
    {
      "epoch": 0.02552548826359321,
      "grad_norm": 3.09375,
      "learning_rate": 4.9937084726778314e-05,
      "loss": 0.9145,
      "step": 1800
    },
    {
      "epoch": 0.025553849917219422,
      "grad_norm": 3.171875,
      "learning_rate": 4.9936926216936905e-05,
      "loss": 1.0309,
      "step": 1802
    },
    {
      "epoch": 0.02558221157084564,
      "grad_norm": 3.359375,
      "learning_rate": 4.99367675079226e-05,
      "loss": 0.9573,
      "step": 1804
    },
    {
      "epoch": 0.025610573224471852,
      "grad_norm": 3.03125,
      "learning_rate": 4.993660859973667e-05,
      "loss": 0.942,
      "step": 1806
    },
    {
      "epoch": 0.02563893487809807,
      "grad_norm": 3.296875,
      "learning_rate": 4.993644949238039e-05,
      "loss": 0.9686,
      "step": 1808
    },
    {
      "epoch": 0.025667296531724282,
      "grad_norm": 3.078125,
      "learning_rate": 4.993629018585502e-05,
      "loss": 0.9753,
      "step": 1810
    },
    {
      "epoch": 0.0256956581853505,
      "grad_norm": 3.046875,
      "learning_rate": 4.993613068016184e-05,
      "loss": 0.9705,
      "step": 1812
    },
    {
      "epoch": 0.025724019838976712,
      "grad_norm": 2.765625,
      "learning_rate": 4.993597097530212e-05,
      "loss": 0.9948,
      "step": 1814
    },
    {
      "epoch": 0.025752381492602925,
      "grad_norm": 2.984375,
      "learning_rate": 4.9935811071277125e-05,
      "loss": 0.9672,
      "step": 1816
    },
    {
      "epoch": 0.02578074314622914,
      "grad_norm": 3.078125,
      "learning_rate": 4.993565096808816e-05,
      "loss": 0.9269,
      "step": 1818
    },
    {
      "epoch": 0.025809104799855355,
      "grad_norm": 3.0,
      "learning_rate": 4.9935490665736484e-05,
      "loss": 1.027,
      "step": 1820
    },
    {
      "epoch": 0.02583746645348157,
      "grad_norm": 3.078125,
      "learning_rate": 4.9935330164223376e-05,
      "loss": 0.9689,
      "step": 1822
    },
    {
      "epoch": 0.025865828107107785,
      "grad_norm": 3.5,
      "learning_rate": 4.993516946355012e-05,
      "loss": 0.9762,
      "step": 1824
    },
    {
      "epoch": 0.025894189760733998,
      "grad_norm": 3.21875,
      "learning_rate": 4.993500856371801e-05,
      "loss": 0.9226,
      "step": 1826
    },
    {
      "epoch": 0.025922551414360215,
      "grad_norm": 2.921875,
      "learning_rate": 4.993484746472832e-05,
      "loss": 0.9227,
      "step": 1828
    },
    {
      "epoch": 0.025950913067986428,
      "grad_norm": 3.03125,
      "learning_rate": 4.993468616658235e-05,
      "loss": 0.9408,
      "step": 1830
    },
    {
      "epoch": 0.025979274721612645,
      "grad_norm": 3.140625,
      "learning_rate": 4.993452466928137e-05,
      "loss": 1.005,
      "step": 1832
    },
    {
      "epoch": 0.026007636375238858,
      "grad_norm": 3.1875,
      "learning_rate": 4.9934362972826674e-05,
      "loss": 0.9385,
      "step": 1834
    },
    {
      "epoch": 0.026035998028865075,
      "grad_norm": 3.28125,
      "learning_rate": 4.993420107721956e-05,
      "loss": 1.0031,
      "step": 1836
    },
    {
      "epoch": 0.026064359682491288,
      "grad_norm": 3.5625,
      "learning_rate": 4.9934038982461315e-05,
      "loss": 0.9925,
      "step": 1838
    },
    {
      "epoch": 0.0260927213361175,
      "grad_norm": 3.03125,
      "learning_rate": 4.9933876688553246e-05,
      "loss": 0.9775,
      "step": 1840
    },
    {
      "epoch": 0.026121082989743718,
      "grad_norm": 3.015625,
      "learning_rate": 4.993371419549664e-05,
      "loss": 0.9831,
      "step": 1842
    },
    {
      "epoch": 0.02614944464336993,
      "grad_norm": 3.15625,
      "learning_rate": 4.99335515032928e-05,
      "loss": 0.9297,
      "step": 1844
    },
    {
      "epoch": 0.026177806296996148,
      "grad_norm": 3.03125,
      "learning_rate": 4.993338861194301e-05,
      "loss": 0.9388,
      "step": 1846
    },
    {
      "epoch": 0.02620616795062236,
      "grad_norm": 2.921875,
      "learning_rate": 4.9933225521448587e-05,
      "loss": 0.9543,
      "step": 1848
    },
    {
      "epoch": 0.026234529604248574,
      "grad_norm": 3.046875,
      "learning_rate": 4.993306223181082e-05,
      "loss": 0.9766,
      "step": 1850
    },
    {
      "epoch": 0.02626289125787479,
      "grad_norm": 3.125,
      "learning_rate": 4.993289874303103e-05,
      "loss": 0.9726,
      "step": 1852
    },
    {
      "epoch": 0.026291252911501004,
      "grad_norm": 3.109375,
      "learning_rate": 4.993273505511052e-05,
      "loss": 1.0022,
      "step": 1854
    },
    {
      "epoch": 0.02631961456512722,
      "grad_norm": 3.03125,
      "learning_rate": 4.993257116805059e-05,
      "loss": 1.0047,
      "step": 1856
    },
    {
      "epoch": 0.026347976218753434,
      "grad_norm": 3.0,
      "learning_rate": 4.993240708185254e-05,
      "loss": 0.9683,
      "step": 1858
    },
    {
      "epoch": 0.026376337872379647,
      "grad_norm": 3.671875,
      "learning_rate": 4.99322427965177e-05,
      "loss": 1.0202,
      "step": 1860
    },
    {
      "epoch": 0.026404699526005864,
      "grad_norm": 3.046875,
      "learning_rate": 4.993207831204738e-05,
      "loss": 0.9677,
      "step": 1862
    },
    {
      "epoch": 0.026433061179632077,
      "grad_norm": 3.546875,
      "learning_rate": 4.993191362844288e-05,
      "loss": 0.9698,
      "step": 1864
    },
    {
      "epoch": 0.026461422833258294,
      "grad_norm": 2.828125,
      "learning_rate": 4.993174874570552e-05,
      "loss": 0.9895,
      "step": 1866
    },
    {
      "epoch": 0.026489784486884507,
      "grad_norm": 3.046875,
      "learning_rate": 4.993158366383662e-05,
      "loss": 0.9978,
      "step": 1868
    },
    {
      "epoch": 0.026518146140510724,
      "grad_norm": 3.453125,
      "learning_rate": 4.993141838283751e-05,
      "loss": 1.0044,
      "step": 1870
    },
    {
      "epoch": 0.026546507794136937,
      "grad_norm": 3.484375,
      "learning_rate": 4.993125290270949e-05,
      "loss": 1.0062,
      "step": 1872
    },
    {
      "epoch": 0.02657486944776315,
      "grad_norm": 3.40625,
      "learning_rate": 4.993108722345389e-05,
      "loss": 0.9589,
      "step": 1874
    },
    {
      "epoch": 0.026603231101389367,
      "grad_norm": 3.0,
      "learning_rate": 4.993092134507203e-05,
      "loss": 1.0257,
      "step": 1876
    },
    {
      "epoch": 0.02663159275501558,
      "grad_norm": 3.015625,
      "learning_rate": 4.993075526756524e-05,
      "loss": 0.9612,
      "step": 1878
    },
    {
      "epoch": 0.026659954408641797,
      "grad_norm": 2.90625,
      "learning_rate": 4.993058899093485e-05,
      "loss": 0.9944,
      "step": 1880
    },
    {
      "epoch": 0.02668831606226801,
      "grad_norm": 2.96875,
      "learning_rate": 4.993042251518218e-05,
      "loss": 0.953,
      "step": 1882
    },
    {
      "epoch": 0.026716677715894224,
      "grad_norm": 2.90625,
      "learning_rate": 4.9930255840308565e-05,
      "loss": 0.9688,
      "step": 1884
    },
    {
      "epoch": 0.02674503936952044,
      "grad_norm": 3.078125,
      "learning_rate": 4.993008896631533e-05,
      "loss": 0.9505,
      "step": 1886
    },
    {
      "epoch": 0.026773401023146653,
      "grad_norm": 2.96875,
      "learning_rate": 4.9929921893203814e-05,
      "loss": 0.9508,
      "step": 1888
    },
    {
      "epoch": 0.02680176267677287,
      "grad_norm": 2.890625,
      "learning_rate": 4.992975462097534e-05,
      "loss": 0.9602,
      "step": 1890
    },
    {
      "epoch": 0.026830124330399083,
      "grad_norm": 3.265625,
      "learning_rate": 4.9929587149631265e-05,
      "loss": 0.9812,
      "step": 1892
    },
    {
      "epoch": 0.0268584859840253,
      "grad_norm": 3.140625,
      "learning_rate": 4.992941947917291e-05,
      "loss": 0.956,
      "step": 1894
    },
    {
      "epoch": 0.026886847637651513,
      "grad_norm": 3.3125,
      "learning_rate": 4.9929251609601625e-05,
      "loss": 0.9646,
      "step": 1896
    },
    {
      "epoch": 0.026915209291277727,
      "grad_norm": 3.375,
      "learning_rate": 4.992908354091874e-05,
      "loss": 1.0219,
      "step": 1898
    },
    {
      "epoch": 0.026943570944903943,
      "grad_norm": 3.15625,
      "learning_rate": 4.99289152731256e-05,
      "loss": 0.9343,
      "step": 1900
    },
    {
      "epoch": 0.026971932598530157,
      "grad_norm": 3.34375,
      "learning_rate": 4.9928746806223545e-05,
      "loss": 0.9743,
      "step": 1902
    },
    {
      "epoch": 0.027000294252156373,
      "grad_norm": 3.140625,
      "learning_rate": 4.992857814021393e-05,
      "loss": 0.9736,
      "step": 1904
    },
    {
      "epoch": 0.027028655905782586,
      "grad_norm": 2.9375,
      "learning_rate": 4.9928409275098106e-05,
      "loss": 0.9568,
      "step": 1906
    },
    {
      "epoch": 0.0270570175594088,
      "grad_norm": 3.109375,
      "learning_rate": 4.992824021087741e-05,
      "loss": 0.9995,
      "step": 1908
    },
    {
      "epoch": 0.027085379213035016,
      "grad_norm": 2.96875,
      "learning_rate": 4.99280709475532e-05,
      "loss": 0.9632,
      "step": 1910
    },
    {
      "epoch": 0.02711374086666123,
      "grad_norm": 2.984375,
      "learning_rate": 4.992790148512682e-05,
      "loss": 0.8786,
      "step": 1912
    },
    {
      "epoch": 0.027142102520287446,
      "grad_norm": 3.140625,
      "learning_rate": 4.992773182359963e-05,
      "loss": 1.0195,
      "step": 1914
    },
    {
      "epoch": 0.02717046417391366,
      "grad_norm": 3.015625,
      "learning_rate": 4.992756196297298e-05,
      "loss": 0.9875,
      "step": 1916
    },
    {
      "epoch": 0.027198825827539876,
      "grad_norm": 3.390625,
      "learning_rate": 4.9927391903248235e-05,
      "loss": 0.9795,
      "step": 1918
    },
    {
      "epoch": 0.02722718748116609,
      "grad_norm": 3.125,
      "learning_rate": 4.992722164442675e-05,
      "loss": 0.9745,
      "step": 1920
    },
    {
      "epoch": 0.027255549134792303,
      "grad_norm": 3.4375,
      "learning_rate": 4.9927051186509876e-05,
      "loss": 0.9001,
      "step": 1922
    },
    {
      "epoch": 0.02728391078841852,
      "grad_norm": 3.046875,
      "learning_rate": 4.992688052949899e-05,
      "loss": 0.9549,
      "step": 1924
    },
    {
      "epoch": 0.027312272442044733,
      "grad_norm": 3.171875,
      "learning_rate": 4.992670967339543e-05,
      "loss": 1.0138,
      "step": 1926
    },
    {
      "epoch": 0.02734063409567095,
      "grad_norm": 3.015625,
      "learning_rate": 4.9926538618200594e-05,
      "loss": 0.9425,
      "step": 1928
    },
    {
      "epoch": 0.027368995749297163,
      "grad_norm": 3.359375,
      "learning_rate": 4.992636736391583e-05,
      "loss": 0.928,
      "step": 1930
    },
    {
      "epoch": 0.027397357402923376,
      "grad_norm": 3.9375,
      "learning_rate": 4.99261959105425e-05,
      "loss": 0.9705,
      "step": 1932
    },
    {
      "epoch": 0.027425719056549593,
      "grad_norm": 3.203125,
      "learning_rate": 4.992602425808198e-05,
      "loss": 0.9711,
      "step": 1934
    },
    {
      "epoch": 0.027454080710175806,
      "grad_norm": 3.203125,
      "learning_rate": 4.992585240653564e-05,
      "loss": 0.9817,
      "step": 1936
    },
    {
      "epoch": 0.027482442363802023,
      "grad_norm": 3.203125,
      "learning_rate": 4.992568035590487e-05,
      "loss": 0.9607,
      "step": 1938
    },
    {
      "epoch": 0.027510804017428236,
      "grad_norm": 3.234375,
      "learning_rate": 4.992550810619101e-05,
      "loss": 0.9138,
      "step": 1940
    },
    {
      "epoch": 0.027539165671054452,
      "grad_norm": 3.3125,
      "learning_rate": 4.992533565739547e-05,
      "loss": 0.9808,
      "step": 1942
    },
    {
      "epoch": 0.027567527324680666,
      "grad_norm": 3.265625,
      "learning_rate": 4.9925163009519607e-05,
      "loss": 0.9404,
      "step": 1944
    },
    {
      "epoch": 0.02759588897830688,
      "grad_norm": 2.953125,
      "learning_rate": 4.992499016256479e-05,
      "loss": 0.9314,
      "step": 1946
    },
    {
      "epoch": 0.027624250631933096,
      "grad_norm": 3.25,
      "learning_rate": 4.992481711653243e-05,
      "loss": 0.9184,
      "step": 1948
    },
    {
      "epoch": 0.02765261228555931,
      "grad_norm": 2.921875,
      "learning_rate": 4.9924643871423895e-05,
      "loss": 0.9711,
      "step": 1950
    },
    {
      "epoch": 0.027680973939185526,
      "grad_norm": 3.109375,
      "learning_rate": 4.9924470427240556e-05,
      "loss": 0.9914,
      "step": 1952
    },
    {
      "epoch": 0.02770933559281174,
      "grad_norm": 3.125,
      "learning_rate": 4.9924296783983814e-05,
      "loss": 1.041,
      "step": 1954
    },
    {
      "epoch": 0.027737697246437952,
      "grad_norm": 2.734375,
      "learning_rate": 4.992412294165505e-05,
      "loss": 0.9332,
      "step": 1956
    },
    {
      "epoch": 0.02776605890006417,
      "grad_norm": 3.046875,
      "learning_rate": 4.9923948900255656e-05,
      "loss": 1.0057,
      "step": 1958
    },
    {
      "epoch": 0.027794420553690382,
      "grad_norm": 3.125,
      "learning_rate": 4.992377465978702e-05,
      "loss": 0.9565,
      "step": 1960
    },
    {
      "epoch": 0.0278227822073166,
      "grad_norm": 3.296875,
      "learning_rate": 4.9923600220250526e-05,
      "loss": 0.9139,
      "step": 1962
    },
    {
      "epoch": 0.027851143860942812,
      "grad_norm": 3.671875,
      "learning_rate": 4.992342558164758e-05,
      "loss": 0.9591,
      "step": 1964
    },
    {
      "epoch": 0.02787950551456903,
      "grad_norm": 3.8125,
      "learning_rate": 4.992325074397957e-05,
      "loss": 0.957,
      "step": 1966
    },
    {
      "epoch": 0.027907867168195242,
      "grad_norm": 3.09375,
      "learning_rate": 4.99230757072479e-05,
      "loss": 0.9193,
      "step": 1968
    },
    {
      "epoch": 0.027936228821821455,
      "grad_norm": 3.109375,
      "learning_rate": 4.992290047145396e-05,
      "loss": 0.9929,
      "step": 1970
    },
    {
      "epoch": 0.027964590475447672,
      "grad_norm": 2.90625,
      "learning_rate": 4.9922725036599146e-05,
      "loss": 0.9628,
      "step": 1972
    },
    {
      "epoch": 0.027992952129073885,
      "grad_norm": 3.296875,
      "learning_rate": 4.992254940268487e-05,
      "loss": 0.9564,
      "step": 1974
    },
    {
      "epoch": 0.0280213137827001,
      "grad_norm": 3.140625,
      "learning_rate": 4.992237356971252e-05,
      "loss": 0.9706,
      "step": 1976
    },
    {
      "epoch": 0.028049675436326315,
      "grad_norm": 3.078125,
      "learning_rate": 4.9922197537683524e-05,
      "loss": 0.9517,
      "step": 1978
    },
    {
      "epoch": 0.028078037089952528,
      "grad_norm": 3.015625,
      "learning_rate": 4.992202130659927e-05,
      "loss": 0.9207,
      "step": 1980
    },
    {
      "epoch": 0.028106398743578745,
      "grad_norm": 3.03125,
      "learning_rate": 4.992184487646116e-05,
      "loss": 0.9569,
      "step": 1982
    },
    {
      "epoch": 0.028134760397204958,
      "grad_norm": 3.65625,
      "learning_rate": 4.992166824727062e-05,
      "loss": 0.9525,
      "step": 1984
    },
    {
      "epoch": 0.028163122050831175,
      "grad_norm": 3.53125,
      "learning_rate": 4.992149141902905e-05,
      "loss": 0.9863,
      "step": 1986
    },
    {
      "epoch": 0.028191483704457388,
      "grad_norm": 3.4375,
      "learning_rate": 4.9921314391737875e-05,
      "loss": 0.9697,
      "step": 1988
    },
    {
      "epoch": 0.028219845358083605,
      "grad_norm": 3.25,
      "learning_rate": 4.992113716539849e-05,
      "loss": 0.9605,
      "step": 1990
    },
    {
      "epoch": 0.028248207011709818,
      "grad_norm": 3.40625,
      "learning_rate": 4.9920959740012326e-05,
      "loss": 0.9802,
      "step": 1992
    },
    {
      "epoch": 0.02827656866533603,
      "grad_norm": 3.03125,
      "learning_rate": 4.9920782115580785e-05,
      "loss": 0.9701,
      "step": 1994
    },
    {
      "epoch": 0.028304930318962248,
      "grad_norm": 2.921875,
      "learning_rate": 4.9920604292105304e-05,
      "loss": 0.9683,
      "step": 1996
    },
    {
      "epoch": 0.02833329197258846,
      "grad_norm": 2.796875,
      "learning_rate": 4.992042626958729e-05,
      "loss": 0.9577,
      "step": 1998
    },
    {
      "epoch": 0.028361653626214678,
      "grad_norm": 3.78125,
      "learning_rate": 4.992024804802817e-05,
      "loss": 0.9432,
      "step": 2000
    },
    {
      "epoch": 0.02839001527984089,
      "grad_norm": 3.3125,
      "learning_rate": 4.9920069627429375e-05,
      "loss": 0.9747,
      "step": 2002
    },
    {
      "epoch": 0.028418376933467104,
      "grad_norm": 3.34375,
      "learning_rate": 4.9919891007792315e-05,
      "loss": 0.9456,
      "step": 2004
    },
    {
      "epoch": 0.02844673858709332,
      "grad_norm": 3.1875,
      "learning_rate": 4.9919712189118424e-05,
      "loss": 1.0218,
      "step": 2006
    },
    {
      "epoch": 0.028475100240719534,
      "grad_norm": 3.609375,
      "learning_rate": 4.991953317140913e-05,
      "loss": 1.0013,
      "step": 2008
    },
    {
      "epoch": 0.02850346189434575,
      "grad_norm": 3.640625,
      "learning_rate": 4.991935395466586e-05,
      "loss": 0.9592,
      "step": 2010
    },
    {
      "epoch": 0.028531823547971964,
      "grad_norm": 3.34375,
      "learning_rate": 4.991917453889004e-05,
      "loss": 0.9389,
      "step": 2012
    },
    {
      "epoch": 0.028560185201598177,
      "grad_norm": 3.3125,
      "learning_rate": 4.991899492408313e-05,
      "loss": 1.0012,
      "step": 2014
    },
    {
      "epoch": 0.028588546855224394,
      "grad_norm": 3.390625,
      "learning_rate": 4.991881511024653e-05,
      "loss": 0.9487,
      "step": 2016
    },
    {
      "epoch": 0.028616908508850607,
      "grad_norm": 3.28125,
      "learning_rate": 4.991863509738169e-05,
      "loss": 1.0159,
      "step": 2018
    },
    {
      "epoch": 0.028645270162476824,
      "grad_norm": 2.96875,
      "learning_rate": 4.991845488549006e-05,
      "loss": 0.9811,
      "step": 2020
    },
    {
      "epoch": 0.028673631816103037,
      "grad_norm": 3.375,
      "learning_rate": 4.991827447457307e-05,
      "loss": 0.9509,
      "step": 2022
    },
    {
      "epoch": 0.028701993469729254,
      "grad_norm": 2.828125,
      "learning_rate": 4.991809386463216e-05,
      "loss": 0.9578,
      "step": 2024
    },
    {
      "epoch": 0.028730355123355467,
      "grad_norm": 3.203125,
      "learning_rate": 4.991791305566876e-05,
      "loss": 0.9471,
      "step": 2026
    },
    {
      "epoch": 0.02875871677698168,
      "grad_norm": 3.109375,
      "learning_rate": 4.991773204768433e-05,
      "loss": 0.9923,
      "step": 2028
    },
    {
      "epoch": 0.028787078430607897,
      "grad_norm": 2.984375,
      "learning_rate": 4.9917550840680317e-05,
      "loss": 0.9028,
      "step": 2030
    },
    {
      "epoch": 0.02881544008423411,
      "grad_norm": 3.09375,
      "learning_rate": 4.991736943465816e-05,
      "loss": 1.0057,
      "step": 2032
    },
    {
      "epoch": 0.028843801737860327,
      "grad_norm": 3.453125,
      "learning_rate": 4.9917187829619313e-05,
      "loss": 1.0544,
      "step": 2034
    },
    {
      "epoch": 0.02887216339148654,
      "grad_norm": 3.5,
      "learning_rate": 4.991700602556523e-05,
      "loss": 0.9443,
      "step": 2036
    },
    {
      "epoch": 0.028900525045112754,
      "grad_norm": 3.4375,
      "learning_rate": 4.9916824022497346e-05,
      "loss": 0.9566,
      "step": 2038
    },
    {
      "epoch": 0.02892888669873897,
      "grad_norm": 3.3125,
      "learning_rate": 4.991664182041714e-05,
      "loss": 1.0628,
      "step": 2040
    },
    {
      "epoch": 0.028957248352365184,
      "grad_norm": 3.53125,
      "learning_rate": 4.991645941932604e-05,
      "loss": 0.9259,
      "step": 2042
    },
    {
      "epoch": 0.0289856100059914,
      "grad_norm": 3.390625,
      "learning_rate": 4.991627681922553e-05,
      "loss": 0.9033,
      "step": 2044
    },
    {
      "epoch": 0.029013971659617614,
      "grad_norm": 3.453125,
      "learning_rate": 4.9916094020117044e-05,
      "loss": 0.9603,
      "step": 2046
    },
    {
      "epoch": 0.02904233331324383,
      "grad_norm": 3.0625,
      "learning_rate": 4.991591102200206e-05,
      "loss": 0.9132,
      "step": 2048
    },
    {
      "epoch": 0.029070694966870043,
      "grad_norm": 3.03125,
      "learning_rate": 4.991572782488203e-05,
      "loss": 0.9205,
      "step": 2050
    },
    {
      "epoch": 0.029099056620496257,
      "grad_norm": 2.921875,
      "learning_rate": 4.991554442875842e-05,
      "loss": 0.9435,
      "step": 2052
    },
    {
      "epoch": 0.029127418274122473,
      "grad_norm": 3.078125,
      "learning_rate": 4.991536083363269e-05,
      "loss": 0.9144,
      "step": 2054
    },
    {
      "epoch": 0.029155779927748687,
      "grad_norm": 3.296875,
      "learning_rate": 4.9915177039506325e-05,
      "loss": 0.9861,
      "step": 2056
    },
    {
      "epoch": 0.029184141581374903,
      "grad_norm": 3.875,
      "learning_rate": 4.991499304638077e-05,
      "loss": 1.0043,
      "step": 2058
    },
    {
      "epoch": 0.029212503235001117,
      "grad_norm": 3.359375,
      "learning_rate": 4.991480885425751e-05,
      "loss": 0.9888,
      "step": 2060
    },
    {
      "epoch": 0.02924086488862733,
      "grad_norm": 3.4375,
      "learning_rate": 4.9914624463138e-05,
      "loss": 0.9761,
      "step": 2062
    },
    {
      "epoch": 0.029269226542253547,
      "grad_norm": 3.25,
      "learning_rate": 4.9914439873023724e-05,
      "loss": 0.968,
      "step": 2064
    },
    {
      "epoch": 0.02929758819587976,
      "grad_norm": 3.046875,
      "learning_rate": 4.991425508391616e-05,
      "loss": 0.9443,
      "step": 2066
    },
    {
      "epoch": 0.029325949849505976,
      "grad_norm": 2.5625,
      "learning_rate": 4.9914070095816775e-05,
      "loss": 0.9272,
      "step": 2068
    },
    {
      "epoch": 0.02935431150313219,
      "grad_norm": 3.140625,
      "learning_rate": 4.991388490872705e-05,
      "loss": 0.9669,
      "step": 2070
    },
    {
      "epoch": 0.029382673156758406,
      "grad_norm": 3.703125,
      "learning_rate": 4.991369952264847e-05,
      "loss": 1.0146,
      "step": 2072
    },
    {
      "epoch": 0.02941103481038462,
      "grad_norm": 2.859375,
      "learning_rate": 4.991351393758251e-05,
      "loss": 0.8965,
      "step": 2074
    },
    {
      "epoch": 0.029439396464010833,
      "grad_norm": 3.234375,
      "learning_rate": 4.991332815353065e-05,
      "loss": 0.9345,
      "step": 2076
    },
    {
      "epoch": 0.02946775811763705,
      "grad_norm": 2.65625,
      "learning_rate": 4.991314217049437e-05,
      "loss": 0.9065,
      "step": 2078
    },
    {
      "epoch": 0.029496119771263263,
      "grad_norm": 3.125,
      "learning_rate": 4.991295598847517e-05,
      "loss": 0.9222,
      "step": 2080
    },
    {
      "epoch": 0.02952448142488948,
      "grad_norm": 3.109375,
      "learning_rate": 4.991276960747452e-05,
      "loss": 0.9114,
      "step": 2082
    },
    {
      "epoch": 0.029552843078515693,
      "grad_norm": 3.359375,
      "learning_rate": 4.991258302749393e-05,
      "loss": 0.9781,
      "step": 2084
    },
    {
      "epoch": 0.029581204732141906,
      "grad_norm": 2.8125,
      "learning_rate": 4.991239624853487e-05,
      "loss": 0.9602,
      "step": 2086
    },
    {
      "epoch": 0.029609566385768123,
      "grad_norm": 2.828125,
      "learning_rate": 4.991220927059884e-05,
      "loss": 0.9674,
      "step": 2088
    },
    {
      "epoch": 0.029637928039394336,
      "grad_norm": 2.9375,
      "learning_rate": 4.991202209368733e-05,
      "loss": 0.9406,
      "step": 2090
    },
    {
      "epoch": 0.029666289693020553,
      "grad_norm": 3.15625,
      "learning_rate": 4.991183471780184e-05,
      "loss": 0.9482,
      "step": 2092
    },
    {
      "epoch": 0.029694651346646766,
      "grad_norm": 3.546875,
      "learning_rate": 4.991164714294387e-05,
      "loss": 0.9941,
      "step": 2094
    },
    {
      "epoch": 0.029723013000272983,
      "grad_norm": 3.28125,
      "learning_rate": 4.991145936911491e-05,
      "loss": 0.9457,
      "step": 2096
    },
    {
      "epoch": 0.029751374653899196,
      "grad_norm": 3.390625,
      "learning_rate": 4.991127139631646e-05,
      "loss": 0.9496,
      "step": 2098
    },
    {
      "epoch": 0.02977973630752541,
      "grad_norm": 3.046875,
      "learning_rate": 4.9911083224550024e-05,
      "loss": 0.9594,
      "step": 2100
    },
    {
      "epoch": 0.029808097961151626,
      "grad_norm": 3.375,
      "learning_rate": 4.9910894853817106e-05,
      "loss": 0.9344,
      "step": 2102
    },
    {
      "epoch": 0.02983645961477784,
      "grad_norm": 3.359375,
      "learning_rate": 4.991070628411921e-05,
      "loss": 0.9193,
      "step": 2104
    },
    {
      "epoch": 0.029864821268404056,
      "grad_norm": 3.078125,
      "learning_rate": 4.991051751545784e-05,
      "loss": 0.9022,
      "step": 2106
    },
    {
      "epoch": 0.02989318292203027,
      "grad_norm": 2.78125,
      "learning_rate": 4.991032854783451e-05,
      "loss": 0.928,
      "step": 2108
    },
    {
      "epoch": 0.029921544575656482,
      "grad_norm": 3.359375,
      "learning_rate": 4.991013938125072e-05,
      "loss": 0.9092,
      "step": 2110
    },
    {
      "epoch": 0.0299499062292827,
      "grad_norm": 3.546875,
      "learning_rate": 4.990995001570798e-05,
      "loss": 0.989,
      "step": 2112
    },
    {
      "epoch": 0.029978267882908912,
      "grad_norm": 3.296875,
      "learning_rate": 4.990976045120781e-05,
      "loss": 0.9392,
      "step": 2114
    },
    {
      "epoch": 0.03000662953653513,
      "grad_norm": 2.984375,
      "learning_rate": 4.990957068775173e-05,
      "loss": 0.9926,
      "step": 2116
    },
    {
      "epoch": 0.030034991190161342,
      "grad_norm": 3.296875,
      "learning_rate": 4.990938072534124e-05,
      "loss": 0.9788,
      "step": 2118
    },
    {
      "epoch": 0.03006335284378756,
      "grad_norm": 3.078125,
      "learning_rate": 4.990919056397786e-05,
      "loss": 0.9279,
      "step": 2120
    },
    {
      "epoch": 0.030091714497413772,
      "grad_norm": 2.84375,
      "learning_rate": 4.990900020366313e-05,
      "loss": 0.9286,
      "step": 2122
    },
    {
      "epoch": 0.030120076151039985,
      "grad_norm": 2.875,
      "learning_rate": 4.990880964439853e-05,
      "loss": 0.8859,
      "step": 2124
    },
    {
      "epoch": 0.030148437804666202,
      "grad_norm": 3.3125,
      "learning_rate": 4.990861888618562e-05,
      "loss": 0.9428,
      "step": 2126
    },
    {
      "epoch": 0.030176799458292415,
      "grad_norm": 2.96875,
      "learning_rate": 4.990842792902591e-05,
      "loss": 1.0067,
      "step": 2128
    },
    {
      "epoch": 0.030205161111918632,
      "grad_norm": 2.71875,
      "learning_rate": 4.990823677292092e-05,
      "loss": 0.9069,
      "step": 2130
    },
    {
      "epoch": 0.030233522765544845,
      "grad_norm": 3.265625,
      "learning_rate": 4.99080454178722e-05,
      "loss": 0.9975,
      "step": 2132
    },
    {
      "epoch": 0.03026188441917106,
      "grad_norm": 3.34375,
      "learning_rate": 4.990785386388124e-05,
      "loss": 0.9642,
      "step": 2134
    },
    {
      "epoch": 0.030290246072797275,
      "grad_norm": 3.65625,
      "learning_rate": 4.990766211094959e-05,
      "loss": 1.0293,
      "step": 2136
    },
    {
      "epoch": 0.030318607726423488,
      "grad_norm": 3.34375,
      "learning_rate": 4.990747015907879e-05,
      "loss": 0.9714,
      "step": 2138
    },
    {
      "epoch": 0.030346969380049705,
      "grad_norm": 2.890625,
      "learning_rate": 4.9907278008270364e-05,
      "loss": 0.9857,
      "step": 2140
    },
    {
      "epoch": 0.030375331033675918,
      "grad_norm": 3.796875,
      "learning_rate": 4.990708565852584e-05,
      "loss": 0.9854,
      "step": 2142
    },
    {
      "epoch": 0.030403692687302135,
      "grad_norm": 3.21875,
      "learning_rate": 4.9906893109846765e-05,
      "loss": 0.9813,
      "step": 2144
    },
    {
      "epoch": 0.030432054340928348,
      "grad_norm": 3.328125,
      "learning_rate": 4.990670036223468e-05,
      "loss": 0.9661,
      "step": 2146
    },
    {
      "epoch": 0.03046041599455456,
      "grad_norm": 3.109375,
      "learning_rate": 4.990650741569111e-05,
      "loss": 0.9802,
      "step": 2148
    },
    {
      "epoch": 0.030488777648180778,
      "grad_norm": 3.109375,
      "learning_rate": 4.99063142702176e-05,
      "loss": 0.9553,
      "step": 2150
    },
    {
      "epoch": 0.03051713930180699,
      "grad_norm": 3.15625,
      "learning_rate": 4.9906120925815706e-05,
      "loss": 0.9449,
      "step": 2152
    },
    {
      "epoch": 0.030545500955433208,
      "grad_norm": 3.03125,
      "learning_rate": 4.990592738248696e-05,
      "loss": 0.9758,
      "step": 2154
    },
    {
      "epoch": 0.03057386260905942,
      "grad_norm": 2.9375,
      "learning_rate": 4.990573364023291e-05,
      "loss": 1.0034,
      "step": 2156
    },
    {
      "epoch": 0.030602224262685634,
      "grad_norm": 3.0,
      "learning_rate": 4.9905539699055104e-05,
      "loss": 0.954,
      "step": 2158
    },
    {
      "epoch": 0.03063058591631185,
      "grad_norm": 3.859375,
      "learning_rate": 4.99053455589551e-05,
      "loss": 0.982,
      "step": 2160
    },
    {
      "epoch": 0.030658947569938064,
      "grad_norm": 3.03125,
      "learning_rate": 4.990515121993442e-05,
      "loss": 0.9485,
      "step": 2162
    },
    {
      "epoch": 0.03068730922356428,
      "grad_norm": 2.796875,
      "learning_rate": 4.990495668199465e-05,
      "loss": 0.9123,
      "step": 2164
    },
    {
      "epoch": 0.030715670877190494,
      "grad_norm": 3.0625,
      "learning_rate": 4.990476194513733e-05,
      "loss": 0.94,
      "step": 2166
    },
    {
      "epoch": 0.030744032530816708,
      "grad_norm": 2.921875,
      "learning_rate": 4.990456700936401e-05,
      "loss": 0.9483,
      "step": 2168
    },
    {
      "epoch": 0.030772394184442924,
      "grad_norm": 3.3125,
      "learning_rate": 4.990437187467626e-05,
      "loss": 0.9642,
      "step": 2170
    },
    {
      "epoch": 0.030800755838069137,
      "grad_norm": 3.53125,
      "learning_rate": 4.990417654107562e-05,
      "loss": 0.9336,
      "step": 2172
    },
    {
      "epoch": 0.030829117491695354,
      "grad_norm": 3.34375,
      "learning_rate": 4.990398100856367e-05,
      "loss": 0.9772,
      "step": 2174
    },
    {
      "epoch": 0.030857479145321567,
      "grad_norm": 3.21875,
      "learning_rate": 4.990378527714196e-05,
      "loss": 0.9597,
      "step": 2176
    },
    {
      "epoch": 0.030885840798947784,
      "grad_norm": 3.0,
      "learning_rate": 4.9903589346812054e-05,
      "loss": 0.9722,
      "step": 2178
    },
    {
      "epoch": 0.030914202452573997,
      "grad_norm": 2.78125,
      "learning_rate": 4.990339321757552e-05,
      "loss": 0.9192,
      "step": 2180
    },
    {
      "epoch": 0.03094256410620021,
      "grad_norm": 3.078125,
      "learning_rate": 4.990319688943392e-05,
      "loss": 0.9073,
      "step": 2182
    },
    {
      "epoch": 0.030970925759826427,
      "grad_norm": 2.78125,
      "learning_rate": 4.990300036238883e-05,
      "loss": 0.9574,
      "step": 2184
    },
    {
      "epoch": 0.03099928741345264,
      "grad_norm": 3.046875,
      "learning_rate": 4.990280363644181e-05,
      "loss": 0.9759,
      "step": 2186
    },
    {
      "epoch": 0.031027649067078857,
      "grad_norm": 3.0625,
      "learning_rate": 4.990260671159444e-05,
      "loss": 0.9791,
      "step": 2188
    },
    {
      "epoch": 0.03105601072070507,
      "grad_norm": 2.953125,
      "learning_rate": 4.990240958784829e-05,
      "loss": 0.944,
      "step": 2190
    },
    {
      "epoch": 0.031084372374331284,
      "grad_norm": 2.796875,
      "learning_rate": 4.990221226520493e-05,
      "loss": 0.9255,
      "step": 2192
    },
    {
      "epoch": 0.0311127340279575,
      "grad_norm": 2.921875,
      "learning_rate": 4.990201474366594e-05,
      "loss": 0.9418,
      "step": 2194
    },
    {
      "epoch": 0.031141095681583714,
      "grad_norm": 3.234375,
      "learning_rate": 4.99018170232329e-05,
      "loss": 0.9401,
      "step": 2196
    },
    {
      "epoch": 0.03116945733520993,
      "grad_norm": 3.0625,
      "learning_rate": 4.990161910390739e-05,
      "loss": 0.9653,
      "step": 2198
    },
    {
      "epoch": 0.031197818988836144,
      "grad_norm": 3.34375,
      "learning_rate": 4.9901420985690985e-05,
      "loss": 0.9955,
      "step": 2200
    },
    {
      "epoch": 0.03122618064246236,
      "grad_norm": 3.484375,
      "learning_rate": 4.9901222668585266e-05,
      "loss": 0.9829,
      "step": 2202
    },
    {
      "epoch": 0.031254542296088574,
      "grad_norm": 3.1875,
      "learning_rate": 4.990102415259182e-05,
      "loss": 0.8901,
      "step": 2204
    },
    {
      "epoch": 0.03128290394971479,
      "grad_norm": 3.546875,
      "learning_rate": 4.990082543771224e-05,
      "loss": 1.0051,
      "step": 2206
    },
    {
      "epoch": 0.031311265603341,
      "grad_norm": 3.140625,
      "learning_rate": 4.99006265239481e-05,
      "loss": 0.9163,
      "step": 2208
    },
    {
      "epoch": 0.03133962725696722,
      "grad_norm": 2.84375,
      "learning_rate": 4.990042741130099e-05,
      "loss": 0.8886,
      "step": 2210
    },
    {
      "epoch": 0.03136798891059343,
      "grad_norm": 3.5,
      "learning_rate": 4.9900228099772516e-05,
      "loss": 0.9822,
      "step": 2212
    },
    {
      "epoch": 0.03139635056421965,
      "grad_norm": 3.03125,
      "learning_rate": 4.9900028589364254e-05,
      "loss": 0.9296,
      "step": 2214
    },
    {
      "epoch": 0.03142471221784586,
      "grad_norm": 3.15625,
      "learning_rate": 4.9899828880077806e-05,
      "loss": 0.9333,
      "step": 2216
    },
    {
      "epoch": 0.03145307387147207,
      "grad_norm": 3.34375,
      "learning_rate": 4.9899628971914764e-05,
      "loss": 0.9307,
      "step": 2218
    },
    {
      "epoch": 0.03148143552509829,
      "grad_norm": 3.40625,
      "learning_rate": 4.989942886487672e-05,
      "loss": 0.9732,
      "step": 2220
    },
    {
      "epoch": 0.031509797178724507,
      "grad_norm": 2.765625,
      "learning_rate": 4.989922855896528e-05,
      "loss": 0.953,
      "step": 2222
    },
    {
      "epoch": 0.03153815883235072,
      "grad_norm": 3.21875,
      "learning_rate": 4.989902805418204e-05,
      "loss": 0.9886,
      "step": 2224
    },
    {
      "epoch": 0.03156652048597693,
      "grad_norm": 3.640625,
      "learning_rate": 4.9898827350528605e-05,
      "loss": 0.9232,
      "step": 2226
    },
    {
      "epoch": 0.031594882139603146,
      "grad_norm": 3.140625,
      "learning_rate": 4.989862644800657e-05,
      "loss": 0.9158,
      "step": 2228
    },
    {
      "epoch": 0.031623243793229366,
      "grad_norm": 3.46875,
      "learning_rate": 4.989842534661755e-05,
      "loss": 0.9896,
      "step": 2230
    },
    {
      "epoch": 0.03165160544685558,
      "grad_norm": 3.40625,
      "learning_rate": 4.989822404636314e-05,
      "loss": 0.9294,
      "step": 2232
    },
    {
      "epoch": 0.03167996710048179,
      "grad_norm": 3.046875,
      "learning_rate": 4.989802254724496e-05,
      "loss": 0.9501,
      "step": 2234
    },
    {
      "epoch": 0.031708328754108006,
      "grad_norm": 3.203125,
      "learning_rate": 4.989782084926461e-05,
      "loss": 0.9218,
      "step": 2236
    },
    {
      "epoch": 0.031736690407734226,
      "grad_norm": 3.03125,
      "learning_rate": 4.989761895242371e-05,
      "loss": 0.9732,
      "step": 2238
    },
    {
      "epoch": 0.03176505206136044,
      "grad_norm": 3.4375,
      "learning_rate": 4.9897416856723856e-05,
      "loss": 0.984,
      "step": 2240
    },
    {
      "epoch": 0.03179341371498665,
      "grad_norm": 3.203125,
      "learning_rate": 4.989721456216668e-05,
      "loss": 0.901,
      "step": 2242
    },
    {
      "epoch": 0.031821775368612866,
      "grad_norm": 3.25,
      "learning_rate": 4.989701206875379e-05,
      "loss": 0.8876,
      "step": 2244
    },
    {
      "epoch": 0.03185013702223908,
      "grad_norm": 3.171875,
      "learning_rate": 4.9896809376486796e-05,
      "loss": 0.9965,
      "step": 2246
    },
    {
      "epoch": 0.0318784986758653,
      "grad_norm": 3.484375,
      "learning_rate": 4.989660648536734e-05,
      "loss": 1.0027,
      "step": 2248
    },
    {
      "epoch": 0.03190686032949151,
      "grad_norm": 2.984375,
      "learning_rate": 4.989640339539702e-05,
      "loss": 0.9797,
      "step": 2250
    },
    {
      "epoch": 0.031935221983117726,
      "grad_norm": 3.0625,
      "learning_rate": 4.9896200106577465e-05,
      "loss": 0.9285,
      "step": 2252
    },
    {
      "epoch": 0.03196358363674394,
      "grad_norm": 2.953125,
      "learning_rate": 4.989599661891029e-05,
      "loss": 0.9713,
      "step": 2254
    },
    {
      "epoch": 0.03199194529037015,
      "grad_norm": 2.78125,
      "learning_rate": 4.989579293239715e-05,
      "loss": 0.947,
      "step": 2256
    },
    {
      "epoch": 0.03202030694399637,
      "grad_norm": 2.84375,
      "learning_rate": 4.989558904703963e-05,
      "loss": 0.9117,
      "step": 2258
    },
    {
      "epoch": 0.032048668597622586,
      "grad_norm": 3.34375,
      "learning_rate": 4.989538496283939e-05,
      "loss": 0.9845,
      "step": 2260
    },
    {
      "epoch": 0.0320770302512488,
      "grad_norm": 3.40625,
      "learning_rate": 4.989518067979805e-05,
      "loss": 0.9599,
      "step": 2262
    },
    {
      "epoch": 0.03210539190487501,
      "grad_norm": 3.125,
      "learning_rate": 4.989497619791724e-05,
      "loss": 0.9601,
      "step": 2264
    },
    {
      "epoch": 0.032133753558501225,
      "grad_norm": 2.9375,
      "learning_rate": 4.98947715171986e-05,
      "loss": 0.9082,
      "step": 2266
    },
    {
      "epoch": 0.032162115212127446,
      "grad_norm": 3.3125,
      "learning_rate": 4.989456663764375e-05,
      "loss": 0.9261,
      "step": 2268
    },
    {
      "epoch": 0.03219047686575366,
      "grad_norm": 3.546875,
      "learning_rate": 4.9894361559254345e-05,
      "loss": 0.925,
      "step": 2270
    },
    {
      "epoch": 0.03221883851937987,
      "grad_norm": 3.703125,
      "learning_rate": 4.9894156282032e-05,
      "loss": 1.0176,
      "step": 2272
    },
    {
      "epoch": 0.032247200173006085,
      "grad_norm": 3.078125,
      "learning_rate": 4.989395080597838e-05,
      "loss": 0.9862,
      "step": 2274
    },
    {
      "epoch": 0.0322755618266323,
      "grad_norm": 3.609375,
      "learning_rate": 4.9893745131095116e-05,
      "loss": 0.9733,
      "step": 2276
    },
    {
      "epoch": 0.03230392348025852,
      "grad_norm": 3.0,
      "learning_rate": 4.989353925738384e-05,
      "loss": 0.9576,
      "step": 2278
    },
    {
      "epoch": 0.03233228513388473,
      "grad_norm": 2.984375,
      "learning_rate": 4.9893333184846214e-05,
      "loss": 0.9234,
      "step": 2280
    },
    {
      "epoch": 0.032360646787510945,
      "grad_norm": 3.390625,
      "learning_rate": 4.989312691348387e-05,
      "loss": 0.9416,
      "step": 2282
    },
    {
      "epoch": 0.03238900844113716,
      "grad_norm": 3.140625,
      "learning_rate": 4.989292044329846e-05,
      "loss": 0.9948,
      "step": 2284
    },
    {
      "epoch": 0.03241737009476338,
      "grad_norm": 3.140625,
      "learning_rate": 4.989271377429164e-05,
      "loss": 0.9116,
      "step": 2286
    },
    {
      "epoch": 0.03244573174838959,
      "grad_norm": 3.34375,
      "learning_rate": 4.9892506906465055e-05,
      "loss": 0.9318,
      "step": 2288
    },
    {
      "epoch": 0.032474093402015805,
      "grad_norm": 3.09375,
      "learning_rate": 4.989229983982035e-05,
      "loss": 1.0087,
      "step": 2290
    },
    {
      "epoch": 0.03250245505564202,
      "grad_norm": 3.421875,
      "learning_rate": 4.989209257435919e-05,
      "loss": 0.9494,
      "step": 2292
    },
    {
      "epoch": 0.03253081670926823,
      "grad_norm": 3.109375,
      "learning_rate": 4.989188511008322e-05,
      "loss": 0.9293,
      "step": 2294
    },
    {
      "epoch": 0.03255917836289445,
      "grad_norm": 2.96875,
      "learning_rate": 4.989167744699411e-05,
      "loss": 0.9368,
      "step": 2296
    },
    {
      "epoch": 0.032587540016520665,
      "grad_norm": 3.359375,
      "learning_rate": 4.9891469585093514e-05,
      "loss": 1.0246,
      "step": 2298
    },
    {
      "epoch": 0.03261590167014688,
      "grad_norm": 3.40625,
      "learning_rate": 4.989126152438309e-05,
      "loss": 0.9942,
      "step": 2300
    },
    {
      "epoch": 0.03264426332377309,
      "grad_norm": 2.984375,
      "learning_rate": 4.98910532648645e-05,
      "loss": 0.9162,
      "step": 2302
    },
    {
      "epoch": 0.032672624977399305,
      "grad_norm": 3.546875,
      "learning_rate": 4.98908448065394e-05,
      "loss": 0.9538,
      "step": 2304
    },
    {
      "epoch": 0.032700986631025525,
      "grad_norm": 3.0625,
      "learning_rate": 4.989063614940946e-05,
      "loss": 0.8911,
      "step": 2306
    },
    {
      "epoch": 0.03272934828465174,
      "grad_norm": 3.015625,
      "learning_rate": 4.9890427293476364e-05,
      "loss": 0.9499,
      "step": 2308
    },
    {
      "epoch": 0.03275770993827795,
      "grad_norm": 3.453125,
      "learning_rate": 4.9890218238741753e-05,
      "loss": 0.9706,
      "step": 2310
    },
    {
      "epoch": 0.032786071591904165,
      "grad_norm": 3.75,
      "learning_rate": 4.989000898520732e-05,
      "loss": 0.9541,
      "step": 2312
    },
    {
      "epoch": 0.03281443324553038,
      "grad_norm": 3.046875,
      "learning_rate": 4.988979953287472e-05,
      "loss": 0.9172,
      "step": 2314
    },
    {
      "epoch": 0.0328427948991566,
      "grad_norm": 3.40625,
      "learning_rate": 4.9889589881745625e-05,
      "loss": 0.9544,
      "step": 2316
    },
    {
      "epoch": 0.03287115655278281,
      "grad_norm": 3.625,
      "learning_rate": 4.9889380031821716e-05,
      "loss": 0.9865,
      "step": 2318
    },
    {
      "epoch": 0.032899518206409024,
      "grad_norm": 3.125,
      "learning_rate": 4.988916998310467e-05,
      "loss": 0.9481,
      "step": 2320
    },
    {
      "epoch": 0.03292787986003524,
      "grad_norm": 3.1875,
      "learning_rate": 4.9888959735596165e-05,
      "loss": 0.9383,
      "step": 2322
    },
    {
      "epoch": 0.03295624151366145,
      "grad_norm": 3.796875,
      "learning_rate": 4.988874928929788e-05,
      "loss": 0.9702,
      "step": 2324
    },
    {
      "epoch": 0.03298460316728767,
      "grad_norm": 3.203125,
      "learning_rate": 4.988853864421149e-05,
      "loss": 0.9261,
      "step": 2326
    },
    {
      "epoch": 0.033012964820913884,
      "grad_norm": 3.609375,
      "learning_rate": 4.9888327800338686e-05,
      "loss": 0.9288,
      "step": 2328
    },
    {
      "epoch": 0.0330413264745401,
      "grad_norm": 3.296875,
      "learning_rate": 4.988811675768115e-05,
      "loss": 0.984,
      "step": 2330
    },
    {
      "epoch": 0.03306968812816631,
      "grad_norm": 3.03125,
      "learning_rate": 4.9887905516240555e-05,
      "loss": 0.9621,
      "step": 2332
    },
    {
      "epoch": 0.033098049781792524,
      "grad_norm": 3.125,
      "learning_rate": 4.98876940760186e-05,
      "loss": 0.9327,
      "step": 2334
    },
    {
      "epoch": 0.033126411435418744,
      "grad_norm": 3.1875,
      "learning_rate": 4.988748243701698e-05,
      "loss": 0.9889,
      "step": 2336
    },
    {
      "epoch": 0.03315477308904496,
      "grad_norm": 3.0625,
      "learning_rate": 4.988727059923738e-05,
      "loss": 0.9402,
      "step": 2338
    },
    {
      "epoch": 0.03318313474267117,
      "grad_norm": 2.921875,
      "learning_rate": 4.9887058562681484e-05,
      "loss": 1.0266,
      "step": 2340
    },
    {
      "epoch": 0.033211496396297384,
      "grad_norm": 2.921875,
      "learning_rate": 4.988684632735099e-05,
      "loss": 0.9287,
      "step": 2342
    },
    {
      "epoch": 0.033239858049923604,
      "grad_norm": 3.171875,
      "learning_rate": 4.9886633893247595e-05,
      "loss": 0.989,
      "step": 2344
    },
    {
      "epoch": 0.03326821970354982,
      "grad_norm": 3.1875,
      "learning_rate": 4.9886421260372996e-05,
      "loss": 0.9689,
      "step": 2346
    },
    {
      "epoch": 0.03329658135717603,
      "grad_norm": 3.328125,
      "learning_rate": 4.9886208428728896e-05,
      "loss": 0.9496,
      "step": 2348
    },
    {
      "epoch": 0.033324943010802244,
      "grad_norm": 2.84375,
      "learning_rate": 4.988599539831698e-05,
      "loss": 0.9549,
      "step": 2350
    },
    {
      "epoch": 0.03335330466442846,
      "grad_norm": 3.140625,
      "learning_rate": 4.988578216913896e-05,
      "loss": 1.0257,
      "step": 2352
    },
    {
      "epoch": 0.03338166631805468,
      "grad_norm": 3.0,
      "learning_rate": 4.988556874119655e-05,
      "loss": 0.9331,
      "step": 2354
    },
    {
      "epoch": 0.03341002797168089,
      "grad_norm": 3.265625,
      "learning_rate": 4.988535511449143e-05,
      "loss": 1.0137,
      "step": 2356
    },
    {
      "epoch": 0.033438389625307104,
      "grad_norm": 3.0625,
      "learning_rate": 4.988514128902532e-05,
      "loss": 0.9388,
      "step": 2358
    },
    {
      "epoch": 0.03346675127893332,
      "grad_norm": 2.875,
      "learning_rate": 4.988492726479993e-05,
      "loss": 0.9213,
      "step": 2360
    },
    {
      "epoch": 0.03349511293255953,
      "grad_norm": 3.578125,
      "learning_rate": 4.988471304181697e-05,
      "loss": 0.9582,
      "step": 2362
    },
    {
      "epoch": 0.03352347458618575,
      "grad_norm": 2.703125,
      "learning_rate": 4.988449862007814e-05,
      "loss": 0.9151,
      "step": 2364
    },
    {
      "epoch": 0.033551836239811964,
      "grad_norm": 3.171875,
      "learning_rate": 4.9884283999585166e-05,
      "loss": 0.9248,
      "step": 2366
    },
    {
      "epoch": 0.03358019789343818,
      "grad_norm": 2.90625,
      "learning_rate": 4.9884069180339746e-05,
      "loss": 0.9681,
      "step": 2368
    },
    {
      "epoch": 0.03360855954706439,
      "grad_norm": 3.015625,
      "learning_rate": 4.9883854162343615e-05,
      "loss": 0.93,
      "step": 2370
    },
    {
      "epoch": 0.0336369212006906,
      "grad_norm": 3.234375,
      "learning_rate": 4.988363894559847e-05,
      "loss": 0.9593,
      "step": 2372
    },
    {
      "epoch": 0.03366528285431682,
      "grad_norm": 3.078125,
      "learning_rate": 4.9883423530106057e-05,
      "loss": 0.9575,
      "step": 2374
    },
    {
      "epoch": 0.03369364450794304,
      "grad_norm": 3.21875,
      "learning_rate": 4.988320791586807e-05,
      "loss": 0.9991,
      "step": 2376
    },
    {
      "epoch": 0.03372200616156925,
      "grad_norm": 3.40625,
      "learning_rate": 4.988299210288625e-05,
      "loss": 0.8904,
      "step": 2378
    },
    {
      "epoch": 0.03375036781519546,
      "grad_norm": 3.078125,
      "learning_rate": 4.9882776091162304e-05,
      "loss": 0.9536,
      "step": 2380
    },
    {
      "epoch": 0.033778729468821676,
      "grad_norm": 3.09375,
      "learning_rate": 4.9882559880697964e-05,
      "loss": 0.966,
      "step": 2382
    },
    {
      "epoch": 0.033807091122447896,
      "grad_norm": 3.421875,
      "learning_rate": 4.988234347149496e-05,
      "loss": 0.9086,
      "step": 2384
    },
    {
      "epoch": 0.03383545277607411,
      "grad_norm": 3.09375,
      "learning_rate": 4.988212686355502e-05,
      "loss": 0.9716,
      "step": 2386
    },
    {
      "epoch": 0.03386381442970032,
      "grad_norm": 2.828125,
      "learning_rate": 4.988191005687988e-05,
      "loss": 0.9497,
      "step": 2388
    },
    {
      "epoch": 0.033892176083326536,
      "grad_norm": 3.328125,
      "learning_rate": 4.988169305147126e-05,
      "loss": 0.912,
      "step": 2390
    },
    {
      "epoch": 0.033920537736952756,
      "grad_norm": 2.984375,
      "learning_rate": 4.988147584733089e-05,
      "loss": 0.9219,
      "step": 2392
    },
    {
      "epoch": 0.03394889939057897,
      "grad_norm": 3.84375,
      "learning_rate": 4.988125844446053e-05,
      "loss": 0.8951,
      "step": 2394
    },
    {
      "epoch": 0.03397726104420518,
      "grad_norm": 3.53125,
      "learning_rate": 4.988104084286189e-05,
      "loss": 1.0228,
      "step": 2396
    },
    {
      "epoch": 0.034005622697831396,
      "grad_norm": 3.328125,
      "learning_rate": 4.9880823042536715e-05,
      "loss": 0.9689,
      "step": 2398
    },
    {
      "epoch": 0.03403398435145761,
      "grad_norm": 3.203125,
      "learning_rate": 4.988060504348675e-05,
      "loss": 0.9503,
      "step": 2400
    },
    {
      "epoch": 0.03406234600508383,
      "grad_norm": 3.03125,
      "learning_rate": 4.988038684571373e-05,
      "loss": 1.0261,
      "step": 2402
    },
    {
      "epoch": 0.03409070765871004,
      "grad_norm": 2.890625,
      "learning_rate": 4.98801684492194e-05,
      "loss": 0.9457,
      "step": 2404
    },
    {
      "epoch": 0.034119069312336256,
      "grad_norm": 2.8125,
      "learning_rate": 4.987994985400551e-05,
      "loss": 0.9852,
      "step": 2406
    },
    {
      "epoch": 0.03414743096596247,
      "grad_norm": 3.171875,
      "learning_rate": 4.98797310600738e-05,
      "loss": 0.9465,
      "step": 2408
    },
    {
      "epoch": 0.03417579261958868,
      "grad_norm": 3.125,
      "learning_rate": 4.987951206742601e-05,
      "loss": 0.979,
      "step": 2410
    },
    {
      "epoch": 0.0342041542732149,
      "grad_norm": 3.125,
      "learning_rate": 4.987929287606391e-05,
      "loss": 0.9343,
      "step": 2412
    },
    {
      "epoch": 0.034232515926841116,
      "grad_norm": 3.171875,
      "learning_rate": 4.9879073485989234e-05,
      "loss": 0.9189,
      "step": 2414
    },
    {
      "epoch": 0.03426087758046733,
      "grad_norm": 3.140625,
      "learning_rate": 4.987885389720374e-05,
      "loss": 0.937,
      "step": 2416
    },
    {
      "epoch": 0.03428923923409354,
      "grad_norm": 3.015625,
      "learning_rate": 4.987863410970917e-05,
      "loss": 0.9374,
      "step": 2418
    },
    {
      "epoch": 0.034317600887719756,
      "grad_norm": 3.203125,
      "learning_rate": 4.98784141235073e-05,
      "loss": 0.9331,
      "step": 2420
    },
    {
      "epoch": 0.034345962541345976,
      "grad_norm": 2.96875,
      "learning_rate": 4.987819393859987e-05,
      "loss": 0.9507,
      "step": 2422
    },
    {
      "epoch": 0.03437432419497219,
      "grad_norm": 3.015625,
      "learning_rate": 4.987797355498866e-05,
      "loss": 0.957,
      "step": 2424
    },
    {
      "epoch": 0.0344026858485984,
      "grad_norm": 3.25,
      "learning_rate": 4.98777529726754e-05,
      "loss": 0.9512,
      "step": 2426
    },
    {
      "epoch": 0.034431047502224615,
      "grad_norm": 3.34375,
      "learning_rate": 4.987753219166188e-05,
      "loss": 0.9439,
      "step": 2428
    },
    {
      "epoch": 0.03445940915585083,
      "grad_norm": 3.25,
      "learning_rate": 4.987731121194984e-05,
      "loss": 0.9095,
      "step": 2430
    },
    {
      "epoch": 0.03448777080947705,
      "grad_norm": 3.15625,
      "learning_rate": 4.9877090033541065e-05,
      "loss": 0.9493,
      "step": 2432
    },
    {
      "epoch": 0.03451613246310326,
      "grad_norm": 3.078125,
      "learning_rate": 4.98768686564373e-05,
      "loss": 0.9257,
      "step": 2434
    },
    {
      "epoch": 0.034544494116729475,
      "grad_norm": 3.203125,
      "learning_rate": 4.987664708064034e-05,
      "loss": 0.9929,
      "step": 2436
    },
    {
      "epoch": 0.03457285577035569,
      "grad_norm": 2.953125,
      "learning_rate": 4.987642530615193e-05,
      "loss": 0.9026,
      "step": 2438
    },
    {
      "epoch": 0.03460121742398191,
      "grad_norm": 3.375,
      "learning_rate": 4.987620333297386e-05,
      "loss": 0.9847,
      "step": 2440
    },
    {
      "epoch": 0.03462957907760812,
      "grad_norm": 3.078125,
      "learning_rate": 4.9875981161107885e-05,
      "loss": 1.0044,
      "step": 2442
    },
    {
      "epoch": 0.034657940731234335,
      "grad_norm": 3.03125,
      "learning_rate": 4.9875758790555805e-05,
      "loss": 0.9583,
      "step": 2444
    },
    {
      "epoch": 0.03468630238486055,
      "grad_norm": 3.015625,
      "learning_rate": 4.987553622131936e-05,
      "loss": 0.9715,
      "step": 2446
    },
    {
      "epoch": 0.03471466403848676,
      "grad_norm": 3.140625,
      "learning_rate": 4.987531345340036e-05,
      "loss": 0.9161,
      "step": 2448
    },
    {
      "epoch": 0.03474302569211298,
      "grad_norm": 3.375,
      "learning_rate": 4.987509048680058e-05,
      "loss": 0.9912,
      "step": 2450
    },
    {
      "epoch": 0.034771387345739195,
      "grad_norm": 2.921875,
      "learning_rate": 4.9874867321521776e-05,
      "loss": 0.9411,
      "step": 2452
    },
    {
      "epoch": 0.03479974899936541,
      "grad_norm": 3.328125,
      "learning_rate": 4.9874643957565756e-05,
      "loss": 0.9868,
      "step": 2454
    },
    {
      "epoch": 0.03482811065299162,
      "grad_norm": 3.28125,
      "learning_rate": 4.9874420394934295e-05,
      "loss": 0.9404,
      "step": 2456
    },
    {
      "epoch": 0.034856472306617835,
      "grad_norm": 3.0,
      "learning_rate": 4.987419663362918e-05,
      "loss": 0.9341,
      "step": 2458
    },
    {
      "epoch": 0.034884833960244055,
      "grad_norm": 2.953125,
      "learning_rate": 4.98739726736522e-05,
      "loss": 0.9619,
      "step": 2460
    },
    {
      "epoch": 0.03491319561387027,
      "grad_norm": 3.46875,
      "learning_rate": 4.9873748515005134e-05,
      "loss": 0.9452,
      "step": 2462
    },
    {
      "epoch": 0.03494155726749648,
      "grad_norm": 3.21875,
      "learning_rate": 4.987352415768979e-05,
      "loss": 0.9458,
      "step": 2464
    },
    {
      "epoch": 0.034969918921122695,
      "grad_norm": 3.453125,
      "learning_rate": 4.987329960170794e-05,
      "loss": 0.9616,
      "step": 2466
    },
    {
      "epoch": 0.03499828057474891,
      "grad_norm": 3.140625,
      "learning_rate": 4.987307484706139e-05,
      "loss": 0.9194,
      "step": 2468
    },
    {
      "epoch": 0.03502664222837513,
      "grad_norm": 3.03125,
      "learning_rate": 4.987284989375193e-05,
      "loss": 0.9696,
      "step": 2470
    },
    {
      "epoch": 0.03505500388200134,
      "grad_norm": 3.125,
      "learning_rate": 4.987262474178136e-05,
      "loss": 0.9721,
      "step": 2472
    },
    {
      "epoch": 0.035083365535627554,
      "grad_norm": 3.125,
      "learning_rate": 4.987239939115148e-05,
      "loss": 1.0022,
      "step": 2474
    },
    {
      "epoch": 0.03511172718925377,
      "grad_norm": 3.0625,
      "learning_rate": 4.987217384186409e-05,
      "loss": 0.9376,
      "step": 2476
    },
    {
      "epoch": 0.03514008884287998,
      "grad_norm": 2.9375,
      "learning_rate": 4.987194809392098e-05,
      "loss": 0.962,
      "step": 2478
    },
    {
      "epoch": 0.0351684504965062,
      "grad_norm": 3.015625,
      "learning_rate": 4.9871722147323965e-05,
      "loss": 0.9192,
      "step": 2480
    },
    {
      "epoch": 0.035196812150132414,
      "grad_norm": 3.515625,
      "learning_rate": 4.9871496002074845e-05,
      "loss": 0.9698,
      "step": 2482
    },
    {
      "epoch": 0.03522517380375863,
      "grad_norm": 3.0,
      "learning_rate": 4.987126965817543e-05,
      "loss": 0.9654,
      "step": 2484
    },
    {
      "epoch": 0.03525353545738484,
      "grad_norm": 2.828125,
      "learning_rate": 4.987104311562753e-05,
      "loss": 0.9292,
      "step": 2486
    },
    {
      "epoch": 0.035281897111011054,
      "grad_norm": 3.078125,
      "learning_rate": 4.987081637443294e-05,
      "loss": 0.9167,
      "step": 2488
    },
    {
      "epoch": 0.035310258764637274,
      "grad_norm": 3.265625,
      "learning_rate": 4.987058943459348e-05,
      "loss": 0.9359,
      "step": 2490
    },
    {
      "epoch": 0.03533862041826349,
      "grad_norm": 3.484375,
      "learning_rate": 4.9870362296110974e-05,
      "loss": 0.9838,
      "step": 2492
    },
    {
      "epoch": 0.0353669820718897,
      "grad_norm": 3.1875,
      "learning_rate": 4.987013495898721e-05,
      "loss": 0.9883,
      "step": 2494
    },
    {
      "epoch": 0.035395343725515914,
      "grad_norm": 3.28125,
      "learning_rate": 4.986990742322404e-05,
      "loss": 0.9159,
      "step": 2496
    },
    {
      "epoch": 0.035423705379142134,
      "grad_norm": 2.640625,
      "learning_rate": 4.986967968882324e-05,
      "loss": 0.957,
      "step": 2498
    },
    {
      "epoch": 0.03545206703276835,
      "grad_norm": 2.984375,
      "learning_rate": 4.986945175578667e-05,
      "loss": 0.9826,
      "step": 2500
    },
    {
      "epoch": 0.03548042868639456,
      "grad_norm": 3.1875,
      "learning_rate": 4.986922362411611e-05,
      "loss": 0.9155,
      "step": 2502
    },
    {
      "epoch": 0.035508790340020774,
      "grad_norm": 2.953125,
      "learning_rate": 4.9868995293813415e-05,
      "loss": 0.9382,
      "step": 2504
    },
    {
      "epoch": 0.03553715199364699,
      "grad_norm": 2.609375,
      "learning_rate": 4.9868766764880394e-05,
      "loss": 0.8946,
      "step": 2506
    },
    {
      "epoch": 0.03556551364727321,
      "grad_norm": 3.0,
      "learning_rate": 4.9868538037318865e-05,
      "loss": 0.9481,
      "step": 2508
    },
    {
      "epoch": 0.03559387530089942,
      "grad_norm": 3.265625,
      "learning_rate": 4.986830911113067e-05,
      "loss": 0.9466,
      "step": 2510
    },
    {
      "epoch": 0.035622236954525634,
      "grad_norm": 3.375,
      "learning_rate": 4.986807998631764e-05,
      "loss": 0.9481,
      "step": 2512
    },
    {
      "epoch": 0.03565059860815185,
      "grad_norm": 2.90625,
      "learning_rate": 4.986785066288159e-05,
      "loss": 0.9291,
      "step": 2514
    },
    {
      "epoch": 0.03567896026177806,
      "grad_norm": 3.15625,
      "learning_rate": 4.986762114082435e-05,
      "loss": 0.9184,
      "step": 2516
    },
    {
      "epoch": 0.03570732191540428,
      "grad_norm": 3.296875,
      "learning_rate": 4.986739142014777e-05,
      "loss": 0.9388,
      "step": 2518
    },
    {
      "epoch": 0.035735683569030494,
      "grad_norm": 2.875,
      "learning_rate": 4.9867161500853674e-05,
      "loss": 0.937,
      "step": 2520
    },
    {
      "epoch": 0.03576404522265671,
      "grad_norm": 3.0,
      "learning_rate": 4.98669313829439e-05,
      "loss": 0.9204,
      "step": 2522
    },
    {
      "epoch": 0.03579240687628292,
      "grad_norm": 3.03125,
      "learning_rate": 4.986670106642028e-05,
      "loss": 0.9359,
      "step": 2524
    },
    {
      "epoch": 0.03582076852990913,
      "grad_norm": 3.40625,
      "learning_rate": 4.986647055128467e-05,
      "loss": 0.9309,
      "step": 2526
    },
    {
      "epoch": 0.03584913018353535,
      "grad_norm": 3.21875,
      "learning_rate": 4.9866239837538894e-05,
      "loss": 0.9727,
      "step": 2528
    },
    {
      "epoch": 0.03587749183716157,
      "grad_norm": 3.265625,
      "learning_rate": 4.986600892518482e-05,
      "loss": 1.0178,
      "step": 2530
    },
    {
      "epoch": 0.03590585349078778,
      "grad_norm": 3.453125,
      "learning_rate": 4.9865777814224254e-05,
      "loss": 0.9966,
      "step": 2532
    },
    {
      "epoch": 0.03593421514441399,
      "grad_norm": 3.34375,
      "learning_rate": 4.9865546504659063e-05,
      "loss": 0.9467,
      "step": 2534
    },
    {
      "epoch": 0.035962576798040206,
      "grad_norm": 3.328125,
      "learning_rate": 4.98653149964911e-05,
      "loss": 1.0089,
      "step": 2536
    },
    {
      "epoch": 0.03599093845166643,
      "grad_norm": 3.375,
      "learning_rate": 4.986508328972221e-05,
      "loss": 0.9364,
      "step": 2538
    },
    {
      "epoch": 0.03601930010529264,
      "grad_norm": 3.359375,
      "learning_rate": 4.986485138435424e-05,
      "loss": 0.9394,
      "step": 2540
    },
    {
      "epoch": 0.03604766175891885,
      "grad_norm": 2.765625,
      "learning_rate": 4.9864619280389036e-05,
      "loss": 0.9471,
      "step": 2542
    },
    {
      "epoch": 0.036076023412545066,
      "grad_norm": 3.65625,
      "learning_rate": 4.9864386977828466e-05,
      "loss": 0.9119,
      "step": 2544
    },
    {
      "epoch": 0.036104385066171286,
      "grad_norm": 3.015625,
      "learning_rate": 4.9864154476674374e-05,
      "loss": 0.8974,
      "step": 2546
    },
    {
      "epoch": 0.0361327467197975,
      "grad_norm": 2.90625,
      "learning_rate": 4.9863921776928624e-05,
      "loss": 0.9453,
      "step": 2548
    },
    {
      "epoch": 0.03616110837342371,
      "grad_norm": 3.28125,
      "learning_rate": 4.986368887859307e-05,
      "loss": 0.9797,
      "step": 2550
    },
    {
      "epoch": 0.036189470027049926,
      "grad_norm": 3.203125,
      "learning_rate": 4.986345578166958e-05,
      "loss": 0.9929,
      "step": 2552
    },
    {
      "epoch": 0.03621783168067614,
      "grad_norm": 3.3125,
      "learning_rate": 4.986322248616001e-05,
      "loss": 0.9148,
      "step": 2554
    },
    {
      "epoch": 0.03624619333430236,
      "grad_norm": 3.109375,
      "learning_rate": 4.986298899206622e-05,
      "loss": 0.9413,
      "step": 2556
    },
    {
      "epoch": 0.03627455498792857,
      "grad_norm": 3.046875,
      "learning_rate": 4.986275529939008e-05,
      "loss": 0.9648,
      "step": 2558
    },
    {
      "epoch": 0.036302916641554786,
      "grad_norm": 2.984375,
      "learning_rate": 4.986252140813346e-05,
      "loss": 0.9752,
      "step": 2560
    },
    {
      "epoch": 0.036331278295181,
      "grad_norm": 3.234375,
      "learning_rate": 4.9862287318298213e-05,
      "loss": 0.9226,
      "step": 2562
    },
    {
      "epoch": 0.03635963994880721,
      "grad_norm": 3.21875,
      "learning_rate": 4.9862053029886226e-05,
      "loss": 0.9403,
      "step": 2564
    },
    {
      "epoch": 0.03638800160243343,
      "grad_norm": 3.15625,
      "learning_rate": 4.986181854289936e-05,
      "loss": 0.994,
      "step": 2566
    },
    {
      "epoch": 0.036416363256059646,
      "grad_norm": 3.1875,
      "learning_rate": 4.9861583857339486e-05,
      "loss": 0.9743,
      "step": 2568
    },
    {
      "epoch": 0.03644472490968586,
      "grad_norm": 2.546875,
      "learning_rate": 4.9861348973208494e-05,
      "loss": 0.9341,
      "step": 2570
    },
    {
      "epoch": 0.03647308656331207,
      "grad_norm": 2.984375,
      "learning_rate": 4.9861113890508246e-05,
      "loss": 0.9457,
      "step": 2572
    },
    {
      "epoch": 0.036501448216938286,
      "grad_norm": 3.625,
      "learning_rate": 4.9860878609240626e-05,
      "loss": 1.0079,
      "step": 2574
    },
    {
      "epoch": 0.036529809870564506,
      "grad_norm": 3.765625,
      "learning_rate": 4.9860643129407504e-05,
      "loss": 0.9638,
      "step": 2576
    },
    {
      "epoch": 0.03655817152419072,
      "grad_norm": 3.015625,
      "learning_rate": 4.986040745101077e-05,
      "loss": 0.9799,
      "step": 2578
    },
    {
      "epoch": 0.03658653317781693,
      "grad_norm": 2.875,
      "learning_rate": 4.98601715740523e-05,
      "loss": 0.9228,
      "step": 2580
    },
    {
      "epoch": 0.036614894831443145,
      "grad_norm": 2.90625,
      "learning_rate": 4.985993549853398e-05,
      "loss": 0.9615,
      "step": 2582
    },
    {
      "epoch": 0.03664325648506936,
      "grad_norm": 3.234375,
      "learning_rate": 4.9859699224457704e-05,
      "loss": 0.9349,
      "step": 2584
    },
    {
      "epoch": 0.03667161813869558,
      "grad_norm": 3.21875,
      "learning_rate": 4.985946275182535e-05,
      "loss": 0.9544,
      "step": 2586
    },
    {
      "epoch": 0.03669997979232179,
      "grad_norm": 3.046875,
      "learning_rate": 4.985922608063881e-05,
      "loss": 0.9708,
      "step": 2588
    },
    {
      "epoch": 0.036728341445948005,
      "grad_norm": 3.296875,
      "learning_rate": 4.985898921089997e-05,
      "loss": 0.9114,
      "step": 2590
    },
    {
      "epoch": 0.03675670309957422,
      "grad_norm": 3.625,
      "learning_rate": 4.985875214261073e-05,
      "loss": 0.9517,
      "step": 2592
    },
    {
      "epoch": 0.03678506475320044,
      "grad_norm": 3.171875,
      "learning_rate": 4.9858514875772974e-05,
      "loss": 0.9283,
      "step": 2594
    },
    {
      "epoch": 0.03681342640682665,
      "grad_norm": 3.109375,
      "learning_rate": 4.98582774103886e-05,
      "loss": 0.9312,
      "step": 2596
    },
    {
      "epoch": 0.036841788060452865,
      "grad_norm": 3.078125,
      "learning_rate": 4.985803974645951e-05,
      "loss": 0.8272,
      "step": 2598
    },
    {
      "epoch": 0.03687014971407908,
      "grad_norm": 2.96875,
      "learning_rate": 4.98578018839876e-05,
      "loss": 0.8745,
      "step": 2600
    },
    {
      "epoch": 0.03689851136770529,
      "grad_norm": 2.921875,
      "learning_rate": 4.985756382297476e-05,
      "loss": 0.9505,
      "step": 2602
    },
    {
      "epoch": 0.03692687302133151,
      "grad_norm": 3.765625,
      "learning_rate": 4.985732556342291e-05,
      "loss": 0.9357,
      "step": 2604
    },
    {
      "epoch": 0.036955234674957725,
      "grad_norm": 3.28125,
      "learning_rate": 4.985708710533394e-05,
      "loss": 0.9069,
      "step": 2606
    },
    {
      "epoch": 0.03698359632858394,
      "grad_norm": 3.15625,
      "learning_rate": 4.985684844870975e-05,
      "loss": 0.9857,
      "step": 2608
    },
    {
      "epoch": 0.03701195798221015,
      "grad_norm": 3.203125,
      "learning_rate": 4.9856609593552264e-05,
      "loss": 0.9657,
      "step": 2610
    },
    {
      "epoch": 0.037040319635836365,
      "grad_norm": 3.140625,
      "learning_rate": 4.985637053986337e-05,
      "loss": 0.9611,
      "step": 2612
    },
    {
      "epoch": 0.037068681289462585,
      "grad_norm": 3.25,
      "learning_rate": 4.985613128764499e-05,
      "loss": 0.9181,
      "step": 2614
    },
    {
      "epoch": 0.0370970429430888,
      "grad_norm": 3.25,
      "learning_rate": 4.985589183689904e-05,
      "loss": 0.9769,
      "step": 2616
    },
    {
      "epoch": 0.03712540459671501,
      "grad_norm": 2.796875,
      "learning_rate": 4.985565218762741e-05,
      "loss": 0.9369,
      "step": 2618
    },
    {
      "epoch": 0.037153766250341225,
      "grad_norm": 3.203125,
      "learning_rate": 4.9855412339832034e-05,
      "loss": 1.0013,
      "step": 2620
    },
    {
      "epoch": 0.03718212790396744,
      "grad_norm": 3.234375,
      "learning_rate": 4.9855172293514826e-05,
      "loss": 0.9634,
      "step": 2622
    },
    {
      "epoch": 0.03721048955759366,
      "grad_norm": 2.9375,
      "learning_rate": 4.985493204867769e-05,
      "loss": 0.8902,
      "step": 2624
    },
    {
      "epoch": 0.03723885121121987,
      "grad_norm": 3.5,
      "learning_rate": 4.9854691605322566e-05,
      "loss": 0.9198,
      "step": 2626
    },
    {
      "epoch": 0.037267212864846085,
      "grad_norm": 3.078125,
      "learning_rate": 4.985445096345135e-05,
      "loss": 0.8597,
      "step": 2628
    },
    {
      "epoch": 0.0372955745184723,
      "grad_norm": 2.96875,
      "learning_rate": 4.985421012306598e-05,
      "loss": 0.8926,
      "step": 2630
    },
    {
      "epoch": 0.03732393617209851,
      "grad_norm": 3.109375,
      "learning_rate": 4.9853969084168386e-05,
      "loss": 0.8964,
      "step": 2632
    },
    {
      "epoch": 0.03735229782572473,
      "grad_norm": 2.9375,
      "learning_rate": 4.985372784676047e-05,
      "loss": 0.9236,
      "step": 2634
    },
    {
      "epoch": 0.037380659479350944,
      "grad_norm": 3.453125,
      "learning_rate": 4.985348641084418e-05,
      "loss": 0.9352,
      "step": 2636
    },
    {
      "epoch": 0.03740902113297716,
      "grad_norm": 3.171875,
      "learning_rate": 4.9853244776421434e-05,
      "loss": 0.9789,
      "step": 2638
    },
    {
      "epoch": 0.03743738278660337,
      "grad_norm": 3.0625,
      "learning_rate": 4.985300294349417e-05,
      "loss": 0.961,
      "step": 2640
    },
    {
      "epoch": 0.037465744440229584,
      "grad_norm": 2.796875,
      "learning_rate": 4.98527609120643e-05,
      "loss": 0.9507,
      "step": 2642
    },
    {
      "epoch": 0.037494106093855804,
      "grad_norm": 3.421875,
      "learning_rate": 4.985251868213377e-05,
      "loss": 0.9313,
      "step": 2644
    },
    {
      "epoch": 0.03752246774748202,
      "grad_norm": 3.1875,
      "learning_rate": 4.985227625370453e-05,
      "loss": 0.9657,
      "step": 2646
    },
    {
      "epoch": 0.03755082940110823,
      "grad_norm": 2.796875,
      "learning_rate": 4.9852033626778496e-05,
      "loss": 0.926,
      "step": 2648
    },
    {
      "epoch": 0.037579191054734444,
      "grad_norm": 3.15625,
      "learning_rate": 4.985179080135761e-05,
      "loss": 0.9057,
      "step": 2650
    },
    {
      "epoch": 0.037607552708360664,
      "grad_norm": 3.34375,
      "learning_rate": 4.985154777744382e-05,
      "loss": 0.9325,
      "step": 2652
    },
    {
      "epoch": 0.03763591436198688,
      "grad_norm": 3.0625,
      "learning_rate": 4.9851304555039046e-05,
      "loss": 0.9544,
      "step": 2654
    },
    {
      "epoch": 0.03766427601561309,
      "grad_norm": 3.21875,
      "learning_rate": 4.985106113414526e-05,
      "loss": 0.9114,
      "step": 2656
    },
    {
      "epoch": 0.037692637669239304,
      "grad_norm": 4.21875,
      "learning_rate": 4.985081751476438e-05,
      "loss": 0.9855,
      "step": 2658
    },
    {
      "epoch": 0.03772099932286552,
      "grad_norm": 3.6875,
      "learning_rate": 4.985057369689837e-05,
      "loss": 0.9693,
      "step": 2660
    },
    {
      "epoch": 0.03774936097649174,
      "grad_norm": 2.90625,
      "learning_rate": 4.985032968054917e-05,
      "loss": 0.9405,
      "step": 2662
    },
    {
      "epoch": 0.03777772263011795,
      "grad_norm": 4.0625,
      "learning_rate": 4.985008546571873e-05,
      "loss": 0.9643,
      "step": 2664
    },
    {
      "epoch": 0.037806084283744164,
      "grad_norm": 2.796875,
      "learning_rate": 4.9849841052408995e-05,
      "loss": 0.9336,
      "step": 2666
    },
    {
      "epoch": 0.03783444593737038,
      "grad_norm": 3.015625,
      "learning_rate": 4.9849596440621925e-05,
      "loss": 0.9628,
      "step": 2668
    },
    {
      "epoch": 0.03786280759099659,
      "grad_norm": 3.109375,
      "learning_rate": 4.984935163035946e-05,
      "loss": 1.0044,
      "step": 2670
    },
    {
      "epoch": 0.03789116924462281,
      "grad_norm": 3.3125,
      "learning_rate": 4.984910662162359e-05,
      "loss": 0.9322,
      "step": 2672
    },
    {
      "epoch": 0.037919530898249024,
      "grad_norm": 3.015625,
      "learning_rate": 4.9848861414416226e-05,
      "loss": 0.9411,
      "step": 2674
    },
    {
      "epoch": 0.03794789255187524,
      "grad_norm": 3.34375,
      "learning_rate": 4.984861600873936e-05,
      "loss": 0.9837,
      "step": 2676
    },
    {
      "epoch": 0.03797625420550145,
      "grad_norm": 2.90625,
      "learning_rate": 4.9848370404594937e-05,
      "loss": 0.9709,
      "step": 2678
    },
    {
      "epoch": 0.03800461585912766,
      "grad_norm": 2.953125,
      "learning_rate": 4.984812460198493e-05,
      "loss": 0.9074,
      "step": 2680
    },
    {
      "epoch": 0.038032977512753884,
      "grad_norm": 3.109375,
      "learning_rate": 4.984787860091128e-05,
      "loss": 0.9626,
      "step": 2682
    },
    {
      "epoch": 0.0380613391663801,
      "grad_norm": 3.203125,
      "learning_rate": 4.984763240137597e-05,
      "loss": 0.9058,
      "step": 2684
    },
    {
      "epoch": 0.03808970082000631,
      "grad_norm": 3.640625,
      "learning_rate": 4.984738600338097e-05,
      "loss": 0.9681,
      "step": 2686
    },
    {
      "epoch": 0.03811806247363252,
      "grad_norm": 3.15625,
      "learning_rate": 4.9847139406928234e-05,
      "loss": 0.9472,
      "step": 2688
    },
    {
      "epoch": 0.038146424127258736,
      "grad_norm": 3.171875,
      "learning_rate": 4.984689261201974e-05,
      "loss": 0.9497,
      "step": 2690
    },
    {
      "epoch": 0.03817478578088496,
      "grad_norm": 3.15625,
      "learning_rate": 4.9846645618657463e-05,
      "loss": 0.8817,
      "step": 2692
    },
    {
      "epoch": 0.03820314743451117,
      "grad_norm": 3.078125,
      "learning_rate": 4.984639842684336e-05,
      "loss": 0.9618,
      "step": 2694
    },
    {
      "epoch": 0.03823150908813738,
      "grad_norm": 3.078125,
      "learning_rate": 4.984615103657942e-05,
      "loss": 0.9814,
      "step": 2696
    },
    {
      "epoch": 0.038259870741763596,
      "grad_norm": 3.40625,
      "learning_rate": 4.984590344786761e-05,
      "loss": 0.9685,
      "step": 2698
    },
    {
      "epoch": 0.03828823239538982,
      "grad_norm": 3.25,
      "learning_rate": 4.9845655660709924e-05,
      "loss": 0.8749,
      "step": 2700
    },
    {
      "epoch": 0.03831659404901603,
      "grad_norm": 3.296875,
      "learning_rate": 4.984540767510833e-05,
      "loss": 0.9677,
      "step": 2702
    },
    {
      "epoch": 0.03834495570264224,
      "grad_norm": 2.90625,
      "learning_rate": 4.9845159491064786e-05,
      "loss": 0.8953,
      "step": 2704
    },
    {
      "epoch": 0.038373317356268456,
      "grad_norm": 3.1875,
      "learning_rate": 4.9844911108581314e-05,
      "loss": 0.9207,
      "step": 2706
    },
    {
      "epoch": 0.03840167900989467,
      "grad_norm": 2.71875,
      "learning_rate": 4.9844662527659876e-05,
      "loss": 0.8936,
      "step": 2708
    },
    {
      "epoch": 0.03843004066352089,
      "grad_norm": 3.078125,
      "learning_rate": 4.984441374830246e-05,
      "loss": 0.9655,
      "step": 2710
    },
    {
      "epoch": 0.0384584023171471,
      "grad_norm": 3.265625,
      "learning_rate": 4.984416477051106e-05,
      "loss": 0.9322,
      "step": 2712
    },
    {
      "epoch": 0.038486763970773316,
      "grad_norm": 2.828125,
      "learning_rate": 4.984391559428765e-05,
      "loss": 0.9132,
      "step": 2714
    },
    {
      "epoch": 0.03851512562439953,
      "grad_norm": 3.53125,
      "learning_rate": 4.984366621963423e-05,
      "loss": 0.9495,
      "step": 2716
    },
    {
      "epoch": 0.03854348727802574,
      "grad_norm": 3.40625,
      "learning_rate": 4.98434166465528e-05,
      "loss": 0.9573,
      "step": 2718
    },
    {
      "epoch": 0.03857184893165196,
      "grad_norm": 3.015625,
      "learning_rate": 4.9843166875045334e-05,
      "loss": 0.9406,
      "step": 2720
    },
    {
      "epoch": 0.038600210585278176,
      "grad_norm": 2.9375,
      "learning_rate": 4.984291690511384e-05,
      "loss": 0.9136,
      "step": 2722
    },
    {
      "epoch": 0.03862857223890439,
      "grad_norm": 3.546875,
      "learning_rate": 4.984266673676031e-05,
      "loss": 0.9374,
      "step": 2724
    },
    {
      "epoch": 0.0386569338925306,
      "grad_norm": 3.140625,
      "learning_rate": 4.984241636998674e-05,
      "loss": 0.9429,
      "step": 2726
    },
    {
      "epoch": 0.038685295546156816,
      "grad_norm": 3.328125,
      "learning_rate": 4.984216580479515e-05,
      "loss": 0.9108,
      "step": 2728
    },
    {
      "epoch": 0.038713657199783036,
      "grad_norm": 3.359375,
      "learning_rate": 4.9841915041187515e-05,
      "loss": 0.9416,
      "step": 2730
    },
    {
      "epoch": 0.03874201885340925,
      "grad_norm": 3.09375,
      "learning_rate": 4.984166407916584e-05,
      "loss": 0.9223,
      "step": 2732
    },
    {
      "epoch": 0.03877038050703546,
      "grad_norm": 2.640625,
      "learning_rate": 4.9841412918732145e-05,
      "loss": 0.9658,
      "step": 2734
    },
    {
      "epoch": 0.038798742160661676,
      "grad_norm": 3.453125,
      "learning_rate": 4.9841161559888426e-05,
      "loss": 1.0167,
      "step": 2736
    },
    {
      "epoch": 0.03882710381428789,
      "grad_norm": 3.328125,
      "learning_rate": 4.98409100026367e-05,
      "loss": 0.9754,
      "step": 2738
    },
    {
      "epoch": 0.03885546546791411,
      "grad_norm": 2.890625,
      "learning_rate": 4.984065824697896e-05,
      "loss": 0.886,
      "step": 2740
    },
    {
      "epoch": 0.03888382712154032,
      "grad_norm": 3.1875,
      "learning_rate": 4.984040629291723e-05,
      "loss": 0.9702,
      "step": 2742
    },
    {
      "epoch": 0.038912188775166535,
      "grad_norm": 2.859375,
      "learning_rate": 4.984015414045352e-05,
      "loss": 0.8762,
      "step": 2744
    },
    {
      "epoch": 0.03894055042879275,
      "grad_norm": 3.53125,
      "learning_rate": 4.9839901789589836e-05,
      "loss": 0.9153,
      "step": 2746
    },
    {
      "epoch": 0.03896891208241897,
      "grad_norm": 2.96875,
      "learning_rate": 4.98396492403282e-05,
      "loss": 0.895,
      "step": 2748
    },
    {
      "epoch": 0.03899727373604518,
      "grad_norm": 3.28125,
      "learning_rate": 4.9839396492670634e-05,
      "loss": 0.9373,
      "step": 2750
    },
    {
      "epoch": 0.039025635389671395,
      "grad_norm": 3.515625,
      "learning_rate": 4.9839143546619146e-05,
      "loss": 0.9833,
      "step": 2752
    },
    {
      "epoch": 0.03905399704329761,
      "grad_norm": 3.109375,
      "learning_rate": 4.9838890402175764e-05,
      "loss": 0.9074,
      "step": 2754
    },
    {
      "epoch": 0.03908235869692382,
      "grad_norm": 3.03125,
      "learning_rate": 4.983863705934251e-05,
      "loss": 0.9346,
      "step": 2756
    },
    {
      "epoch": 0.03911072035055004,
      "grad_norm": 2.9375,
      "learning_rate": 4.98383835181214e-05,
      "loss": 0.933,
      "step": 2758
    },
    {
      "epoch": 0.039139082004176255,
      "grad_norm": 3.09375,
      "learning_rate": 4.9838129778514466e-05,
      "loss": 0.9368,
      "step": 2760
    },
    {
      "epoch": 0.03916744365780247,
      "grad_norm": 2.984375,
      "learning_rate": 4.9837875840523734e-05,
      "loss": 0.9744,
      "step": 2762
    },
    {
      "epoch": 0.03919580531142868,
      "grad_norm": 3.203125,
      "learning_rate": 4.9837621704151224e-05,
      "loss": 0.9635,
      "step": 2764
    },
    {
      "epoch": 0.039224166965054895,
      "grad_norm": 3.140625,
      "learning_rate": 4.9837367369398984e-05,
      "loss": 1.0056,
      "step": 2766
    },
    {
      "epoch": 0.039252528618681115,
      "grad_norm": 2.921875,
      "learning_rate": 4.9837112836269026e-05,
      "loss": 0.9289,
      "step": 2768
    },
    {
      "epoch": 0.03928089027230733,
      "grad_norm": 3.046875,
      "learning_rate": 4.9836858104763395e-05,
      "loss": 0.9339,
      "step": 2770
    },
    {
      "epoch": 0.03930925192593354,
      "grad_norm": 2.84375,
      "learning_rate": 4.983660317488412e-05,
      "loss": 0.9718,
      "step": 2772
    },
    {
      "epoch": 0.039337613579559755,
      "grad_norm": 2.6875,
      "learning_rate": 4.9836348046633244e-05,
      "loss": 0.9478,
      "step": 2774
    },
    {
      "epoch": 0.03936597523318597,
      "grad_norm": 3.25,
      "learning_rate": 4.9836092720012795e-05,
      "loss": 0.9915,
      "step": 2776
    },
    {
      "epoch": 0.03939433688681219,
      "grad_norm": 3.125,
      "learning_rate": 4.983583719502482e-05,
      "loss": 0.8977,
      "step": 2778
    },
    {
      "epoch": 0.0394226985404384,
      "grad_norm": 3.265625,
      "learning_rate": 4.9835581471671356e-05,
      "loss": 0.979,
      "step": 2780
    },
    {
      "epoch": 0.039451060194064615,
      "grad_norm": 2.8125,
      "learning_rate": 4.9835325549954446e-05,
      "loss": 0.9441,
      "step": 2782
    },
    {
      "epoch": 0.03947942184769083,
      "grad_norm": 2.9375,
      "learning_rate": 4.983506942987613e-05,
      "loss": 0.9377,
      "step": 2784
    },
    {
      "epoch": 0.03950778350131704,
      "grad_norm": 3.375,
      "learning_rate": 4.983481311143846e-05,
      "loss": 0.9055,
      "step": 2786
    },
    {
      "epoch": 0.03953614515494326,
      "grad_norm": 4.125,
      "learning_rate": 4.983455659464348e-05,
      "loss": 0.9925,
      "step": 2788
    },
    {
      "epoch": 0.039564506808569475,
      "grad_norm": 3.03125,
      "learning_rate": 4.983429987949324e-05,
      "loss": 0.9112,
      "step": 2790
    },
    {
      "epoch": 0.03959286846219569,
      "grad_norm": 2.765625,
      "learning_rate": 4.983404296598979e-05,
      "loss": 0.9294,
      "step": 2792
    },
    {
      "epoch": 0.0396212301158219,
      "grad_norm": 3.234375,
      "learning_rate": 4.983378585413519e-05,
      "loss": 0.9394,
      "step": 2794
    },
    {
      "epoch": 0.039649591769448114,
      "grad_norm": 3.234375,
      "learning_rate": 4.983352854393148e-05,
      "loss": 0.9037,
      "step": 2796
    },
    {
      "epoch": 0.039677953423074334,
      "grad_norm": 3.0,
      "learning_rate": 4.9833271035380726e-05,
      "loss": 0.9349,
      "step": 2798
    },
    {
      "epoch": 0.03970631507670055,
      "grad_norm": 3.5,
      "learning_rate": 4.9833013328484965e-05,
      "loss": 0.9215,
      "step": 2800
    },
    {
      "epoch": 0.03973467673032676,
      "grad_norm": 3.140625,
      "learning_rate": 4.983275542324629e-05,
      "loss": 0.9314,
      "step": 2802
    },
    {
      "epoch": 0.039763038383952974,
      "grad_norm": 2.90625,
      "learning_rate": 4.9832497319666724e-05,
      "loss": 0.9567,
      "step": 2804
    },
    {
      "epoch": 0.039791400037579194,
      "grad_norm": 2.96875,
      "learning_rate": 4.983223901774836e-05,
      "loss": 0.9873,
      "step": 2806
    },
    {
      "epoch": 0.03981976169120541,
      "grad_norm": 3.1875,
      "learning_rate": 4.983198051749324e-05,
      "loss": 0.8804,
      "step": 2808
    },
    {
      "epoch": 0.03984812334483162,
      "grad_norm": 2.921875,
      "learning_rate": 4.9831721818903434e-05,
      "loss": 0.9592,
      "step": 2810
    },
    {
      "epoch": 0.039876484998457834,
      "grad_norm": 4.09375,
      "learning_rate": 4.983146292198101e-05,
      "loss": 0.9304,
      "step": 2812
    },
    {
      "epoch": 0.03990484665208405,
      "grad_norm": 3.4375,
      "learning_rate": 4.9831203826728036e-05,
      "loss": 0.9416,
      "step": 2814
    },
    {
      "epoch": 0.03993320830571027,
      "grad_norm": 3.21875,
      "learning_rate": 4.9830944533146584e-05,
      "loss": 1.0525,
      "step": 2816
    },
    {
      "epoch": 0.03996156995933648,
      "grad_norm": 2.828125,
      "learning_rate": 4.983068504123871e-05,
      "loss": 0.9605,
      "step": 2818
    },
    {
      "epoch": 0.039989931612962694,
      "grad_norm": 3.515625,
      "learning_rate": 4.9830425351006515e-05,
      "loss": 0.9575,
      "step": 2820
    },
    {
      "epoch": 0.04001829326658891,
      "grad_norm": 2.84375,
      "learning_rate": 4.983016546245205e-05,
      "loss": 0.9281,
      "step": 2822
    },
    {
      "epoch": 0.04004665492021512,
      "grad_norm": 3.03125,
      "learning_rate": 4.982990537557739e-05,
      "loss": 0.9023,
      "step": 2824
    },
    {
      "epoch": 0.04007501657384134,
      "grad_norm": 3.546875,
      "learning_rate": 4.982964509038463e-05,
      "loss": 0.9411,
      "step": 2826
    },
    {
      "epoch": 0.040103378227467554,
      "grad_norm": 3.359375,
      "learning_rate": 4.982938460687583e-05,
      "loss": 0.8976,
      "step": 2828
    },
    {
      "epoch": 0.04013173988109377,
      "grad_norm": 3.296875,
      "learning_rate": 4.982912392505308e-05,
      "loss": 0.9631,
      "step": 2830
    },
    {
      "epoch": 0.04016010153471998,
      "grad_norm": 3.015625,
      "learning_rate": 4.982886304491847e-05,
      "loss": 0.9476,
      "step": 2832
    },
    {
      "epoch": 0.04018846318834619,
      "grad_norm": 2.953125,
      "learning_rate": 4.982860196647406e-05,
      "loss": 0.9361,
      "step": 2834
    },
    {
      "epoch": 0.040216824841972414,
      "grad_norm": 2.859375,
      "learning_rate": 4.982834068972196e-05,
      "loss": 0.9101,
      "step": 2836
    },
    {
      "epoch": 0.04024518649559863,
      "grad_norm": 2.953125,
      "learning_rate": 4.9828079214664245e-05,
      "loss": 0.9587,
      "step": 2838
    },
    {
      "epoch": 0.04027354814922484,
      "grad_norm": 3.28125,
      "learning_rate": 4.982781754130301e-05,
      "loss": 0.896,
      "step": 2840
    },
    {
      "epoch": 0.04030190980285105,
      "grad_norm": 3.71875,
      "learning_rate": 4.9827555669640335e-05,
      "loss": 0.9158,
      "step": 2842
    },
    {
      "epoch": 0.04033027145647727,
      "grad_norm": 3.375,
      "learning_rate": 4.982729359967832e-05,
      "loss": 0.9883,
      "step": 2844
    },
    {
      "epoch": 0.04035863311010349,
      "grad_norm": 3.328125,
      "learning_rate": 4.9827031331419056e-05,
      "loss": 0.9299,
      "step": 2846
    },
    {
      "epoch": 0.0403869947637297,
      "grad_norm": 3.203125,
      "learning_rate": 4.982676886486463e-05,
      "loss": 0.9702,
      "step": 2848
    },
    {
      "epoch": 0.04041535641735591,
      "grad_norm": 2.828125,
      "learning_rate": 4.982650620001715e-05,
      "loss": 0.9169,
      "step": 2850
    },
    {
      "epoch": 0.040443718070982126,
      "grad_norm": 2.75,
      "learning_rate": 4.982624333687871e-05,
      "loss": 0.9169,
      "step": 2852
    },
    {
      "epoch": 0.04047207972460835,
      "grad_norm": 2.90625,
      "learning_rate": 4.982598027545141e-05,
      "loss": 0.9319,
      "step": 2854
    },
    {
      "epoch": 0.04050044137823456,
      "grad_norm": 3.078125,
      "learning_rate": 4.9825717015737336e-05,
      "loss": 0.9809,
      "step": 2856
    },
    {
      "epoch": 0.04052880303186077,
      "grad_norm": 3.015625,
      "learning_rate": 4.982545355773862e-05,
      "loss": 0.938,
      "step": 2858
    },
    {
      "epoch": 0.040557164685486986,
      "grad_norm": 3.890625,
      "learning_rate": 4.982518990145735e-05,
      "loss": 0.8894,
      "step": 2860
    },
    {
      "epoch": 0.0405855263391132,
      "grad_norm": 3.25,
      "learning_rate": 4.982492604689562e-05,
      "loss": 0.931,
      "step": 2862
    },
    {
      "epoch": 0.04061388799273942,
      "grad_norm": 2.953125,
      "learning_rate": 4.9824661994055564e-05,
      "loss": 0.9195,
      "step": 2864
    },
    {
      "epoch": 0.04064224964636563,
      "grad_norm": 3.109375,
      "learning_rate": 4.982439774293927e-05,
      "loss": 0.9341,
      "step": 2866
    },
    {
      "epoch": 0.040670611299991846,
      "grad_norm": 3.484375,
      "learning_rate": 4.982413329354885e-05,
      "loss": 0.9446,
      "step": 2868
    },
    {
      "epoch": 0.04069897295361806,
      "grad_norm": 3.421875,
      "learning_rate": 4.982386864588643e-05,
      "loss": 0.9015,
      "step": 2870
    },
    {
      "epoch": 0.04072733460724427,
      "grad_norm": 3.125,
      "learning_rate": 4.982360379995411e-05,
      "loss": 0.9191,
      "step": 2872
    },
    {
      "epoch": 0.04075569626087049,
      "grad_norm": 2.953125,
      "learning_rate": 4.982333875575401e-05,
      "loss": 0.9032,
      "step": 2874
    },
    {
      "epoch": 0.040784057914496706,
      "grad_norm": 3.265625,
      "learning_rate": 4.982307351328825e-05,
      "loss": 0.9839,
      "step": 2876
    },
    {
      "epoch": 0.04081241956812292,
      "grad_norm": 3.5,
      "learning_rate": 4.982280807255895e-05,
      "loss": 0.9874,
      "step": 2878
    },
    {
      "epoch": 0.04084078122174913,
      "grad_norm": 3.1875,
      "learning_rate": 4.982254243356823e-05,
      "loss": 0.9909,
      "step": 2880
    },
    {
      "epoch": 0.040869142875375346,
      "grad_norm": 3.34375,
      "learning_rate": 4.9822276596318195e-05,
      "loss": 0.9119,
      "step": 2882
    },
    {
      "epoch": 0.040897504529001566,
      "grad_norm": 2.96875,
      "learning_rate": 4.9822010560810986e-05,
      "loss": 0.8848,
      "step": 2884
    },
    {
      "epoch": 0.04092586618262778,
      "grad_norm": 3.03125,
      "learning_rate": 4.982174432704872e-05,
      "loss": 0.9456,
      "step": 2886
    },
    {
      "epoch": 0.04095422783625399,
      "grad_norm": 3.421875,
      "learning_rate": 4.982147789503353e-05,
      "loss": 0.9198,
      "step": 2888
    },
    {
      "epoch": 0.040982589489880206,
      "grad_norm": 3.34375,
      "learning_rate": 4.982121126476755e-05,
      "loss": 0.9682,
      "step": 2890
    },
    {
      "epoch": 0.04101095114350642,
      "grad_norm": 3.5625,
      "learning_rate": 4.982094443625289e-05,
      "loss": 1.0083,
      "step": 2892
    },
    {
      "epoch": 0.04103931279713264,
      "grad_norm": 2.625,
      "learning_rate": 4.982067740949169e-05,
      "loss": 0.9441,
      "step": 2894
    },
    {
      "epoch": 0.04106767445075885,
      "grad_norm": 3.25,
      "learning_rate": 4.982041018448609e-05,
      "loss": 0.9469,
      "step": 2896
    },
    {
      "epoch": 0.041096036104385066,
      "grad_norm": 3.125,
      "learning_rate": 4.982014276123821e-05,
      "loss": 0.9347,
      "step": 2898
    },
    {
      "epoch": 0.04112439775801128,
      "grad_norm": 3.109375,
      "learning_rate": 4.98198751397502e-05,
      "loss": 0.9864,
      "step": 2900
    },
    {
      "epoch": 0.0411527594116375,
      "grad_norm": 2.96875,
      "learning_rate": 4.981960732002419e-05,
      "loss": 0.9041,
      "step": 2902
    },
    {
      "epoch": 0.04118112106526371,
      "grad_norm": 3.015625,
      "learning_rate": 4.981933930206232e-05,
      "loss": 0.912,
      "step": 2904
    },
    {
      "epoch": 0.041209482718889925,
      "grad_norm": 2.984375,
      "learning_rate": 4.981907108586674e-05,
      "loss": 0.8821,
      "step": 2906
    },
    {
      "epoch": 0.04123784437251614,
      "grad_norm": 3.46875,
      "learning_rate": 4.981880267143957e-05,
      "loss": 0.9636,
      "step": 2908
    },
    {
      "epoch": 0.04126620602614235,
      "grad_norm": 2.828125,
      "learning_rate": 4.981853405878298e-05,
      "loss": 0.9409,
      "step": 2910
    },
    {
      "epoch": 0.04129456767976857,
      "grad_norm": 3.0,
      "learning_rate": 4.98182652478991e-05,
      "loss": 0.9295,
      "step": 2912
    },
    {
      "epoch": 0.041322929333394785,
      "grad_norm": 3.0625,
      "learning_rate": 4.981799623879008e-05,
      "loss": 0.9139,
      "step": 2914
    },
    {
      "epoch": 0.041351290987021,
      "grad_norm": 3.390625,
      "learning_rate": 4.981772703145806e-05,
      "loss": 0.8897,
      "step": 2916
    },
    {
      "epoch": 0.04137965264064721,
      "grad_norm": 3.125,
      "learning_rate": 4.981745762590521e-05,
      "loss": 0.9803,
      "step": 2918
    },
    {
      "epoch": 0.041408014294273425,
      "grad_norm": 3.140625,
      "learning_rate": 4.981718802213366e-05,
      "loss": 0.9084,
      "step": 2920
    },
    {
      "epoch": 0.041436375947899645,
      "grad_norm": 2.875,
      "learning_rate": 4.981691822014558e-05,
      "loss": 0.8994,
      "step": 2922
    },
    {
      "epoch": 0.04146473760152586,
      "grad_norm": 2.984375,
      "learning_rate": 4.981664821994312e-05,
      "loss": 0.9365,
      "step": 2924
    },
    {
      "epoch": 0.04149309925515207,
      "grad_norm": 3.90625,
      "learning_rate": 4.981637802152843e-05,
      "loss": 0.9405,
      "step": 2926
    },
    {
      "epoch": 0.041521460908778285,
      "grad_norm": 3.1875,
      "learning_rate": 4.981610762490368e-05,
      "loss": 0.9418,
      "step": 2928
    },
    {
      "epoch": 0.0415498225624045,
      "grad_norm": 3.3125,
      "learning_rate": 4.9815837030071024e-05,
      "loss": 0.9672,
      "step": 2930
    },
    {
      "epoch": 0.04157818421603072,
      "grad_norm": 3.234375,
      "learning_rate": 4.981556623703262e-05,
      "loss": 0.9216,
      "step": 2932
    },
    {
      "epoch": 0.04160654586965693,
      "grad_norm": 3.46875,
      "learning_rate": 4.981529524579063e-05,
      "loss": 0.9303,
      "step": 2934
    },
    {
      "epoch": 0.041634907523283145,
      "grad_norm": 3.34375,
      "learning_rate": 4.981502405634722e-05,
      "loss": 0.9041,
      "step": 2936
    },
    {
      "epoch": 0.04166326917690936,
      "grad_norm": 3.109375,
      "learning_rate": 4.9814752668704565e-05,
      "loss": 0.9281,
      "step": 2938
    },
    {
      "epoch": 0.04169163083053557,
      "grad_norm": 3.015625,
      "learning_rate": 4.981448108286482e-05,
      "loss": 0.9555,
      "step": 2940
    },
    {
      "epoch": 0.04171999248416179,
      "grad_norm": 3.1875,
      "learning_rate": 4.981420929883016e-05,
      "loss": 0.9433,
      "step": 2942
    },
    {
      "epoch": 0.041748354137788005,
      "grad_norm": 2.859375,
      "learning_rate": 4.981393731660276e-05,
      "loss": 0.9179,
      "step": 2944
    },
    {
      "epoch": 0.04177671579141422,
      "grad_norm": 3.609375,
      "learning_rate": 4.981366513618478e-05,
      "loss": 0.9292,
      "step": 2946
    },
    {
      "epoch": 0.04180507744504043,
      "grad_norm": 3.1875,
      "learning_rate": 4.9813392757578405e-05,
      "loss": 0.9108,
      "step": 2948
    },
    {
      "epoch": 0.041833439098666644,
      "grad_norm": 3.21875,
      "learning_rate": 4.9813120180785814e-05,
      "loss": 1.0379,
      "step": 2950
    },
    {
      "epoch": 0.041861800752292865,
      "grad_norm": 3.234375,
      "learning_rate": 4.981284740580916e-05,
      "loss": 0.9573,
      "step": 2952
    },
    {
      "epoch": 0.04189016240591908,
      "grad_norm": 3.09375,
      "learning_rate": 4.9812574432650654e-05,
      "loss": 0.9472,
      "step": 2954
    },
    {
      "epoch": 0.04191852405954529,
      "grad_norm": 3.328125,
      "learning_rate": 4.9812301261312456e-05,
      "loss": 0.9576,
      "step": 2956
    },
    {
      "epoch": 0.041946885713171504,
      "grad_norm": 3.078125,
      "learning_rate": 4.981202789179675e-05,
      "loss": 0.8965,
      "step": 2958
    },
    {
      "epoch": 0.041975247366797724,
      "grad_norm": 2.84375,
      "learning_rate": 4.981175432410573e-05,
      "loss": 0.8836,
      "step": 2960
    },
    {
      "epoch": 0.04200360902042394,
      "grad_norm": 3.375,
      "learning_rate": 4.981148055824157e-05,
      "loss": 0.8897,
      "step": 2962
    },
    {
      "epoch": 0.04203197067405015,
      "grad_norm": 3.375,
      "learning_rate": 4.981120659420646e-05,
      "loss": 0.9476,
      "step": 2964
    },
    {
      "epoch": 0.042060332327676364,
      "grad_norm": 3.421875,
      "learning_rate": 4.981093243200258e-05,
      "loss": 0.9885,
      "step": 2966
    },
    {
      "epoch": 0.04208869398130258,
      "grad_norm": 3.015625,
      "learning_rate": 4.981065807163214e-05,
      "loss": 0.9165,
      "step": 2968
    },
    {
      "epoch": 0.0421170556349288,
      "grad_norm": 2.984375,
      "learning_rate": 4.9810383513097316e-05,
      "loss": 0.9329,
      "step": 2970
    },
    {
      "epoch": 0.04214541728855501,
      "grad_norm": 3.28125,
      "learning_rate": 4.9810108756400294e-05,
      "loss": 0.9247,
      "step": 2972
    },
    {
      "epoch": 0.042173778942181224,
      "grad_norm": 2.921875,
      "learning_rate": 4.980983380154329e-05,
      "loss": 0.8931,
      "step": 2974
    },
    {
      "epoch": 0.04220214059580744,
      "grad_norm": 3.25,
      "learning_rate": 4.980955864852848e-05,
      "loss": 0.9464,
      "step": 2976
    },
    {
      "epoch": 0.04223050224943365,
      "grad_norm": 3.15625,
      "learning_rate": 4.9809283297358076e-05,
      "loss": 0.9112,
      "step": 2978
    },
    {
      "epoch": 0.04225886390305987,
      "grad_norm": 3.1875,
      "learning_rate": 4.980900774803427e-05,
      "loss": 0.9306,
      "step": 2980
    },
    {
      "epoch": 0.042287225556686084,
      "grad_norm": 3.90625,
      "learning_rate": 4.980873200055927e-05,
      "loss": 0.9246,
      "step": 2982
    },
    {
      "epoch": 0.0423155872103123,
      "grad_norm": 3.078125,
      "learning_rate": 4.980845605493526e-05,
      "loss": 0.9789,
      "step": 2984
    },
    {
      "epoch": 0.04234394886393851,
      "grad_norm": 3.21875,
      "learning_rate": 4.9808179911164466e-05,
      "loss": 0.9263,
      "step": 2986
    },
    {
      "epoch": 0.042372310517564724,
      "grad_norm": 3.53125,
      "learning_rate": 4.9807903569249085e-05,
      "loss": 0.9466,
      "step": 2988
    },
    {
      "epoch": 0.042400672171190944,
      "grad_norm": 3.296875,
      "learning_rate": 4.980762702919132e-05,
      "loss": 0.9118,
      "step": 2990
    },
    {
      "epoch": 0.04242903382481716,
      "grad_norm": 2.921875,
      "learning_rate": 4.980735029099338e-05,
      "loss": 0.9017,
      "step": 2992
    },
    {
      "epoch": 0.04245739547844337,
      "grad_norm": 3.328125,
      "learning_rate": 4.9807073354657485e-05,
      "loss": 0.9818,
      "step": 2994
    },
    {
      "epoch": 0.04248575713206958,
      "grad_norm": 3.828125,
      "learning_rate": 4.9806796220185836e-05,
      "loss": 0.9899,
      "step": 2996
    },
    {
      "epoch": 0.0425141187856958,
      "grad_norm": 2.953125,
      "learning_rate": 4.980651888758065e-05,
      "loss": 0.967,
      "step": 2998
    },
    {
      "epoch": 0.04254248043932202,
      "grad_norm": 3.09375,
      "learning_rate": 4.9806241356844155e-05,
      "loss": 0.9032,
      "step": 3000
    },
    {
      "epoch": 0.04257084209294823,
      "grad_norm": 2.703125,
      "learning_rate": 4.980596362797855e-05,
      "loss": 0.9328,
      "step": 3002
    },
    {
      "epoch": 0.04259920374657444,
      "grad_norm": 3.0625,
      "learning_rate": 4.980568570098605e-05,
      "loss": 0.9167,
      "step": 3004
    },
    {
      "epoch": 0.04262756540020066,
      "grad_norm": 3.03125,
      "learning_rate": 4.980540757586889e-05,
      "loss": 0.9344,
      "step": 3006
    },
    {
      "epoch": 0.04265592705382688,
      "grad_norm": 3.25,
      "learning_rate": 4.980512925262928e-05,
      "loss": 0.9981,
      "step": 3008
    },
    {
      "epoch": 0.04268428870745309,
      "grad_norm": 3.515625,
      "learning_rate": 4.980485073126946e-05,
      "loss": 0.9373,
      "step": 3010
    },
    {
      "epoch": 0.0427126503610793,
      "grad_norm": 3.046875,
      "learning_rate": 4.980457201179163e-05,
      "loss": 0.9236,
      "step": 3012
    },
    {
      "epoch": 0.042741012014705516,
      "grad_norm": 2.9375,
      "learning_rate": 4.9804293094198034e-05,
      "loss": 0.8755,
      "step": 3014
    },
    {
      "epoch": 0.04276937366833173,
      "grad_norm": 3.0,
      "learning_rate": 4.980401397849089e-05,
      "loss": 0.9276,
      "step": 3016
    },
    {
      "epoch": 0.04279773532195795,
      "grad_norm": 3.15625,
      "learning_rate": 4.9803734664672444e-05,
      "loss": 0.9319,
      "step": 3018
    },
    {
      "epoch": 0.04282609697558416,
      "grad_norm": 3.4375,
      "learning_rate": 4.9803455152744905e-05,
      "loss": 0.9213,
      "step": 3020
    },
    {
      "epoch": 0.042854458629210376,
      "grad_norm": 3.5,
      "learning_rate": 4.980317544271052e-05,
      "loss": 0.961,
      "step": 3022
    },
    {
      "epoch": 0.04288282028283659,
      "grad_norm": 3.328125,
      "learning_rate": 4.9802895534571516e-05,
      "loss": 0.9421,
      "step": 3024
    },
    {
      "epoch": 0.0429111819364628,
      "grad_norm": 3.34375,
      "learning_rate": 4.980261542833013e-05,
      "loss": 0.963,
      "step": 3026
    },
    {
      "epoch": 0.04293954359008902,
      "grad_norm": 2.703125,
      "learning_rate": 4.98023351239886e-05,
      "loss": 0.9187,
      "step": 3028
    },
    {
      "epoch": 0.042967905243715236,
      "grad_norm": 2.984375,
      "learning_rate": 4.980205462154917e-05,
      "loss": 0.9213,
      "step": 3030
    },
    {
      "epoch": 0.04299626689734145,
      "grad_norm": 3.125,
      "learning_rate": 4.980177392101407e-05,
      "loss": 0.915,
      "step": 3032
    },
    {
      "epoch": 0.04302462855096766,
      "grad_norm": 3.140625,
      "learning_rate": 4.980149302238555e-05,
      "loss": 0.9377,
      "step": 3034
    },
    {
      "epoch": 0.043052990204593876,
      "grad_norm": 2.765625,
      "learning_rate": 4.980121192566585e-05,
      "loss": 0.945,
      "step": 3036
    },
    {
      "epoch": 0.043081351858220096,
      "grad_norm": 3.03125,
      "learning_rate": 4.980093063085722e-05,
      "loss": 0.9242,
      "step": 3038
    },
    {
      "epoch": 0.04310971351184631,
      "grad_norm": 2.828125,
      "learning_rate": 4.9800649137961894e-05,
      "loss": 0.9368,
      "step": 3040
    },
    {
      "epoch": 0.04313807516547252,
      "grad_norm": 2.984375,
      "learning_rate": 4.980036744698214e-05,
      "loss": 0.9094,
      "step": 3042
    },
    {
      "epoch": 0.043166436819098736,
      "grad_norm": 3.5625,
      "learning_rate": 4.980008555792019e-05,
      "loss": 0.9681,
      "step": 3044
    },
    {
      "epoch": 0.04319479847272495,
      "grad_norm": 2.84375,
      "learning_rate": 4.979980347077831e-05,
      "loss": 0.8958,
      "step": 3046
    },
    {
      "epoch": 0.04322316012635117,
      "grad_norm": 3.109375,
      "learning_rate": 4.979952118555873e-05,
      "loss": 0.9428,
      "step": 3048
    },
    {
      "epoch": 0.04325152177997738,
      "grad_norm": 3.3125,
      "learning_rate": 4.979923870226373e-05,
      "loss": 0.8998,
      "step": 3050
    },
    {
      "epoch": 0.043279883433603596,
      "grad_norm": 3.046875,
      "learning_rate": 4.979895602089556e-05,
      "loss": 0.9338,
      "step": 3052
    },
    {
      "epoch": 0.04330824508722981,
      "grad_norm": 2.90625,
      "learning_rate": 4.979867314145646e-05,
      "loss": 0.908,
      "step": 3054
    },
    {
      "epoch": 0.04333660674085603,
      "grad_norm": 2.953125,
      "learning_rate": 4.979839006394872e-05,
      "loss": 0.9123,
      "step": 3056
    },
    {
      "epoch": 0.04336496839448224,
      "grad_norm": 3.015625,
      "learning_rate": 4.979810678837458e-05,
      "loss": 0.9776,
      "step": 3058
    },
    {
      "epoch": 0.043393330048108456,
      "grad_norm": 3.046875,
      "learning_rate": 4.9797823314736304e-05,
      "loss": 0.8977,
      "step": 3060
    },
    {
      "epoch": 0.04342169170173467,
      "grad_norm": 2.96875,
      "learning_rate": 4.979753964303616e-05,
      "loss": 1.0023,
      "step": 3062
    },
    {
      "epoch": 0.04345005335536088,
      "grad_norm": 3.078125,
      "learning_rate": 4.979725577327641e-05,
      "loss": 0.9269,
      "step": 3064
    },
    {
      "epoch": 0.0434784150089871,
      "grad_norm": 2.921875,
      "learning_rate": 4.979697170545933e-05,
      "loss": 0.9094,
      "step": 3066
    },
    {
      "epoch": 0.043506776662613315,
      "grad_norm": 3.28125,
      "learning_rate": 4.9796687439587184e-05,
      "loss": 0.97,
      "step": 3068
    },
    {
      "epoch": 0.04353513831623953,
      "grad_norm": 2.984375,
      "learning_rate": 4.9796402975662234e-05,
      "loss": 0.9724,
      "step": 3070
    },
    {
      "epoch": 0.04356349996986574,
      "grad_norm": 3.078125,
      "learning_rate": 4.9796118313686765e-05,
      "loss": 0.917,
      "step": 3072
    },
    {
      "epoch": 0.043591861623491955,
      "grad_norm": 3.421875,
      "learning_rate": 4.9795833453663046e-05,
      "loss": 0.9411,
      "step": 3074
    },
    {
      "epoch": 0.043620223277118175,
      "grad_norm": 3.421875,
      "learning_rate": 4.979554839559335e-05,
      "loss": 0.8897,
      "step": 3076
    },
    {
      "epoch": 0.04364858493074439,
      "grad_norm": 3.140625,
      "learning_rate": 4.9795263139479946e-05,
      "loss": 0.9741,
      "step": 3078
    },
    {
      "epoch": 0.0436769465843706,
      "grad_norm": 3.21875,
      "learning_rate": 4.9794977685325126e-05,
      "loss": 0.9598,
      "step": 3080
    },
    {
      "epoch": 0.043705308237996815,
      "grad_norm": 3.15625,
      "learning_rate": 4.9794692033131176e-05,
      "loss": 0.8977,
      "step": 3082
    },
    {
      "epoch": 0.04373366989162303,
      "grad_norm": 3.09375,
      "learning_rate": 4.9794406182900355e-05,
      "loss": 0.8904,
      "step": 3084
    },
    {
      "epoch": 0.04376203154524925,
      "grad_norm": 2.734375,
      "learning_rate": 4.9794120134634965e-05,
      "loss": 0.9249,
      "step": 3086
    },
    {
      "epoch": 0.04379039319887546,
      "grad_norm": 3.59375,
      "learning_rate": 4.9793833888337284e-05,
      "loss": 0.9548,
      "step": 3088
    },
    {
      "epoch": 0.043818754852501675,
      "grad_norm": 3.078125,
      "learning_rate": 4.9793547444009594e-05,
      "loss": 0.9262,
      "step": 3090
    },
    {
      "epoch": 0.04384711650612789,
      "grad_norm": 3.03125,
      "learning_rate": 4.979326080165419e-05,
      "loss": 0.9289,
      "step": 3092
    },
    {
      "epoch": 0.0438754781597541,
      "grad_norm": 2.9375,
      "learning_rate": 4.979297396127335e-05,
      "loss": 0.9124,
      "step": 3094
    },
    {
      "epoch": 0.04390383981338032,
      "grad_norm": 3.328125,
      "learning_rate": 4.979268692286938e-05,
      "loss": 0.905,
      "step": 3096
    },
    {
      "epoch": 0.043932201467006535,
      "grad_norm": 3.453125,
      "learning_rate": 4.9792399686444566e-05,
      "loss": 0.949,
      "step": 3098
    },
    {
      "epoch": 0.04396056312063275,
      "grad_norm": 3.390625,
      "learning_rate": 4.97921122520012e-05,
      "loss": 0.9305,
      "step": 3100
    },
    {
      "epoch": 0.04398892477425896,
      "grad_norm": 2.984375,
      "learning_rate": 4.979182461954158e-05,
      "loss": 0.9357,
      "step": 3102
    },
    {
      "epoch": 0.044017286427885174,
      "grad_norm": 2.953125,
      "learning_rate": 4.979153678906801e-05,
      "loss": 0.9737,
      "step": 3104
    },
    {
      "epoch": 0.044045648081511395,
      "grad_norm": 3.09375,
      "learning_rate": 4.9791248760582776e-05,
      "loss": 0.9552,
      "step": 3106
    },
    {
      "epoch": 0.04407400973513761,
      "grad_norm": 3.234375,
      "learning_rate": 4.9790960534088184e-05,
      "loss": 0.8867,
      "step": 3108
    },
    {
      "epoch": 0.04410237138876382,
      "grad_norm": 3.0625,
      "learning_rate": 4.9790672109586535e-05,
      "loss": 0.9771,
      "step": 3110
    },
    {
      "epoch": 0.044130733042390034,
      "grad_norm": 3.40625,
      "learning_rate": 4.979038348708014e-05,
      "loss": 0.961,
      "step": 3112
    },
    {
      "epoch": 0.044159094696016254,
      "grad_norm": 3.0,
      "learning_rate": 4.97900946665713e-05,
      "loss": 0.9418,
      "step": 3114
    },
    {
      "epoch": 0.04418745634964247,
      "grad_norm": 4.03125,
      "learning_rate": 4.978980564806231e-05,
      "loss": 0.9424,
      "step": 3116
    },
    {
      "epoch": 0.04421581800326868,
      "grad_norm": 3.078125,
      "learning_rate": 4.978951643155549e-05,
      "loss": 0.9673,
      "step": 3118
    },
    {
      "epoch": 0.044244179656894894,
      "grad_norm": 3.078125,
      "learning_rate": 4.978922701705316e-05,
      "loss": 0.9116,
      "step": 3120
    },
    {
      "epoch": 0.04427254131052111,
      "grad_norm": 3.15625,
      "learning_rate": 4.9788937404557615e-05,
      "loss": 0.9493,
      "step": 3122
    },
    {
      "epoch": 0.04430090296414733,
      "grad_norm": 2.96875,
      "learning_rate": 4.978864759407117e-05,
      "loss": 0.9975,
      "step": 3124
    },
    {
      "epoch": 0.04432926461777354,
      "grad_norm": 2.921875,
      "learning_rate": 4.978835758559615e-05,
      "loss": 0.9271,
      "step": 3126
    },
    {
      "epoch": 0.044357626271399754,
      "grad_norm": 3.171875,
      "learning_rate": 4.978806737913485e-05,
      "loss": 0.9636,
      "step": 3128
    },
    {
      "epoch": 0.04438598792502597,
      "grad_norm": 3.09375,
      "learning_rate": 4.9787776974689614e-05,
      "loss": 0.907,
      "step": 3130
    },
    {
      "epoch": 0.04441434957865218,
      "grad_norm": 3.140625,
      "learning_rate": 4.9787486372262746e-05,
      "loss": 0.9014,
      "step": 3132
    },
    {
      "epoch": 0.0444427112322784,
      "grad_norm": 2.859375,
      "learning_rate": 4.9787195571856584e-05,
      "loss": 0.9399,
      "step": 3134
    },
    {
      "epoch": 0.044471072885904614,
      "grad_norm": 3.28125,
      "learning_rate": 4.9786904573473425e-05,
      "loss": 0.907,
      "step": 3136
    },
    {
      "epoch": 0.04449943453953083,
      "grad_norm": 3.09375,
      "learning_rate": 4.9786613377115606e-05,
      "loss": 0.909,
      "step": 3138
    },
    {
      "epoch": 0.04452779619315704,
      "grad_norm": 3.359375,
      "learning_rate": 4.978632198278545e-05,
      "loss": 0.9499,
      "step": 3140
    },
    {
      "epoch": 0.044556157846783254,
      "grad_norm": 3.078125,
      "learning_rate": 4.9786030390485295e-05,
      "loss": 0.9809,
      "step": 3142
    },
    {
      "epoch": 0.044584519500409474,
      "grad_norm": 3.421875,
      "learning_rate": 4.978573860021746e-05,
      "loss": 0.9246,
      "step": 3144
    },
    {
      "epoch": 0.04461288115403569,
      "grad_norm": 3.140625,
      "learning_rate": 4.978544661198428e-05,
      "loss": 0.9493,
      "step": 3146
    },
    {
      "epoch": 0.0446412428076619,
      "grad_norm": 2.828125,
      "learning_rate": 4.9785154425788074e-05,
      "loss": 0.849,
      "step": 3148
    },
    {
      "epoch": 0.044669604461288114,
      "grad_norm": 3.09375,
      "learning_rate": 4.9784862041631196e-05,
      "loss": 0.893,
      "step": 3150
    },
    {
      "epoch": 0.04469796611491433,
      "grad_norm": 3.15625,
      "learning_rate": 4.978456945951597e-05,
      "loss": 0.9135,
      "step": 3152
    },
    {
      "epoch": 0.04472632776854055,
      "grad_norm": 3.421875,
      "learning_rate": 4.978427667944473e-05,
      "loss": 0.9909,
      "step": 3154
    },
    {
      "epoch": 0.04475468942216676,
      "grad_norm": 2.859375,
      "learning_rate": 4.978398370141982e-05,
      "loss": 0.9353,
      "step": 3156
    },
    {
      "epoch": 0.04478305107579297,
      "grad_norm": 2.96875,
      "learning_rate": 4.978369052544358e-05,
      "loss": 0.9553,
      "step": 3158
    },
    {
      "epoch": 0.04481141272941919,
      "grad_norm": 2.828125,
      "learning_rate": 4.9783397151518365e-05,
      "loss": 0.9367,
      "step": 3160
    },
    {
      "epoch": 0.04483977438304541,
      "grad_norm": 3.4375,
      "learning_rate": 4.978310357964648e-05,
      "loss": 0.93,
      "step": 3162
    },
    {
      "epoch": 0.04486813603667162,
      "grad_norm": 3.859375,
      "learning_rate": 4.978280980983031e-05,
      "loss": 0.9286,
      "step": 3164
    },
    {
      "epoch": 0.04489649769029783,
      "grad_norm": 3.875,
      "learning_rate": 4.978251584207218e-05,
      "loss": 0.919,
      "step": 3166
    },
    {
      "epoch": 0.044924859343924047,
      "grad_norm": 3.140625,
      "learning_rate": 4.978222167637444e-05,
      "loss": 0.9463,
      "step": 3168
    },
    {
      "epoch": 0.04495322099755026,
      "grad_norm": 3.140625,
      "learning_rate": 4.978192731273945e-05,
      "loss": 0.9604,
      "step": 3170
    },
    {
      "epoch": 0.04498158265117648,
      "grad_norm": 3.96875,
      "learning_rate": 4.978163275116955e-05,
      "loss": 1.0044,
      "step": 3172
    },
    {
      "epoch": 0.04500994430480269,
      "grad_norm": 3.25,
      "learning_rate": 4.9781337991667094e-05,
      "loss": 0.9188,
      "step": 3174
    },
    {
      "epoch": 0.045038305958428906,
      "grad_norm": 3.25,
      "learning_rate": 4.9781043034234443e-05,
      "loss": 0.9387,
      "step": 3176
    },
    {
      "epoch": 0.04506666761205512,
      "grad_norm": 3.4375,
      "learning_rate": 4.9780747878873944e-05,
      "loss": 0.8993,
      "step": 3178
    },
    {
      "epoch": 0.04509502926568133,
      "grad_norm": 2.765625,
      "learning_rate": 4.978045252558796e-05,
      "loss": 0.9147,
      "step": 3180
    },
    {
      "epoch": 0.04512339091930755,
      "grad_norm": 3.09375,
      "learning_rate": 4.978015697437884e-05,
      "loss": 0.9594,
      "step": 3182
    },
    {
      "epoch": 0.045151752572933766,
      "grad_norm": 3.53125,
      "learning_rate": 4.977986122524897e-05,
      "loss": 0.9976,
      "step": 3184
    },
    {
      "epoch": 0.04518011422655998,
      "grad_norm": 2.890625,
      "learning_rate": 4.977956527820069e-05,
      "loss": 0.913,
      "step": 3186
    },
    {
      "epoch": 0.04520847588018619,
      "grad_norm": 3.25,
      "learning_rate": 4.977926913323636e-05,
      "loss": 0.9059,
      "step": 3188
    },
    {
      "epoch": 0.045236837533812406,
      "grad_norm": 2.875,
      "learning_rate": 4.977897279035837e-05,
      "loss": 0.8823,
      "step": 3190
    },
    {
      "epoch": 0.045265199187438626,
      "grad_norm": 3.296875,
      "learning_rate": 4.9778676249569055e-05,
      "loss": 0.9317,
      "step": 3192
    },
    {
      "epoch": 0.04529356084106484,
      "grad_norm": 3.140625,
      "learning_rate": 4.977837951087081e-05,
      "loss": 0.9011,
      "step": 3194
    },
    {
      "epoch": 0.04532192249469105,
      "grad_norm": 3.328125,
      "learning_rate": 4.977808257426599e-05,
      "loss": 0.9612,
      "step": 3196
    },
    {
      "epoch": 0.045350284148317266,
      "grad_norm": 2.859375,
      "learning_rate": 4.977778543975698e-05,
      "loss": 0.9206,
      "step": 3198
    },
    {
      "epoch": 0.04537864580194348,
      "grad_norm": 3.5,
      "learning_rate": 4.9777488107346135e-05,
      "loss": 0.9314,
      "step": 3200
    },
    {
      "epoch": 0.0454070074555697,
      "grad_norm": 2.765625,
      "learning_rate": 4.977719057703585e-05,
      "loss": 0.9239,
      "step": 3202
    },
    {
      "epoch": 0.04543536910919591,
      "grad_norm": 3.25,
      "learning_rate": 4.977689284882849e-05,
      "loss": 0.931,
      "step": 3204
    },
    {
      "epoch": 0.045463730762822126,
      "grad_norm": 3.078125,
      "learning_rate": 4.9776594922726424e-05,
      "loss": 0.9285,
      "step": 3206
    },
    {
      "epoch": 0.04549209241644834,
      "grad_norm": 2.96875,
      "learning_rate": 4.9776296798732054e-05,
      "loss": 0.9376,
      "step": 3208
    },
    {
      "epoch": 0.04552045407007456,
      "grad_norm": 3.5625,
      "learning_rate": 4.977599847684774e-05,
      "loss": 0.9655,
      "step": 3210
    },
    {
      "epoch": 0.04554881572370077,
      "grad_norm": 3.296875,
      "learning_rate": 4.977569995707588e-05,
      "loss": 0.9456,
      "step": 3212
    },
    {
      "epoch": 0.045577177377326986,
      "grad_norm": 3.125,
      "learning_rate": 4.9775401239418844e-05,
      "loss": 0.8899,
      "step": 3214
    },
    {
      "epoch": 0.0456055390309532,
      "grad_norm": 3.125,
      "learning_rate": 4.977510232387903e-05,
      "loss": 0.9059,
      "step": 3216
    },
    {
      "epoch": 0.04563390068457941,
      "grad_norm": 3.484375,
      "learning_rate": 4.977480321045882e-05,
      "loss": 0.9738,
      "step": 3218
    },
    {
      "epoch": 0.04566226233820563,
      "grad_norm": 3.296875,
      "learning_rate": 4.9774503899160606e-05,
      "loss": 0.9015,
      "step": 3220
    },
    {
      "epoch": 0.045690623991831845,
      "grad_norm": 2.71875,
      "learning_rate": 4.9774204389986776e-05,
      "loss": 0.9299,
      "step": 3222
    },
    {
      "epoch": 0.04571898564545806,
      "grad_norm": 3.03125,
      "learning_rate": 4.977390468293973e-05,
      "loss": 0.9277,
      "step": 3224
    },
    {
      "epoch": 0.04574734729908427,
      "grad_norm": 3.28125,
      "learning_rate": 4.9773604778021844e-05,
      "loss": 0.9769,
      "step": 3226
    },
    {
      "epoch": 0.045775708952710485,
      "grad_norm": 3.25,
      "learning_rate": 4.9773304675235534e-05,
      "loss": 0.891,
      "step": 3228
    },
    {
      "epoch": 0.045804070606336705,
      "grad_norm": 3.234375,
      "learning_rate": 4.9773004374583174e-05,
      "loss": 0.9185,
      "step": 3230
    },
    {
      "epoch": 0.04583243225996292,
      "grad_norm": 3.1875,
      "learning_rate": 4.977270387606719e-05,
      "loss": 0.9706,
      "step": 3232
    },
    {
      "epoch": 0.04586079391358913,
      "grad_norm": 3.125,
      "learning_rate": 4.977240317968996e-05,
      "loss": 0.9086,
      "step": 3234
    },
    {
      "epoch": 0.045889155567215345,
      "grad_norm": 3.65625,
      "learning_rate": 4.9772102285453894e-05,
      "loss": 0.9956,
      "step": 3236
    },
    {
      "epoch": 0.04591751722084156,
      "grad_norm": 2.890625,
      "learning_rate": 4.97718011933614e-05,
      "loss": 0.9056,
      "step": 3238
    },
    {
      "epoch": 0.04594587887446778,
      "grad_norm": 3.1875,
      "learning_rate": 4.977149990341487e-05,
      "loss": 0.8912,
      "step": 3240
    },
    {
      "epoch": 0.04597424052809399,
      "grad_norm": 2.96875,
      "learning_rate": 4.977119841561672e-05,
      "loss": 0.9426,
      "step": 3242
    },
    {
      "epoch": 0.046002602181720205,
      "grad_norm": 2.75,
      "learning_rate": 4.9770896729969366e-05,
      "loss": 0.9448,
      "step": 3244
    },
    {
      "epoch": 0.04603096383534642,
      "grad_norm": 3.109375,
      "learning_rate": 4.977059484647519e-05,
      "loss": 0.9121,
      "step": 3246
    },
    {
      "epoch": 0.04605932548897263,
      "grad_norm": 3.09375,
      "learning_rate": 4.977029276513663e-05,
      "loss": 0.9288,
      "step": 3248
    },
    {
      "epoch": 0.04608768714259885,
      "grad_norm": 3.03125,
      "learning_rate": 4.976999048595609e-05,
      "loss": 0.9436,
      "step": 3250
    },
    {
      "epoch": 0.046116048796225065,
      "grad_norm": 3.21875,
      "learning_rate": 4.976968800893598e-05,
      "loss": 0.9197,
      "step": 3252
    },
    {
      "epoch": 0.04614441044985128,
      "grad_norm": 3.0625,
      "learning_rate": 4.976938533407873e-05,
      "loss": 0.9303,
      "step": 3254
    },
    {
      "epoch": 0.04617277210347749,
      "grad_norm": 3.0,
      "learning_rate": 4.9769082461386736e-05,
      "loss": 0.9191,
      "step": 3256
    },
    {
      "epoch": 0.046201133757103705,
      "grad_norm": 3.21875,
      "learning_rate": 4.976877939086243e-05,
      "loss": 0.976,
      "step": 3258
    },
    {
      "epoch": 0.046229495410729925,
      "grad_norm": 3.046875,
      "learning_rate": 4.9768476122508236e-05,
      "loss": 0.9114,
      "step": 3260
    },
    {
      "epoch": 0.04625785706435614,
      "grad_norm": 3.265625,
      "learning_rate": 4.976817265632657e-05,
      "loss": 0.907,
      "step": 3262
    },
    {
      "epoch": 0.04628621871798235,
      "grad_norm": 2.875,
      "learning_rate": 4.9767868992319854e-05,
      "loss": 0.9185,
      "step": 3264
    },
    {
      "epoch": 0.046314580371608564,
      "grad_norm": 3.015625,
      "learning_rate": 4.976756513049051e-05,
      "loss": 0.8994,
      "step": 3266
    },
    {
      "epoch": 0.046342942025234785,
      "grad_norm": 3.109375,
      "learning_rate": 4.9767261070840984e-05,
      "loss": 0.9589,
      "step": 3268
    },
    {
      "epoch": 0.046371303678861,
      "grad_norm": 3.09375,
      "learning_rate": 4.976695681337369e-05,
      "loss": 0.8996,
      "step": 3270
    },
    {
      "epoch": 0.04639966533248721,
      "grad_norm": 3.09375,
      "learning_rate": 4.9766652358091054e-05,
      "loss": 0.9569,
      "step": 3272
    },
    {
      "epoch": 0.046428026986113424,
      "grad_norm": 3.1875,
      "learning_rate": 4.976634770499552e-05,
      "loss": 0.9386,
      "step": 3274
    },
    {
      "epoch": 0.04645638863973964,
      "grad_norm": 3.0625,
      "learning_rate": 4.97660428540895e-05,
      "loss": 0.926,
      "step": 3276
    },
    {
      "epoch": 0.04648475029336586,
      "grad_norm": 3.09375,
      "learning_rate": 4.9765737805375465e-05,
      "loss": 0.9133,
      "step": 3278
    },
    {
      "epoch": 0.04651311194699207,
      "grad_norm": 3.4375,
      "learning_rate": 4.9765432558855815e-05,
      "loss": 0.9125,
      "step": 3280
    },
    {
      "epoch": 0.046541473600618284,
      "grad_norm": 3.046875,
      "learning_rate": 4.976512711453301e-05,
      "loss": 0.9468,
      "step": 3282
    },
    {
      "epoch": 0.0465698352542445,
      "grad_norm": 2.90625,
      "learning_rate": 4.9764821472409484e-05,
      "loss": 0.8747,
      "step": 3284
    },
    {
      "epoch": 0.04659819690787071,
      "grad_norm": 3.140625,
      "learning_rate": 4.9764515632487676e-05,
      "loss": 0.9263,
      "step": 3286
    },
    {
      "epoch": 0.04662655856149693,
      "grad_norm": 3.09375,
      "learning_rate": 4.976420959477002e-05,
      "loss": 0.9107,
      "step": 3288
    },
    {
      "epoch": 0.046654920215123144,
      "grad_norm": 3.09375,
      "learning_rate": 4.976390335925898e-05,
      "loss": 0.961,
      "step": 3290
    },
    {
      "epoch": 0.04668328186874936,
      "grad_norm": 2.984375,
      "learning_rate": 4.976359692595699e-05,
      "loss": 0.8902,
      "step": 3292
    },
    {
      "epoch": 0.04671164352237557,
      "grad_norm": 3.375,
      "learning_rate": 4.9763290294866505e-05,
      "loss": 0.9315,
      "step": 3294
    },
    {
      "epoch": 0.046740005176001784,
      "grad_norm": 3.34375,
      "learning_rate": 4.976298346598996e-05,
      "loss": 0.9785,
      "step": 3296
    },
    {
      "epoch": 0.046768366829628004,
      "grad_norm": 3.046875,
      "learning_rate": 4.976267643932982e-05,
      "loss": 0.9046,
      "step": 3298
    },
    {
      "epoch": 0.04679672848325422,
      "grad_norm": 3.296875,
      "learning_rate": 4.976236921488853e-05,
      "loss": 0.967,
      "step": 3300
    },
    {
      "epoch": 0.04682509013688043,
      "grad_norm": 3.375,
      "learning_rate": 4.9762061792668546e-05,
      "loss": 0.8963,
      "step": 3302
    },
    {
      "epoch": 0.046853451790506644,
      "grad_norm": 2.921875,
      "learning_rate": 4.976175417267232e-05,
      "loss": 0.9218,
      "step": 3304
    },
    {
      "epoch": 0.04688181344413286,
      "grad_norm": 3.078125,
      "learning_rate": 4.976144635490231e-05,
      "loss": 0.96,
      "step": 3306
    },
    {
      "epoch": 0.04691017509775908,
      "grad_norm": 2.890625,
      "learning_rate": 4.976113833936098e-05,
      "loss": 0.9446,
      "step": 3308
    },
    {
      "epoch": 0.04693853675138529,
      "grad_norm": 3.078125,
      "learning_rate": 4.9760830126050796e-05,
      "loss": 0.892,
      "step": 3310
    },
    {
      "epoch": 0.046966898405011503,
      "grad_norm": 3.046875,
      "learning_rate": 4.97605217149742e-05,
      "loss": 0.9317,
      "step": 3312
    },
    {
      "epoch": 0.04699526005863772,
      "grad_norm": 3.21875,
      "learning_rate": 4.976021310613366e-05,
      "loss": 0.918,
      "step": 3314
    },
    {
      "epoch": 0.04702362171226394,
      "grad_norm": 3.046875,
      "learning_rate": 4.9759904299531655e-05,
      "loss": 0.889,
      "step": 3316
    },
    {
      "epoch": 0.04705198336589015,
      "grad_norm": 3.40625,
      "learning_rate": 4.975959529517064e-05,
      "loss": 0.9074,
      "step": 3318
    },
    {
      "epoch": 0.04708034501951636,
      "grad_norm": 2.984375,
      "learning_rate": 4.9759286093053086e-05,
      "loss": 0.91,
      "step": 3320
    },
    {
      "epoch": 0.04710870667314258,
      "grad_norm": 2.96875,
      "learning_rate": 4.9758976693181464e-05,
      "loss": 0.9077,
      "step": 3322
    },
    {
      "epoch": 0.04713706832676879,
      "grad_norm": 3.140625,
      "learning_rate": 4.9758667095558244e-05,
      "loss": 0.9751,
      "step": 3324
    },
    {
      "epoch": 0.04716542998039501,
      "grad_norm": 3.15625,
      "learning_rate": 4.97583573001859e-05,
      "loss": 0.9422,
      "step": 3326
    },
    {
      "epoch": 0.04719379163402122,
      "grad_norm": 3.09375,
      "learning_rate": 4.97580473070669e-05,
      "loss": 0.9031,
      "step": 3328
    },
    {
      "epoch": 0.047222153287647436,
      "grad_norm": 2.890625,
      "learning_rate": 4.9757737116203726e-05,
      "loss": 0.9305,
      "step": 3330
    },
    {
      "epoch": 0.04725051494127365,
      "grad_norm": 3.390625,
      "learning_rate": 4.975742672759885e-05,
      "loss": 0.9085,
      "step": 3332
    },
    {
      "epoch": 0.04727887659489986,
      "grad_norm": 3.0,
      "learning_rate": 4.975711614125477e-05,
      "loss": 0.9191,
      "step": 3334
    },
    {
      "epoch": 0.04730723824852608,
      "grad_norm": 3.015625,
      "learning_rate": 4.9756805357173944e-05,
      "loss": 0.8889,
      "step": 3336
    },
    {
      "epoch": 0.047335599902152296,
      "grad_norm": 3.109375,
      "learning_rate": 4.975649437535885e-05,
      "loss": 0.9321,
      "step": 3338
    },
    {
      "epoch": 0.04736396155577851,
      "grad_norm": 2.96875,
      "learning_rate": 4.9756183195812e-05,
      "loss": 0.884,
      "step": 3340
    },
    {
      "epoch": 0.04739232320940472,
      "grad_norm": 3.078125,
      "learning_rate": 4.975587181853586e-05,
      "loss": 0.9119,
      "step": 3342
    },
    {
      "epoch": 0.047420684863030936,
      "grad_norm": 3.390625,
      "learning_rate": 4.9755560243532915e-05,
      "loss": 0.9104,
      "step": 3344
    },
    {
      "epoch": 0.047449046516657156,
      "grad_norm": 2.765625,
      "learning_rate": 4.9755248470805665e-05,
      "loss": 0.9129,
      "step": 3346
    },
    {
      "epoch": 0.04747740817028337,
      "grad_norm": 2.734375,
      "learning_rate": 4.97549365003566e-05,
      "loss": 0.9714,
      "step": 3348
    },
    {
      "epoch": 0.04750576982390958,
      "grad_norm": 3.390625,
      "learning_rate": 4.9754624332188194e-05,
      "loss": 0.905,
      "step": 3350
    },
    {
      "epoch": 0.047534131477535796,
      "grad_norm": 3.0,
      "learning_rate": 4.975431196630296e-05,
      "loss": 0.8572,
      "step": 3352
    },
    {
      "epoch": 0.04756249313116201,
      "grad_norm": 3.046875,
      "learning_rate": 4.975399940270338e-05,
      "loss": 0.9366,
      "step": 3354
    },
    {
      "epoch": 0.04759085478478823,
      "grad_norm": 3.171875,
      "learning_rate": 4.9753686641391955e-05,
      "loss": 0.9513,
      "step": 3356
    },
    {
      "epoch": 0.04761921643841444,
      "grad_norm": 2.921875,
      "learning_rate": 4.975337368237118e-05,
      "loss": 0.9167,
      "step": 3358
    },
    {
      "epoch": 0.047647578092040656,
      "grad_norm": 2.828125,
      "learning_rate": 4.975306052564357e-05,
      "loss": 0.8736,
      "step": 3360
    },
    {
      "epoch": 0.04767593974566687,
      "grad_norm": 3.203125,
      "learning_rate": 4.9752747171211614e-05,
      "loss": 0.9614,
      "step": 3362
    },
    {
      "epoch": 0.04770430139929309,
      "grad_norm": 2.625,
      "learning_rate": 4.975243361907781e-05,
      "loss": 0.9364,
      "step": 3364
    },
    {
      "epoch": 0.0477326630529193,
      "grad_norm": 2.765625,
      "learning_rate": 4.9752119869244665e-05,
      "loss": 0.8855,
      "step": 3366
    },
    {
      "epoch": 0.047761024706545516,
      "grad_norm": 3.21875,
      "learning_rate": 4.975180592171469e-05,
      "loss": 0.9412,
      "step": 3368
    },
    {
      "epoch": 0.04778938636017173,
      "grad_norm": 3.96875,
      "learning_rate": 4.975149177649039e-05,
      "loss": 0.9166,
      "step": 3370
    },
    {
      "epoch": 0.04781774801379794,
      "grad_norm": 2.984375,
      "learning_rate": 4.975117743357428e-05,
      "loss": 0.9002,
      "step": 3372
    },
    {
      "epoch": 0.04784610966742416,
      "grad_norm": 3.203125,
      "learning_rate": 4.975086289296886e-05,
      "loss": 0.9112,
      "step": 3374
    },
    {
      "epoch": 0.047874471321050376,
      "grad_norm": 3.21875,
      "learning_rate": 4.975054815467665e-05,
      "loss": 0.9567,
      "step": 3376
    },
    {
      "epoch": 0.04790283297467659,
      "grad_norm": 3.3125,
      "learning_rate": 4.975023321870015e-05,
      "loss": 0.9662,
      "step": 3378
    },
    {
      "epoch": 0.0479311946283028,
      "grad_norm": 2.84375,
      "learning_rate": 4.97499180850419e-05,
      "loss": 0.9169,
      "step": 3380
    },
    {
      "epoch": 0.047959556281929015,
      "grad_norm": 3.28125,
      "learning_rate": 4.974960275370439e-05,
      "loss": 0.9122,
      "step": 3382
    },
    {
      "epoch": 0.047987917935555235,
      "grad_norm": 3.359375,
      "learning_rate": 4.9749287224690166e-05,
      "loss": 0.9505,
      "step": 3384
    },
    {
      "epoch": 0.04801627958918145,
      "grad_norm": 3.328125,
      "learning_rate": 4.974897149800173e-05,
      "loss": 0.9609,
      "step": 3386
    },
    {
      "epoch": 0.04804464124280766,
      "grad_norm": 2.875,
      "learning_rate": 4.974865557364161e-05,
      "loss": 0.909,
      "step": 3388
    },
    {
      "epoch": 0.048073002896433875,
      "grad_norm": 2.734375,
      "learning_rate": 4.974833945161232e-05,
      "loss": 0.875,
      "step": 3390
    },
    {
      "epoch": 0.04810136455006009,
      "grad_norm": 3.09375,
      "learning_rate": 4.97480231319164e-05,
      "loss": 0.9045,
      "step": 3392
    },
    {
      "epoch": 0.04812972620368631,
      "grad_norm": 3.40625,
      "learning_rate": 4.974770661455636e-05,
      "loss": 0.9215,
      "step": 3394
    },
    {
      "epoch": 0.04815808785731252,
      "grad_norm": 3.75,
      "learning_rate": 4.974738989953474e-05,
      "loss": 0.9425,
      "step": 3396
    },
    {
      "epoch": 0.048186449510938735,
      "grad_norm": 2.8125,
      "learning_rate": 4.974707298685407e-05,
      "loss": 0.9353,
      "step": 3398
    },
    {
      "epoch": 0.04821481116456495,
      "grad_norm": 3.109375,
      "learning_rate": 4.974675587651687e-05,
      "loss": 0.9566,
      "step": 3400
    },
    {
      "epoch": 0.04824317281819116,
      "grad_norm": 3.4375,
      "learning_rate": 4.974643856852569e-05,
      "loss": 0.9053,
      "step": 3402
    },
    {
      "epoch": 0.04827153447181738,
      "grad_norm": 4.15625,
      "learning_rate": 4.974612106288305e-05,
      "loss": 0.9251,
      "step": 3404
    },
    {
      "epoch": 0.048299896125443595,
      "grad_norm": 3.75,
      "learning_rate": 4.974580335959148e-05,
      "loss": 0.8812,
      "step": 3406
    },
    {
      "epoch": 0.04832825777906981,
      "grad_norm": 2.9375,
      "learning_rate": 4.974548545865354e-05,
      "loss": 0.8967,
      "step": 3408
    },
    {
      "epoch": 0.04835661943269602,
      "grad_norm": 3.15625,
      "learning_rate": 4.974516736007176e-05,
      "loss": 0.8755,
      "step": 3410
    },
    {
      "epoch": 0.048384981086322235,
      "grad_norm": 3.15625,
      "learning_rate": 4.974484906384867e-05,
      "loss": 0.908,
      "step": 3412
    },
    {
      "epoch": 0.048413342739948455,
      "grad_norm": 3.21875,
      "learning_rate": 4.974453056998683e-05,
      "loss": 0.9206,
      "step": 3414
    },
    {
      "epoch": 0.04844170439357467,
      "grad_norm": 3.765625,
      "learning_rate": 4.9744211878488756e-05,
      "loss": 0.9799,
      "step": 3416
    },
    {
      "epoch": 0.04847006604720088,
      "grad_norm": 3.390625,
      "learning_rate": 4.974389298935702e-05,
      "loss": 0.9143,
      "step": 3418
    },
    {
      "epoch": 0.048498427700827094,
      "grad_norm": 2.859375,
      "learning_rate": 4.9743573902594164e-05,
      "loss": 0.886,
      "step": 3420
    },
    {
      "epoch": 0.048526789354453315,
      "grad_norm": 3.515625,
      "learning_rate": 4.974325461820273e-05,
      "loss": 0.9103,
      "step": 3422
    },
    {
      "epoch": 0.04855515100807953,
      "grad_norm": 3.34375,
      "learning_rate": 4.9742935136185276e-05,
      "loss": 0.9356,
      "step": 3424
    },
    {
      "epoch": 0.04858351266170574,
      "grad_norm": 3.140625,
      "learning_rate": 4.9742615456544336e-05,
      "loss": 0.9706,
      "step": 3426
    },
    {
      "epoch": 0.048611874315331954,
      "grad_norm": 3.171875,
      "learning_rate": 4.974229557928249e-05,
      "loss": 0.9055,
      "step": 3428
    },
    {
      "epoch": 0.04864023596895817,
      "grad_norm": 3.1875,
      "learning_rate": 4.974197550440227e-05,
      "loss": 0.9383,
      "step": 3430
    },
    {
      "epoch": 0.04866859762258439,
      "grad_norm": 3.1875,
      "learning_rate": 4.9741655231906246e-05,
      "loss": 0.9055,
      "step": 3432
    },
    {
      "epoch": 0.0486969592762106,
      "grad_norm": 3.234375,
      "learning_rate": 4.9741334761796974e-05,
      "loss": 0.8819,
      "step": 3434
    },
    {
      "epoch": 0.048725320929836814,
      "grad_norm": 3.125,
      "learning_rate": 4.9741014094077004e-05,
      "loss": 0.9568,
      "step": 3436
    },
    {
      "epoch": 0.04875368258346303,
      "grad_norm": 3.375,
      "learning_rate": 4.974069322874891e-05,
      "loss": 0.9157,
      "step": 3438
    },
    {
      "epoch": 0.04878204423708924,
      "grad_norm": 3.078125,
      "learning_rate": 4.974037216581524e-05,
      "loss": 0.9767,
      "step": 3440
    },
    {
      "epoch": 0.04881040589071546,
      "grad_norm": 3.265625,
      "learning_rate": 4.9740050905278577e-05,
      "loss": 0.9535,
      "step": 3442
    },
    {
      "epoch": 0.048838767544341674,
      "grad_norm": 2.96875,
      "learning_rate": 4.9739729447141476e-05,
      "loss": 0.8733,
      "step": 3444
    },
    {
      "epoch": 0.04886712919796789,
      "grad_norm": 3.109375,
      "learning_rate": 4.97394077914065e-05,
      "loss": 0.935,
      "step": 3446
    },
    {
      "epoch": 0.0488954908515941,
      "grad_norm": 3.171875,
      "learning_rate": 4.973908593807623e-05,
      "loss": 0.9171,
      "step": 3448
    },
    {
      "epoch": 0.048923852505220314,
      "grad_norm": 3.421875,
      "learning_rate": 4.973876388715323e-05,
      "loss": 0.9399,
      "step": 3450
    },
    {
      "epoch": 0.048952214158846534,
      "grad_norm": 3.171875,
      "learning_rate": 4.9738441638640064e-05,
      "loss": 0.9318,
      "step": 3452
    },
    {
      "epoch": 0.04898057581247275,
      "grad_norm": 3.421875,
      "learning_rate": 4.9738119192539326e-05,
      "loss": 0.8993,
      "step": 3454
    },
    {
      "epoch": 0.04900893746609896,
      "grad_norm": 2.859375,
      "learning_rate": 4.973779654885358e-05,
      "loss": 0.9413,
      "step": 3456
    },
    {
      "epoch": 0.049037299119725174,
      "grad_norm": 3.21875,
      "learning_rate": 4.9737473707585394e-05,
      "loss": 0.9562,
      "step": 3458
    },
    {
      "epoch": 0.04906566077335139,
      "grad_norm": 3.328125,
      "learning_rate": 4.9737150668737356e-05,
      "loss": 0.8997,
      "step": 3460
    },
    {
      "epoch": 0.04909402242697761,
      "grad_norm": 3.390625,
      "learning_rate": 4.973682743231205e-05,
      "loss": 0.9577,
      "step": 3462
    },
    {
      "epoch": 0.04912238408060382,
      "grad_norm": 3.203125,
      "learning_rate": 4.973650399831206e-05,
      "loss": 0.9161,
      "step": 3464
    },
    {
      "epoch": 0.049150745734230034,
      "grad_norm": 3.0,
      "learning_rate": 4.973618036673995e-05,
      "loss": 0.9951,
      "step": 3466
    },
    {
      "epoch": 0.04917910738785625,
      "grad_norm": 2.90625,
      "learning_rate": 4.973585653759832e-05,
      "loss": 0.9502,
      "step": 3468
    },
    {
      "epoch": 0.04920746904148247,
      "grad_norm": 3.03125,
      "learning_rate": 4.973553251088976e-05,
      "loss": 0.8939,
      "step": 3470
    },
    {
      "epoch": 0.04923583069510868,
      "grad_norm": 3.21875,
      "learning_rate": 4.973520828661684e-05,
      "loss": 0.9495,
      "step": 3472
    },
    {
      "epoch": 0.04926419234873489,
      "grad_norm": 2.953125,
      "learning_rate": 4.973488386478218e-05,
      "loss": 0.9154,
      "step": 3474
    },
    {
      "epoch": 0.04929255400236111,
      "grad_norm": 3.171875,
      "learning_rate": 4.9734559245388336e-05,
      "loss": 0.8997,
      "step": 3476
    },
    {
      "epoch": 0.04932091565598732,
      "grad_norm": 2.921875,
      "learning_rate": 4.973423442843793e-05,
      "loss": 0.8961,
      "step": 3478
    },
    {
      "epoch": 0.04934927730961354,
      "grad_norm": 2.828125,
      "learning_rate": 4.9733909413933535e-05,
      "loss": 0.9454,
      "step": 3480
    },
    {
      "epoch": 0.04937763896323975,
      "grad_norm": 3.453125,
      "learning_rate": 4.973358420187776e-05,
      "loss": 0.9108,
      "step": 3482
    },
    {
      "epoch": 0.04940600061686597,
      "grad_norm": 3.1875,
      "learning_rate": 4.9733258792273194e-05,
      "loss": 0.9783,
      "step": 3484
    },
    {
      "epoch": 0.04943436227049218,
      "grad_norm": 2.90625,
      "learning_rate": 4.973293318512244e-05,
      "loss": 0.9166,
      "step": 3486
    },
    {
      "epoch": 0.04946272392411839,
      "grad_norm": 3.125,
      "learning_rate": 4.973260738042811e-05,
      "loss": 0.9123,
      "step": 3488
    },
    {
      "epoch": 0.04949108557774461,
      "grad_norm": 3.25,
      "learning_rate": 4.973228137819278e-05,
      "loss": 0.9085,
      "step": 3490
    },
    {
      "epoch": 0.049519447231370826,
      "grad_norm": 2.890625,
      "learning_rate": 4.9731955178419075e-05,
      "loss": 0.9062,
      "step": 3492
    },
    {
      "epoch": 0.04954780888499704,
      "grad_norm": 3.28125,
      "learning_rate": 4.97316287811096e-05,
      "loss": 0.9447,
      "step": 3494
    },
    {
      "epoch": 0.04957617053862325,
      "grad_norm": 3.125,
      "learning_rate": 4.973130218626695e-05,
      "loss": 0.9568,
      "step": 3496
    },
    {
      "epoch": 0.049604532192249466,
      "grad_norm": 3.5,
      "learning_rate": 4.973097539389374e-05,
      "loss": 0.933,
      "step": 3498
    },
    {
      "epoch": 0.049632893845875686,
      "grad_norm": 3.0,
      "learning_rate": 4.973064840399259e-05,
      "loss": 0.9869,
      "step": 3500
    },
    {
      "epoch": 0.0496612554995019,
      "grad_norm": 2.875,
      "learning_rate": 4.973032121656609e-05,
      "loss": 0.9109,
      "step": 3502
    },
    {
      "epoch": 0.04968961715312811,
      "grad_norm": 3.296875,
      "learning_rate": 4.9729993831616874e-05,
      "loss": 0.9435,
      "step": 3504
    },
    {
      "epoch": 0.049717978806754326,
      "grad_norm": 3.25,
      "learning_rate": 4.972966624914753e-05,
      "loss": 0.9058,
      "step": 3506
    },
    {
      "epoch": 0.04974634046038054,
      "grad_norm": 3.296875,
      "learning_rate": 4.972933846916072e-05,
      "loss": 0.9087,
      "step": 3508
    },
    {
      "epoch": 0.04977470211400676,
      "grad_norm": 3.28125,
      "learning_rate": 4.972901049165901e-05,
      "loss": 0.9131,
      "step": 3510
    },
    {
      "epoch": 0.04980306376763297,
      "grad_norm": 3.09375,
      "learning_rate": 4.972868231664505e-05,
      "loss": 0.9046,
      "step": 3512
    },
    {
      "epoch": 0.049831425421259186,
      "grad_norm": 3.15625,
      "learning_rate": 4.972835394412146e-05,
      "loss": 0.9063,
      "step": 3514
    },
    {
      "epoch": 0.0498597870748854,
      "grad_norm": 2.984375,
      "learning_rate": 4.972802537409085e-05,
      "loss": 0.939,
      "step": 3516
    },
    {
      "epoch": 0.04988814872851162,
      "grad_norm": 3.484375,
      "learning_rate": 4.972769660655586e-05,
      "loss": 0.9446,
      "step": 3518
    },
    {
      "epoch": 0.04991651038213783,
      "grad_norm": 3.0,
      "learning_rate": 4.972736764151911e-05,
      "loss": 0.9604,
      "step": 3520
    },
    {
      "epoch": 0.049944872035764046,
      "grad_norm": 2.921875,
      "learning_rate": 4.972703847898321e-05,
      "loss": 0.913,
      "step": 3522
    },
    {
      "epoch": 0.04997323368939026,
      "grad_norm": 3.078125,
      "learning_rate": 4.972670911895082e-05,
      "loss": 0.9591,
      "step": 3524
    },
    {
      "epoch": 0.05000159534301647,
      "grad_norm": 3.171875,
      "learning_rate": 4.972637956142455e-05,
      "loss": 0.9043,
      "step": 3526
    },
    {
      "epoch": 0.05002995699664269,
      "grad_norm": 3.234375,
      "learning_rate": 4.972604980640704e-05,
      "loss": 0.8965,
      "step": 3528
    },
    {
      "epoch": 0.050058318650268906,
      "grad_norm": 3.109375,
      "learning_rate": 4.972571985390091e-05,
      "loss": 0.9392,
      "step": 3530
    },
    {
      "epoch": 0.05008668030389512,
      "grad_norm": 2.875,
      "learning_rate": 4.972538970390882e-05,
      "loss": 0.912,
      "step": 3532
    },
    {
      "epoch": 0.05011504195752133,
      "grad_norm": 3.390625,
      "learning_rate": 4.9725059356433386e-05,
      "loss": 0.9663,
      "step": 3534
    },
    {
      "epoch": 0.050143403611147545,
      "grad_norm": 2.78125,
      "learning_rate": 4.972472881147725e-05,
      "loss": 0.8937,
      "step": 3536
    },
    {
      "epoch": 0.050171765264773766,
      "grad_norm": 3.421875,
      "learning_rate": 4.972439806904307e-05,
      "loss": 0.9355,
      "step": 3538
    },
    {
      "epoch": 0.05020012691839998,
      "grad_norm": 3.671875,
      "learning_rate": 4.9724067129133456e-05,
      "loss": 0.8814,
      "step": 3540
    },
    {
      "epoch": 0.05022848857202619,
      "grad_norm": 3.234375,
      "learning_rate": 4.9723735991751076e-05,
      "loss": 0.9808,
      "step": 3542
    },
    {
      "epoch": 0.050256850225652405,
      "grad_norm": 3.34375,
      "learning_rate": 4.972340465689857e-05,
      "loss": 0.9412,
      "step": 3544
    },
    {
      "epoch": 0.05028521187927862,
      "grad_norm": 3.109375,
      "learning_rate": 4.972307312457858e-05,
      "loss": 0.9444,
      "step": 3546
    },
    {
      "epoch": 0.05031357353290484,
      "grad_norm": 3.296875,
      "learning_rate": 4.972274139479376e-05,
      "loss": 0.9428,
      "step": 3548
    },
    {
      "epoch": 0.05034193518653105,
      "grad_norm": 3.0625,
      "learning_rate": 4.972240946754675e-05,
      "loss": 0.9166,
      "step": 3550
    },
    {
      "epoch": 0.050370296840157265,
      "grad_norm": 3.203125,
      "learning_rate": 4.9722077342840204e-05,
      "loss": 0.9303,
      "step": 3552
    },
    {
      "epoch": 0.05039865849378348,
      "grad_norm": 3.234375,
      "learning_rate": 4.9721745020676784e-05,
      "loss": 0.9084,
      "step": 3554
    },
    {
      "epoch": 0.05042702014740969,
      "grad_norm": 2.828125,
      "learning_rate": 4.972141250105914e-05,
      "loss": 0.9475,
      "step": 3556
    },
    {
      "epoch": 0.05045538180103591,
      "grad_norm": 3.140625,
      "learning_rate": 4.9721079783989915e-05,
      "loss": 0.957,
      "step": 3558
    },
    {
      "epoch": 0.050483743454662125,
      "grad_norm": 2.90625,
      "learning_rate": 4.9720746869471784e-05,
      "loss": 0.9195,
      "step": 3560
    },
    {
      "epoch": 0.05051210510828834,
      "grad_norm": 3.03125,
      "learning_rate": 4.97204137575074e-05,
      "loss": 0.8872,
      "step": 3562
    },
    {
      "epoch": 0.05054046676191455,
      "grad_norm": 3.453125,
      "learning_rate": 4.972008044809941e-05,
      "loss": 0.8812,
      "step": 3564
    },
    {
      "epoch": 0.050568828415540765,
      "grad_norm": 3.53125,
      "learning_rate": 4.9719746941250506e-05,
      "loss": 0.9426,
      "step": 3566
    },
    {
      "epoch": 0.050597190069166985,
      "grad_norm": 3.359375,
      "learning_rate": 4.971941323696333e-05,
      "loss": 0.95,
      "step": 3568
    },
    {
      "epoch": 0.0506255517227932,
      "grad_norm": 3.453125,
      "learning_rate": 4.971907933524055e-05,
      "loss": 0.9044,
      "step": 3570
    },
    {
      "epoch": 0.05065391337641941,
      "grad_norm": 3.140625,
      "learning_rate": 4.971874523608483e-05,
      "loss": 0.9305,
      "step": 3572
    },
    {
      "epoch": 0.050682275030045625,
      "grad_norm": 3.125,
      "learning_rate": 4.971841093949885e-05,
      "loss": 0.8895,
      "step": 3574
    },
    {
      "epoch": 0.050710636683671845,
      "grad_norm": 3.015625,
      "learning_rate": 4.9718076445485264e-05,
      "loss": 0.9302,
      "step": 3576
    },
    {
      "epoch": 0.05073899833729806,
      "grad_norm": 3.03125,
      "learning_rate": 4.971774175404675e-05,
      "loss": 0.9504,
      "step": 3578
    },
    {
      "epoch": 0.05076735999092427,
      "grad_norm": 3.171875,
      "learning_rate": 4.9717406865186e-05,
      "loss": 0.9453,
      "step": 3580
    },
    {
      "epoch": 0.050795721644550484,
      "grad_norm": 3.0,
      "learning_rate": 4.9717071778905667e-05,
      "loss": 0.8936,
      "step": 3582
    },
    {
      "epoch": 0.0508240832981767,
      "grad_norm": 3.609375,
      "learning_rate": 4.9716736495208424e-05,
      "loss": 0.9045,
      "step": 3584
    },
    {
      "epoch": 0.05085244495180292,
      "grad_norm": 2.953125,
      "learning_rate": 4.9716401014096966e-05,
      "loss": 0.9282,
      "step": 3586
    },
    {
      "epoch": 0.05088080660542913,
      "grad_norm": 3.515625,
      "learning_rate": 4.971606533557396e-05,
      "loss": 0.9448,
      "step": 3588
    },
    {
      "epoch": 0.050909168259055344,
      "grad_norm": 2.8125,
      "learning_rate": 4.97157294596421e-05,
      "loss": 0.8974,
      "step": 3590
    },
    {
      "epoch": 0.05093752991268156,
      "grad_norm": 3.109375,
      "learning_rate": 4.9715393386304056e-05,
      "loss": 0.9332,
      "step": 3592
    },
    {
      "epoch": 0.05096589156630777,
      "grad_norm": 3.109375,
      "learning_rate": 4.971505711556251e-05,
      "loss": 0.8801,
      "step": 3594
    },
    {
      "epoch": 0.05099425321993399,
      "grad_norm": 3.25,
      "learning_rate": 4.971472064742016e-05,
      "loss": 0.9357,
      "step": 3596
    },
    {
      "epoch": 0.051022614873560204,
      "grad_norm": 3.078125,
      "learning_rate": 4.9714383981879684e-05,
      "loss": 0.9387,
      "step": 3598
    },
    {
      "epoch": 0.05105097652718642,
      "grad_norm": 2.859375,
      "learning_rate": 4.971404711894378e-05,
      "loss": 0.9277,
      "step": 3600
    },
    {
      "epoch": 0.05107933818081263,
      "grad_norm": 3.4375,
      "learning_rate": 4.971371005861514e-05,
      "loss": 0.901,
      "step": 3602
    },
    {
      "epoch": 0.051107699834438844,
      "grad_norm": 2.96875,
      "learning_rate": 4.971337280089643e-05,
      "loss": 0.9528,
      "step": 3604
    },
    {
      "epoch": 0.051136061488065064,
      "grad_norm": 3.171875,
      "learning_rate": 4.971303534579038e-05,
      "loss": 0.9334,
      "step": 3606
    },
    {
      "epoch": 0.05116442314169128,
      "grad_norm": 2.859375,
      "learning_rate": 4.971269769329966e-05,
      "loss": 0.8398,
      "step": 3608
    },
    {
      "epoch": 0.05119278479531749,
      "grad_norm": 3.046875,
      "learning_rate": 4.9712359843426975e-05,
      "loss": 0.9769,
      "step": 3610
    },
    {
      "epoch": 0.051221146448943704,
      "grad_norm": 2.953125,
      "learning_rate": 4.971202179617502e-05,
      "loss": 0.9377,
      "step": 3612
    },
    {
      "epoch": 0.05124950810256992,
      "grad_norm": 3.359375,
      "learning_rate": 4.9711683551546514e-05,
      "loss": 0.985,
      "step": 3614
    },
    {
      "epoch": 0.05127786975619614,
      "grad_norm": 3.046875,
      "learning_rate": 4.9711345109544124e-05,
      "loss": 0.9326,
      "step": 3616
    },
    {
      "epoch": 0.05130623140982235,
      "grad_norm": 3.3125,
      "learning_rate": 4.971100647017058e-05,
      "loss": 0.8943,
      "step": 3618
    },
    {
      "epoch": 0.051334593063448564,
      "grad_norm": 3.40625,
      "learning_rate": 4.9710667633428587e-05,
      "loss": 0.9193,
      "step": 3620
    },
    {
      "epoch": 0.05136295471707478,
      "grad_norm": 2.90625,
      "learning_rate": 4.9710328599320846e-05,
      "loss": 0.9574,
      "step": 3622
    },
    {
      "epoch": 0.051391316370701,
      "grad_norm": 3.3125,
      "learning_rate": 4.9709989367850044e-05,
      "loss": 0.9447,
      "step": 3624
    },
    {
      "epoch": 0.05141967802432721,
      "grad_norm": 3.25,
      "learning_rate": 4.970964993901892e-05,
      "loss": 0.8921,
      "step": 3626
    },
    {
      "epoch": 0.051448039677953424,
      "grad_norm": 3.125,
      "learning_rate": 4.970931031283017e-05,
      "loss": 0.9269,
      "step": 3628
    },
    {
      "epoch": 0.05147640133157964,
      "grad_norm": 3.453125,
      "learning_rate": 4.9708970489286514e-05,
      "loss": 0.9078,
      "step": 3630
    },
    {
      "epoch": 0.05150476298520585,
      "grad_norm": 3.15625,
      "learning_rate": 4.970863046839066e-05,
      "loss": 0.8863,
      "step": 3632
    },
    {
      "epoch": 0.05153312463883207,
      "grad_norm": 3.140625,
      "learning_rate": 4.970829025014533e-05,
      "loss": 0.9152,
      "step": 3634
    },
    {
      "epoch": 0.05156148629245828,
      "grad_norm": 3.8125,
      "learning_rate": 4.970794983455324e-05,
      "loss": 0.9653,
      "step": 3636
    },
    {
      "epoch": 0.0515898479460845,
      "grad_norm": 3.375,
      "learning_rate": 4.97076092216171e-05,
      "loss": 0.8976,
      "step": 3638
    },
    {
      "epoch": 0.05161820959971071,
      "grad_norm": 2.796875,
      "learning_rate": 4.970726841133963e-05,
      "loss": 0.8729,
      "step": 3640
    },
    {
      "epoch": 0.05164657125333692,
      "grad_norm": 3.1875,
      "learning_rate": 4.9706927403723574e-05,
      "loss": 0.98,
      "step": 3642
    },
    {
      "epoch": 0.05167493290696314,
      "grad_norm": 3.140625,
      "learning_rate": 4.9706586198771634e-05,
      "loss": 0.9131,
      "step": 3644
    },
    {
      "epoch": 0.051703294560589357,
      "grad_norm": 2.859375,
      "learning_rate": 4.970624479648654e-05,
      "loss": 0.8685,
      "step": 3646
    },
    {
      "epoch": 0.05173165621421557,
      "grad_norm": 2.984375,
      "learning_rate": 4.9705903196871026e-05,
      "loss": 0.8985,
      "step": 3648
    },
    {
      "epoch": 0.05176001786784178,
      "grad_norm": 2.828125,
      "learning_rate": 4.970556139992781e-05,
      "loss": 0.9154,
      "step": 3650
    },
    {
      "epoch": 0.051788379521467996,
      "grad_norm": 3.125,
      "learning_rate": 4.9705219405659635e-05,
      "loss": 0.9176,
      "step": 3652
    },
    {
      "epoch": 0.051816741175094216,
      "grad_norm": 3.4375,
      "learning_rate": 4.9704877214069223e-05,
      "loss": 0.9383,
      "step": 3654
    },
    {
      "epoch": 0.05184510282872043,
      "grad_norm": 3.078125,
      "learning_rate": 4.97045348251593e-05,
      "loss": 0.922,
      "step": 3656
    },
    {
      "epoch": 0.05187346448234664,
      "grad_norm": 3.078125,
      "learning_rate": 4.970419223893262e-05,
      "loss": 0.8998,
      "step": 3658
    },
    {
      "epoch": 0.051901826135972856,
      "grad_norm": 3.125,
      "learning_rate": 4.9703849455391906e-05,
      "loss": 0.9507,
      "step": 3660
    },
    {
      "epoch": 0.05193018778959907,
      "grad_norm": 2.828125,
      "learning_rate": 4.9703506474539894e-05,
      "loss": 0.8821,
      "step": 3662
    },
    {
      "epoch": 0.05195854944322529,
      "grad_norm": 3.640625,
      "learning_rate": 4.970316329637933e-05,
      "loss": 0.9471,
      "step": 3664
    },
    {
      "epoch": 0.0519869110968515,
      "grad_norm": 3.328125,
      "learning_rate": 4.9702819920912956e-05,
      "loss": 0.9287,
      "step": 3666
    },
    {
      "epoch": 0.052015272750477716,
      "grad_norm": 3.1875,
      "learning_rate": 4.970247634814351e-05,
      "loss": 0.9355,
      "step": 3668
    },
    {
      "epoch": 0.05204363440410393,
      "grad_norm": 3.125,
      "learning_rate": 4.970213257807374e-05,
      "loss": 0.9212,
      "step": 3670
    },
    {
      "epoch": 0.05207199605773015,
      "grad_norm": 3.140625,
      "learning_rate": 4.9701788610706384e-05,
      "loss": 0.9113,
      "step": 3672
    },
    {
      "epoch": 0.05210035771135636,
      "grad_norm": 3.625,
      "learning_rate": 4.97014444460442e-05,
      "loss": 0.94,
      "step": 3674
    },
    {
      "epoch": 0.052128719364982576,
      "grad_norm": 2.84375,
      "learning_rate": 4.970110008408993e-05,
      "loss": 0.8996,
      "step": 3676
    },
    {
      "epoch": 0.05215708101860879,
      "grad_norm": 3.3125,
      "learning_rate": 4.970075552484632e-05,
      "loss": 0.95,
      "step": 3678
    },
    {
      "epoch": 0.052185442672235,
      "grad_norm": 3.109375,
      "learning_rate": 4.970041076831614e-05,
      "loss": 0.9065,
      "step": 3680
    },
    {
      "epoch": 0.05221380432586122,
      "grad_norm": 2.9375,
      "learning_rate": 4.9700065814502125e-05,
      "loss": 0.8558,
      "step": 3682
    },
    {
      "epoch": 0.052242165979487436,
      "grad_norm": 2.90625,
      "learning_rate": 4.9699720663407035e-05,
      "loss": 0.9405,
      "step": 3684
    },
    {
      "epoch": 0.05227052763311365,
      "grad_norm": 3.140625,
      "learning_rate": 4.969937531503363e-05,
      "loss": 0.8957,
      "step": 3686
    },
    {
      "epoch": 0.05229888928673986,
      "grad_norm": 2.90625,
      "learning_rate": 4.9699029769384663e-05,
      "loss": 0.9366,
      "step": 3688
    },
    {
      "epoch": 0.052327250940366075,
      "grad_norm": 3.046875,
      "learning_rate": 4.9698684026462904e-05,
      "loss": 0.9426,
      "step": 3690
    },
    {
      "epoch": 0.052355612593992296,
      "grad_norm": 3.09375,
      "learning_rate": 4.9698338086271114e-05,
      "loss": 0.9271,
      "step": 3692
    },
    {
      "epoch": 0.05238397424761851,
      "grad_norm": 3.421875,
      "learning_rate": 4.969799194881204e-05,
      "loss": 0.9229,
      "step": 3694
    },
    {
      "epoch": 0.05241233590124472,
      "grad_norm": 3.078125,
      "learning_rate": 4.969764561408846e-05,
      "loss": 0.9287,
      "step": 3696
    },
    {
      "epoch": 0.052440697554870935,
      "grad_norm": 3.1875,
      "learning_rate": 4.9697299082103134e-05,
      "loss": 0.8985,
      "step": 3698
    },
    {
      "epoch": 0.05246905920849715,
      "grad_norm": 3.15625,
      "learning_rate": 4.969695235285884e-05,
      "loss": 0.8843,
      "step": 3700
    },
    {
      "epoch": 0.05249742086212337,
      "grad_norm": 2.828125,
      "learning_rate": 4.969660542635833e-05,
      "loss": 0.9025,
      "step": 3702
    },
    {
      "epoch": 0.05252578251574958,
      "grad_norm": 2.984375,
      "learning_rate": 4.969625830260439e-05,
      "loss": 0.9403,
      "step": 3704
    },
    {
      "epoch": 0.052554144169375795,
      "grad_norm": 3.234375,
      "learning_rate": 4.969591098159979e-05,
      "loss": 0.9139,
      "step": 3706
    },
    {
      "epoch": 0.05258250582300201,
      "grad_norm": 2.859375,
      "learning_rate": 4.969556346334731e-05,
      "loss": 0.8745,
      "step": 3708
    },
    {
      "epoch": 0.05261086747662822,
      "grad_norm": 3.421875,
      "learning_rate": 4.96952157478497e-05,
      "loss": 0.9437,
      "step": 3710
    },
    {
      "epoch": 0.05263922913025444,
      "grad_norm": 3.234375,
      "learning_rate": 4.969486783510976e-05,
      "loss": 0.924,
      "step": 3712
    },
    {
      "epoch": 0.052667590783880655,
      "grad_norm": 2.828125,
      "learning_rate": 4.9694519725130265e-05,
      "loss": 0.8821,
      "step": 3714
    },
    {
      "epoch": 0.05269595243750687,
      "grad_norm": 3.203125,
      "learning_rate": 4.9694171417914e-05,
      "loss": 0.8983,
      "step": 3716
    },
    {
      "epoch": 0.05272431409113308,
      "grad_norm": 2.984375,
      "learning_rate": 4.969382291346373e-05,
      "loss": 0.8759,
      "step": 3718
    },
    {
      "epoch": 0.052752675744759295,
      "grad_norm": 3.453125,
      "learning_rate": 4.969347421178226e-05,
      "loss": 0.9228,
      "step": 3720
    },
    {
      "epoch": 0.052781037398385515,
      "grad_norm": 3.375,
      "learning_rate": 4.9693125312872356e-05,
      "loss": 0.9237,
      "step": 3722
    },
    {
      "epoch": 0.05280939905201173,
      "grad_norm": 3.59375,
      "learning_rate": 4.969277621673682e-05,
      "loss": 0.9282,
      "step": 3724
    },
    {
      "epoch": 0.05283776070563794,
      "grad_norm": 2.984375,
      "learning_rate": 4.9692426923378436e-05,
      "loss": 0.9286,
      "step": 3726
    },
    {
      "epoch": 0.052866122359264155,
      "grad_norm": 3.109375,
      "learning_rate": 4.969207743279998e-05,
      "loss": 0.8906,
      "step": 3728
    },
    {
      "epoch": 0.052894484012890375,
      "grad_norm": 3.328125,
      "learning_rate": 4.9691727745004255e-05,
      "loss": 0.9362,
      "step": 3730
    },
    {
      "epoch": 0.05292284566651659,
      "grad_norm": 3.0,
      "learning_rate": 4.9691377859994056e-05,
      "loss": 0.921,
      "step": 3732
    },
    {
      "epoch": 0.0529512073201428,
      "grad_norm": 3.046875,
      "learning_rate": 4.9691027777772184e-05,
      "loss": 0.901,
      "step": 3734
    },
    {
      "epoch": 0.052979568973769015,
      "grad_norm": 3.125,
      "learning_rate": 4.969067749834142e-05,
      "loss": 0.9084,
      "step": 3736
    },
    {
      "epoch": 0.05300793062739523,
      "grad_norm": 3.015625,
      "learning_rate": 4.9690327021704563e-05,
      "loss": 0.9764,
      "step": 3738
    },
    {
      "epoch": 0.05303629228102145,
      "grad_norm": 3.328125,
      "learning_rate": 4.968997634786442e-05,
      "loss": 0.9496,
      "step": 3740
    },
    {
      "epoch": 0.05306465393464766,
      "grad_norm": 3.546875,
      "learning_rate": 4.9689625476823795e-05,
      "loss": 0.8638,
      "step": 3742
    },
    {
      "epoch": 0.053093015588273874,
      "grad_norm": 2.8125,
      "learning_rate": 4.968927440858548e-05,
      "loss": 0.9089,
      "step": 3744
    },
    {
      "epoch": 0.05312137724190009,
      "grad_norm": 3.859375,
      "learning_rate": 4.968892314315228e-05,
      "loss": 0.9581,
      "step": 3746
    },
    {
      "epoch": 0.0531497388955263,
      "grad_norm": 3.6875,
      "learning_rate": 4.968857168052701e-05,
      "loss": 0.9375,
      "step": 3748
    },
    {
      "epoch": 0.05317810054915252,
      "grad_norm": 3.015625,
      "learning_rate": 4.968822002071247e-05,
      "loss": 0.9311,
      "step": 3750
    },
    {
      "epoch": 0.053206462202778734,
      "grad_norm": 3.3125,
      "learning_rate": 4.9687868163711474e-05,
      "loss": 0.9389,
      "step": 3752
    },
    {
      "epoch": 0.05323482385640495,
      "grad_norm": 2.921875,
      "learning_rate": 4.9687516109526814e-05,
      "loss": 0.9258,
      "step": 3754
    },
    {
      "epoch": 0.05326318551003116,
      "grad_norm": 2.671875,
      "learning_rate": 4.968716385816133e-05,
      "loss": 0.9011,
      "step": 3756
    },
    {
      "epoch": 0.053291547163657374,
      "grad_norm": 3.171875,
      "learning_rate": 4.968681140961782e-05,
      "loss": 0.908,
      "step": 3758
    },
    {
      "epoch": 0.053319908817283594,
      "grad_norm": 3.21875,
      "learning_rate": 4.9686458763899094e-05,
      "loss": 0.9229,
      "step": 3760
    },
    {
      "epoch": 0.05334827047090981,
      "grad_norm": 3.40625,
      "learning_rate": 4.968610592100797e-05,
      "loss": 0.913,
      "step": 3762
    },
    {
      "epoch": 0.05337663212453602,
      "grad_norm": 3.171875,
      "learning_rate": 4.968575288094729e-05,
      "loss": 0.9435,
      "step": 3764
    },
    {
      "epoch": 0.053404993778162234,
      "grad_norm": 3.09375,
      "learning_rate": 4.968539964371984e-05,
      "loss": 0.9707,
      "step": 3766
    },
    {
      "epoch": 0.05343335543178845,
      "grad_norm": 2.875,
      "learning_rate": 4.968504620932847e-05,
      "loss": 0.9046,
      "step": 3768
    },
    {
      "epoch": 0.05346171708541467,
      "grad_norm": 2.84375,
      "learning_rate": 4.968469257777598e-05,
      "loss": 0.9455,
      "step": 3770
    },
    {
      "epoch": 0.05349007873904088,
      "grad_norm": 2.765625,
      "learning_rate": 4.96843387490652e-05,
      "loss": 0.9176,
      "step": 3772
    },
    {
      "epoch": 0.053518440392667094,
      "grad_norm": 2.9375,
      "learning_rate": 4.968398472319897e-05,
      "loss": 0.8523,
      "step": 3774
    },
    {
      "epoch": 0.05354680204629331,
      "grad_norm": 2.890625,
      "learning_rate": 4.96836305001801e-05,
      "loss": 0.8885,
      "step": 3776
    },
    {
      "epoch": 0.05357516369991953,
      "grad_norm": 3.171875,
      "learning_rate": 4.968327608001143e-05,
      "loss": 0.9748,
      "step": 3778
    },
    {
      "epoch": 0.05360352535354574,
      "grad_norm": 3.3125,
      "learning_rate": 4.968292146269579e-05,
      "loss": 0.9641,
      "step": 3780
    },
    {
      "epoch": 0.053631887007171954,
      "grad_norm": 3.375,
      "learning_rate": 4.9682566648236007e-05,
      "loss": 0.9717,
      "step": 3782
    },
    {
      "epoch": 0.05366024866079817,
      "grad_norm": 3.15625,
      "learning_rate": 4.968221163663491e-05,
      "loss": 0.9269,
      "step": 3784
    },
    {
      "epoch": 0.05368861031442438,
      "grad_norm": 3.578125,
      "learning_rate": 4.968185642789536e-05,
      "loss": 0.8923,
      "step": 3786
    },
    {
      "epoch": 0.0537169719680506,
      "grad_norm": 2.875,
      "learning_rate": 4.968150102202016e-05,
      "loss": 0.9235,
      "step": 3788
    },
    {
      "epoch": 0.053745333621676813,
      "grad_norm": 2.96875,
      "learning_rate": 4.9681145419012173e-05,
      "loss": 0.9232,
      "step": 3790
    },
    {
      "epoch": 0.05377369527530303,
      "grad_norm": 3.03125,
      "learning_rate": 4.968078961887423e-05,
      "loss": 0.9261,
      "step": 3792
    },
    {
      "epoch": 0.05380205692892924,
      "grad_norm": 3.40625,
      "learning_rate": 4.968043362160917e-05,
      "loss": 0.9291,
      "step": 3794
    },
    {
      "epoch": 0.05383041858255545,
      "grad_norm": 3.171875,
      "learning_rate": 4.9680077427219853e-05,
      "loss": 0.916,
      "step": 3796
    },
    {
      "epoch": 0.05385878023618167,
      "grad_norm": 3.390625,
      "learning_rate": 4.9679721035709095e-05,
      "loss": 0.9227,
      "step": 3798
    },
    {
      "epoch": 0.05388714188980789,
      "grad_norm": 3.65625,
      "learning_rate": 4.967936444707977e-05,
      "loss": 0.9122,
      "step": 3800
    },
    {
      "epoch": 0.0539155035434341,
      "grad_norm": 3.21875,
      "learning_rate": 4.967900766133471e-05,
      "loss": 0.9252,
      "step": 3802
    },
    {
      "epoch": 0.05394386519706031,
      "grad_norm": 2.921875,
      "learning_rate": 4.967865067847677e-05,
      "loss": 0.9387,
      "step": 3804
    },
    {
      "epoch": 0.053972226850686526,
      "grad_norm": 3.046875,
      "learning_rate": 4.967829349850881e-05,
      "loss": 0.9127,
      "step": 3806
    },
    {
      "epoch": 0.054000588504312746,
      "grad_norm": 3.453125,
      "learning_rate": 4.9677936121433665e-05,
      "loss": 0.8928,
      "step": 3808
    },
    {
      "epoch": 0.05402895015793896,
      "grad_norm": 2.890625,
      "learning_rate": 4.96775785472542e-05,
      "loss": 0.9075,
      "step": 3810
    },
    {
      "epoch": 0.05405731181156517,
      "grad_norm": 2.703125,
      "learning_rate": 4.967722077597327e-05,
      "loss": 0.8678,
      "step": 3812
    },
    {
      "epoch": 0.054085673465191386,
      "grad_norm": 2.78125,
      "learning_rate": 4.9676862807593726e-05,
      "loss": 0.9472,
      "step": 3814
    },
    {
      "epoch": 0.0541140351188176,
      "grad_norm": 2.921875,
      "learning_rate": 4.967650464211844e-05,
      "loss": 0.9258,
      "step": 3816
    },
    {
      "epoch": 0.05414239677244382,
      "grad_norm": 3.046875,
      "learning_rate": 4.9676146279550264e-05,
      "loss": 0.9811,
      "step": 3818
    },
    {
      "epoch": 0.05417075842607003,
      "grad_norm": 2.828125,
      "learning_rate": 4.9675787719892056e-05,
      "loss": 0.8911,
      "step": 3820
    },
    {
      "epoch": 0.054199120079696246,
      "grad_norm": 3.09375,
      "learning_rate": 4.967542896314669e-05,
      "loss": 0.876,
      "step": 3822
    },
    {
      "epoch": 0.05422748173332246,
      "grad_norm": 2.953125,
      "learning_rate": 4.967507000931702e-05,
      "loss": 0.8905,
      "step": 3824
    },
    {
      "epoch": 0.05425584338694868,
      "grad_norm": 3.296875,
      "learning_rate": 4.967471085840594e-05,
      "loss": 0.9459,
      "step": 3826
    },
    {
      "epoch": 0.05428420504057489,
      "grad_norm": 2.875,
      "learning_rate": 4.967435151041629e-05,
      "loss": 0.9338,
      "step": 3828
    },
    {
      "epoch": 0.054312566694201106,
      "grad_norm": 3.078125,
      "learning_rate": 4.9673991965350935e-05,
      "loss": 0.8941,
      "step": 3830
    },
    {
      "epoch": 0.05434092834782732,
      "grad_norm": 2.765625,
      "learning_rate": 4.967363222321277e-05,
      "loss": 0.8853,
      "step": 3832
    },
    {
      "epoch": 0.05436929000145353,
      "grad_norm": 3.03125,
      "learning_rate": 4.967327228400466e-05,
      "loss": 0.9733,
      "step": 3834
    },
    {
      "epoch": 0.05439765165507975,
      "grad_norm": 2.859375,
      "learning_rate": 4.967291214772948e-05,
      "loss": 0.9177,
      "step": 3836
    },
    {
      "epoch": 0.054426013308705966,
      "grad_norm": 2.96875,
      "learning_rate": 4.96725518143901e-05,
      "loss": 0.9368,
      "step": 3838
    },
    {
      "epoch": 0.05445437496233218,
      "grad_norm": 2.9375,
      "learning_rate": 4.96721912839894e-05,
      "loss": 0.9043,
      "step": 3840
    },
    {
      "epoch": 0.05448273661595839,
      "grad_norm": 2.671875,
      "learning_rate": 4.967183055653027e-05,
      "loss": 0.9427,
      "step": 3842
    },
    {
      "epoch": 0.054511098269584606,
      "grad_norm": 3.40625,
      "learning_rate": 4.9671469632015574e-05,
      "loss": 0.9332,
      "step": 3844
    },
    {
      "epoch": 0.054539459923210826,
      "grad_norm": 3.234375,
      "learning_rate": 4.967110851044822e-05,
      "loss": 0.9451,
      "step": 3846
    },
    {
      "epoch": 0.05456782157683704,
      "grad_norm": 3.609375,
      "learning_rate": 4.967074719183107e-05,
      "loss": 0.8989,
      "step": 3848
    },
    {
      "epoch": 0.05459618323046325,
      "grad_norm": 3.5,
      "learning_rate": 4.9670385676167005e-05,
      "loss": 0.9446,
      "step": 3850
    },
    {
      "epoch": 0.054624544884089465,
      "grad_norm": 2.796875,
      "learning_rate": 4.967002396345894e-05,
      "loss": 0.831,
      "step": 3852
    },
    {
      "epoch": 0.05465290653771568,
      "grad_norm": 2.984375,
      "learning_rate": 4.966966205370974e-05,
      "loss": 0.9183,
      "step": 3854
    },
    {
      "epoch": 0.0546812681913419,
      "grad_norm": 3.125,
      "learning_rate": 4.966929994692231e-05,
      "loss": 0.9089,
      "step": 3856
    },
    {
      "epoch": 0.05470962984496811,
      "grad_norm": 4.03125,
      "learning_rate": 4.966893764309953e-05,
      "loss": 0.9306,
      "step": 3858
    },
    {
      "epoch": 0.054737991498594325,
      "grad_norm": 3.203125,
      "learning_rate": 4.96685751422443e-05,
      "loss": 0.9314,
      "step": 3860
    },
    {
      "epoch": 0.05476635315222054,
      "grad_norm": 3.203125,
      "learning_rate": 4.966821244435952e-05,
      "loss": 0.8726,
      "step": 3862
    },
    {
      "epoch": 0.05479471480584675,
      "grad_norm": 3.0625,
      "learning_rate": 4.9667849549448074e-05,
      "loss": 0.9141,
      "step": 3864
    },
    {
      "epoch": 0.05482307645947297,
      "grad_norm": 3.75,
      "learning_rate": 4.966748645751287e-05,
      "loss": 0.9364,
      "step": 3866
    },
    {
      "epoch": 0.054851438113099185,
      "grad_norm": 3.28125,
      "learning_rate": 4.966712316855681e-05,
      "loss": 0.8801,
      "step": 3868
    },
    {
      "epoch": 0.0548797997667254,
      "grad_norm": 3.75,
      "learning_rate": 4.966675968258279e-05,
      "loss": 0.9289,
      "step": 3870
    },
    {
      "epoch": 0.05490816142035161,
      "grad_norm": 3.3125,
      "learning_rate": 4.966639599959372e-05,
      "loss": 0.9783,
      "step": 3872
    },
    {
      "epoch": 0.054936523073977825,
      "grad_norm": 3.40625,
      "learning_rate": 4.966603211959249e-05,
      "loss": 0.9019,
      "step": 3874
    },
    {
      "epoch": 0.054964884727604045,
      "grad_norm": 3.296875,
      "learning_rate": 4.966566804258203e-05,
      "loss": 0.9049,
      "step": 3876
    },
    {
      "epoch": 0.05499324638123026,
      "grad_norm": 2.875,
      "learning_rate": 4.966530376856522e-05,
      "loss": 0.9076,
      "step": 3878
    },
    {
      "epoch": 0.05502160803485647,
      "grad_norm": 3.03125,
      "learning_rate": 4.966493929754499e-05,
      "loss": 0.9655,
      "step": 3880
    },
    {
      "epoch": 0.055049969688482685,
      "grad_norm": 2.875,
      "learning_rate": 4.966457462952424e-05,
      "loss": 0.8643,
      "step": 3882
    },
    {
      "epoch": 0.055078331342108905,
      "grad_norm": 2.875,
      "learning_rate": 4.96642097645059e-05,
      "loss": 0.8833,
      "step": 3884
    },
    {
      "epoch": 0.05510669299573512,
      "grad_norm": 2.875,
      "learning_rate": 4.966384470249286e-05,
      "loss": 0.8916,
      "step": 3886
    },
    {
      "epoch": 0.05513505464936133,
      "grad_norm": 3.109375,
      "learning_rate": 4.9663479443488046e-05,
      "loss": 0.9021,
      "step": 3888
    },
    {
      "epoch": 0.055163416302987545,
      "grad_norm": 2.875,
      "learning_rate": 4.966311398749438e-05,
      "loss": 0.8903,
      "step": 3890
    },
    {
      "epoch": 0.05519177795661376,
      "grad_norm": 3.140625,
      "learning_rate": 4.9662748334514784e-05,
      "loss": 0.908,
      "step": 3892
    },
    {
      "epoch": 0.05522013961023998,
      "grad_norm": 3.703125,
      "learning_rate": 4.966238248455216e-05,
      "loss": 0.9403,
      "step": 3894
    },
    {
      "epoch": 0.05524850126386619,
      "grad_norm": 3.125,
      "learning_rate": 4.966201643760945e-05,
      "loss": 0.9352,
      "step": 3896
    },
    {
      "epoch": 0.055276862917492404,
      "grad_norm": 2.84375,
      "learning_rate": 4.966165019368957e-05,
      "loss": 0.9373,
      "step": 3898
    },
    {
      "epoch": 0.05530522457111862,
      "grad_norm": 3.5625,
      "learning_rate": 4.966128375279545e-05,
      "loss": 0.9486,
      "step": 3900
    },
    {
      "epoch": 0.05533358622474483,
      "grad_norm": 3.015625,
      "learning_rate": 4.966091711493e-05,
      "loss": 0.8962,
      "step": 3902
    },
    {
      "epoch": 0.05536194787837105,
      "grad_norm": 3.359375,
      "learning_rate": 4.9660550280096164e-05,
      "loss": 0.9285,
      "step": 3904
    },
    {
      "epoch": 0.055390309531997264,
      "grad_norm": 3.5625,
      "learning_rate": 4.966018324829687e-05,
      "loss": 0.8523,
      "step": 3906
    },
    {
      "epoch": 0.05541867118562348,
      "grad_norm": 2.984375,
      "learning_rate": 4.9659816019535045e-05,
      "loss": 0.9025,
      "step": 3908
    },
    {
      "epoch": 0.05544703283924969,
      "grad_norm": 3.15625,
      "learning_rate": 4.9659448593813626e-05,
      "loss": 0.9078,
      "step": 3910
    },
    {
      "epoch": 0.055475394492875904,
      "grad_norm": 3.34375,
      "learning_rate": 4.965908097113555e-05,
      "loss": 0.8764,
      "step": 3912
    },
    {
      "epoch": 0.055503756146502124,
      "grad_norm": 3.296875,
      "learning_rate": 4.9658713151503743e-05,
      "loss": 0.8835,
      "step": 3914
    },
    {
      "epoch": 0.05553211780012834,
      "grad_norm": 2.953125,
      "learning_rate": 4.9658345134921156e-05,
      "loss": 0.9239,
      "step": 3916
    },
    {
      "epoch": 0.05556047945375455,
      "grad_norm": 3.125,
      "learning_rate": 4.9657976921390716e-05,
      "loss": 0.9468,
      "step": 3918
    },
    {
      "epoch": 0.055588841107380764,
      "grad_norm": 3.546875,
      "learning_rate": 4.965760851091538e-05,
      "loss": 0.886,
      "step": 3920
    },
    {
      "epoch": 0.05561720276100698,
      "grad_norm": 3.09375,
      "learning_rate": 4.9657239903498064e-05,
      "loss": 0.9122,
      "step": 3922
    },
    {
      "epoch": 0.0556455644146332,
      "grad_norm": 3.5625,
      "learning_rate": 4.965687109914173e-05,
      "loss": 0.9522,
      "step": 3924
    },
    {
      "epoch": 0.05567392606825941,
      "grad_norm": 3.046875,
      "learning_rate": 4.9656502097849334e-05,
      "loss": 0.9198,
      "step": 3926
    },
    {
      "epoch": 0.055702287721885624,
      "grad_norm": 3.515625,
      "learning_rate": 4.965613289962381e-05,
      "loss": 0.9477,
      "step": 3928
    },
    {
      "epoch": 0.05573064937551184,
      "grad_norm": 2.890625,
      "learning_rate": 4.96557635044681e-05,
      "loss": 0.8735,
      "step": 3930
    },
    {
      "epoch": 0.05575901102913806,
      "grad_norm": 3.125,
      "learning_rate": 4.9655393912385164e-05,
      "loss": 0.9669,
      "step": 3932
    },
    {
      "epoch": 0.05578737268276427,
      "grad_norm": 3.140625,
      "learning_rate": 4.9655024123377956e-05,
      "loss": 0.9301,
      "step": 3934
    },
    {
      "epoch": 0.055815734336390484,
      "grad_norm": 3.03125,
      "learning_rate": 4.965465413744942e-05,
      "loss": 0.8876,
      "step": 3936
    },
    {
      "epoch": 0.0558440959900167,
      "grad_norm": 3.125,
      "learning_rate": 4.965428395460252e-05,
      "loss": 0.931,
      "step": 3938
    },
    {
      "epoch": 0.05587245764364291,
      "grad_norm": 3.140625,
      "learning_rate": 4.965391357484021e-05,
      "loss": 0.9693,
      "step": 3940
    },
    {
      "epoch": 0.05590081929726913,
      "grad_norm": 2.921875,
      "learning_rate": 4.965354299816545e-05,
      "loss": 0.9041,
      "step": 3942
    },
    {
      "epoch": 0.055929180950895344,
      "grad_norm": 3.296875,
      "learning_rate": 4.9653172224581195e-05,
      "loss": 0.9071,
      "step": 3944
    },
    {
      "epoch": 0.05595754260452156,
      "grad_norm": 3.46875,
      "learning_rate": 4.9652801254090405e-05,
      "loss": 0.9234,
      "step": 3946
    },
    {
      "epoch": 0.05598590425814777,
      "grad_norm": 2.984375,
      "learning_rate": 4.965243008669605e-05,
      "loss": 0.9222,
      "step": 3948
    },
    {
      "epoch": 0.05601426591177398,
      "grad_norm": 3.328125,
      "learning_rate": 4.96520587224011e-05,
      "loss": 0.9363,
      "step": 3950
    },
    {
      "epoch": 0.0560426275654002,
      "grad_norm": 3.484375,
      "learning_rate": 4.9651687161208505e-05,
      "loss": 0.9262,
      "step": 3952
    },
    {
      "epoch": 0.05607098921902642,
      "grad_norm": 3.421875,
      "learning_rate": 4.9651315403121246e-05,
      "loss": 0.9375,
      "step": 3954
    },
    {
      "epoch": 0.05609935087265263,
      "grad_norm": 2.796875,
      "learning_rate": 4.9650943448142276e-05,
      "loss": 0.882,
      "step": 3956
    },
    {
      "epoch": 0.05612771252627884,
      "grad_norm": 2.6875,
      "learning_rate": 4.9650571296274584e-05,
      "loss": 0.9223,
      "step": 3958
    },
    {
      "epoch": 0.056156074179905056,
      "grad_norm": 3.25,
      "learning_rate": 4.965019894752113e-05,
      "loss": 0.9141,
      "step": 3960
    },
    {
      "epoch": 0.05618443583353128,
      "grad_norm": 3.328125,
      "learning_rate": 4.9649826401884904e-05,
      "loss": 0.9411,
      "step": 3962
    },
    {
      "epoch": 0.05621279748715749,
      "grad_norm": 3.34375,
      "learning_rate": 4.964945365936886e-05,
      "loss": 0.9714,
      "step": 3964
    },
    {
      "epoch": 0.0562411591407837,
      "grad_norm": 3.046875,
      "learning_rate": 4.964908071997598e-05,
      "loss": 0.9085,
      "step": 3966
    },
    {
      "epoch": 0.056269520794409916,
      "grad_norm": 3.46875,
      "learning_rate": 4.9648707583709266e-05,
      "loss": 0.9553,
      "step": 3968
    },
    {
      "epoch": 0.05629788244803613,
      "grad_norm": 2.734375,
      "learning_rate": 4.964833425057167e-05,
      "loss": 0.9194,
      "step": 3970
    },
    {
      "epoch": 0.05632624410166235,
      "grad_norm": 2.796875,
      "learning_rate": 4.964796072056618e-05,
      "loss": 0.9152,
      "step": 3972
    },
    {
      "epoch": 0.05635460575528856,
      "grad_norm": 3.28125,
      "learning_rate": 4.964758699369579e-05,
      "loss": 0.9586,
      "step": 3974
    },
    {
      "epoch": 0.056382967408914776,
      "grad_norm": 2.78125,
      "learning_rate": 4.9647213069963475e-05,
      "loss": 0.9352,
      "step": 3976
    },
    {
      "epoch": 0.05641132906254099,
      "grad_norm": 2.921875,
      "learning_rate": 4.964683894937223e-05,
      "loss": 0.9609,
      "step": 3978
    },
    {
      "epoch": 0.05643969071616721,
      "grad_norm": 3.375,
      "learning_rate": 4.9646464631925035e-05,
      "loss": 0.8815,
      "step": 3980
    },
    {
      "epoch": 0.05646805236979342,
      "grad_norm": 2.890625,
      "learning_rate": 4.964609011762488e-05,
      "loss": 0.888,
      "step": 3982
    },
    {
      "epoch": 0.056496414023419636,
      "grad_norm": 2.90625,
      "learning_rate": 4.9645715406474766e-05,
      "loss": 0.9233,
      "step": 3984
    },
    {
      "epoch": 0.05652477567704585,
      "grad_norm": 2.90625,
      "learning_rate": 4.9645340498477676e-05,
      "loss": 0.9163,
      "step": 3986
    },
    {
      "epoch": 0.05655313733067206,
      "grad_norm": 2.859375,
      "learning_rate": 4.964496539363661e-05,
      "loss": 0.9275,
      "step": 3988
    },
    {
      "epoch": 0.05658149898429828,
      "grad_norm": 2.515625,
      "learning_rate": 4.9644590091954556e-05,
      "loss": 0.8584,
      "step": 3990
    },
    {
      "epoch": 0.056609860637924496,
      "grad_norm": 2.828125,
      "learning_rate": 4.964421459343452e-05,
      "loss": 0.8798,
      "step": 3992
    },
    {
      "epoch": 0.05663822229155071,
      "grad_norm": 3.15625,
      "learning_rate": 4.96438388980795e-05,
      "loss": 0.9352,
      "step": 3994
    },
    {
      "epoch": 0.05666658394517692,
      "grad_norm": 3.078125,
      "learning_rate": 4.9643463005892495e-05,
      "loss": 0.9005,
      "step": 3996
    },
    {
      "epoch": 0.056694945598803136,
      "grad_norm": 3.46875,
      "learning_rate": 4.96430869168765e-05,
      "loss": 0.9313,
      "step": 3998
    },
    {
      "epoch": 0.056723307252429356,
      "grad_norm": 3.1875,
      "learning_rate": 4.964271063103454e-05,
      "loss": 0.9333,
      "step": 4000
    },
    {
      "epoch": 0.05675166890605557,
      "grad_norm": 3.5,
      "learning_rate": 4.9642334148369595e-05,
      "loss": 0.898,
      "step": 4002
    },
    {
      "epoch": 0.05678003055968178,
      "grad_norm": 3.421875,
      "learning_rate": 4.964195746888469e-05,
      "loss": 0.9046,
      "step": 4004
    },
    {
      "epoch": 0.056808392213307995,
      "grad_norm": 2.953125,
      "learning_rate": 4.964158059258282e-05,
      "loss": 0.9282,
      "step": 4006
    },
    {
      "epoch": 0.05683675386693421,
      "grad_norm": 3.25,
      "learning_rate": 4.964120351946701e-05,
      "loss": 0.9305,
      "step": 4008
    },
    {
      "epoch": 0.05686511552056043,
      "grad_norm": 3.34375,
      "learning_rate": 4.964082624954026e-05,
      "loss": 0.9904,
      "step": 4010
    },
    {
      "epoch": 0.05689347717418664,
      "grad_norm": 2.78125,
      "learning_rate": 4.964044878280558e-05,
      "loss": 0.8854,
      "step": 4012
    },
    {
      "epoch": 0.056921838827812855,
      "grad_norm": 3.140625,
      "learning_rate": 4.9640071119266007e-05,
      "loss": 0.9607,
      "step": 4014
    },
    {
      "epoch": 0.05695020048143907,
      "grad_norm": 3.0625,
      "learning_rate": 4.9639693258924534e-05,
      "loss": 0.9621,
      "step": 4016
    },
    {
      "epoch": 0.05697856213506528,
      "grad_norm": 3.734375,
      "learning_rate": 4.9639315201784185e-05,
      "loss": 0.9136,
      "step": 4018
    },
    {
      "epoch": 0.0570069237886915,
      "grad_norm": 2.921875,
      "learning_rate": 4.963893694784799e-05,
      "loss": 0.8716,
      "step": 4020
    },
    {
      "epoch": 0.057035285442317715,
      "grad_norm": 2.875,
      "learning_rate": 4.9638558497118956e-05,
      "loss": 0.8394,
      "step": 4022
    },
    {
      "epoch": 0.05706364709594393,
      "grad_norm": 3.171875,
      "learning_rate": 4.9638179849600105e-05,
      "loss": 0.9747,
      "step": 4024
    },
    {
      "epoch": 0.05709200874957014,
      "grad_norm": 3.671875,
      "learning_rate": 4.963780100529448e-05,
      "loss": 0.9131,
      "step": 4026
    },
    {
      "epoch": 0.057120370403196355,
      "grad_norm": 3.203125,
      "learning_rate": 4.963742196420509e-05,
      "loss": 0.882,
      "step": 4028
    },
    {
      "epoch": 0.057148732056822575,
      "grad_norm": 3.25,
      "learning_rate": 4.9637042726334975e-05,
      "loss": 0.9566,
      "step": 4030
    },
    {
      "epoch": 0.05717709371044879,
      "grad_norm": 3.296875,
      "learning_rate": 4.963666329168715e-05,
      "loss": 0.9162,
      "step": 4032
    },
    {
      "epoch": 0.057205455364075,
      "grad_norm": 3.21875,
      "learning_rate": 4.963628366026465e-05,
      "loss": 0.962,
      "step": 4034
    },
    {
      "epoch": 0.057233817017701215,
      "grad_norm": 3.21875,
      "learning_rate": 4.9635903832070515e-05,
      "loss": 0.9595,
      "step": 4036
    },
    {
      "epoch": 0.057262178671327435,
      "grad_norm": 3.03125,
      "learning_rate": 4.9635523807107764e-05,
      "loss": 0.9642,
      "step": 4038
    },
    {
      "epoch": 0.05729054032495365,
      "grad_norm": 2.578125,
      "learning_rate": 4.963514358537945e-05,
      "loss": 0.8164,
      "step": 4040
    },
    {
      "epoch": 0.05731890197857986,
      "grad_norm": 3.453125,
      "learning_rate": 4.96347631668886e-05,
      "loss": 0.9387,
      "step": 4042
    },
    {
      "epoch": 0.057347263632206075,
      "grad_norm": 3.125,
      "learning_rate": 4.963438255163825e-05,
      "loss": 0.9071,
      "step": 4044
    },
    {
      "epoch": 0.05737562528583229,
      "grad_norm": 3.265625,
      "learning_rate": 4.963400173963144e-05,
      "loss": 0.8781,
      "step": 4046
    },
    {
      "epoch": 0.05740398693945851,
      "grad_norm": 2.875,
      "learning_rate": 4.963362073087122e-05,
      "loss": 0.9219,
      "step": 4048
    },
    {
      "epoch": 0.05743234859308472,
      "grad_norm": 3.1875,
      "learning_rate": 4.963323952536063e-05,
      "loss": 0.9128,
      "step": 4050
    },
    {
      "epoch": 0.057460710246710935,
      "grad_norm": 3.140625,
      "learning_rate": 4.963285812310271e-05,
      "loss": 0.9502,
      "step": 4052
    },
    {
      "epoch": 0.05748907190033715,
      "grad_norm": 2.96875,
      "learning_rate": 4.9632476524100515e-05,
      "loss": 0.9305,
      "step": 4054
    },
    {
      "epoch": 0.05751743355396336,
      "grad_norm": 3.015625,
      "learning_rate": 4.9632094728357084e-05,
      "loss": 0.9069,
      "step": 4056
    },
    {
      "epoch": 0.05754579520758958,
      "grad_norm": 3.421875,
      "learning_rate": 4.963171273587546e-05,
      "loss": 0.9272,
      "step": 4058
    },
    {
      "epoch": 0.057574156861215794,
      "grad_norm": 3.390625,
      "learning_rate": 4.963133054665871e-05,
      "loss": 0.9138,
      "step": 4060
    },
    {
      "epoch": 0.05760251851484201,
      "grad_norm": 3.203125,
      "learning_rate": 4.963094816070988e-05,
      "loss": 0.966,
      "step": 4062
    },
    {
      "epoch": 0.05763088016846822,
      "grad_norm": 3.140625,
      "learning_rate": 4.963056557803203e-05,
      "loss": 0.9608,
      "step": 4064
    },
    {
      "epoch": 0.057659241822094434,
      "grad_norm": 3.03125,
      "learning_rate": 4.96301827986282e-05,
      "loss": 0.9095,
      "step": 4066
    },
    {
      "epoch": 0.057687603475720654,
      "grad_norm": 3.125,
      "learning_rate": 4.9629799822501465e-05,
      "loss": 0.9448,
      "step": 4068
    },
    {
      "epoch": 0.05771596512934687,
      "grad_norm": 3.046875,
      "learning_rate": 4.962941664965487e-05,
      "loss": 0.9514,
      "step": 4070
    },
    {
      "epoch": 0.05774432678297308,
      "grad_norm": 3.109375,
      "learning_rate": 4.962903328009149e-05,
      "loss": 0.9531,
      "step": 4072
    },
    {
      "epoch": 0.057772688436599294,
      "grad_norm": 3.421875,
      "learning_rate": 4.9628649713814365e-05,
      "loss": 0.878,
      "step": 4074
    },
    {
      "epoch": 0.05780105009022551,
      "grad_norm": 2.96875,
      "learning_rate": 4.962826595082658e-05,
      "loss": 0.899,
      "step": 4076
    },
    {
      "epoch": 0.05782941174385173,
      "grad_norm": 3.59375,
      "learning_rate": 4.962788199113119e-05,
      "loss": 0.884,
      "step": 4078
    },
    {
      "epoch": 0.05785777339747794,
      "grad_norm": 2.828125,
      "learning_rate": 4.962749783473127e-05,
      "loss": 0.902,
      "step": 4080
    },
    {
      "epoch": 0.057886135051104154,
      "grad_norm": 3.203125,
      "learning_rate": 4.962711348162987e-05,
      "loss": 0.9249,
      "step": 4082
    },
    {
      "epoch": 0.05791449670473037,
      "grad_norm": 3.078125,
      "learning_rate": 4.962672893183009e-05,
      "loss": 0.9337,
      "step": 4084
    },
    {
      "epoch": 0.05794285835835659,
      "grad_norm": 3.296875,
      "learning_rate": 4.962634418533497e-05,
      "loss": 0.8881,
      "step": 4086
    },
    {
      "epoch": 0.0579712200119828,
      "grad_norm": 3.328125,
      "learning_rate": 4.96259592421476e-05,
      "loss": 0.888,
      "step": 4088
    },
    {
      "epoch": 0.057999581665609014,
      "grad_norm": 2.953125,
      "learning_rate": 4.9625574102271046e-05,
      "loss": 0.8787,
      "step": 4090
    },
    {
      "epoch": 0.05802794331923523,
      "grad_norm": 2.828125,
      "learning_rate": 4.96251887657084e-05,
      "loss": 0.8999,
      "step": 4092
    },
    {
      "epoch": 0.05805630497286144,
      "grad_norm": 2.984375,
      "learning_rate": 4.962480323246272e-05,
      "loss": 0.9003,
      "step": 4094
    },
    {
      "epoch": 0.05808466662648766,
      "grad_norm": 2.796875,
      "learning_rate": 4.962441750253709e-05,
      "loss": 0.8583,
      "step": 4096
    },
    {
      "epoch": 0.058113028280113874,
      "grad_norm": 3.046875,
      "learning_rate": 4.962403157593461e-05,
      "loss": 0.9175,
      "step": 4098
    },
    {
      "epoch": 0.05814138993374009,
      "grad_norm": 3.25,
      "learning_rate": 4.962364545265834e-05,
      "loss": 0.9352,
      "step": 4100
    },
    {
      "epoch": 0.0581697515873663,
      "grad_norm": 2.84375,
      "learning_rate": 4.9623259132711365e-05,
      "loss": 0.8515,
      "step": 4102
    },
    {
      "epoch": 0.05819811324099251,
      "grad_norm": 2.84375,
      "learning_rate": 4.9622872616096786e-05,
      "loss": 0.872,
      "step": 4104
    },
    {
      "epoch": 0.058226474894618734,
      "grad_norm": 2.921875,
      "learning_rate": 4.962248590281767e-05,
      "loss": 0.8794,
      "step": 4106
    },
    {
      "epoch": 0.05825483654824495,
      "grad_norm": 3.109375,
      "learning_rate": 4.962209899287713e-05,
      "loss": 0.8974,
      "step": 4108
    },
    {
      "epoch": 0.05828319820187116,
      "grad_norm": 3.296875,
      "learning_rate": 4.962171188627823e-05,
      "loss": 0.9136,
      "step": 4110
    },
    {
      "epoch": 0.05831155985549737,
      "grad_norm": 3.015625,
      "learning_rate": 4.9621324583024085e-05,
      "loss": 0.9142,
      "step": 4112
    },
    {
      "epoch": 0.058339921509123586,
      "grad_norm": 3.203125,
      "learning_rate": 4.962093708311778e-05,
      "loss": 0.8969,
      "step": 4114
    },
    {
      "epoch": 0.05836828316274981,
      "grad_norm": 3.1875,
      "learning_rate": 4.9620549386562396e-05,
      "loss": 0.9435,
      "step": 4116
    },
    {
      "epoch": 0.05839664481637602,
      "grad_norm": 3.015625,
      "learning_rate": 4.962016149336105e-05,
      "loss": 0.922,
      "step": 4118
    },
    {
      "epoch": 0.05842500647000223,
      "grad_norm": 3.0625,
      "learning_rate": 4.9619773403516835e-05,
      "loss": 0.8736,
      "step": 4120
    },
    {
      "epoch": 0.058453368123628446,
      "grad_norm": 3.046875,
      "learning_rate": 4.961938511703284e-05,
      "loss": 0.9218,
      "step": 4122
    },
    {
      "epoch": 0.05848172977725466,
      "grad_norm": 3.203125,
      "learning_rate": 4.9618996633912176e-05,
      "loss": 0.907,
      "step": 4124
    },
    {
      "epoch": 0.05851009143088088,
      "grad_norm": 3.140625,
      "learning_rate": 4.961860795415795e-05,
      "loss": 0.9015,
      "step": 4126
    },
    {
      "epoch": 0.05853845308450709,
      "grad_norm": 2.96875,
      "learning_rate": 4.9618219077773255e-05,
      "loss": 0.935,
      "step": 4128
    },
    {
      "epoch": 0.058566814738133306,
      "grad_norm": 2.984375,
      "learning_rate": 4.96178300047612e-05,
      "loss": 0.9105,
      "step": 4130
    },
    {
      "epoch": 0.05859517639175952,
      "grad_norm": 3.140625,
      "learning_rate": 4.96174407351249e-05,
      "loss": 0.9218,
      "step": 4132
    },
    {
      "epoch": 0.05862353804538574,
      "grad_norm": 3.25,
      "learning_rate": 4.961705126886746e-05,
      "loss": 0.9189,
      "step": 4134
    },
    {
      "epoch": 0.05865189969901195,
      "grad_norm": 3.078125,
      "learning_rate": 4.961666160599198e-05,
      "loss": 0.8609,
      "step": 4136
    },
    {
      "epoch": 0.058680261352638166,
      "grad_norm": 3.09375,
      "learning_rate": 4.9616271746501584e-05,
      "loss": 0.9286,
      "step": 4138
    },
    {
      "epoch": 0.05870862300626438,
      "grad_norm": 2.921875,
      "learning_rate": 4.961588169039939e-05,
      "loss": 0.9292,
      "step": 4140
    },
    {
      "epoch": 0.05873698465989059,
      "grad_norm": 3.015625,
      "learning_rate": 4.96154914376885e-05,
      "loss": 0.9492,
      "step": 4142
    },
    {
      "epoch": 0.05876534631351681,
      "grad_norm": 3.03125,
      "learning_rate": 4.961510098837205e-05,
      "loss": 0.9309,
      "step": 4144
    },
    {
      "epoch": 0.058793707967143026,
      "grad_norm": 3.140625,
      "learning_rate": 4.961471034245314e-05,
      "loss": 0.9205,
      "step": 4146
    },
    {
      "epoch": 0.05882206962076924,
      "grad_norm": 3.125,
      "learning_rate": 4.9614319499934894e-05,
      "loss": 0.9451,
      "step": 4148
    },
    {
      "epoch": 0.05885043127439545,
      "grad_norm": 3.703125,
      "learning_rate": 4.9613928460820434e-05,
      "loss": 0.9254,
      "step": 4150
    },
    {
      "epoch": 0.058878792928021666,
      "grad_norm": 3.328125,
      "learning_rate": 4.9613537225112893e-05,
      "loss": 0.9407,
      "step": 4152
    },
    {
      "epoch": 0.058907154581647886,
      "grad_norm": 3.125,
      "learning_rate": 4.961314579281538e-05,
      "loss": 0.8709,
      "step": 4154
    },
    {
      "epoch": 0.0589355162352741,
      "grad_norm": 3.171875,
      "learning_rate": 4.9612754163931044e-05,
      "loss": 0.8752,
      "step": 4156
    },
    {
      "epoch": 0.05896387788890031,
      "grad_norm": 3.015625,
      "learning_rate": 4.9612362338462984e-05,
      "loss": 0.9443,
      "step": 4158
    },
    {
      "epoch": 0.058992239542526526,
      "grad_norm": 3.25,
      "learning_rate": 4.961197031641435e-05,
      "loss": 0.9955,
      "step": 4160
    },
    {
      "epoch": 0.05902060119615274,
      "grad_norm": 3.15625,
      "learning_rate": 4.961157809778827e-05,
      "loss": 0.9343,
      "step": 4162
    },
    {
      "epoch": 0.05904896284977896,
      "grad_norm": 3.078125,
      "learning_rate": 4.9611185682587866e-05,
      "loss": 0.9154,
      "step": 4164
    },
    {
      "epoch": 0.05907732450340517,
      "grad_norm": 3.53125,
      "learning_rate": 4.9610793070816295e-05,
      "loss": 0.8669,
      "step": 4166
    },
    {
      "epoch": 0.059105686157031385,
      "grad_norm": 3.453125,
      "learning_rate": 4.961040026247666e-05,
      "loss": 0.9062,
      "step": 4168
    },
    {
      "epoch": 0.0591340478106576,
      "grad_norm": 3.015625,
      "learning_rate": 4.9610007257572135e-05,
      "loss": 0.9228,
      "step": 4170
    },
    {
      "epoch": 0.05916240946428381,
      "grad_norm": 2.6875,
      "learning_rate": 4.960961405610582e-05,
      "loss": 0.8419,
      "step": 4172
    },
    {
      "epoch": 0.05919077111791003,
      "grad_norm": 3.390625,
      "learning_rate": 4.960922065808089e-05,
      "loss": 0.8924,
      "step": 4174
    },
    {
      "epoch": 0.059219132771536245,
      "grad_norm": 3.296875,
      "learning_rate": 4.960882706350047e-05,
      "loss": 0.9452,
      "step": 4176
    },
    {
      "epoch": 0.05924749442516246,
      "grad_norm": 3.1875,
      "learning_rate": 4.9608433272367714e-05,
      "loss": 0.8585,
      "step": 4178
    },
    {
      "epoch": 0.05927585607878867,
      "grad_norm": 3.125,
      "learning_rate": 4.960803928468575e-05,
      "loss": 0.8603,
      "step": 4180
    },
    {
      "epoch": 0.059304217732414885,
      "grad_norm": 2.890625,
      "learning_rate": 4.960764510045774e-05,
      "loss": 0.898,
      "step": 4182
    },
    {
      "epoch": 0.059332579386041105,
      "grad_norm": 3.03125,
      "learning_rate": 4.960725071968683e-05,
      "loss": 0.8364,
      "step": 4184
    },
    {
      "epoch": 0.05936094103966732,
      "grad_norm": 3.1875,
      "learning_rate": 4.9606856142376156e-05,
      "loss": 0.921,
      "step": 4186
    },
    {
      "epoch": 0.05938930269329353,
      "grad_norm": 3.03125,
      "learning_rate": 4.960646136852889e-05,
      "loss": 0.9831,
      "step": 4188
    },
    {
      "epoch": 0.059417664346919745,
      "grad_norm": 2.890625,
      "learning_rate": 4.960606639814817e-05,
      "loss": 0.8861,
      "step": 4190
    },
    {
      "epoch": 0.059446026000545965,
      "grad_norm": 3.1875,
      "learning_rate": 4.960567123123716e-05,
      "loss": 0.9198,
      "step": 4192
    },
    {
      "epoch": 0.05947438765417218,
      "grad_norm": 3.25,
      "learning_rate": 4.960527586779902e-05,
      "loss": 0.9375,
      "step": 4194
    },
    {
      "epoch": 0.05950274930779839,
      "grad_norm": 2.984375,
      "learning_rate": 4.960488030783689e-05,
      "loss": 0.9685,
      "step": 4196
    },
    {
      "epoch": 0.059531110961424605,
      "grad_norm": 2.890625,
      "learning_rate": 4.960448455135394e-05,
      "loss": 0.9174,
      "step": 4198
    },
    {
      "epoch": 0.05955947261505082,
      "grad_norm": 3.125,
      "learning_rate": 4.960408859835333e-05,
      "loss": 0.9629,
      "step": 4200
    },
    {
      "epoch": 0.05958783426867704,
      "grad_norm": 3.375,
      "learning_rate": 4.960369244883823e-05,
      "loss": 0.9274,
      "step": 4202
    },
    {
      "epoch": 0.05961619592230325,
      "grad_norm": 2.828125,
      "learning_rate": 4.960329610281179e-05,
      "loss": 0.8886,
      "step": 4204
    },
    {
      "epoch": 0.059644557575929465,
      "grad_norm": 3.078125,
      "learning_rate": 4.9602899560277197e-05,
      "loss": 0.8974,
      "step": 4206
    },
    {
      "epoch": 0.05967291922955568,
      "grad_norm": 2.984375,
      "learning_rate": 4.9602502821237586e-05,
      "loss": 0.9225,
      "step": 4208
    },
    {
      "epoch": 0.05970128088318189,
      "grad_norm": 3.203125,
      "learning_rate": 4.960210588569616e-05,
      "loss": 0.9008,
      "step": 4210
    },
    {
      "epoch": 0.05972964253680811,
      "grad_norm": 2.796875,
      "learning_rate": 4.960170875365606e-05,
      "loss": 0.8873,
      "step": 4212
    },
    {
      "epoch": 0.059758004190434325,
      "grad_norm": 2.9375,
      "learning_rate": 4.9601311425120475e-05,
      "loss": 0.9013,
      "step": 4214
    },
    {
      "epoch": 0.05978636584406054,
      "grad_norm": 3.4375,
      "learning_rate": 4.960091390009258e-05,
      "loss": 0.9161,
      "step": 4216
    },
    {
      "epoch": 0.05981472749768675,
      "grad_norm": 3.0,
      "learning_rate": 4.960051617857554e-05,
      "loss": 0.8981,
      "step": 4218
    },
    {
      "epoch": 0.059843089151312964,
      "grad_norm": 2.828125,
      "learning_rate": 4.960011826057254e-05,
      "loss": 0.8686,
      "step": 4220
    },
    {
      "epoch": 0.059871450804939184,
      "grad_norm": 3.265625,
      "learning_rate": 4.959972014608675e-05,
      "loss": 0.8942,
      "step": 4222
    },
    {
      "epoch": 0.0598998124585654,
      "grad_norm": 3.46875,
      "learning_rate": 4.9599321835121355e-05,
      "loss": 0.8953,
      "step": 4224
    },
    {
      "epoch": 0.05992817411219161,
      "grad_norm": 3.828125,
      "learning_rate": 4.9598923327679534e-05,
      "loss": 0.8942,
      "step": 4226
    },
    {
      "epoch": 0.059956535765817824,
      "grad_norm": 2.765625,
      "learning_rate": 4.9598524623764476e-05,
      "loss": 0.9078,
      "step": 4228
    },
    {
      "epoch": 0.05998489741944404,
      "grad_norm": 2.84375,
      "learning_rate": 4.959812572337936e-05,
      "loss": 0.916,
      "step": 4230
    },
    {
      "epoch": 0.06001325907307026,
      "grad_norm": 3.0,
      "learning_rate": 4.959772662652737e-05,
      "loss": 0.9056,
      "step": 4232
    },
    {
      "epoch": 0.06004162072669647,
      "grad_norm": 3.140625,
      "learning_rate": 4.9597327333211703e-05,
      "loss": 0.9418,
      "step": 4234
    },
    {
      "epoch": 0.060069982380322684,
      "grad_norm": 2.84375,
      "learning_rate": 4.959692784343554e-05,
      "loss": 0.9352,
      "step": 4236
    },
    {
      "epoch": 0.0600983440339489,
      "grad_norm": 3.390625,
      "learning_rate": 4.959652815720207e-05,
      "loss": 0.9008,
      "step": 4238
    },
    {
      "epoch": 0.06012670568757512,
      "grad_norm": 3.328125,
      "learning_rate": 4.959612827451449e-05,
      "loss": 0.906,
      "step": 4240
    },
    {
      "epoch": 0.06015506734120133,
      "grad_norm": 2.953125,
      "learning_rate": 4.9595728195375996e-05,
      "loss": 0.9324,
      "step": 4242
    },
    {
      "epoch": 0.060183428994827544,
      "grad_norm": 3.265625,
      "learning_rate": 4.959532791978978e-05,
      "loss": 0.9592,
      "step": 4244
    },
    {
      "epoch": 0.06021179064845376,
      "grad_norm": 3.0625,
      "learning_rate": 4.959492744775904e-05,
      "loss": 0.9036,
      "step": 4246
    },
    {
      "epoch": 0.06024015230207997,
      "grad_norm": 3.359375,
      "learning_rate": 4.9594526779286966e-05,
      "loss": 0.9199,
      "step": 4248
    },
    {
      "epoch": 0.06026851395570619,
      "grad_norm": 3.375,
      "learning_rate": 4.959412591437678e-05,
      "loss": 0.9158,
      "step": 4250
    },
    {
      "epoch": 0.060296875609332404,
      "grad_norm": 3.328125,
      "learning_rate": 4.959372485303165e-05,
      "loss": 0.9309,
      "step": 4252
    },
    {
      "epoch": 0.06032523726295862,
      "grad_norm": 3.34375,
      "learning_rate": 4.959332359525482e-05,
      "loss": 0.9032,
      "step": 4254
    },
    {
      "epoch": 0.06035359891658483,
      "grad_norm": 3.109375,
      "learning_rate": 4.959292214104946e-05,
      "loss": 0.9427,
      "step": 4256
    },
    {
      "epoch": 0.06038196057021104,
      "grad_norm": 3.21875,
      "learning_rate": 4.9592520490418806e-05,
      "loss": 0.9504,
      "step": 4258
    },
    {
      "epoch": 0.060410322223837264,
      "grad_norm": 2.9375,
      "learning_rate": 4.9592118643366035e-05,
      "loss": 0.8932,
      "step": 4260
    },
    {
      "epoch": 0.06043868387746348,
      "grad_norm": 2.984375,
      "learning_rate": 4.959171659989438e-05,
      "loss": 0.9152,
      "step": 4262
    },
    {
      "epoch": 0.06046704553108969,
      "grad_norm": 2.71875,
      "learning_rate": 4.959131436000703e-05,
      "loss": 0.9027,
      "step": 4264
    },
    {
      "epoch": 0.0604954071847159,
      "grad_norm": 2.890625,
      "learning_rate": 4.959091192370723e-05,
      "loss": 0.8978,
      "step": 4266
    },
    {
      "epoch": 0.06052376883834212,
      "grad_norm": 2.84375,
      "learning_rate": 4.959050929099817e-05,
      "loss": 0.8886,
      "step": 4268
    },
    {
      "epoch": 0.06055213049196834,
      "grad_norm": 3.484375,
      "learning_rate": 4.9590106461883076e-05,
      "loss": 0.9047,
      "step": 4270
    },
    {
      "epoch": 0.06058049214559455,
      "grad_norm": 3.03125,
      "learning_rate": 4.9589703436365156e-05,
      "loss": 0.8836,
      "step": 4272
    },
    {
      "epoch": 0.06060885379922076,
      "grad_norm": 3.515625,
      "learning_rate": 4.958930021444763e-05,
      "loss": 0.9398,
      "step": 4274
    },
    {
      "epoch": 0.060637215452846976,
      "grad_norm": 3.640625,
      "learning_rate": 4.958889679613373e-05,
      "loss": 0.9166,
      "step": 4276
    },
    {
      "epoch": 0.06066557710647319,
      "grad_norm": 3.53125,
      "learning_rate": 4.9588493181426674e-05,
      "loss": 0.9233,
      "step": 4278
    },
    {
      "epoch": 0.06069393876009941,
      "grad_norm": 3.078125,
      "learning_rate": 4.958808937032968e-05,
      "loss": 0.8999,
      "step": 4280
    },
    {
      "epoch": 0.06072230041372562,
      "grad_norm": 3.046875,
      "learning_rate": 4.958768536284597e-05,
      "loss": 0.8856,
      "step": 4282
    },
    {
      "epoch": 0.060750662067351836,
      "grad_norm": 3.0,
      "learning_rate": 4.958728115897878e-05,
      "loss": 0.842,
      "step": 4284
    },
    {
      "epoch": 0.06077902372097805,
      "grad_norm": 3.328125,
      "learning_rate": 4.958687675873134e-05,
      "loss": 0.924,
      "step": 4286
    },
    {
      "epoch": 0.06080738537460427,
      "grad_norm": 2.625,
      "learning_rate": 4.958647216210688e-05,
      "loss": 0.8743,
      "step": 4288
    },
    {
      "epoch": 0.06083574702823048,
      "grad_norm": 3.265625,
      "learning_rate": 4.958606736910861e-05,
      "loss": 0.9053,
      "step": 4290
    },
    {
      "epoch": 0.060864108681856696,
      "grad_norm": 3.015625,
      "learning_rate": 4.9585662379739796e-05,
      "loss": 0.9297,
      "step": 4292
    },
    {
      "epoch": 0.06089247033548291,
      "grad_norm": 3.078125,
      "learning_rate": 4.958525719400365e-05,
      "loss": 0.9139,
      "step": 4294
    },
    {
      "epoch": 0.06092083198910912,
      "grad_norm": 2.859375,
      "learning_rate": 4.9584851811903414e-05,
      "loss": 0.8685,
      "step": 4296
    },
    {
      "epoch": 0.06094919364273534,
      "grad_norm": 3.015625,
      "learning_rate": 4.9584446233442326e-05,
      "loss": 0.9324,
      "step": 4298
    },
    {
      "epoch": 0.060977555296361556,
      "grad_norm": 3.15625,
      "learning_rate": 4.958404045862363e-05,
      "loss": 0.899,
      "step": 4300
    },
    {
      "epoch": 0.06100591694998777,
      "grad_norm": 2.9375,
      "learning_rate": 4.9583634487450565e-05,
      "loss": 0.9426,
      "step": 4302
    },
    {
      "epoch": 0.06103427860361398,
      "grad_norm": 3.0625,
      "learning_rate": 4.958322831992637e-05,
      "loss": 0.9031,
      "step": 4304
    },
    {
      "epoch": 0.061062640257240196,
      "grad_norm": 3.375,
      "learning_rate": 4.958282195605428e-05,
      "loss": 0.9681,
      "step": 4306
    },
    {
      "epoch": 0.061091001910866416,
      "grad_norm": 3.03125,
      "learning_rate": 4.958241539583757e-05,
      "loss": 0.9221,
      "step": 4308
    },
    {
      "epoch": 0.06111936356449263,
      "grad_norm": 3.046875,
      "learning_rate": 4.958200863927945e-05,
      "loss": 0.8804,
      "step": 4310
    },
    {
      "epoch": 0.06114772521811884,
      "grad_norm": 3.0,
      "learning_rate": 4.95816016863832e-05,
      "loss": 0.9227,
      "step": 4312
    },
    {
      "epoch": 0.061176086871745056,
      "grad_norm": 2.890625,
      "learning_rate": 4.958119453715205e-05,
      "loss": 0.9391,
      "step": 4314
    },
    {
      "epoch": 0.06120444852537127,
      "grad_norm": 4.0625,
      "learning_rate": 4.958078719158927e-05,
      "loss": 0.9791,
      "step": 4316
    },
    {
      "epoch": 0.06123281017899749,
      "grad_norm": 3.0,
      "learning_rate": 4.9580379649698096e-05,
      "loss": 0.9057,
      "step": 4318
    },
    {
      "epoch": 0.0612611718326237,
      "grad_norm": 2.859375,
      "learning_rate": 4.957997191148179e-05,
      "loss": 0.9394,
      "step": 4320
    },
    {
      "epoch": 0.061289533486249916,
      "grad_norm": 2.796875,
      "learning_rate": 4.957956397694361e-05,
      "loss": 0.8939,
      "step": 4322
    },
    {
      "epoch": 0.06131789513987613,
      "grad_norm": 3.234375,
      "learning_rate": 4.957915584608681e-05,
      "loss": 0.9503,
      "step": 4324
    },
    {
      "epoch": 0.06134625679350234,
      "grad_norm": 3.15625,
      "learning_rate": 4.957874751891466e-05,
      "loss": 0.8942,
      "step": 4326
    },
    {
      "epoch": 0.06137461844712856,
      "grad_norm": 3.265625,
      "learning_rate": 4.9578338995430406e-05,
      "loss": 0.9176,
      "step": 4328
    },
    {
      "epoch": 0.061402980100754775,
      "grad_norm": 3.859375,
      "learning_rate": 4.957793027563732e-05,
      "loss": 0.8954,
      "step": 4330
    },
    {
      "epoch": 0.06143134175438099,
      "grad_norm": 3.25,
      "learning_rate": 4.957752135953867e-05,
      "loss": 0.8904,
      "step": 4332
    },
    {
      "epoch": 0.0614597034080072,
      "grad_norm": 3.03125,
      "learning_rate": 4.9577112247137715e-05,
      "loss": 0.8943,
      "step": 4334
    },
    {
      "epoch": 0.061488065061633415,
      "grad_norm": 3.71875,
      "learning_rate": 4.9576702938437724e-05,
      "loss": 0.96,
      "step": 4336
    },
    {
      "epoch": 0.061516426715259635,
      "grad_norm": 3.46875,
      "learning_rate": 4.9576293433441974e-05,
      "loss": 0.9093,
      "step": 4338
    },
    {
      "epoch": 0.06154478836888585,
      "grad_norm": 3.4375,
      "learning_rate": 4.9575883732153725e-05,
      "loss": 0.937,
      "step": 4340
    },
    {
      "epoch": 0.06157315002251206,
      "grad_norm": 3.234375,
      "learning_rate": 4.957547383457625e-05,
      "loss": 0.9304,
      "step": 4342
    },
    {
      "epoch": 0.061601511676138275,
      "grad_norm": 3.1875,
      "learning_rate": 4.957506374071283e-05,
      "loss": 0.917,
      "step": 4344
    },
    {
      "epoch": 0.061629873329764495,
      "grad_norm": 3.296875,
      "learning_rate": 4.957465345056673e-05,
      "loss": 0.9286,
      "step": 4346
    },
    {
      "epoch": 0.06165823498339071,
      "grad_norm": 2.90625,
      "learning_rate": 4.957424296414125e-05,
      "loss": 0.9119,
      "step": 4348
    },
    {
      "epoch": 0.06168659663701692,
      "grad_norm": 3.296875,
      "learning_rate": 4.9573832281439636e-05,
      "loss": 0.9454,
      "step": 4350
    },
    {
      "epoch": 0.061714958290643135,
      "grad_norm": 2.890625,
      "learning_rate": 4.957342140246519e-05,
      "loss": 0.9039,
      "step": 4352
    },
    {
      "epoch": 0.06174331994426935,
      "grad_norm": 3.03125,
      "learning_rate": 4.957301032722119e-05,
      "loss": 0.8884,
      "step": 4354
    },
    {
      "epoch": 0.06177168159789557,
      "grad_norm": 2.953125,
      "learning_rate": 4.957259905571092e-05,
      "loss": 0.8839,
      "step": 4356
    },
    {
      "epoch": 0.06180004325152178,
      "grad_norm": 3.28125,
      "learning_rate": 4.9572187587937655e-05,
      "loss": 0.913,
      "step": 4358
    },
    {
      "epoch": 0.061828404905147995,
      "grad_norm": 3.265625,
      "learning_rate": 4.9571775923904694e-05,
      "loss": 0.899,
      "step": 4360
    },
    {
      "epoch": 0.06185676655877421,
      "grad_norm": 2.796875,
      "learning_rate": 4.957136406361532e-05,
      "loss": 0.9114,
      "step": 4362
    },
    {
      "epoch": 0.06188512821240042,
      "grad_norm": 2.921875,
      "learning_rate": 4.957095200707281e-05,
      "loss": 0.8548,
      "step": 4364
    },
    {
      "epoch": 0.06191348986602664,
      "grad_norm": 3.0,
      "learning_rate": 4.9570539754280484e-05,
      "loss": 0.9058,
      "step": 4366
    },
    {
      "epoch": 0.061941851519652855,
      "grad_norm": 2.921875,
      "learning_rate": 4.9570127305241606e-05,
      "loss": 0.8778,
      "step": 4368
    },
    {
      "epoch": 0.06197021317327907,
      "grad_norm": 3.46875,
      "learning_rate": 4.956971465995949e-05,
      "loss": 0.9196,
      "step": 4370
    },
    {
      "epoch": 0.06199857482690528,
      "grad_norm": 3.390625,
      "learning_rate": 4.956930181843742e-05,
      "loss": 0.8792,
      "step": 4372
    },
    {
      "epoch": 0.062026936480531494,
      "grad_norm": 2.96875,
      "learning_rate": 4.95688887806787e-05,
      "loss": 0.8845,
      "step": 4374
    },
    {
      "epoch": 0.062055298134157715,
      "grad_norm": 2.859375,
      "learning_rate": 4.9568475546686634e-05,
      "loss": 0.8971,
      "step": 4376
    },
    {
      "epoch": 0.06208365978778393,
      "grad_norm": 3.125,
      "learning_rate": 4.956806211646451e-05,
      "loss": 0.9152,
      "step": 4378
    },
    {
      "epoch": 0.06211202144141014,
      "grad_norm": 2.96875,
      "learning_rate": 4.956764849001563e-05,
      "loss": 0.8933,
      "step": 4380
    },
    {
      "epoch": 0.062140383095036354,
      "grad_norm": 3.390625,
      "learning_rate": 4.9567234667343305e-05,
      "loss": 0.9274,
      "step": 4382
    },
    {
      "epoch": 0.06216874474866257,
      "grad_norm": 3.3125,
      "learning_rate": 4.956682064845084e-05,
      "loss": 0.9549,
      "step": 4384
    },
    {
      "epoch": 0.06219710640228879,
      "grad_norm": 3.046875,
      "learning_rate": 4.956640643334154e-05,
      "loss": 0.8492,
      "step": 4386
    },
    {
      "epoch": 0.062225468055915,
      "grad_norm": 3.21875,
      "learning_rate": 4.95659920220187e-05,
      "loss": 0.9271,
      "step": 4388
    },
    {
      "epoch": 0.062253829709541214,
      "grad_norm": 2.8125,
      "learning_rate": 4.956557741448566e-05,
      "loss": 0.8903,
      "step": 4390
    },
    {
      "epoch": 0.06228219136316743,
      "grad_norm": 2.90625,
      "learning_rate": 4.956516261074571e-05,
      "loss": 0.8773,
      "step": 4392
    },
    {
      "epoch": 0.06231055301679365,
      "grad_norm": 3.046875,
      "learning_rate": 4.9564747610802165e-05,
      "loss": 0.9306,
      "step": 4394
    },
    {
      "epoch": 0.06233891467041986,
      "grad_norm": 2.984375,
      "learning_rate": 4.956433241465835e-05,
      "loss": 0.8718,
      "step": 4396
    },
    {
      "epoch": 0.062367276324046074,
      "grad_norm": 2.6875,
      "learning_rate": 4.9563917022317565e-05,
      "loss": 0.9326,
      "step": 4398
    },
    {
      "epoch": 0.06239563797767229,
      "grad_norm": 2.96875,
      "learning_rate": 4.956350143378315e-05,
      "loss": 0.8871,
      "step": 4400
    },
    {
      "epoch": 0.0624239996312985,
      "grad_norm": 2.984375,
      "learning_rate": 4.9563085649058395e-05,
      "loss": 0.8899,
      "step": 4402
    },
    {
      "epoch": 0.06245236128492472,
      "grad_norm": 3.109375,
      "learning_rate": 4.956266966814663e-05,
      "loss": 0.8899,
      "step": 4404
    },
    {
      "epoch": 0.062480722938550934,
      "grad_norm": 3.015625,
      "learning_rate": 4.95622534910512e-05,
      "loss": 0.8481,
      "step": 4406
    },
    {
      "epoch": 0.06250908459217715,
      "grad_norm": 3.15625,
      "learning_rate": 4.956183711777541e-05,
      "loss": 0.8748,
      "step": 4408
    },
    {
      "epoch": 0.06253744624580336,
      "grad_norm": 3.09375,
      "learning_rate": 4.956142054832259e-05,
      "loss": 0.8775,
      "step": 4410
    },
    {
      "epoch": 0.06256580789942957,
      "grad_norm": 3.28125,
      "learning_rate": 4.9561003782696055e-05,
      "loss": 0.8933,
      "step": 4412
    },
    {
      "epoch": 0.06259416955305579,
      "grad_norm": 3.125,
      "learning_rate": 4.956058682089916e-05,
      "loss": 0.9504,
      "step": 4414
    },
    {
      "epoch": 0.062622531206682,
      "grad_norm": 4.1875,
      "learning_rate": 4.956016966293521e-05,
      "loss": 0.9071,
      "step": 4416
    },
    {
      "epoch": 0.06265089286030823,
      "grad_norm": 3.125,
      "learning_rate": 4.955975230880755e-05,
      "loss": 0.8703,
      "step": 4418
    },
    {
      "epoch": 0.06267925451393444,
      "grad_norm": 3.078125,
      "learning_rate": 4.955933475851951e-05,
      "loss": 0.9153,
      "step": 4420
    },
    {
      "epoch": 0.06270761616756065,
      "grad_norm": 3.078125,
      "learning_rate": 4.9558917012074425e-05,
      "loss": 0.8569,
      "step": 4422
    },
    {
      "epoch": 0.06273597782118687,
      "grad_norm": 3.515625,
      "learning_rate": 4.955849906947563e-05,
      "loss": 0.8946,
      "step": 4424
    },
    {
      "epoch": 0.06276433947481308,
      "grad_norm": 3.3125,
      "learning_rate": 4.955808093072647e-05,
      "loss": 0.9502,
      "step": 4426
    },
    {
      "epoch": 0.0627927011284393,
      "grad_norm": 2.984375,
      "learning_rate": 4.9557662595830274e-05,
      "loss": 0.8673,
      "step": 4428
    },
    {
      "epoch": 0.0628210627820655,
      "grad_norm": 3.234375,
      "learning_rate": 4.9557244064790384e-05,
      "loss": 0.9455,
      "step": 4430
    },
    {
      "epoch": 0.06284942443569172,
      "grad_norm": 3.0625,
      "learning_rate": 4.9556825337610156e-05,
      "loss": 0.8943,
      "step": 4432
    },
    {
      "epoch": 0.06287778608931793,
      "grad_norm": 3.671875,
      "learning_rate": 4.955640641429293e-05,
      "loss": 0.8992,
      "step": 4434
    },
    {
      "epoch": 0.06290614774294415,
      "grad_norm": 3.0,
      "learning_rate": 4.955598729484204e-05,
      "loss": 0.9285,
      "step": 4436
    },
    {
      "epoch": 0.06293450939657037,
      "grad_norm": 3.046875,
      "learning_rate": 4.955556797926084e-05,
      "loss": 0.9537,
      "step": 4438
    },
    {
      "epoch": 0.06296287105019659,
      "grad_norm": 2.953125,
      "learning_rate": 4.955514846755269e-05,
      "loss": 0.9474,
      "step": 4440
    },
    {
      "epoch": 0.0629912327038228,
      "grad_norm": 3.25,
      "learning_rate": 4.9554728759720925e-05,
      "loss": 0.8976,
      "step": 4442
    },
    {
      "epoch": 0.06301959435744901,
      "grad_norm": 3.296875,
      "learning_rate": 4.95543088557689e-05,
      "loss": 0.899,
      "step": 4444
    },
    {
      "epoch": 0.06304795601107523,
      "grad_norm": 3.375,
      "learning_rate": 4.955388875569998e-05,
      "loss": 0.9235,
      "step": 4446
    },
    {
      "epoch": 0.06307631766470144,
      "grad_norm": 3.8125,
      "learning_rate": 4.9553468459517513e-05,
      "loss": 0.9637,
      "step": 4448
    },
    {
      "epoch": 0.06310467931832765,
      "grad_norm": 3.40625,
      "learning_rate": 4.9553047967224856e-05,
      "loss": 0.9181,
      "step": 4450
    },
    {
      "epoch": 0.06313304097195387,
      "grad_norm": 3.109375,
      "learning_rate": 4.955262727882536e-05,
      "loss": 0.8548,
      "step": 4452
    },
    {
      "epoch": 0.06316140262558008,
      "grad_norm": 3.109375,
      "learning_rate": 4.955220639432239e-05,
      "loss": 0.9494,
      "step": 4454
    },
    {
      "epoch": 0.06318976427920629,
      "grad_norm": 2.953125,
      "learning_rate": 4.955178531371932e-05,
      "loss": 0.8628,
      "step": 4456
    },
    {
      "epoch": 0.06321812593283252,
      "grad_norm": 3.265625,
      "learning_rate": 4.9551364037019496e-05,
      "loss": 0.8835,
      "step": 4458
    },
    {
      "epoch": 0.06324648758645873,
      "grad_norm": 2.9375,
      "learning_rate": 4.955094256422629e-05,
      "loss": 0.939,
      "step": 4460
    },
    {
      "epoch": 0.06327484924008495,
      "grad_norm": 3.515625,
      "learning_rate": 4.955052089534308e-05,
      "loss": 0.9424,
      "step": 4462
    },
    {
      "epoch": 0.06330321089371116,
      "grad_norm": 3.25,
      "learning_rate": 4.955009903037321e-05,
      "loss": 0.8945,
      "step": 4464
    },
    {
      "epoch": 0.06333157254733737,
      "grad_norm": 3.25,
      "learning_rate": 4.9549676969320055e-05,
      "loss": 0.9261,
      "step": 4466
    },
    {
      "epoch": 0.06335993420096359,
      "grad_norm": 2.875,
      "learning_rate": 4.954925471218701e-05,
      "loss": 0.9343,
      "step": 4468
    },
    {
      "epoch": 0.0633882958545898,
      "grad_norm": 3.296875,
      "learning_rate": 4.954883225897742e-05,
      "loss": 0.9182,
      "step": 4470
    },
    {
      "epoch": 0.06341665750821601,
      "grad_norm": 3.359375,
      "learning_rate": 4.954840960969467e-05,
      "loss": 0.8781,
      "step": 4472
    },
    {
      "epoch": 0.06344501916184223,
      "grad_norm": 3.59375,
      "learning_rate": 4.954798676434214e-05,
      "loss": 0.9237,
      "step": 4474
    },
    {
      "epoch": 0.06347338081546845,
      "grad_norm": 3.265625,
      "learning_rate": 4.95475637229232e-05,
      "loss": 0.8915,
      "step": 4476
    },
    {
      "epoch": 0.06350174246909467,
      "grad_norm": 3.234375,
      "learning_rate": 4.9547140485441224e-05,
      "loss": 0.9183,
      "step": 4478
    },
    {
      "epoch": 0.06353010412272088,
      "grad_norm": 3.296875,
      "learning_rate": 4.9546717051899604e-05,
      "loss": 0.9093,
      "step": 4480
    },
    {
      "epoch": 0.06355846577634709,
      "grad_norm": 3.375,
      "learning_rate": 4.9546293422301724e-05,
      "loss": 0.8853,
      "step": 4482
    },
    {
      "epoch": 0.0635868274299733,
      "grad_norm": 2.90625,
      "learning_rate": 4.9545869596650955e-05,
      "loss": 0.8812,
      "step": 4484
    },
    {
      "epoch": 0.06361518908359952,
      "grad_norm": 3.109375,
      "learning_rate": 4.954544557495069e-05,
      "loss": 0.869,
      "step": 4486
    },
    {
      "epoch": 0.06364355073722573,
      "grad_norm": 2.90625,
      "learning_rate": 4.9545021357204316e-05,
      "loss": 0.9197,
      "step": 4488
    },
    {
      "epoch": 0.06367191239085195,
      "grad_norm": 2.59375,
      "learning_rate": 4.954459694341521e-05,
      "loss": 0.852,
      "step": 4490
    },
    {
      "epoch": 0.06370027404447816,
      "grad_norm": 3.203125,
      "learning_rate": 4.954417233358678e-05,
      "loss": 0.8924,
      "step": 4492
    },
    {
      "epoch": 0.06372863569810437,
      "grad_norm": 3.03125,
      "learning_rate": 4.9543747527722404e-05,
      "loss": 0.8952,
      "step": 4494
    },
    {
      "epoch": 0.0637569973517306,
      "grad_norm": 3.0,
      "learning_rate": 4.954332252582549e-05,
      "loss": 0.8673,
      "step": 4496
    },
    {
      "epoch": 0.06378535900535681,
      "grad_norm": 3.125,
      "learning_rate": 4.954289732789941e-05,
      "loss": 0.9252,
      "step": 4498
    },
    {
      "epoch": 0.06381372065898303,
      "grad_norm": 2.953125,
      "learning_rate": 4.954247193394758e-05,
      "loss": 0.8773,
      "step": 4500
    },
    {
      "epoch": 0.06384208231260924,
      "grad_norm": 3.1875,
      "learning_rate": 4.954204634397338e-05,
      "loss": 0.9478,
      "step": 4502
    },
    {
      "epoch": 0.06387044396623545,
      "grad_norm": 3.109375,
      "learning_rate": 4.954162055798023e-05,
      "loss": 0.8976,
      "step": 4504
    },
    {
      "epoch": 0.06389880561986166,
      "grad_norm": 3.03125,
      "learning_rate": 4.954119457597151e-05,
      "loss": 0.9254,
      "step": 4506
    },
    {
      "epoch": 0.06392716727348788,
      "grad_norm": 2.609375,
      "learning_rate": 4.954076839795064e-05,
      "loss": 0.8699,
      "step": 4508
    },
    {
      "epoch": 0.06395552892711409,
      "grad_norm": 2.890625,
      "learning_rate": 4.954034202392101e-05,
      "loss": 0.9503,
      "step": 4510
    },
    {
      "epoch": 0.0639838905807403,
      "grad_norm": 2.953125,
      "learning_rate": 4.953991545388603e-05,
      "loss": 0.9343,
      "step": 4512
    },
    {
      "epoch": 0.06401225223436652,
      "grad_norm": 3.5,
      "learning_rate": 4.953948868784911e-05,
      "loss": 0.9285,
      "step": 4514
    },
    {
      "epoch": 0.06404061388799275,
      "grad_norm": 2.90625,
      "learning_rate": 4.9539061725813664e-05,
      "loss": 0.9171,
      "step": 4516
    },
    {
      "epoch": 0.06406897554161896,
      "grad_norm": 3.265625,
      "learning_rate": 4.953863456778309e-05,
      "loss": 0.8988,
      "step": 4518
    },
    {
      "epoch": 0.06409733719524517,
      "grad_norm": 3.640625,
      "learning_rate": 4.9538207213760804e-05,
      "loss": 0.9067,
      "step": 4520
    },
    {
      "epoch": 0.06412569884887138,
      "grad_norm": 3.15625,
      "learning_rate": 4.9537779663750225e-05,
      "loss": 0.8961,
      "step": 4522
    },
    {
      "epoch": 0.0641540605024976,
      "grad_norm": 2.625,
      "learning_rate": 4.953735191775476e-05,
      "loss": 0.9187,
      "step": 4524
    },
    {
      "epoch": 0.06418242215612381,
      "grad_norm": 2.890625,
      "learning_rate": 4.953692397577783e-05,
      "loss": 0.8512,
      "step": 4526
    },
    {
      "epoch": 0.06421078380975002,
      "grad_norm": 3.0,
      "learning_rate": 4.953649583782284e-05,
      "loss": 0.8694,
      "step": 4528
    },
    {
      "epoch": 0.06423914546337624,
      "grad_norm": 3.21875,
      "learning_rate": 4.953606750389324e-05,
      "loss": 0.8567,
      "step": 4530
    },
    {
      "epoch": 0.06426750711700245,
      "grad_norm": 3.109375,
      "learning_rate": 4.9535638973992416e-05,
      "loss": 0.8811,
      "step": 4532
    },
    {
      "epoch": 0.06429586877062868,
      "grad_norm": 3.15625,
      "learning_rate": 4.953521024812382e-05,
      "loss": 0.9173,
      "step": 4534
    },
    {
      "epoch": 0.06432423042425489,
      "grad_norm": 2.984375,
      "learning_rate": 4.9534781326290854e-05,
      "loss": 0.8845,
      "step": 4536
    },
    {
      "epoch": 0.0643525920778811,
      "grad_norm": 3.265625,
      "learning_rate": 4.953435220849696e-05,
      "loss": 0.8952,
      "step": 4538
    },
    {
      "epoch": 0.06438095373150732,
      "grad_norm": 3.578125,
      "learning_rate": 4.953392289474555e-05,
      "loss": 0.9113,
      "step": 4540
    },
    {
      "epoch": 0.06440931538513353,
      "grad_norm": 3.0,
      "learning_rate": 4.9533493385040067e-05,
      "loss": 0.921,
      "step": 4542
    },
    {
      "epoch": 0.06443767703875974,
      "grad_norm": 3.140625,
      "learning_rate": 4.953306367938393e-05,
      "loss": 0.9076,
      "step": 4544
    },
    {
      "epoch": 0.06446603869238596,
      "grad_norm": 3.140625,
      "learning_rate": 4.953263377778059e-05,
      "loss": 0.9131,
      "step": 4546
    },
    {
      "epoch": 0.06449440034601217,
      "grad_norm": 2.921875,
      "learning_rate": 4.953220368023346e-05,
      "loss": 0.9501,
      "step": 4548
    },
    {
      "epoch": 0.06452276199963838,
      "grad_norm": 3.3125,
      "learning_rate": 4.9531773386745986e-05,
      "loss": 0.9396,
      "step": 4550
    },
    {
      "epoch": 0.0645511236532646,
      "grad_norm": 3.203125,
      "learning_rate": 4.953134289732159e-05,
      "loss": 0.9094,
      "step": 4552
    },
    {
      "epoch": 0.06457948530689082,
      "grad_norm": 3.171875,
      "learning_rate": 4.953091221196373e-05,
      "loss": 0.8433,
      "step": 4554
    },
    {
      "epoch": 0.06460784696051704,
      "grad_norm": 2.84375,
      "learning_rate": 4.953048133067584e-05,
      "loss": 0.901,
      "step": 4556
    },
    {
      "epoch": 0.06463620861414325,
      "grad_norm": 3.359375,
      "learning_rate": 4.953005025346136e-05,
      "loss": 0.9377,
      "step": 4558
    },
    {
      "epoch": 0.06466457026776946,
      "grad_norm": 3.234375,
      "learning_rate": 4.952961898032373e-05,
      "loss": 0.8444,
      "step": 4560
    },
    {
      "epoch": 0.06469293192139568,
      "grad_norm": 3.15625,
      "learning_rate": 4.9529187511266395e-05,
      "loss": 0.9473,
      "step": 4562
    },
    {
      "epoch": 0.06472129357502189,
      "grad_norm": 2.921875,
      "learning_rate": 4.9528755846292806e-05,
      "loss": 0.8601,
      "step": 4564
    },
    {
      "epoch": 0.0647496552286481,
      "grad_norm": 3.09375,
      "learning_rate": 4.952832398540641e-05,
      "loss": 0.9027,
      "step": 4566
    },
    {
      "epoch": 0.06477801688227432,
      "grad_norm": 3.03125,
      "learning_rate": 4.952789192861065e-05,
      "loss": 0.8995,
      "step": 4568
    },
    {
      "epoch": 0.06480637853590053,
      "grad_norm": 3.15625,
      "learning_rate": 4.9527459675908986e-05,
      "loss": 0.8699,
      "step": 4570
    },
    {
      "epoch": 0.06483474018952676,
      "grad_norm": 3.1875,
      "learning_rate": 4.952702722730486e-05,
      "loss": 0.9362,
      "step": 4572
    },
    {
      "epoch": 0.06486310184315297,
      "grad_norm": 3.1875,
      "learning_rate": 4.9526594582801736e-05,
      "loss": 0.9343,
      "step": 4574
    },
    {
      "epoch": 0.06489146349677918,
      "grad_norm": 3.0,
      "learning_rate": 4.952616174240306e-05,
      "loss": 0.862,
      "step": 4576
    },
    {
      "epoch": 0.0649198251504054,
      "grad_norm": 3.1875,
      "learning_rate": 4.9525728706112295e-05,
      "loss": 0.8915,
      "step": 4578
    },
    {
      "epoch": 0.06494818680403161,
      "grad_norm": 2.78125,
      "learning_rate": 4.95252954739329e-05,
      "loss": 0.8781,
      "step": 4580
    },
    {
      "epoch": 0.06497654845765782,
      "grad_norm": 3.53125,
      "learning_rate": 4.952486204586834e-05,
      "loss": 0.8758,
      "step": 4582
    },
    {
      "epoch": 0.06500491011128404,
      "grad_norm": 3.171875,
      "learning_rate": 4.9524428421922066e-05,
      "loss": 0.9401,
      "step": 4584
    },
    {
      "epoch": 0.06503327176491025,
      "grad_norm": 3.5,
      "learning_rate": 4.952399460209755e-05,
      "loss": 0.8709,
      "step": 4586
    },
    {
      "epoch": 0.06506163341853646,
      "grad_norm": 2.859375,
      "learning_rate": 4.952356058639825e-05,
      "loss": 0.8392,
      "step": 4588
    },
    {
      "epoch": 0.06508999507216268,
      "grad_norm": 3.15625,
      "learning_rate": 4.9523126374827635e-05,
      "loss": 0.8782,
      "step": 4590
    },
    {
      "epoch": 0.0651183567257889,
      "grad_norm": 3.0,
      "learning_rate": 4.9522691967389175e-05,
      "loss": 0.9035,
      "step": 4592
    },
    {
      "epoch": 0.06514671837941512,
      "grad_norm": 3.09375,
      "learning_rate": 4.952225736408634e-05,
      "loss": 0.8941,
      "step": 4594
    },
    {
      "epoch": 0.06517508003304133,
      "grad_norm": 2.671875,
      "learning_rate": 4.952182256492259e-05,
      "loss": 0.8849,
      "step": 4596
    },
    {
      "epoch": 0.06520344168666754,
      "grad_norm": 3.359375,
      "learning_rate": 4.952138756990142e-05,
      "loss": 0.9264,
      "step": 4598
    },
    {
      "epoch": 0.06523180334029376,
      "grad_norm": 3.59375,
      "learning_rate": 4.952095237902629e-05,
      "loss": 0.9522,
      "step": 4600
    },
    {
      "epoch": 0.06526016499391997,
      "grad_norm": 3.4375,
      "learning_rate": 4.9520516992300675e-05,
      "loss": 0.8927,
      "step": 4602
    },
    {
      "epoch": 0.06528852664754618,
      "grad_norm": 3.046875,
      "learning_rate": 4.952008140972806e-05,
      "loss": 0.9275,
      "step": 4604
    },
    {
      "epoch": 0.0653168883011724,
      "grad_norm": 3.171875,
      "learning_rate": 4.951964563131191e-05,
      "loss": 0.8411,
      "step": 4606
    },
    {
      "epoch": 0.06534524995479861,
      "grad_norm": 3.09375,
      "learning_rate": 4.951920965705572e-05,
      "loss": 0.9227,
      "step": 4608
    },
    {
      "epoch": 0.06537361160842482,
      "grad_norm": 3.53125,
      "learning_rate": 4.9518773486962965e-05,
      "loss": 0.9514,
      "step": 4610
    },
    {
      "epoch": 0.06540197326205105,
      "grad_norm": 3.25,
      "learning_rate": 4.951833712103714e-05,
      "loss": 0.8834,
      "step": 4612
    },
    {
      "epoch": 0.06543033491567726,
      "grad_norm": 3.21875,
      "learning_rate": 4.951790055928171e-05,
      "loss": 0.9361,
      "step": 4614
    },
    {
      "epoch": 0.06545869656930348,
      "grad_norm": 3.515625,
      "learning_rate": 4.9517463801700185e-05,
      "loss": 0.8768,
      "step": 4616
    },
    {
      "epoch": 0.06548705822292969,
      "grad_norm": 3.140625,
      "learning_rate": 4.951702684829602e-05,
      "loss": 0.8992,
      "step": 4618
    },
    {
      "epoch": 0.0655154198765559,
      "grad_norm": 3.3125,
      "learning_rate": 4.951658969907275e-05,
      "loss": 0.9516,
      "step": 4620
    },
    {
      "epoch": 0.06554378153018212,
      "grad_norm": 2.90625,
      "learning_rate": 4.9516152354033826e-05,
      "loss": 0.8974,
      "step": 4622
    },
    {
      "epoch": 0.06557214318380833,
      "grad_norm": 3.078125,
      "learning_rate": 4.951571481318277e-05,
      "loss": 0.8681,
      "step": 4624
    },
    {
      "epoch": 0.06560050483743454,
      "grad_norm": 3.15625,
      "learning_rate": 4.951527707652305e-05,
      "loss": 0.8831,
      "step": 4626
    },
    {
      "epoch": 0.06562886649106076,
      "grad_norm": 2.984375,
      "learning_rate": 4.951483914405819e-05,
      "loss": 0.8918,
      "step": 4628
    },
    {
      "epoch": 0.06565722814468698,
      "grad_norm": 3.0,
      "learning_rate": 4.9514401015791666e-05,
      "loss": 0.9445,
      "step": 4630
    },
    {
      "epoch": 0.0656855897983132,
      "grad_norm": 3.28125,
      "learning_rate": 4.9513962691726986e-05,
      "loss": 0.9211,
      "step": 4632
    },
    {
      "epoch": 0.06571395145193941,
      "grad_norm": 3.515625,
      "learning_rate": 4.9513524171867666e-05,
      "loss": 0.965,
      "step": 4634
    },
    {
      "epoch": 0.06574231310556562,
      "grad_norm": 3.375,
      "learning_rate": 4.951308545621718e-05,
      "loss": 0.9273,
      "step": 4636
    },
    {
      "epoch": 0.06577067475919184,
      "grad_norm": 3.328125,
      "learning_rate": 4.951264654477905e-05,
      "loss": 0.8892,
      "step": 4638
    },
    {
      "epoch": 0.06579903641281805,
      "grad_norm": 3.09375,
      "learning_rate": 4.9512207437556766e-05,
      "loss": 0.8915,
      "step": 4640
    },
    {
      "epoch": 0.06582739806644426,
      "grad_norm": 3.140625,
      "learning_rate": 4.951176813455386e-05,
      "loss": 0.9026,
      "step": 4642
    },
    {
      "epoch": 0.06585575972007048,
      "grad_norm": 3.25,
      "learning_rate": 4.951132863577382e-05,
      "loss": 0.9051,
      "step": 4644
    },
    {
      "epoch": 0.06588412137369669,
      "grad_norm": 2.671875,
      "learning_rate": 4.9510888941220166e-05,
      "loss": 0.8719,
      "step": 4646
    },
    {
      "epoch": 0.0659124830273229,
      "grad_norm": 3.265625,
      "learning_rate": 4.951044905089641e-05,
      "loss": 0.8735,
      "step": 4648
    },
    {
      "epoch": 0.06594084468094913,
      "grad_norm": 3.53125,
      "learning_rate": 4.9510008964806055e-05,
      "loss": 0.9887,
      "step": 4650
    },
    {
      "epoch": 0.06596920633457534,
      "grad_norm": 2.90625,
      "learning_rate": 4.9509568682952627e-05,
      "loss": 0.8866,
      "step": 4652
    },
    {
      "epoch": 0.06599756798820156,
      "grad_norm": 2.9375,
      "learning_rate": 4.950912820533964e-05,
      "loss": 0.9101,
      "step": 4654
    },
    {
      "epoch": 0.06602592964182777,
      "grad_norm": 2.90625,
      "learning_rate": 4.950868753197061e-05,
      "loss": 0.9354,
      "step": 4656
    },
    {
      "epoch": 0.06605429129545398,
      "grad_norm": 3.828125,
      "learning_rate": 4.9508246662849055e-05,
      "loss": 0.9042,
      "step": 4658
    },
    {
      "epoch": 0.0660826529490802,
      "grad_norm": 3.21875,
      "learning_rate": 4.950780559797851e-05,
      "loss": 0.8734,
      "step": 4660
    },
    {
      "epoch": 0.06611101460270641,
      "grad_norm": 3.46875,
      "learning_rate": 4.950736433736248e-05,
      "loss": 0.9635,
      "step": 4662
    },
    {
      "epoch": 0.06613937625633262,
      "grad_norm": 2.71875,
      "learning_rate": 4.950692288100449e-05,
      "loss": 0.8971,
      "step": 4664
    },
    {
      "epoch": 0.06616773790995883,
      "grad_norm": 3.296875,
      "learning_rate": 4.9506481228908085e-05,
      "loss": 0.9066,
      "step": 4666
    },
    {
      "epoch": 0.06619609956358505,
      "grad_norm": 3.09375,
      "learning_rate": 4.950603938107677e-05,
      "loss": 0.9691,
      "step": 4668
    },
    {
      "epoch": 0.06622446121721128,
      "grad_norm": 3.296875,
      "learning_rate": 4.950559733751409e-05,
      "loss": 0.8895,
      "step": 4670
    },
    {
      "epoch": 0.06625282287083749,
      "grad_norm": 3.21875,
      "learning_rate": 4.9505155098223565e-05,
      "loss": 0.8744,
      "step": 4672
    },
    {
      "epoch": 0.0662811845244637,
      "grad_norm": 3.796875,
      "learning_rate": 4.950471266320873e-05,
      "loss": 0.931,
      "step": 4674
    },
    {
      "epoch": 0.06630954617808991,
      "grad_norm": 3.4375,
      "learning_rate": 4.9504270032473127e-05,
      "loss": 0.9097,
      "step": 4676
    },
    {
      "epoch": 0.06633790783171613,
      "grad_norm": 3.46875,
      "learning_rate": 4.950382720602028e-05,
      "loss": 0.9375,
      "step": 4678
    },
    {
      "epoch": 0.06636626948534234,
      "grad_norm": 2.78125,
      "learning_rate": 4.950338418385373e-05,
      "loss": 0.8558,
      "step": 4680
    },
    {
      "epoch": 0.06639463113896855,
      "grad_norm": 2.96875,
      "learning_rate": 4.9502940965977026e-05,
      "loss": 0.8984,
      "step": 4682
    },
    {
      "epoch": 0.06642299279259477,
      "grad_norm": 3.015625,
      "learning_rate": 4.950249755239369e-05,
      "loss": 0.8891,
      "step": 4684
    },
    {
      "epoch": 0.06645135444622098,
      "grad_norm": 3.453125,
      "learning_rate": 4.950205394310727e-05,
      "loss": 0.9366,
      "step": 4686
    },
    {
      "epoch": 0.06647971609984721,
      "grad_norm": 3.234375,
      "learning_rate": 4.9501610138121314e-05,
      "loss": 0.8813,
      "step": 4688
    },
    {
      "epoch": 0.06650807775347342,
      "grad_norm": 2.96875,
      "learning_rate": 4.950116613743936e-05,
      "loss": 0.9062,
      "step": 4690
    },
    {
      "epoch": 0.06653643940709963,
      "grad_norm": 2.875,
      "learning_rate": 4.9500721941064964e-05,
      "loss": 0.862,
      "step": 4692
    },
    {
      "epoch": 0.06656480106072585,
      "grad_norm": 3.109375,
      "learning_rate": 4.950027754900166e-05,
      "loss": 0.9169,
      "step": 4694
    },
    {
      "epoch": 0.06659316271435206,
      "grad_norm": 3.109375,
      "learning_rate": 4.9499832961253013e-05,
      "loss": 0.9028,
      "step": 4696
    },
    {
      "epoch": 0.06662152436797827,
      "grad_norm": 3.21875,
      "learning_rate": 4.949938817782256e-05,
      "loss": 0.8751,
      "step": 4698
    },
    {
      "epoch": 0.06664988602160449,
      "grad_norm": 3.234375,
      "learning_rate": 4.949894319871387e-05,
      "loss": 0.8997,
      "step": 4700
    },
    {
      "epoch": 0.0666782476752307,
      "grad_norm": 3.46875,
      "learning_rate": 4.949849802393047e-05,
      "loss": 0.909,
      "step": 4702
    },
    {
      "epoch": 0.06670660932885691,
      "grad_norm": 2.953125,
      "learning_rate": 4.9498052653475946e-05,
      "loss": 0.8715,
      "step": 4704
    },
    {
      "epoch": 0.06673497098248313,
      "grad_norm": 3.125,
      "learning_rate": 4.9497607087353835e-05,
      "loss": 0.9047,
      "step": 4706
    },
    {
      "epoch": 0.06676333263610935,
      "grad_norm": 3.125,
      "learning_rate": 4.94971613255677e-05,
      "loss": 0.9103,
      "step": 4708
    },
    {
      "epoch": 0.06679169428973557,
      "grad_norm": 3.21875,
      "learning_rate": 4.949671536812111e-05,
      "loss": 0.9101,
      "step": 4710
    },
    {
      "epoch": 0.06682005594336178,
      "grad_norm": 3.125,
      "learning_rate": 4.9496269215017624e-05,
      "loss": 0.8696,
      "step": 4712
    },
    {
      "epoch": 0.066848417596988,
      "grad_norm": 2.921875,
      "learning_rate": 4.9495822866260786e-05,
      "loss": 0.9196,
      "step": 4714
    },
    {
      "epoch": 0.06687677925061421,
      "grad_norm": 3.296875,
      "learning_rate": 4.949537632185419e-05,
      "loss": 0.9438,
      "step": 4716
    },
    {
      "epoch": 0.06690514090424042,
      "grad_norm": 3.4375,
      "learning_rate": 4.9494929581801384e-05,
      "loss": 0.9332,
      "step": 4718
    },
    {
      "epoch": 0.06693350255786663,
      "grad_norm": 2.859375,
      "learning_rate": 4.949448264610595e-05,
      "loss": 0.86,
      "step": 4720
    },
    {
      "epoch": 0.06696186421149285,
      "grad_norm": 2.96875,
      "learning_rate": 4.949403551477144e-05,
      "loss": 0.9432,
      "step": 4722
    },
    {
      "epoch": 0.06699022586511906,
      "grad_norm": 3.5,
      "learning_rate": 4.949358818780143e-05,
      "loss": 0.8539,
      "step": 4724
    },
    {
      "epoch": 0.06701858751874529,
      "grad_norm": 3.1875,
      "learning_rate": 4.949314066519951e-05,
      "loss": 0.8974,
      "step": 4726
    },
    {
      "epoch": 0.0670469491723715,
      "grad_norm": 3.3125,
      "learning_rate": 4.9492692946969234e-05,
      "loss": 0.9282,
      "step": 4728
    },
    {
      "epoch": 0.06707531082599771,
      "grad_norm": 3.203125,
      "learning_rate": 4.949224503311419e-05,
      "loss": 0.9234,
      "step": 4730
    },
    {
      "epoch": 0.06710367247962393,
      "grad_norm": 3.015625,
      "learning_rate": 4.9491796923637945e-05,
      "loss": 0.923,
      "step": 4732
    },
    {
      "epoch": 0.06713203413325014,
      "grad_norm": 3.21875,
      "learning_rate": 4.9491348618544085e-05,
      "loss": 0.9557,
      "step": 4734
    },
    {
      "epoch": 0.06716039578687635,
      "grad_norm": 3.1875,
      "learning_rate": 4.94909001178362e-05,
      "loss": 0.8622,
      "step": 4736
    },
    {
      "epoch": 0.06718875744050257,
      "grad_norm": 3.09375,
      "learning_rate": 4.949045142151785e-05,
      "loss": 0.8679,
      "step": 4738
    },
    {
      "epoch": 0.06721711909412878,
      "grad_norm": 3.140625,
      "learning_rate": 4.949000252959264e-05,
      "loss": 0.9219,
      "step": 4740
    },
    {
      "epoch": 0.067245480747755,
      "grad_norm": 3.03125,
      "learning_rate": 4.948955344206414e-05,
      "loss": 0.9315,
      "step": 4742
    },
    {
      "epoch": 0.0672738424013812,
      "grad_norm": 3.15625,
      "learning_rate": 4.948910415893595e-05,
      "loss": 0.9132,
      "step": 4744
    },
    {
      "epoch": 0.06730220405500743,
      "grad_norm": 3.34375,
      "learning_rate": 4.9488654680211645e-05,
      "loss": 0.9293,
      "step": 4746
    },
    {
      "epoch": 0.06733056570863365,
      "grad_norm": 3.140625,
      "learning_rate": 4.9488205005894826e-05,
      "loss": 0.9094,
      "step": 4748
    },
    {
      "epoch": 0.06735892736225986,
      "grad_norm": 3.078125,
      "learning_rate": 4.948775513598908e-05,
      "loss": 0.8796,
      "step": 4750
    },
    {
      "epoch": 0.06738728901588607,
      "grad_norm": 3.046875,
      "learning_rate": 4.9487305070498e-05,
      "loss": 0.8686,
      "step": 4752
    },
    {
      "epoch": 0.06741565066951229,
      "grad_norm": 3.03125,
      "learning_rate": 4.9486854809425175e-05,
      "loss": 0.8891,
      "step": 4754
    },
    {
      "epoch": 0.0674440123231385,
      "grad_norm": 3.078125,
      "learning_rate": 4.948640435277422e-05,
      "loss": 0.9095,
      "step": 4756
    },
    {
      "epoch": 0.06747237397676471,
      "grad_norm": 3.234375,
      "learning_rate": 4.948595370054871e-05,
      "loss": 0.9015,
      "step": 4758
    },
    {
      "epoch": 0.06750073563039093,
      "grad_norm": 3.109375,
      "learning_rate": 4.9485502852752254e-05,
      "loss": 0.8796,
      "step": 4760
    },
    {
      "epoch": 0.06752909728401714,
      "grad_norm": 3.359375,
      "learning_rate": 4.948505180938846e-05,
      "loss": 0.8553,
      "step": 4762
    },
    {
      "epoch": 0.06755745893764335,
      "grad_norm": 3.09375,
      "learning_rate": 4.948460057046092e-05,
      "loss": 0.9223,
      "step": 4764
    },
    {
      "epoch": 0.06758582059126958,
      "grad_norm": 3.078125,
      "learning_rate": 4.948414913597325e-05,
      "loss": 0.9168,
      "step": 4766
    },
    {
      "epoch": 0.06761418224489579,
      "grad_norm": 3.25,
      "learning_rate": 4.948369750592903e-05,
      "loss": 0.9121,
      "step": 4768
    },
    {
      "epoch": 0.067642543898522,
      "grad_norm": 3.109375,
      "learning_rate": 4.94832456803319e-05,
      "loss": 0.9086,
      "step": 4770
    },
    {
      "epoch": 0.06767090555214822,
      "grad_norm": 3.109375,
      "learning_rate": 4.948279365918544e-05,
      "loss": 0.9111,
      "step": 4772
    },
    {
      "epoch": 0.06769926720577443,
      "grad_norm": 3.296875,
      "learning_rate": 4.948234144249329e-05,
      "loss": 0.873,
      "step": 4774
    },
    {
      "epoch": 0.06772762885940065,
      "grad_norm": 3.421875,
      "learning_rate": 4.948188903025904e-05,
      "loss": 0.8919,
      "step": 4776
    },
    {
      "epoch": 0.06775599051302686,
      "grad_norm": 3.234375,
      "learning_rate": 4.948143642248631e-05,
      "loss": 0.8923,
      "step": 4778
    },
    {
      "epoch": 0.06778435216665307,
      "grad_norm": 3.5625,
      "learning_rate": 4.948098361917871e-05,
      "loss": 0.9039,
      "step": 4780
    },
    {
      "epoch": 0.06781271382027929,
      "grad_norm": 2.984375,
      "learning_rate": 4.948053062033986e-05,
      "loss": 0.9143,
      "step": 4782
    },
    {
      "epoch": 0.06784107547390551,
      "grad_norm": 3.15625,
      "learning_rate": 4.9480077425973396e-05,
      "loss": 0.9118,
      "step": 4784
    },
    {
      "epoch": 0.06786943712753173,
      "grad_norm": 3.484375,
      "learning_rate": 4.94796240360829e-05,
      "loss": 0.8754,
      "step": 4786
    },
    {
      "epoch": 0.06789779878115794,
      "grad_norm": 2.796875,
      "learning_rate": 4.947917045067202e-05,
      "loss": 0.89,
      "step": 4788
    },
    {
      "epoch": 0.06792616043478415,
      "grad_norm": 3.09375,
      "learning_rate": 4.947871666974437e-05,
      "loss": 0.9133,
      "step": 4790
    },
    {
      "epoch": 0.06795452208841037,
      "grad_norm": 3.375,
      "learning_rate": 4.947826269330359e-05,
      "loss": 0.8694,
      "step": 4792
    },
    {
      "epoch": 0.06798288374203658,
      "grad_norm": 3.140625,
      "learning_rate": 4.947780852135328e-05,
      "loss": 0.9042,
      "step": 4794
    },
    {
      "epoch": 0.06801124539566279,
      "grad_norm": 3.359375,
      "learning_rate": 4.9477354153897093e-05,
      "loss": 0.9627,
      "step": 4796
    },
    {
      "epoch": 0.068039607049289,
      "grad_norm": 3.078125,
      "learning_rate": 4.947689959093864e-05,
      "loss": 0.9197,
      "step": 4798
    },
    {
      "epoch": 0.06806796870291522,
      "grad_norm": 2.828125,
      "learning_rate": 4.947644483248155e-05,
      "loss": 0.8959,
      "step": 4800
    },
    {
      "epoch": 0.06809633035654143,
      "grad_norm": 3.4375,
      "learning_rate": 4.947598987852947e-05,
      "loss": 0.899,
      "step": 4802
    },
    {
      "epoch": 0.06812469201016766,
      "grad_norm": 3.359375,
      "learning_rate": 4.9475534729086026e-05,
      "loss": 0.9141,
      "step": 4804
    },
    {
      "epoch": 0.06815305366379387,
      "grad_norm": 2.953125,
      "learning_rate": 4.947507938415485e-05,
      "loss": 0.8935,
      "step": 4806
    },
    {
      "epoch": 0.06818141531742009,
      "grad_norm": 3.1875,
      "learning_rate": 4.947462384373959e-05,
      "loss": 0.8559,
      "step": 4808
    },
    {
      "epoch": 0.0682097769710463,
      "grad_norm": 3.140625,
      "learning_rate": 4.947416810784386e-05,
      "loss": 0.9231,
      "step": 4810
    },
    {
      "epoch": 0.06823813862467251,
      "grad_norm": 3.078125,
      "learning_rate": 4.947371217647133e-05,
      "loss": 0.9026,
      "step": 4812
    },
    {
      "epoch": 0.06826650027829873,
      "grad_norm": 3.328125,
      "learning_rate": 4.9473256049625627e-05,
      "loss": 0.8978,
      "step": 4814
    },
    {
      "epoch": 0.06829486193192494,
      "grad_norm": 3.125,
      "learning_rate": 4.94727997273104e-05,
      "loss": 0.9079,
      "step": 4816
    },
    {
      "epoch": 0.06832322358555115,
      "grad_norm": 3.09375,
      "learning_rate": 4.947234320952928e-05,
      "loss": 0.9053,
      "step": 4818
    },
    {
      "epoch": 0.06835158523917736,
      "grad_norm": 2.8125,
      "learning_rate": 4.947188649628593e-05,
      "loss": 0.9019,
      "step": 4820
    },
    {
      "epoch": 0.06837994689280358,
      "grad_norm": 3.015625,
      "learning_rate": 4.9471429587583985e-05,
      "loss": 0.9115,
      "step": 4822
    },
    {
      "epoch": 0.0684083085464298,
      "grad_norm": 3.578125,
      "learning_rate": 4.9470972483427105e-05,
      "loss": 0.8846,
      "step": 4824
    },
    {
      "epoch": 0.06843667020005602,
      "grad_norm": 2.9375,
      "learning_rate": 4.947051518381893e-05,
      "loss": 0.8998,
      "step": 4826
    },
    {
      "epoch": 0.06846503185368223,
      "grad_norm": 2.96875,
      "learning_rate": 4.947005768876312e-05,
      "loss": 0.8873,
      "step": 4828
    },
    {
      "epoch": 0.06849339350730844,
      "grad_norm": 3.1875,
      "learning_rate": 4.946959999826333e-05,
      "loss": 0.9077,
      "step": 4830
    },
    {
      "epoch": 0.06852175516093466,
      "grad_norm": 2.890625,
      "learning_rate": 4.946914211232321e-05,
      "loss": 0.8898,
      "step": 4832
    },
    {
      "epoch": 0.06855011681456087,
      "grad_norm": 2.96875,
      "learning_rate": 4.946868403094642e-05,
      "loss": 0.8703,
      "step": 4834
    },
    {
      "epoch": 0.06857847846818708,
      "grad_norm": 4.125,
      "learning_rate": 4.946822575413662e-05,
      "loss": 0.9202,
      "step": 4836
    },
    {
      "epoch": 0.0686068401218133,
      "grad_norm": 3.046875,
      "learning_rate": 4.946776728189746e-05,
      "loss": 0.9301,
      "step": 4838
    },
    {
      "epoch": 0.06863520177543951,
      "grad_norm": 3.21875,
      "learning_rate": 4.946730861423262e-05,
      "loss": 0.9239,
      "step": 4840
    },
    {
      "epoch": 0.06866356342906574,
      "grad_norm": 3.0,
      "learning_rate": 4.9466849751145754e-05,
      "loss": 0.8734,
      "step": 4842
    },
    {
      "epoch": 0.06869192508269195,
      "grad_norm": 3.734375,
      "learning_rate": 4.946639069264053e-05,
      "loss": 0.9039,
      "step": 4844
    },
    {
      "epoch": 0.06872028673631816,
      "grad_norm": 3.390625,
      "learning_rate": 4.94659314387206e-05,
      "loss": 0.9275,
      "step": 4846
    },
    {
      "epoch": 0.06874864838994438,
      "grad_norm": 2.828125,
      "learning_rate": 4.946547198938966e-05,
      "loss": 0.9069,
      "step": 4848
    },
    {
      "epoch": 0.06877701004357059,
      "grad_norm": 3.46875,
      "learning_rate": 4.946501234465135e-05,
      "loss": 0.8998,
      "step": 4850
    },
    {
      "epoch": 0.0688053716971968,
      "grad_norm": 3.265625,
      "learning_rate": 4.9464552504509353e-05,
      "loss": 0.8747,
      "step": 4852
    },
    {
      "epoch": 0.06883373335082302,
      "grad_norm": 3.5,
      "learning_rate": 4.946409246896735e-05,
      "loss": 0.9337,
      "step": 4854
    },
    {
      "epoch": 0.06886209500444923,
      "grad_norm": 3.296875,
      "learning_rate": 4.946363223802901e-05,
      "loss": 0.8646,
      "step": 4856
    },
    {
      "epoch": 0.06889045665807544,
      "grad_norm": 3.1875,
      "learning_rate": 4.9463171811698004e-05,
      "loss": 0.9242,
      "step": 4858
    },
    {
      "epoch": 0.06891881831170166,
      "grad_norm": 3.640625,
      "learning_rate": 4.946271118997802e-05,
      "loss": 0.9542,
      "step": 4860
    },
    {
      "epoch": 0.06894717996532788,
      "grad_norm": 3.328125,
      "learning_rate": 4.946225037287272e-05,
      "loss": 0.8801,
      "step": 4862
    },
    {
      "epoch": 0.0689755416189541,
      "grad_norm": 3.171875,
      "learning_rate": 4.9461789360385805e-05,
      "loss": 0.9439,
      "step": 4864
    },
    {
      "epoch": 0.06900390327258031,
      "grad_norm": 3.328125,
      "learning_rate": 4.946132815252094e-05,
      "loss": 0.9156,
      "step": 4866
    },
    {
      "epoch": 0.06903226492620652,
      "grad_norm": 3.484375,
      "learning_rate": 4.946086674928181e-05,
      "loss": 0.9218,
      "step": 4868
    },
    {
      "epoch": 0.06906062657983274,
      "grad_norm": 3.171875,
      "learning_rate": 4.946040515067211e-05,
      "loss": 0.8954,
      "step": 4870
    },
    {
      "epoch": 0.06908898823345895,
      "grad_norm": 3.515625,
      "learning_rate": 4.945994335669552e-05,
      "loss": 0.8171,
      "step": 4872
    },
    {
      "epoch": 0.06911734988708516,
      "grad_norm": 3.234375,
      "learning_rate": 4.945948136735574e-05,
      "loss": 0.9116,
      "step": 4874
    },
    {
      "epoch": 0.06914571154071138,
      "grad_norm": 3.34375,
      "learning_rate": 4.945901918265645e-05,
      "loss": 0.9184,
      "step": 4876
    },
    {
      "epoch": 0.06917407319433759,
      "grad_norm": 3.140625,
      "learning_rate": 4.945855680260133e-05,
      "loss": 0.8684,
      "step": 4878
    },
    {
      "epoch": 0.06920243484796382,
      "grad_norm": 2.578125,
      "learning_rate": 4.94580942271941e-05,
      "loss": 0.791,
      "step": 4880
    },
    {
      "epoch": 0.06923079650159003,
      "grad_norm": 3.140625,
      "learning_rate": 4.945763145643844e-05,
      "loss": 0.8931,
      "step": 4882
    },
    {
      "epoch": 0.06925915815521624,
      "grad_norm": 2.828125,
      "learning_rate": 4.945716849033803e-05,
      "loss": 0.9043,
      "step": 4884
    },
    {
      "epoch": 0.06928751980884246,
      "grad_norm": 3.3125,
      "learning_rate": 4.94567053288966e-05,
      "loss": 0.8478,
      "step": 4886
    },
    {
      "epoch": 0.06931588146246867,
      "grad_norm": 2.859375,
      "learning_rate": 4.945624197211782e-05,
      "loss": 0.8988,
      "step": 4888
    },
    {
      "epoch": 0.06934424311609488,
      "grad_norm": 2.984375,
      "learning_rate": 4.9455778420005415e-05,
      "loss": 0.8681,
      "step": 4890
    },
    {
      "epoch": 0.0693726047697211,
      "grad_norm": 3.0625,
      "learning_rate": 4.945531467256307e-05,
      "loss": 0.9221,
      "step": 4892
    },
    {
      "epoch": 0.06940096642334731,
      "grad_norm": 3.078125,
      "learning_rate": 4.9454850729794503e-05,
      "loss": 0.9234,
      "step": 4894
    },
    {
      "epoch": 0.06942932807697352,
      "grad_norm": 3.28125,
      "learning_rate": 4.9454386591703404e-05,
      "loss": 0.8899,
      "step": 4896
    },
    {
      "epoch": 0.06945768973059974,
      "grad_norm": 3.125,
      "learning_rate": 4.945392225829349e-05,
      "loss": 0.9384,
      "step": 4898
    },
    {
      "epoch": 0.06948605138422596,
      "grad_norm": 3.03125,
      "learning_rate": 4.945345772956848e-05,
      "loss": 0.912,
      "step": 4900
    },
    {
      "epoch": 0.06951441303785218,
      "grad_norm": 2.828125,
      "learning_rate": 4.945299300553206e-05,
      "loss": 0.9086,
      "step": 4902
    },
    {
      "epoch": 0.06954277469147839,
      "grad_norm": 2.65625,
      "learning_rate": 4.945252808618795e-05,
      "loss": 0.8864,
      "step": 4904
    },
    {
      "epoch": 0.0695711363451046,
      "grad_norm": 3.140625,
      "learning_rate": 4.9452062971539875e-05,
      "loss": 0.9153,
      "step": 4906
    },
    {
      "epoch": 0.06959949799873082,
      "grad_norm": 3.0625,
      "learning_rate": 4.945159766159154e-05,
      "loss": 0.8683,
      "step": 4908
    },
    {
      "epoch": 0.06962785965235703,
      "grad_norm": 2.90625,
      "learning_rate": 4.945113215634666e-05,
      "loss": 0.8259,
      "step": 4910
    },
    {
      "epoch": 0.06965622130598324,
      "grad_norm": 3.328125,
      "learning_rate": 4.9450666455808965e-05,
      "loss": 0.919,
      "step": 4912
    },
    {
      "epoch": 0.06968458295960946,
      "grad_norm": 3.0625,
      "learning_rate": 4.9450200559982165e-05,
      "loss": 0.8744,
      "step": 4914
    },
    {
      "epoch": 0.06971294461323567,
      "grad_norm": 3.328125,
      "learning_rate": 4.9449734468869973e-05,
      "loss": 0.8782,
      "step": 4916
    },
    {
      "epoch": 0.06974130626686188,
      "grad_norm": 2.90625,
      "learning_rate": 4.9449268182476125e-05,
      "loss": 0.9095,
      "step": 4918
    },
    {
      "epoch": 0.06976966792048811,
      "grad_norm": 3.28125,
      "learning_rate": 4.9448801700804346e-05,
      "loss": 0.884,
      "step": 4920
    },
    {
      "epoch": 0.06979802957411432,
      "grad_norm": 2.890625,
      "learning_rate": 4.944833502385835e-05,
      "loss": 0.8555,
      "step": 4922
    },
    {
      "epoch": 0.06982639122774054,
      "grad_norm": 3.140625,
      "learning_rate": 4.944786815164187e-05,
      "loss": 0.8914,
      "step": 4924
    },
    {
      "epoch": 0.06985475288136675,
      "grad_norm": 3.296875,
      "learning_rate": 4.944740108415865e-05,
      "loss": 0.8677,
      "step": 4926
    },
    {
      "epoch": 0.06988311453499296,
      "grad_norm": 3.421875,
      "learning_rate": 4.94469338214124e-05,
      "loss": 0.9808,
      "step": 4928
    },
    {
      "epoch": 0.06991147618861918,
      "grad_norm": 3.6875,
      "learning_rate": 4.944646636340685e-05,
      "loss": 0.924,
      "step": 4930
    },
    {
      "epoch": 0.06993983784224539,
      "grad_norm": 3.1875,
      "learning_rate": 4.944599871014576e-05,
      "loss": 0.9308,
      "step": 4932
    },
    {
      "epoch": 0.0699681994958716,
      "grad_norm": 2.953125,
      "learning_rate": 4.9445530861632836e-05,
      "loss": 0.9063,
      "step": 4934
    },
    {
      "epoch": 0.06999656114949782,
      "grad_norm": 2.75,
      "learning_rate": 4.944506281787182e-05,
      "loss": 0.8603,
      "step": 4936
    },
    {
      "epoch": 0.07002492280312404,
      "grad_norm": 2.765625,
      "learning_rate": 4.9444594578866464e-05,
      "loss": 0.8926,
      "step": 4938
    },
    {
      "epoch": 0.07005328445675026,
      "grad_norm": 3.09375,
      "learning_rate": 4.944412614462051e-05,
      "loss": 0.9773,
      "step": 4940
    },
    {
      "epoch": 0.07008164611037647,
      "grad_norm": 2.96875,
      "learning_rate": 4.9443657515137674e-05,
      "loss": 0.8809,
      "step": 4942
    },
    {
      "epoch": 0.07011000776400268,
      "grad_norm": 3.125,
      "learning_rate": 4.944318869042173e-05,
      "loss": 0.8797,
      "step": 4944
    },
    {
      "epoch": 0.0701383694176289,
      "grad_norm": 3.03125,
      "learning_rate": 4.9442719670476396e-05,
      "loss": 0.851,
      "step": 4946
    },
    {
      "epoch": 0.07016673107125511,
      "grad_norm": 3.015625,
      "learning_rate": 4.944225045530543e-05,
      "loss": 0.9138,
      "step": 4948
    },
    {
      "epoch": 0.07019509272488132,
      "grad_norm": 3.171875,
      "learning_rate": 4.944178104491258e-05,
      "loss": 0.8986,
      "step": 4950
    },
    {
      "epoch": 0.07022345437850754,
      "grad_norm": 3.140625,
      "learning_rate": 4.944131143930161e-05,
      "loss": 0.9003,
      "step": 4952
    },
    {
      "epoch": 0.07025181603213375,
      "grad_norm": 3.0,
      "learning_rate": 4.944084163847624e-05,
      "loss": 0.9021,
      "step": 4954
    },
    {
      "epoch": 0.07028017768575996,
      "grad_norm": 3.09375,
      "learning_rate": 4.944037164244024e-05,
      "loss": 0.9215,
      "step": 4956
    },
    {
      "epoch": 0.07030853933938619,
      "grad_norm": 3.0625,
      "learning_rate": 4.943990145119736e-05,
      "loss": 0.9127,
      "step": 4958
    },
    {
      "epoch": 0.0703369009930124,
      "grad_norm": 3.109375,
      "learning_rate": 4.943943106475136e-05,
      "loss": 0.8826,
      "step": 4960
    },
    {
      "epoch": 0.07036526264663862,
      "grad_norm": 3.3125,
      "learning_rate": 4.943896048310599e-05,
      "loss": 0.8451,
      "step": 4962
    },
    {
      "epoch": 0.07039362430026483,
      "grad_norm": 3.015625,
      "learning_rate": 4.943848970626502e-05,
      "loss": 0.9363,
      "step": 4964
    },
    {
      "epoch": 0.07042198595389104,
      "grad_norm": 3.140625,
      "learning_rate": 4.943801873423219e-05,
      "loss": 0.9009,
      "step": 4966
    },
    {
      "epoch": 0.07045034760751726,
      "grad_norm": 3.21875,
      "learning_rate": 4.9437547567011286e-05,
      "loss": 0.9169,
      "step": 4968
    },
    {
      "epoch": 0.07047870926114347,
      "grad_norm": 3.015625,
      "learning_rate": 4.9437076204606056e-05,
      "loss": 0.8814,
      "step": 4970
    },
    {
      "epoch": 0.07050707091476968,
      "grad_norm": 3.296875,
      "learning_rate": 4.943660464702027e-05,
      "loss": 0.8545,
      "step": 4972
    },
    {
      "epoch": 0.0705354325683959,
      "grad_norm": 3.59375,
      "learning_rate": 4.943613289425769e-05,
      "loss": 0.9296,
      "step": 4974
    },
    {
      "epoch": 0.07056379422202211,
      "grad_norm": 3.40625,
      "learning_rate": 4.943566094632209e-05,
      "loss": 0.8731,
      "step": 4976
    },
    {
      "epoch": 0.07059215587564834,
      "grad_norm": 3.15625,
      "learning_rate": 4.943518880321723e-05,
      "loss": 0.9074,
      "step": 4978
    },
    {
      "epoch": 0.07062051752927455,
      "grad_norm": 3.6875,
      "learning_rate": 4.9434716464946895e-05,
      "loss": 0.8793,
      "step": 4980
    },
    {
      "epoch": 0.07064887918290076,
      "grad_norm": 3.125,
      "learning_rate": 4.943424393151485e-05,
      "loss": 0.9093,
      "step": 4982
    },
    {
      "epoch": 0.07067724083652697,
      "grad_norm": 3.40625,
      "learning_rate": 4.9433771202924864e-05,
      "loss": 0.9451,
      "step": 4984
    },
    {
      "epoch": 0.07070560249015319,
      "grad_norm": 3.1875,
      "learning_rate": 4.943329827918071e-05,
      "loss": 0.9089,
      "step": 4986
    },
    {
      "epoch": 0.0707339641437794,
      "grad_norm": 3.15625,
      "learning_rate": 4.9432825160286186e-05,
      "loss": 0.8978,
      "step": 4988
    },
    {
      "epoch": 0.07076232579740561,
      "grad_norm": 3.171875,
      "learning_rate": 4.943235184624505e-05,
      "loss": 0.9008,
      "step": 4990
    },
    {
      "epoch": 0.07079068745103183,
      "grad_norm": 2.921875,
      "learning_rate": 4.943187833706109e-05,
      "loss": 0.8975,
      "step": 4992
    },
    {
      "epoch": 0.07081904910465804,
      "grad_norm": 3.21875,
      "learning_rate": 4.94314046327381e-05,
      "loss": 0.9021,
      "step": 4994
    },
    {
      "epoch": 0.07084741075828427,
      "grad_norm": 3.21875,
      "learning_rate": 4.943093073327984e-05,
      "loss": 0.95,
      "step": 4996
    },
    {
      "epoch": 0.07087577241191048,
      "grad_norm": 3.1875,
      "learning_rate": 4.943045663869011e-05,
      "loss": 0.9109,
      "step": 4998
    },
    {
      "epoch": 0.0709041340655367,
      "grad_norm": 2.84375,
      "learning_rate": 4.942998234897269e-05,
      "loss": 0.8498,
      "step": 5000
    },
    {
      "epoch": 0.07093249571916291,
      "grad_norm": 3.265625,
      "learning_rate": 4.9429507864131375e-05,
      "loss": 0.9046,
      "step": 5002
    },
    {
      "epoch": 0.07096085737278912,
      "grad_norm": 3.34375,
      "learning_rate": 4.942903318416995e-05,
      "loss": 0.9088,
      "step": 5004
    },
    {
      "epoch": 0.07098921902641533,
      "grad_norm": 3.203125,
      "learning_rate": 4.942855830909221e-05,
      "loss": 0.841,
      "step": 5006
    },
    {
      "epoch": 0.07101758068004155,
      "grad_norm": 3.359375,
      "learning_rate": 4.9428083238901945e-05,
      "loss": 0.9115,
      "step": 5008
    },
    {
      "epoch": 0.07104594233366776,
      "grad_norm": 2.96875,
      "learning_rate": 4.9427607973602946e-05,
      "loss": 0.8759,
      "step": 5010
    },
    {
      "epoch": 0.07107430398729397,
      "grad_norm": 3.359375,
      "learning_rate": 4.9427132513199015e-05,
      "loss": 0.923,
      "step": 5012
    },
    {
      "epoch": 0.07110266564092019,
      "grad_norm": 3.578125,
      "learning_rate": 4.9426656857693946e-05,
      "loss": 0.8979,
      "step": 5014
    },
    {
      "epoch": 0.07113102729454641,
      "grad_norm": 3.359375,
      "learning_rate": 4.942618100709154e-05,
      "loss": 0.9151,
      "step": 5016
    },
    {
      "epoch": 0.07115938894817263,
      "grad_norm": 3.359375,
      "learning_rate": 4.94257049613956e-05,
      "loss": 0.9702,
      "step": 5018
    },
    {
      "epoch": 0.07118775060179884,
      "grad_norm": 3.015625,
      "learning_rate": 4.942522872060993e-05,
      "loss": 0.8755,
      "step": 5020
    },
    {
      "epoch": 0.07121611225542505,
      "grad_norm": 3.015625,
      "learning_rate": 4.942475228473832e-05,
      "loss": 0.9218,
      "step": 5022
    },
    {
      "epoch": 0.07124447390905127,
      "grad_norm": 3.359375,
      "learning_rate": 4.942427565378458e-05,
      "loss": 0.8962,
      "step": 5024
    },
    {
      "epoch": 0.07127283556267748,
      "grad_norm": 3.3125,
      "learning_rate": 4.942379882775253e-05,
      "loss": 0.9124,
      "step": 5026
    },
    {
      "epoch": 0.0713011972163037,
      "grad_norm": 2.71875,
      "learning_rate": 4.942332180664597e-05,
      "loss": 0.8774,
      "step": 5028
    },
    {
      "epoch": 0.07132955886992991,
      "grad_norm": 2.84375,
      "learning_rate": 4.942284459046871e-05,
      "loss": 0.8982,
      "step": 5030
    },
    {
      "epoch": 0.07135792052355612,
      "grad_norm": 3.40625,
      "learning_rate": 4.9422367179224555e-05,
      "loss": 0.9087,
      "step": 5032
    },
    {
      "epoch": 0.07138628217718235,
      "grad_norm": 3.640625,
      "learning_rate": 4.942188957291732e-05,
      "loss": 0.9145,
      "step": 5034
    },
    {
      "epoch": 0.07141464383080856,
      "grad_norm": 3.625,
      "learning_rate": 4.9421411771550844e-05,
      "loss": 0.9258,
      "step": 5036
    },
    {
      "epoch": 0.07144300548443477,
      "grad_norm": 2.8125,
      "learning_rate": 4.9420933775128907e-05,
      "loss": 0.9102,
      "step": 5038
    },
    {
      "epoch": 0.07147136713806099,
      "grad_norm": 3.1875,
      "learning_rate": 4.9420455583655346e-05,
      "loss": 0.8685,
      "step": 5040
    },
    {
      "epoch": 0.0714997287916872,
      "grad_norm": 3.203125,
      "learning_rate": 4.9419977197133984e-05,
      "loss": 0.9087,
      "step": 5042
    },
    {
      "epoch": 0.07152809044531341,
      "grad_norm": 3.03125,
      "learning_rate": 4.941949861556863e-05,
      "loss": 0.9242,
      "step": 5044
    },
    {
      "epoch": 0.07155645209893963,
      "grad_norm": 2.953125,
      "learning_rate": 4.94190198389631e-05,
      "loss": 0.8829,
      "step": 5046
    },
    {
      "epoch": 0.07158481375256584,
      "grad_norm": 3.15625,
      "learning_rate": 4.941854086732125e-05,
      "loss": 0.8801,
      "step": 5048
    },
    {
      "epoch": 0.07161317540619205,
      "grad_norm": 3.34375,
      "learning_rate": 4.941806170064687e-05,
      "loss": 0.9401,
      "step": 5050
    },
    {
      "epoch": 0.07164153705981827,
      "grad_norm": 3.203125,
      "learning_rate": 4.9417582338943815e-05,
      "loss": 0.921,
      "step": 5052
    },
    {
      "epoch": 0.0716698987134445,
      "grad_norm": 2.875,
      "learning_rate": 4.9417102782215896e-05,
      "loss": 0.927,
      "step": 5054
    },
    {
      "epoch": 0.0716982603670707,
      "grad_norm": 3.078125,
      "learning_rate": 4.941662303046695e-05,
      "loss": 0.9085,
      "step": 5056
    },
    {
      "epoch": 0.07172662202069692,
      "grad_norm": 2.828125,
      "learning_rate": 4.94161430837008e-05,
      "loss": 0.8364,
      "step": 5058
    },
    {
      "epoch": 0.07175498367432313,
      "grad_norm": 3.703125,
      "learning_rate": 4.941566294192129e-05,
      "loss": 0.9031,
      "step": 5060
    },
    {
      "epoch": 0.07178334532794935,
      "grad_norm": 3.734375,
      "learning_rate": 4.9415182605132255e-05,
      "loss": 0.9796,
      "step": 5062
    },
    {
      "epoch": 0.07181170698157556,
      "grad_norm": 3.015625,
      "learning_rate": 4.941470207333753e-05,
      "loss": 0.8777,
      "step": 5064
    },
    {
      "epoch": 0.07184006863520177,
      "grad_norm": 3.328125,
      "learning_rate": 4.941422134654095e-05,
      "loss": 0.9068,
      "step": 5066
    },
    {
      "epoch": 0.07186843028882799,
      "grad_norm": 2.953125,
      "learning_rate": 4.9413740424746345e-05,
      "loss": 0.8382,
      "step": 5068
    },
    {
      "epoch": 0.0718967919424542,
      "grad_norm": 2.96875,
      "learning_rate": 4.9413259307957576e-05,
      "loss": 0.8776,
      "step": 5070
    },
    {
      "epoch": 0.07192515359608041,
      "grad_norm": 2.875,
      "learning_rate": 4.9412777996178474e-05,
      "loss": 0.909,
      "step": 5072
    },
    {
      "epoch": 0.07195351524970664,
      "grad_norm": 2.953125,
      "learning_rate": 4.941229648941289e-05,
      "loss": 0.9247,
      "step": 5074
    },
    {
      "epoch": 0.07198187690333285,
      "grad_norm": 2.78125,
      "learning_rate": 4.9411814787664664e-05,
      "loss": 0.9028,
      "step": 5076
    },
    {
      "epoch": 0.07201023855695907,
      "grad_norm": 3.390625,
      "learning_rate": 4.9411332890937646e-05,
      "loss": 0.9285,
      "step": 5078
    },
    {
      "epoch": 0.07203860021058528,
      "grad_norm": 2.953125,
      "learning_rate": 4.941085079923567e-05,
      "loss": 0.9397,
      "step": 5080
    },
    {
      "epoch": 0.07206696186421149,
      "grad_norm": 3.25,
      "learning_rate": 4.9410368512562624e-05,
      "loss": 0.8836,
      "step": 5082
    },
    {
      "epoch": 0.0720953235178377,
      "grad_norm": 3.015625,
      "learning_rate": 4.940988603092232e-05,
      "loss": 0.9514,
      "step": 5084
    },
    {
      "epoch": 0.07212368517146392,
      "grad_norm": 3.234375,
      "learning_rate": 4.940940335431863e-05,
      "loss": 0.8837,
      "step": 5086
    },
    {
      "epoch": 0.07215204682509013,
      "grad_norm": 3.265625,
      "learning_rate": 4.940892048275542e-05,
      "loss": 0.8616,
      "step": 5088
    },
    {
      "epoch": 0.07218040847871635,
      "grad_norm": 3.3125,
      "learning_rate": 4.940843741623652e-05,
      "loss": 0.8934,
      "step": 5090
    },
    {
      "epoch": 0.07220877013234257,
      "grad_norm": 3.265625,
      "learning_rate": 4.94079541547658e-05,
      "loss": 0.95,
      "step": 5092
    },
    {
      "epoch": 0.07223713178596879,
      "grad_norm": 2.875,
      "learning_rate": 4.940747069834713e-05,
      "loss": 0.8857,
      "step": 5094
    },
    {
      "epoch": 0.072265493439595,
      "grad_norm": 3.0,
      "learning_rate": 4.940698704698436e-05,
      "loss": 0.8629,
      "step": 5096
    },
    {
      "epoch": 0.07229385509322121,
      "grad_norm": 3.765625,
      "learning_rate": 4.940650320068137e-05,
      "loss": 0.8864,
      "step": 5098
    },
    {
      "epoch": 0.07232221674684743,
      "grad_norm": 3.15625,
      "learning_rate": 4.940601915944199e-05,
      "loss": 0.9356,
      "step": 5100
    },
    {
      "epoch": 0.07235057840047364,
      "grad_norm": 3.265625,
      "learning_rate": 4.940553492327012e-05,
      "loss": 0.9046,
      "step": 5102
    },
    {
      "epoch": 0.07237894005409985,
      "grad_norm": 3.3125,
      "learning_rate": 4.9405050492169614e-05,
      "loss": 0.9554,
      "step": 5104
    },
    {
      "epoch": 0.07240730170772607,
      "grad_norm": 3.15625,
      "learning_rate": 4.940456586614434e-05,
      "loss": 0.8741,
      "step": 5106
    },
    {
      "epoch": 0.07243566336135228,
      "grad_norm": 3.0625,
      "learning_rate": 4.9404081045198164e-05,
      "loss": 0.9147,
      "step": 5108
    },
    {
      "epoch": 0.07246402501497849,
      "grad_norm": 2.859375,
      "learning_rate": 4.940359602933498e-05,
      "loss": 0.8676,
      "step": 5110
    },
    {
      "epoch": 0.07249238666860472,
      "grad_norm": 2.9375,
      "learning_rate": 4.940311081855863e-05,
      "loss": 0.8993,
      "step": 5112
    },
    {
      "epoch": 0.07252074832223093,
      "grad_norm": 3.15625,
      "learning_rate": 4.940262541287302e-05,
      "loss": 0.9169,
      "step": 5114
    },
    {
      "epoch": 0.07254910997585715,
      "grad_norm": 3.3125,
      "learning_rate": 4.9402139812282e-05,
      "loss": 0.9215,
      "step": 5116
    },
    {
      "epoch": 0.07257747162948336,
      "grad_norm": 3.484375,
      "learning_rate": 4.9401654016789466e-05,
      "loss": 0.8859,
      "step": 5118
    },
    {
      "epoch": 0.07260583328310957,
      "grad_norm": 3.296875,
      "learning_rate": 4.94011680263993e-05,
      "loss": 0.9126,
      "step": 5120
    },
    {
      "epoch": 0.07263419493673579,
      "grad_norm": 3.21875,
      "learning_rate": 4.940068184111537e-05,
      "loss": 0.8635,
      "step": 5122
    },
    {
      "epoch": 0.072662556590362,
      "grad_norm": 3.328125,
      "learning_rate": 4.9400195460941576e-05,
      "loss": 0.8939,
      "step": 5124
    },
    {
      "epoch": 0.07269091824398821,
      "grad_norm": 3.375,
      "learning_rate": 4.9399708885881784e-05,
      "loss": 0.9239,
      "step": 5126
    },
    {
      "epoch": 0.07271927989761443,
      "grad_norm": 3.171875,
      "learning_rate": 4.939922211593989e-05,
      "loss": 0.9314,
      "step": 5128
    },
    {
      "epoch": 0.07274764155124064,
      "grad_norm": 2.953125,
      "learning_rate": 4.939873515111979e-05,
      "loss": 0.88,
      "step": 5130
    },
    {
      "epoch": 0.07277600320486687,
      "grad_norm": 3.390625,
      "learning_rate": 4.939824799142536e-05,
      "loss": 0.9403,
      "step": 5132
    },
    {
      "epoch": 0.07280436485849308,
      "grad_norm": 3.234375,
      "learning_rate": 4.939776063686049e-05,
      "loss": 0.866,
      "step": 5134
    },
    {
      "epoch": 0.07283272651211929,
      "grad_norm": 3.015625,
      "learning_rate": 4.939727308742908e-05,
      "loss": 0.9141,
      "step": 5136
    },
    {
      "epoch": 0.0728610881657455,
      "grad_norm": 3.015625,
      "learning_rate": 4.939678534313503e-05,
      "loss": 0.8756,
      "step": 5138
    },
    {
      "epoch": 0.07288944981937172,
      "grad_norm": 3.59375,
      "learning_rate": 4.9396297403982225e-05,
      "loss": 0.8975,
      "step": 5140
    },
    {
      "epoch": 0.07291781147299793,
      "grad_norm": 3.46875,
      "learning_rate": 4.939580926997457e-05,
      "loss": 0.9002,
      "step": 5142
    },
    {
      "epoch": 0.07294617312662414,
      "grad_norm": 3.75,
      "learning_rate": 4.939532094111595e-05,
      "loss": 0.9568,
      "step": 5144
    },
    {
      "epoch": 0.07297453478025036,
      "grad_norm": 3.53125,
      "learning_rate": 4.9394832417410285e-05,
      "loss": 0.9316,
      "step": 5146
    },
    {
      "epoch": 0.07300289643387657,
      "grad_norm": 3.359375,
      "learning_rate": 4.9394343698861453e-05,
      "loss": 0.9398,
      "step": 5148
    },
    {
      "epoch": 0.0730312580875028,
      "grad_norm": 2.890625,
      "learning_rate": 4.939385478547338e-05,
      "loss": 0.9241,
      "step": 5150
    },
    {
      "epoch": 0.07305961974112901,
      "grad_norm": 3.0625,
      "learning_rate": 4.939336567724996e-05,
      "loss": 0.883,
      "step": 5152
    },
    {
      "epoch": 0.07308798139475522,
      "grad_norm": 3.203125,
      "learning_rate": 4.939287637419511e-05,
      "loss": 0.9274,
      "step": 5154
    },
    {
      "epoch": 0.07311634304838144,
      "grad_norm": 3.1875,
      "learning_rate": 4.9392386876312715e-05,
      "loss": 0.8555,
      "step": 5156
    },
    {
      "epoch": 0.07314470470200765,
      "grad_norm": 2.984375,
      "learning_rate": 4.9391897183606706e-05,
      "loss": 0.8862,
      "step": 5158
    },
    {
      "epoch": 0.07317306635563386,
      "grad_norm": 3.03125,
      "learning_rate": 4.9391407296080985e-05,
      "loss": 0.9889,
      "step": 5160
    },
    {
      "epoch": 0.07320142800926008,
      "grad_norm": 3.125,
      "learning_rate": 4.939091721373946e-05,
      "loss": 0.9273,
      "step": 5162
    },
    {
      "epoch": 0.07322978966288629,
      "grad_norm": 2.96875,
      "learning_rate": 4.9390426936586066e-05,
      "loss": 0.8659,
      "step": 5164
    },
    {
      "epoch": 0.0732581513165125,
      "grad_norm": 3.5625,
      "learning_rate": 4.9389936464624695e-05,
      "loss": 0.9335,
      "step": 5166
    },
    {
      "epoch": 0.07328651297013872,
      "grad_norm": 3.578125,
      "learning_rate": 4.938944579785928e-05,
      "loss": 0.8202,
      "step": 5168
    },
    {
      "epoch": 0.07331487462376494,
      "grad_norm": 2.984375,
      "learning_rate": 4.938895493629374e-05,
      "loss": 0.8634,
      "step": 5170
    },
    {
      "epoch": 0.07334323627739116,
      "grad_norm": 2.984375,
      "learning_rate": 4.938846387993198e-05,
      "loss": 0.899,
      "step": 5172
    },
    {
      "epoch": 0.07337159793101737,
      "grad_norm": 3.21875,
      "learning_rate": 4.938797262877793e-05,
      "loss": 0.927,
      "step": 5174
    },
    {
      "epoch": 0.07339995958464358,
      "grad_norm": 3.5625,
      "learning_rate": 4.938748118283553e-05,
      "loss": 0.8834,
      "step": 5176
    },
    {
      "epoch": 0.0734283212382698,
      "grad_norm": 2.8125,
      "learning_rate": 4.938698954210868e-05,
      "loss": 0.8675,
      "step": 5178
    },
    {
      "epoch": 0.07345668289189601,
      "grad_norm": 2.71875,
      "learning_rate": 4.938649770660132e-05,
      "loss": 0.8624,
      "step": 5180
    },
    {
      "epoch": 0.07348504454552222,
      "grad_norm": 3.0625,
      "learning_rate": 4.9386005676317385e-05,
      "loss": 0.8336,
      "step": 5182
    },
    {
      "epoch": 0.07351340619914844,
      "grad_norm": 2.984375,
      "learning_rate": 4.938551345126078e-05,
      "loss": 0.9037,
      "step": 5184
    },
    {
      "epoch": 0.07354176785277465,
      "grad_norm": 2.9375,
      "learning_rate": 4.9385021031435466e-05,
      "loss": 0.8866,
      "step": 5186
    },
    {
      "epoch": 0.07357012950640088,
      "grad_norm": 3.34375,
      "learning_rate": 4.9384528416845355e-05,
      "loss": 0.9415,
      "step": 5188
    },
    {
      "epoch": 0.07359849116002709,
      "grad_norm": 3.15625,
      "learning_rate": 4.938403560749439e-05,
      "loss": 0.9414,
      "step": 5190
    },
    {
      "epoch": 0.0736268528136533,
      "grad_norm": 3.359375,
      "learning_rate": 4.938354260338651e-05,
      "loss": 0.9028,
      "step": 5192
    },
    {
      "epoch": 0.07365521446727952,
      "grad_norm": 2.671875,
      "learning_rate": 4.938304940452564e-05,
      "loss": 0.8778,
      "step": 5194
    },
    {
      "epoch": 0.07368357612090573,
      "grad_norm": 3.0625,
      "learning_rate": 4.938255601091574e-05,
      "loss": 0.8695,
      "step": 5196
    },
    {
      "epoch": 0.07371193777453194,
      "grad_norm": 3.375,
      "learning_rate": 4.9382062422560735e-05,
      "loss": 0.9239,
      "step": 5198
    },
    {
      "epoch": 0.07374029942815816,
      "grad_norm": 3.046875,
      "learning_rate": 4.938156863946456e-05,
      "loss": 0.8914,
      "step": 5200
    },
    {
      "epoch": 0.07376866108178437,
      "grad_norm": 2.71875,
      "learning_rate": 4.938107466163119e-05,
      "loss": 0.9011,
      "step": 5202
    },
    {
      "epoch": 0.07379702273541058,
      "grad_norm": 3.65625,
      "learning_rate": 4.938058048906453e-05,
      "loss": 0.8847,
      "step": 5204
    },
    {
      "epoch": 0.0738253843890368,
      "grad_norm": 2.875,
      "learning_rate": 4.938008612176856e-05,
      "loss": 0.882,
      "step": 5206
    },
    {
      "epoch": 0.07385374604266302,
      "grad_norm": 2.90625,
      "learning_rate": 4.9379591559747215e-05,
      "loss": 0.8947,
      "step": 5208
    },
    {
      "epoch": 0.07388210769628924,
      "grad_norm": 3.234375,
      "learning_rate": 4.9379096803004444e-05,
      "loss": 0.8932,
      "step": 5210
    },
    {
      "epoch": 0.07391046934991545,
      "grad_norm": 3.5625,
      "learning_rate": 4.93786018515442e-05,
      "loss": 0.9078,
      "step": 5212
    },
    {
      "epoch": 0.07393883100354166,
      "grad_norm": 2.921875,
      "learning_rate": 4.9378106705370445e-05,
      "loss": 0.8919,
      "step": 5214
    },
    {
      "epoch": 0.07396719265716788,
      "grad_norm": 2.953125,
      "learning_rate": 4.937761136448711e-05,
      "loss": 0.9291,
      "step": 5216
    },
    {
      "epoch": 0.07399555431079409,
      "grad_norm": 3.46875,
      "learning_rate": 4.937711582889818e-05,
      "loss": 0.9132,
      "step": 5218
    },
    {
      "epoch": 0.0740239159644203,
      "grad_norm": 3.296875,
      "learning_rate": 4.93766200986076e-05,
      "loss": 0.9557,
      "step": 5220
    },
    {
      "epoch": 0.07405227761804652,
      "grad_norm": 2.953125,
      "learning_rate": 4.937612417361932e-05,
      "loss": 0.9311,
      "step": 5222
    },
    {
      "epoch": 0.07408063927167273,
      "grad_norm": 3.09375,
      "learning_rate": 4.937562805393731e-05,
      "loss": 0.8904,
      "step": 5224
    },
    {
      "epoch": 0.07410900092529894,
      "grad_norm": 3.09375,
      "learning_rate": 4.937513173956554e-05,
      "loss": 0.8949,
      "step": 5226
    },
    {
      "epoch": 0.07413736257892517,
      "grad_norm": 3.171875,
      "learning_rate": 4.9374635230507964e-05,
      "loss": 0.8684,
      "step": 5228
    },
    {
      "epoch": 0.07416572423255138,
      "grad_norm": 2.96875,
      "learning_rate": 4.937413852676855e-05,
      "loss": 0.9395,
      "step": 5230
    },
    {
      "epoch": 0.0741940858861776,
      "grad_norm": 3.03125,
      "learning_rate": 4.937364162835127e-05,
      "loss": 0.8616,
      "step": 5232
    },
    {
      "epoch": 0.07422244753980381,
      "grad_norm": 3.109375,
      "learning_rate": 4.9373144535260086e-05,
      "loss": 0.875,
      "step": 5234
    },
    {
      "epoch": 0.07425080919343002,
      "grad_norm": 2.546875,
      "learning_rate": 4.937264724749897e-05,
      "loss": 0.89,
      "step": 5236
    },
    {
      "epoch": 0.07427917084705624,
      "grad_norm": 3.234375,
      "learning_rate": 4.937214976507189e-05,
      "loss": 0.9526,
      "step": 5238
    },
    {
      "epoch": 0.07430753250068245,
      "grad_norm": 3.171875,
      "learning_rate": 4.937165208798283e-05,
      "loss": 0.8587,
      "step": 5240
    },
    {
      "epoch": 0.07433589415430866,
      "grad_norm": 3.109375,
      "learning_rate": 4.937115421623577e-05,
      "loss": 0.8753,
      "step": 5242
    },
    {
      "epoch": 0.07436425580793488,
      "grad_norm": 2.71875,
      "learning_rate": 4.937065614983466e-05,
      "loss": 0.8933,
      "step": 5244
    },
    {
      "epoch": 0.0743926174615611,
      "grad_norm": 2.984375,
      "learning_rate": 4.93701578887835e-05,
      "loss": 0.898,
      "step": 5246
    },
    {
      "epoch": 0.07442097911518732,
      "grad_norm": 3.375,
      "learning_rate": 4.9369659433086256e-05,
      "loss": 0.8563,
      "step": 5248
    },
    {
      "epoch": 0.07444934076881353,
      "grad_norm": 3.015625,
      "learning_rate": 4.9369160782746927e-05,
      "loss": 0.8778,
      "step": 5250
    },
    {
      "epoch": 0.07447770242243974,
      "grad_norm": 2.96875,
      "learning_rate": 4.9368661937769475e-05,
      "loss": 0.8942,
      "step": 5252
    },
    {
      "epoch": 0.07450606407606596,
      "grad_norm": 2.890625,
      "learning_rate": 4.936816289815791e-05,
      "loss": 0.874,
      "step": 5254
    },
    {
      "epoch": 0.07453442572969217,
      "grad_norm": 3.25,
      "learning_rate": 4.936766366391619e-05,
      "loss": 0.9194,
      "step": 5256
    },
    {
      "epoch": 0.07456278738331838,
      "grad_norm": 3.0625,
      "learning_rate": 4.936716423504832e-05,
      "loss": 0.8969,
      "step": 5258
    },
    {
      "epoch": 0.0745911490369446,
      "grad_norm": 3.140625,
      "learning_rate": 4.936666461155828e-05,
      "loss": 0.9198,
      "step": 5260
    },
    {
      "epoch": 0.07461951069057081,
      "grad_norm": 3.265625,
      "learning_rate": 4.9366164793450066e-05,
      "loss": 0.9038,
      "step": 5262
    },
    {
      "epoch": 0.07464787234419702,
      "grad_norm": 3.1875,
      "learning_rate": 4.936566478072767e-05,
      "loss": 0.8808,
      "step": 5264
    },
    {
      "epoch": 0.07467623399782325,
      "grad_norm": 3.046875,
      "learning_rate": 4.9365164573395086e-05,
      "loss": 0.872,
      "step": 5266
    },
    {
      "epoch": 0.07470459565144946,
      "grad_norm": 2.875,
      "learning_rate": 4.93646641714563e-05,
      "loss": 0.8713,
      "step": 5268
    },
    {
      "epoch": 0.07473295730507568,
      "grad_norm": 3.140625,
      "learning_rate": 4.936416357491532e-05,
      "loss": 0.8961,
      "step": 5270
    },
    {
      "epoch": 0.07476131895870189,
      "grad_norm": 3.265625,
      "learning_rate": 4.9363662783776146e-05,
      "loss": 0.9058,
      "step": 5272
    },
    {
      "epoch": 0.0747896806123281,
      "grad_norm": 3.65625,
      "learning_rate": 4.9363161798042766e-05,
      "loss": 0.8826,
      "step": 5274
    },
    {
      "epoch": 0.07481804226595432,
      "grad_norm": 3.09375,
      "learning_rate": 4.936266061771919e-05,
      "loss": 0.8865,
      "step": 5276
    },
    {
      "epoch": 0.07484640391958053,
      "grad_norm": 2.890625,
      "learning_rate": 4.9362159242809415e-05,
      "loss": 0.8901,
      "step": 5278
    },
    {
      "epoch": 0.07487476557320674,
      "grad_norm": 2.703125,
      "learning_rate": 4.9361657673317455e-05,
      "loss": 0.9303,
      "step": 5280
    },
    {
      "epoch": 0.07490312722683296,
      "grad_norm": 2.9375,
      "learning_rate": 4.93611559092473e-05,
      "loss": 0.8695,
      "step": 5282
    },
    {
      "epoch": 0.07493148888045917,
      "grad_norm": 3.328125,
      "learning_rate": 4.936065395060298e-05,
      "loss": 0.9216,
      "step": 5284
    },
    {
      "epoch": 0.0749598505340854,
      "grad_norm": 2.8125,
      "learning_rate": 4.936015179738849e-05,
      "loss": 0.88,
      "step": 5286
    },
    {
      "epoch": 0.07498821218771161,
      "grad_norm": 3.0,
      "learning_rate": 4.9359649449607834e-05,
      "loss": 0.9641,
      "step": 5288
    },
    {
      "epoch": 0.07501657384133782,
      "grad_norm": 3.140625,
      "learning_rate": 4.935914690726504e-05,
      "loss": 0.8961,
      "step": 5290
    },
    {
      "epoch": 0.07504493549496404,
      "grad_norm": 2.796875,
      "learning_rate": 4.935864417036412e-05,
      "loss": 0.8868,
      "step": 5292
    },
    {
      "epoch": 0.07507329714859025,
      "grad_norm": 3.296875,
      "learning_rate": 4.9358141238909075e-05,
      "loss": 0.8834,
      "step": 5294
    },
    {
      "epoch": 0.07510165880221646,
      "grad_norm": 2.640625,
      "learning_rate": 4.9357638112903936e-05,
      "loss": 0.8888,
      "step": 5296
    },
    {
      "epoch": 0.07513002045584267,
      "grad_norm": 3.296875,
      "learning_rate": 4.935713479235271e-05,
      "loss": 0.9013,
      "step": 5298
    },
    {
      "epoch": 0.07515838210946889,
      "grad_norm": 2.859375,
      "learning_rate": 4.9356631277259436e-05,
      "loss": 0.8759,
      "step": 5300
    },
    {
      "epoch": 0.0751867437630951,
      "grad_norm": 2.96875,
      "learning_rate": 4.935612756762811e-05,
      "loss": 0.9136,
      "step": 5302
    },
    {
      "epoch": 0.07521510541672133,
      "grad_norm": 2.859375,
      "learning_rate": 4.935562366346278e-05,
      "loss": 0.9295,
      "step": 5304
    },
    {
      "epoch": 0.07524346707034754,
      "grad_norm": 3.5625,
      "learning_rate": 4.935511956476745e-05,
      "loss": 0.8987,
      "step": 5306
    },
    {
      "epoch": 0.07527182872397375,
      "grad_norm": 3.046875,
      "learning_rate": 4.9354615271546164e-05,
      "loss": 0.859,
      "step": 5308
    },
    {
      "epoch": 0.07530019037759997,
      "grad_norm": 3.09375,
      "learning_rate": 4.935411078380293e-05,
      "loss": 0.9349,
      "step": 5310
    },
    {
      "epoch": 0.07532855203122618,
      "grad_norm": 3.078125,
      "learning_rate": 4.93536061015418e-05,
      "loss": 0.9032,
      "step": 5312
    },
    {
      "epoch": 0.0753569136848524,
      "grad_norm": 3.015625,
      "learning_rate": 4.935310122476679e-05,
      "loss": 0.8784,
      "step": 5314
    },
    {
      "epoch": 0.07538527533847861,
      "grad_norm": 3.1875,
      "learning_rate": 4.935259615348193e-05,
      "loss": 0.8633,
      "step": 5316
    },
    {
      "epoch": 0.07541363699210482,
      "grad_norm": 3.09375,
      "learning_rate": 4.935209088769127e-05,
      "loss": 0.8513,
      "step": 5318
    },
    {
      "epoch": 0.07544199864573103,
      "grad_norm": 3.0625,
      "learning_rate": 4.9351585427398835e-05,
      "loss": 0.907,
      "step": 5320
    },
    {
      "epoch": 0.07547036029935725,
      "grad_norm": 3.328125,
      "learning_rate": 4.935107977260865e-05,
      "loss": 0.8551,
      "step": 5322
    },
    {
      "epoch": 0.07549872195298347,
      "grad_norm": 2.984375,
      "learning_rate": 4.9350573923324786e-05,
      "loss": 0.89,
      "step": 5324
    },
    {
      "epoch": 0.07552708360660969,
      "grad_norm": 3.15625,
      "learning_rate": 4.935006787955126e-05,
      "loss": 0.8992,
      "step": 5326
    },
    {
      "epoch": 0.0755554452602359,
      "grad_norm": 3.140625,
      "learning_rate": 4.93495616412921e-05,
      "loss": 0.9355,
      "step": 5328
    },
    {
      "epoch": 0.07558380691386211,
      "grad_norm": 3.0,
      "learning_rate": 4.934905520855138e-05,
      "loss": 0.8623,
      "step": 5330
    },
    {
      "epoch": 0.07561216856748833,
      "grad_norm": 3.1875,
      "learning_rate": 4.934854858133313e-05,
      "loss": 0.9072,
      "step": 5332
    },
    {
      "epoch": 0.07564053022111454,
      "grad_norm": 3.140625,
      "learning_rate": 4.93480417596414e-05,
      "loss": 0.8938,
      "step": 5334
    },
    {
      "epoch": 0.07566889187474075,
      "grad_norm": 3.03125,
      "learning_rate": 4.9347534743480236e-05,
      "loss": 0.9031,
      "step": 5336
    },
    {
      "epoch": 0.07569725352836697,
      "grad_norm": 3.09375,
      "learning_rate": 4.934702753285368e-05,
      "loss": 0.962,
      "step": 5338
    },
    {
      "epoch": 0.07572561518199318,
      "grad_norm": 2.671875,
      "learning_rate": 4.934652012776581e-05,
      "loss": 0.8733,
      "step": 5340
    },
    {
      "epoch": 0.07575397683561941,
      "grad_norm": 3.21875,
      "learning_rate": 4.934601252822064e-05,
      "loss": 0.9047,
      "step": 5342
    },
    {
      "epoch": 0.07578233848924562,
      "grad_norm": 3.375,
      "learning_rate": 4.934550473422226e-05,
      "loss": 0.9506,
      "step": 5344
    },
    {
      "epoch": 0.07581070014287183,
      "grad_norm": 3.171875,
      "learning_rate": 4.93449967457747e-05,
      "loss": 0.8966,
      "step": 5346
    },
    {
      "epoch": 0.07583906179649805,
      "grad_norm": 3.078125,
      "learning_rate": 4.934448856288203e-05,
      "loss": 0.8488,
      "step": 5348
    },
    {
      "epoch": 0.07586742345012426,
      "grad_norm": 2.953125,
      "learning_rate": 4.93439801855483e-05,
      "loss": 0.8768,
      "step": 5350
    },
    {
      "epoch": 0.07589578510375047,
      "grad_norm": 3.109375,
      "learning_rate": 4.9343471613777584e-05,
      "loss": 0.8928,
      "step": 5352
    },
    {
      "epoch": 0.07592414675737669,
      "grad_norm": 2.90625,
      "learning_rate": 4.934296284757394e-05,
      "loss": 0.8584,
      "step": 5354
    },
    {
      "epoch": 0.0759525084110029,
      "grad_norm": 3.3125,
      "learning_rate": 4.934245388694142e-05,
      "loss": 0.9175,
      "step": 5356
    },
    {
      "epoch": 0.07598087006462911,
      "grad_norm": 2.9375,
      "learning_rate": 4.934194473188409e-05,
      "loss": 0.8755,
      "step": 5358
    },
    {
      "epoch": 0.07600923171825533,
      "grad_norm": 3.171875,
      "learning_rate": 4.9341435382406045e-05,
      "loss": 0.859,
      "step": 5360
    },
    {
      "epoch": 0.07603759337188155,
      "grad_norm": 2.765625,
      "learning_rate": 4.934092583851132e-05,
      "loss": 0.8689,
      "step": 5362
    },
    {
      "epoch": 0.07606595502550777,
      "grad_norm": 2.71875,
      "learning_rate": 4.934041610020399e-05,
      "loss": 0.866,
      "step": 5364
    },
    {
      "epoch": 0.07609431667913398,
      "grad_norm": 3.296875,
      "learning_rate": 4.933990616748814e-05,
      "loss": 0.832,
      "step": 5366
    },
    {
      "epoch": 0.0761226783327602,
      "grad_norm": 3.03125,
      "learning_rate": 4.9339396040367835e-05,
      "loss": 0.9287,
      "step": 5368
    },
    {
      "epoch": 0.0761510399863864,
      "grad_norm": 3.359375,
      "learning_rate": 4.933888571884716e-05,
      "loss": 0.8963,
      "step": 5370
    },
    {
      "epoch": 0.07617940164001262,
      "grad_norm": 2.953125,
      "learning_rate": 4.933837520293017e-05,
      "loss": 0.8709,
      "step": 5372
    },
    {
      "epoch": 0.07620776329363883,
      "grad_norm": 3.625,
      "learning_rate": 4.9337864492620954e-05,
      "loss": 0.9416,
      "step": 5374
    },
    {
      "epoch": 0.07623612494726505,
      "grad_norm": 3.09375,
      "learning_rate": 4.9337353587923594e-05,
      "loss": 0.8525,
      "step": 5376
    },
    {
      "epoch": 0.07626448660089126,
      "grad_norm": 3.421875,
      "learning_rate": 4.9336842488842175e-05,
      "loss": 0.8782,
      "step": 5378
    },
    {
      "epoch": 0.07629284825451747,
      "grad_norm": 3.671875,
      "learning_rate": 4.9336331195380755e-05,
      "loss": 0.8989,
      "step": 5380
    },
    {
      "epoch": 0.0763212099081437,
      "grad_norm": 3.125,
      "learning_rate": 4.933581970754345e-05,
      "loss": 0.8819,
      "step": 5382
    },
    {
      "epoch": 0.07634957156176991,
      "grad_norm": 3.484375,
      "learning_rate": 4.933530802533432e-05,
      "loss": 0.9443,
      "step": 5384
    },
    {
      "epoch": 0.07637793321539613,
      "grad_norm": 3.21875,
      "learning_rate": 4.933479614875747e-05,
      "loss": 0.8655,
      "step": 5386
    },
    {
      "epoch": 0.07640629486902234,
      "grad_norm": 3.1875,
      "learning_rate": 4.9334284077816974e-05,
      "loss": 0.8881,
      "step": 5388
    },
    {
      "epoch": 0.07643465652264855,
      "grad_norm": 3.015625,
      "learning_rate": 4.933377181251693e-05,
      "loss": 0.8895,
      "step": 5390
    },
    {
      "epoch": 0.07646301817627477,
      "grad_norm": 3.40625,
      "learning_rate": 4.933325935286142e-05,
      "loss": 0.9179,
      "step": 5392
    },
    {
      "epoch": 0.07649137982990098,
      "grad_norm": 3.1875,
      "learning_rate": 4.933274669885456e-05,
      "loss": 0.8599,
      "step": 5394
    },
    {
      "epoch": 0.07651974148352719,
      "grad_norm": 3.0625,
      "learning_rate": 4.933223385050041e-05,
      "loss": 0.9354,
      "step": 5396
    },
    {
      "epoch": 0.0765481031371534,
      "grad_norm": 3.5625,
      "learning_rate": 4.933172080780309e-05,
      "loss": 0.92,
      "step": 5398
    },
    {
      "epoch": 0.07657646479077963,
      "grad_norm": 3.296875,
      "learning_rate": 4.9331207570766706e-05,
      "loss": 0.8878,
      "step": 5400
    },
    {
      "epoch": 0.07660482644440585,
      "grad_norm": 3.078125,
      "learning_rate": 4.933069413939534e-05,
      "loss": 0.8723,
      "step": 5402
    },
    {
      "epoch": 0.07663318809803206,
      "grad_norm": 2.953125,
      "learning_rate": 4.9330180513693095e-05,
      "loss": 0.8998,
      "step": 5404
    },
    {
      "epoch": 0.07666154975165827,
      "grad_norm": 3.265625,
      "learning_rate": 4.932966669366407e-05,
      "loss": 0.893,
      "step": 5406
    },
    {
      "epoch": 0.07668991140528449,
      "grad_norm": 2.890625,
      "learning_rate": 4.932915267931237e-05,
      "loss": 0.8777,
      "step": 5408
    },
    {
      "epoch": 0.0767182730589107,
      "grad_norm": 3.109375,
      "learning_rate": 4.932863847064212e-05,
      "loss": 0.9067,
      "step": 5410
    },
    {
      "epoch": 0.07674663471253691,
      "grad_norm": 3.265625,
      "learning_rate": 4.9328124067657406e-05,
      "loss": 0.8636,
      "step": 5412
    },
    {
      "epoch": 0.07677499636616313,
      "grad_norm": 2.984375,
      "learning_rate": 4.932760947036234e-05,
      "loss": 0.9323,
      "step": 5414
    },
    {
      "epoch": 0.07680335801978934,
      "grad_norm": 2.984375,
      "learning_rate": 4.932709467876104e-05,
      "loss": 0.9058,
      "step": 5416
    },
    {
      "epoch": 0.07683171967341555,
      "grad_norm": 2.953125,
      "learning_rate": 4.932657969285761e-05,
      "loss": 0.878,
      "step": 5418
    },
    {
      "epoch": 0.07686008132704178,
      "grad_norm": 2.796875,
      "learning_rate": 4.932606451265616e-05,
      "loss": 0.9243,
      "step": 5420
    },
    {
      "epoch": 0.07688844298066799,
      "grad_norm": 2.96875,
      "learning_rate": 4.932554913816081e-05,
      "loss": 0.8622,
      "step": 5422
    },
    {
      "epoch": 0.0769168046342942,
      "grad_norm": 2.921875,
      "learning_rate": 4.932503356937568e-05,
      "loss": 0.8482,
      "step": 5424
    },
    {
      "epoch": 0.07694516628792042,
      "grad_norm": 3.3125,
      "learning_rate": 4.9324517806304884e-05,
      "loss": 0.9288,
      "step": 5426
    },
    {
      "epoch": 0.07697352794154663,
      "grad_norm": 3.015625,
      "learning_rate": 4.932400184895254e-05,
      "loss": 0.9244,
      "step": 5428
    },
    {
      "epoch": 0.07700188959517285,
      "grad_norm": 2.84375,
      "learning_rate": 4.932348569732278e-05,
      "loss": 0.8565,
      "step": 5430
    },
    {
      "epoch": 0.07703025124879906,
      "grad_norm": 3.34375,
      "learning_rate": 4.932296935141971e-05,
      "loss": 0.8736,
      "step": 5432
    },
    {
      "epoch": 0.07705861290242527,
      "grad_norm": 2.96875,
      "learning_rate": 4.9322452811247455e-05,
      "loss": 0.8823,
      "step": 5434
    },
    {
      "epoch": 0.07708697455605149,
      "grad_norm": 2.9375,
      "learning_rate": 4.932193607681016e-05,
      "loss": 0.8742,
      "step": 5436
    },
    {
      "epoch": 0.0771153362096777,
      "grad_norm": 3.015625,
      "learning_rate": 4.932141914811193e-05,
      "loss": 0.8538,
      "step": 5438
    },
    {
      "epoch": 0.07714369786330393,
      "grad_norm": 3.1875,
      "learning_rate": 4.932090202515691e-05,
      "loss": 0.911,
      "step": 5440
    },
    {
      "epoch": 0.07717205951693014,
      "grad_norm": 2.890625,
      "learning_rate": 4.932038470794922e-05,
      "loss": 0.8644,
      "step": 5442
    },
    {
      "epoch": 0.07720042117055635,
      "grad_norm": 2.890625,
      "learning_rate": 4.931986719649299e-05,
      "loss": 0.8782,
      "step": 5444
    },
    {
      "epoch": 0.07722878282418257,
      "grad_norm": 3.1875,
      "learning_rate": 4.9319349490792365e-05,
      "loss": 0.8774,
      "step": 5446
    },
    {
      "epoch": 0.07725714447780878,
      "grad_norm": 3.140625,
      "learning_rate": 4.931883159085147e-05,
      "loss": 0.9193,
      "step": 5448
    },
    {
      "epoch": 0.07728550613143499,
      "grad_norm": 3.109375,
      "learning_rate": 4.9318313496674446e-05,
      "loss": 0.8686,
      "step": 5450
    },
    {
      "epoch": 0.0773138677850612,
      "grad_norm": 2.75,
      "learning_rate": 4.931779520826543e-05,
      "loss": 0.9091,
      "step": 5452
    },
    {
      "epoch": 0.07734222943868742,
      "grad_norm": 3.5,
      "learning_rate": 4.9317276725628566e-05,
      "loss": 0.8835,
      "step": 5454
    },
    {
      "epoch": 0.07737059109231363,
      "grad_norm": 3.078125,
      "learning_rate": 4.931675804876798e-05,
      "loss": 0.9199,
      "step": 5456
    },
    {
      "epoch": 0.07739895274593986,
      "grad_norm": 3.515625,
      "learning_rate": 4.931623917768784e-05,
      "loss": 0.9874,
      "step": 5458
    },
    {
      "epoch": 0.07742731439956607,
      "grad_norm": 3.015625,
      "learning_rate": 4.931572011239227e-05,
      "loss": 0.8394,
      "step": 5460
    },
    {
      "epoch": 0.07745567605319228,
      "grad_norm": 2.875,
      "learning_rate": 4.9315200852885415e-05,
      "loss": 0.8789,
      "step": 5462
    },
    {
      "epoch": 0.0774840377068185,
      "grad_norm": 3.40625,
      "learning_rate": 4.931468139917144e-05,
      "loss": 0.9248,
      "step": 5464
    },
    {
      "epoch": 0.07751239936044471,
      "grad_norm": 3.125,
      "learning_rate": 4.9314161751254476e-05,
      "loss": 0.9036,
      "step": 5466
    },
    {
      "epoch": 0.07754076101407092,
      "grad_norm": 2.765625,
      "learning_rate": 4.931364190913868e-05,
      "loss": 0.9053,
      "step": 5468
    },
    {
      "epoch": 0.07756912266769714,
      "grad_norm": 3.109375,
      "learning_rate": 4.93131218728282e-05,
      "loss": 0.8805,
      "step": 5470
    },
    {
      "epoch": 0.07759748432132335,
      "grad_norm": 3.3125,
      "learning_rate": 4.93126016423272e-05,
      "loss": 0.9209,
      "step": 5472
    },
    {
      "epoch": 0.07762584597494956,
      "grad_norm": 3.25,
      "learning_rate": 4.931208121763983e-05,
      "loss": 0.8672,
      "step": 5474
    },
    {
      "epoch": 0.07765420762857578,
      "grad_norm": 2.828125,
      "learning_rate": 4.931156059877025e-05,
      "loss": 0.8389,
      "step": 5476
    },
    {
      "epoch": 0.077682569282202,
      "grad_norm": 3.046875,
      "learning_rate": 4.9311039785722605e-05,
      "loss": 0.9165,
      "step": 5478
    },
    {
      "epoch": 0.07771093093582822,
      "grad_norm": 3.15625,
      "learning_rate": 4.931051877850106e-05,
      "loss": 0.9001,
      "step": 5480
    },
    {
      "epoch": 0.07773929258945443,
      "grad_norm": 3.578125,
      "learning_rate": 4.930999757710979e-05,
      "loss": 0.916,
      "step": 5482
    },
    {
      "epoch": 0.07776765424308064,
      "grad_norm": 3.21875,
      "learning_rate": 4.930947618155294e-05,
      "loss": 0.9157,
      "step": 5484
    },
    {
      "epoch": 0.07779601589670686,
      "grad_norm": 3.03125,
      "learning_rate": 4.9308954591834686e-05,
      "loss": 0.9036,
      "step": 5486
    },
    {
      "epoch": 0.07782437755033307,
      "grad_norm": 3.015625,
      "learning_rate": 4.930843280795919e-05,
      "loss": 0.9086,
      "step": 5488
    },
    {
      "epoch": 0.07785273920395928,
      "grad_norm": 3.03125,
      "learning_rate": 4.9307910829930624e-05,
      "loss": 0.9022,
      "step": 5490
    },
    {
      "epoch": 0.0778811008575855,
      "grad_norm": 3.046875,
      "learning_rate": 4.930738865775315e-05,
      "loss": 0.9016,
      "step": 5492
    },
    {
      "epoch": 0.07790946251121171,
      "grad_norm": 3.53125,
      "learning_rate": 4.930686629143094e-05,
      "loss": 0.9501,
      "step": 5494
    },
    {
      "epoch": 0.07793782416483794,
      "grad_norm": 2.953125,
      "learning_rate": 4.930634373096817e-05,
      "loss": 0.8992,
      "step": 5496
    },
    {
      "epoch": 0.07796618581846415,
      "grad_norm": 2.953125,
      "learning_rate": 4.930582097636901e-05,
      "loss": 0.8981,
      "step": 5498
    },
    {
      "epoch": 0.07799454747209036,
      "grad_norm": 3.3125,
      "learning_rate": 4.930529802763764e-05,
      "loss": 0.8843,
      "step": 5500
    },
    {
      "epoch": 0.07802290912571658,
      "grad_norm": 3.09375,
      "learning_rate": 4.9304774884778224e-05,
      "loss": 0.9155,
      "step": 5502
    },
    {
      "epoch": 0.07805127077934279,
      "grad_norm": 3.234375,
      "learning_rate": 4.930425154779496e-05,
      "loss": 0.9095,
      "step": 5504
    },
    {
      "epoch": 0.078079632432969,
      "grad_norm": 2.921875,
      "learning_rate": 4.930372801669202e-05,
      "loss": 0.8991,
      "step": 5506
    },
    {
      "epoch": 0.07810799408659522,
      "grad_norm": 2.640625,
      "learning_rate": 4.930320429147357e-05,
      "loss": 0.8325,
      "step": 5508
    },
    {
      "epoch": 0.07813635574022143,
      "grad_norm": 3.359375,
      "learning_rate": 4.9302680372143814e-05,
      "loss": 0.9053,
      "step": 5510
    },
    {
      "epoch": 0.07816471739384764,
      "grad_norm": 3.390625,
      "learning_rate": 4.930215625870693e-05,
      "loss": 0.8721,
      "step": 5512
    },
    {
      "epoch": 0.07819307904747386,
      "grad_norm": 3.203125,
      "learning_rate": 4.9301631951167104e-05,
      "loss": 0.8696,
      "step": 5514
    },
    {
      "epoch": 0.07822144070110008,
      "grad_norm": 2.890625,
      "learning_rate": 4.930110744952852e-05,
      "loss": 0.8564,
      "step": 5516
    },
    {
      "epoch": 0.0782498023547263,
      "grad_norm": 3.03125,
      "learning_rate": 4.9300582753795366e-05,
      "loss": 0.8688,
      "step": 5518
    },
    {
      "epoch": 0.07827816400835251,
      "grad_norm": 3.25,
      "learning_rate": 4.9300057863971846e-05,
      "loss": 0.9093,
      "step": 5520
    },
    {
      "epoch": 0.07830652566197872,
      "grad_norm": 3.203125,
      "learning_rate": 4.929953278006213e-05,
      "loss": 0.893,
      "step": 5522
    },
    {
      "epoch": 0.07833488731560494,
      "grad_norm": 3.078125,
      "learning_rate": 4.929900750207044e-05,
      "loss": 0.8534,
      "step": 5524
    },
    {
      "epoch": 0.07836324896923115,
      "grad_norm": 3.359375,
      "learning_rate": 4.9298482030000945e-05,
      "loss": 0.9451,
      "step": 5526
    },
    {
      "epoch": 0.07839161062285736,
      "grad_norm": 3.203125,
      "learning_rate": 4.929795636385786e-05,
      "loss": 0.8788,
      "step": 5528
    },
    {
      "epoch": 0.07841997227648358,
      "grad_norm": 3.265625,
      "learning_rate": 4.929743050364537e-05,
      "loss": 0.9086,
      "step": 5530
    },
    {
      "epoch": 0.07844833393010979,
      "grad_norm": 2.734375,
      "learning_rate": 4.9296904449367685e-05,
      "loss": 0.8937,
      "step": 5532
    },
    {
      "epoch": 0.078476695583736,
      "grad_norm": 3.46875,
      "learning_rate": 4.9296378201029004e-05,
      "loss": 0.8557,
      "step": 5534
    },
    {
      "epoch": 0.07850505723736223,
      "grad_norm": 3.21875,
      "learning_rate": 4.9295851758633536e-05,
      "loss": 0.8467,
      "step": 5536
    },
    {
      "epoch": 0.07853341889098844,
      "grad_norm": 2.96875,
      "learning_rate": 4.929532512218547e-05,
      "loss": 0.8489,
      "step": 5538
    },
    {
      "epoch": 0.07856178054461466,
      "grad_norm": 3.65625,
      "learning_rate": 4.9294798291689025e-05,
      "loss": 0.8713,
      "step": 5540
    },
    {
      "epoch": 0.07859014219824087,
      "grad_norm": 3.0625,
      "learning_rate": 4.9294271267148405e-05,
      "loss": 0.9089,
      "step": 5542
    },
    {
      "epoch": 0.07861850385186708,
      "grad_norm": 3.359375,
      "learning_rate": 4.9293744048567825e-05,
      "loss": 0.9075,
      "step": 5544
    },
    {
      "epoch": 0.0786468655054933,
      "grad_norm": 3.578125,
      "learning_rate": 4.9293216635951484e-05,
      "loss": 0.9247,
      "step": 5546
    },
    {
      "epoch": 0.07867522715911951,
      "grad_norm": 2.78125,
      "learning_rate": 4.9292689029303606e-05,
      "loss": 0.86,
      "step": 5548
    },
    {
      "epoch": 0.07870358881274572,
      "grad_norm": 3.03125,
      "learning_rate": 4.92921612286284e-05,
      "loss": 0.9092,
      "step": 5550
    },
    {
      "epoch": 0.07873195046637194,
      "grad_norm": 3.0,
      "learning_rate": 4.929163323393008e-05,
      "loss": 0.9006,
      "step": 5552
    },
    {
      "epoch": 0.07876031211999816,
      "grad_norm": 3.0,
      "learning_rate": 4.929110504521286e-05,
      "loss": 0.863,
      "step": 5554
    },
    {
      "epoch": 0.07878867377362438,
      "grad_norm": 3.109375,
      "learning_rate": 4.929057666248097e-05,
      "loss": 0.89,
      "step": 5556
    },
    {
      "epoch": 0.07881703542725059,
      "grad_norm": 3.125,
      "learning_rate": 4.929004808573863e-05,
      "loss": 0.8854,
      "step": 5558
    },
    {
      "epoch": 0.0788453970808768,
      "grad_norm": 3.28125,
      "learning_rate": 4.928951931499004e-05,
      "loss": 0.908,
      "step": 5560
    },
    {
      "epoch": 0.07887375873450302,
      "grad_norm": 2.984375,
      "learning_rate": 4.928899035023945e-05,
      "loss": 0.935,
      "step": 5562
    },
    {
      "epoch": 0.07890212038812923,
      "grad_norm": 3.15625,
      "learning_rate": 4.928846119149108e-05,
      "loss": 0.9024,
      "step": 5564
    },
    {
      "epoch": 0.07893048204175544,
      "grad_norm": 3.015625,
      "learning_rate": 4.928793183874913e-05,
      "loss": 0.8569,
      "step": 5566
    },
    {
      "epoch": 0.07895884369538166,
      "grad_norm": 3.21875,
      "learning_rate": 4.928740229201787e-05,
      "loss": 0.8635,
      "step": 5568
    },
    {
      "epoch": 0.07898720534900787,
      "grad_norm": 2.875,
      "learning_rate": 4.92868725513015e-05,
      "loss": 0.8701,
      "step": 5570
    },
    {
      "epoch": 0.07901556700263408,
      "grad_norm": 3.015625,
      "learning_rate": 4.928634261660425e-05,
      "loss": 0.8604,
      "step": 5572
    },
    {
      "epoch": 0.07904392865626031,
      "grad_norm": 3.0,
      "learning_rate": 4.928581248793037e-05,
      "loss": 0.8806,
      "step": 5574
    },
    {
      "epoch": 0.07907229030988652,
      "grad_norm": 3.4375,
      "learning_rate": 4.9285282165284085e-05,
      "loss": 0.9289,
      "step": 5576
    },
    {
      "epoch": 0.07910065196351274,
      "grad_norm": 2.828125,
      "learning_rate": 4.9284751648669634e-05,
      "loss": 0.9016,
      "step": 5578
    },
    {
      "epoch": 0.07912901361713895,
      "grad_norm": 3.1875,
      "learning_rate": 4.928422093809125e-05,
      "loss": 0.8545,
      "step": 5580
    },
    {
      "epoch": 0.07915737527076516,
      "grad_norm": 3.078125,
      "learning_rate": 4.9283690033553174e-05,
      "loss": 0.8787,
      "step": 5582
    },
    {
      "epoch": 0.07918573692439138,
      "grad_norm": 2.890625,
      "learning_rate": 4.928315893505965e-05,
      "loss": 0.8473,
      "step": 5584
    },
    {
      "epoch": 0.07921409857801759,
      "grad_norm": 3.0,
      "learning_rate": 4.928262764261491e-05,
      "loss": 0.8591,
      "step": 5586
    },
    {
      "epoch": 0.0792424602316438,
      "grad_norm": 3.140625,
      "learning_rate": 4.92820961562232e-05,
      "loss": 0.8527,
      "step": 5588
    },
    {
      "epoch": 0.07927082188527002,
      "grad_norm": 3.375,
      "learning_rate": 4.928156447588877e-05,
      "loss": 0.8503,
      "step": 5590
    },
    {
      "epoch": 0.07929918353889623,
      "grad_norm": 3.359375,
      "learning_rate": 4.928103260161587e-05,
      "loss": 0.8904,
      "step": 5592
    },
    {
      "epoch": 0.07932754519252246,
      "grad_norm": 2.90625,
      "learning_rate": 4.928050053340874e-05,
      "loss": 0.9232,
      "step": 5594
    },
    {
      "epoch": 0.07935590684614867,
      "grad_norm": 3.078125,
      "learning_rate": 4.9279968271271634e-05,
      "loss": 0.8704,
      "step": 5596
    },
    {
      "epoch": 0.07938426849977488,
      "grad_norm": 3.078125,
      "learning_rate": 4.9279435815208805e-05,
      "loss": 0.9102,
      "step": 5598
    },
    {
      "epoch": 0.0794126301534011,
      "grad_norm": 2.96875,
      "learning_rate": 4.927890316522449e-05,
      "loss": 0.8681,
      "step": 5600
    },
    {
      "epoch": 0.07944099180702731,
      "grad_norm": 2.984375,
      "learning_rate": 4.927837032132297e-05,
      "loss": 0.888,
      "step": 5602
    },
    {
      "epoch": 0.07946935346065352,
      "grad_norm": 3.140625,
      "learning_rate": 4.927783728350848e-05,
      "loss": 0.9215,
      "step": 5604
    },
    {
      "epoch": 0.07949771511427974,
      "grad_norm": 3.125,
      "learning_rate": 4.927730405178529e-05,
      "loss": 0.8644,
      "step": 5606
    },
    {
      "epoch": 0.07952607676790595,
      "grad_norm": 3.28125,
      "learning_rate": 4.927677062615765e-05,
      "loss": 0.9114,
      "step": 5608
    },
    {
      "epoch": 0.07955443842153216,
      "grad_norm": 2.671875,
      "learning_rate": 4.9276237006629824e-05,
      "loss": 0.8623,
      "step": 5610
    },
    {
      "epoch": 0.07958280007515839,
      "grad_norm": 2.890625,
      "learning_rate": 4.927570319320607e-05,
      "loss": 0.879,
      "step": 5612
    },
    {
      "epoch": 0.0796111617287846,
      "grad_norm": 2.75,
      "learning_rate": 4.927516918589066e-05,
      "loss": 0.8618,
      "step": 5614
    },
    {
      "epoch": 0.07963952338241082,
      "grad_norm": 3.25,
      "learning_rate": 4.927463498468785e-05,
      "loss": 0.9158,
      "step": 5616
    },
    {
      "epoch": 0.07966788503603703,
      "grad_norm": 3.46875,
      "learning_rate": 4.927410058960191e-05,
      "loss": 0.8627,
      "step": 5618
    },
    {
      "epoch": 0.07969624668966324,
      "grad_norm": 3.609375,
      "learning_rate": 4.927356600063712e-05,
      "loss": 0.9368,
      "step": 5620
    },
    {
      "epoch": 0.07972460834328945,
      "grad_norm": 2.9375,
      "learning_rate": 4.927303121779773e-05,
      "loss": 0.9188,
      "step": 5622
    },
    {
      "epoch": 0.07975296999691567,
      "grad_norm": 2.90625,
      "learning_rate": 4.927249624108802e-05,
      "loss": 0.9131,
      "step": 5624
    },
    {
      "epoch": 0.07978133165054188,
      "grad_norm": 2.75,
      "learning_rate": 4.927196107051227e-05,
      "loss": 0.8504,
      "step": 5626
    },
    {
      "epoch": 0.0798096933041681,
      "grad_norm": 3.40625,
      "learning_rate": 4.9271425706074744e-05,
      "loss": 0.9109,
      "step": 5628
    },
    {
      "epoch": 0.07983805495779431,
      "grad_norm": 3.109375,
      "learning_rate": 4.9270890147779726e-05,
      "loss": 0.8672,
      "step": 5630
    },
    {
      "epoch": 0.07986641661142053,
      "grad_norm": 3.34375,
      "learning_rate": 4.927035439563149e-05,
      "loss": 0.8931,
      "step": 5632
    },
    {
      "epoch": 0.07989477826504675,
      "grad_norm": 3.21875,
      "learning_rate": 4.926981844963431e-05,
      "loss": 0.8786,
      "step": 5634
    },
    {
      "epoch": 0.07992313991867296,
      "grad_norm": 3.0,
      "learning_rate": 4.9269282309792465e-05,
      "loss": 0.8745,
      "step": 5636
    },
    {
      "epoch": 0.07995150157229917,
      "grad_norm": 3.03125,
      "learning_rate": 4.926874597611025e-05,
      "loss": 0.8981,
      "step": 5638
    },
    {
      "epoch": 0.07997986322592539,
      "grad_norm": 2.703125,
      "learning_rate": 4.9268209448591944e-05,
      "loss": 0.8774,
      "step": 5640
    },
    {
      "epoch": 0.0800082248795516,
      "grad_norm": 2.734375,
      "learning_rate": 4.9267672727241834e-05,
      "loss": 0.8884,
      "step": 5642
    },
    {
      "epoch": 0.08003658653317781,
      "grad_norm": 2.84375,
      "learning_rate": 4.92671358120642e-05,
      "loss": 0.8539,
      "step": 5644
    },
    {
      "epoch": 0.08006494818680403,
      "grad_norm": 3.125,
      "learning_rate": 4.926659870306333e-05,
      "loss": 0.8138,
      "step": 5646
    },
    {
      "epoch": 0.08009330984043024,
      "grad_norm": 3.53125,
      "learning_rate": 4.926606140024352e-05,
      "loss": 0.9178,
      "step": 5648
    },
    {
      "epoch": 0.08012167149405647,
      "grad_norm": 3.078125,
      "learning_rate": 4.926552390360906e-05,
      "loss": 0.8908,
      "step": 5650
    },
    {
      "epoch": 0.08015003314768268,
      "grad_norm": 2.71875,
      "learning_rate": 4.9264986213164235e-05,
      "loss": 0.883,
      "step": 5652
    },
    {
      "epoch": 0.0801783948013089,
      "grad_norm": 3.421875,
      "learning_rate": 4.9264448328913356e-05,
      "loss": 0.8965,
      "step": 5654
    },
    {
      "epoch": 0.08020675645493511,
      "grad_norm": 2.96875,
      "learning_rate": 4.92639102508607e-05,
      "loss": 0.9029,
      "step": 5656
    },
    {
      "epoch": 0.08023511810856132,
      "grad_norm": 3.09375,
      "learning_rate": 4.926337197901059e-05,
      "loss": 0.9045,
      "step": 5658
    },
    {
      "epoch": 0.08026347976218753,
      "grad_norm": 3.890625,
      "learning_rate": 4.92628335133673e-05,
      "loss": 0.9001,
      "step": 5660
    },
    {
      "epoch": 0.08029184141581375,
      "grad_norm": 2.890625,
      "learning_rate": 4.926229485393513e-05,
      "loss": 0.8524,
      "step": 5662
    },
    {
      "epoch": 0.08032020306943996,
      "grad_norm": 3.03125,
      "learning_rate": 4.926175600071841e-05,
      "loss": 0.9297,
      "step": 5664
    },
    {
      "epoch": 0.08034856472306617,
      "grad_norm": 2.90625,
      "learning_rate": 4.9261216953721415e-05,
      "loss": 0.8889,
      "step": 5666
    },
    {
      "epoch": 0.08037692637669239,
      "grad_norm": 2.984375,
      "learning_rate": 4.926067771294846e-05,
      "loss": 0.8874,
      "step": 5668
    },
    {
      "epoch": 0.08040528803031861,
      "grad_norm": 3.5625,
      "learning_rate": 4.926013827840387e-05,
      "loss": 0.9031,
      "step": 5670
    },
    {
      "epoch": 0.08043364968394483,
      "grad_norm": 3.546875,
      "learning_rate": 4.925959865009193e-05,
      "loss": 0.9212,
      "step": 5672
    },
    {
      "epoch": 0.08046201133757104,
      "grad_norm": 3.234375,
      "learning_rate": 4.925905882801696e-05,
      "loss": 0.8823,
      "step": 5674
    },
    {
      "epoch": 0.08049037299119725,
      "grad_norm": 3.296875,
      "learning_rate": 4.9258518812183266e-05,
      "loss": 0.9034,
      "step": 5676
    },
    {
      "epoch": 0.08051873464482347,
      "grad_norm": 3.0,
      "learning_rate": 4.925797860259516e-05,
      "loss": 0.9158,
      "step": 5678
    },
    {
      "epoch": 0.08054709629844968,
      "grad_norm": 2.9375,
      "learning_rate": 4.9257438199256956e-05,
      "loss": 0.946,
      "step": 5680
    },
    {
      "epoch": 0.0805754579520759,
      "grad_norm": 3.3125,
      "learning_rate": 4.9256897602172986e-05,
      "loss": 0.8964,
      "step": 5682
    },
    {
      "epoch": 0.0806038196057021,
      "grad_norm": 3.296875,
      "learning_rate": 4.9256356811347555e-05,
      "loss": 0.9,
      "step": 5684
    },
    {
      "epoch": 0.08063218125932832,
      "grad_norm": 3.171875,
      "learning_rate": 4.925581582678498e-05,
      "loss": 0.9061,
      "step": 5686
    },
    {
      "epoch": 0.08066054291295453,
      "grad_norm": 3.46875,
      "learning_rate": 4.925527464848959e-05,
      "loss": 0.9439,
      "step": 5688
    },
    {
      "epoch": 0.08068890456658076,
      "grad_norm": 2.796875,
      "learning_rate": 4.9254733276465704e-05,
      "loss": 0.8705,
      "step": 5690
    },
    {
      "epoch": 0.08071726622020697,
      "grad_norm": 3.28125,
      "learning_rate": 4.925419171071765e-05,
      "loss": 0.859,
      "step": 5692
    },
    {
      "epoch": 0.08074562787383319,
      "grad_norm": 3.25,
      "learning_rate": 4.925364995124974e-05,
      "loss": 0.9201,
      "step": 5694
    },
    {
      "epoch": 0.0807739895274594,
      "grad_norm": 3.5625,
      "learning_rate": 4.9253107998066305e-05,
      "loss": 0.9282,
      "step": 5696
    },
    {
      "epoch": 0.08080235118108561,
      "grad_norm": 3.03125,
      "learning_rate": 4.9252565851171694e-05,
      "loss": 0.9211,
      "step": 5698
    },
    {
      "epoch": 0.08083071283471183,
      "grad_norm": 3.078125,
      "learning_rate": 4.925202351057021e-05,
      "loss": 0.928,
      "step": 5700
    },
    {
      "epoch": 0.08085907448833804,
      "grad_norm": 3.046875,
      "learning_rate": 4.925148097626621e-05,
      "loss": 0.8961,
      "step": 5702
    },
    {
      "epoch": 0.08088743614196425,
      "grad_norm": 2.875,
      "learning_rate": 4.9250938248264e-05,
      "loss": 0.898,
      "step": 5704
    },
    {
      "epoch": 0.08091579779559047,
      "grad_norm": 3.109375,
      "learning_rate": 4.9250395326567934e-05,
      "loss": 0.9004,
      "step": 5706
    },
    {
      "epoch": 0.0809441594492167,
      "grad_norm": 3.21875,
      "learning_rate": 4.924985221118234e-05,
      "loss": 0.7934,
      "step": 5708
    },
    {
      "epoch": 0.0809725211028429,
      "grad_norm": 3.09375,
      "learning_rate": 4.924930890211156e-05,
      "loss": 0.9151,
      "step": 5710
    },
    {
      "epoch": 0.08100088275646912,
      "grad_norm": 3.734375,
      "learning_rate": 4.9248765399359934e-05,
      "loss": 0.9909,
      "step": 5712
    },
    {
      "epoch": 0.08102924441009533,
      "grad_norm": 3.28125,
      "learning_rate": 4.9248221702931804e-05,
      "loss": 0.916,
      "step": 5714
    },
    {
      "epoch": 0.08105760606372155,
      "grad_norm": 3.359375,
      "learning_rate": 4.924767781283151e-05,
      "loss": 0.9348,
      "step": 5716
    },
    {
      "epoch": 0.08108596771734776,
      "grad_norm": 3.203125,
      "learning_rate": 4.924713372906339e-05,
      "loss": 0.8824,
      "step": 5718
    },
    {
      "epoch": 0.08111432937097397,
      "grad_norm": 3.1875,
      "learning_rate": 4.9246589451631795e-05,
      "loss": 0.8308,
      "step": 5720
    },
    {
      "epoch": 0.08114269102460019,
      "grad_norm": 3.0625,
      "learning_rate": 4.9246044980541084e-05,
      "loss": 0.8767,
      "step": 5722
    },
    {
      "epoch": 0.0811710526782264,
      "grad_norm": 2.734375,
      "learning_rate": 4.924550031579558e-05,
      "loss": 0.9068,
      "step": 5724
    },
    {
      "epoch": 0.08119941433185261,
      "grad_norm": 3.53125,
      "learning_rate": 4.924495545739966e-05,
      "loss": 0.8897,
      "step": 5726
    },
    {
      "epoch": 0.08122777598547884,
      "grad_norm": 3.171875,
      "learning_rate": 4.924441040535765e-05,
      "loss": 0.8774,
      "step": 5728
    },
    {
      "epoch": 0.08125613763910505,
      "grad_norm": 3.015625,
      "learning_rate": 4.924386515967393e-05,
      "loss": 0.916,
      "step": 5730
    },
    {
      "epoch": 0.08128449929273127,
      "grad_norm": 2.765625,
      "learning_rate": 4.924331972035284e-05,
      "loss": 0.8465,
      "step": 5732
    },
    {
      "epoch": 0.08131286094635748,
      "grad_norm": 3.015625,
      "learning_rate": 4.9242774087398734e-05,
      "loss": 0.8574,
      "step": 5734
    },
    {
      "epoch": 0.08134122259998369,
      "grad_norm": 3.21875,
      "learning_rate": 4.9242228260815984e-05,
      "loss": 0.9077,
      "step": 5736
    },
    {
      "epoch": 0.0813695842536099,
      "grad_norm": 3.171875,
      "learning_rate": 4.924168224060893e-05,
      "loss": 0.8847,
      "step": 5738
    },
    {
      "epoch": 0.08139794590723612,
      "grad_norm": 2.90625,
      "learning_rate": 4.9241136026781945e-05,
      "loss": 0.8297,
      "step": 5740
    },
    {
      "epoch": 0.08142630756086233,
      "grad_norm": 3.3125,
      "learning_rate": 4.924058961933939e-05,
      "loss": 0.8788,
      "step": 5742
    },
    {
      "epoch": 0.08145466921448855,
      "grad_norm": 3.078125,
      "learning_rate": 4.924004301828563e-05,
      "loss": 0.9296,
      "step": 5744
    },
    {
      "epoch": 0.08148303086811476,
      "grad_norm": 3.09375,
      "learning_rate": 4.9239496223625024e-05,
      "loss": 0.8882,
      "step": 5746
    },
    {
      "epoch": 0.08151139252174099,
      "grad_norm": 3.453125,
      "learning_rate": 4.9238949235361956e-05,
      "loss": 0.8954,
      "step": 5748
    },
    {
      "epoch": 0.0815397541753672,
      "grad_norm": 3.015625,
      "learning_rate": 4.923840205350077e-05,
      "loss": 0.8805,
      "step": 5750
    },
    {
      "epoch": 0.08156811582899341,
      "grad_norm": 2.640625,
      "learning_rate": 4.9237854678045855e-05,
      "loss": 0.8748,
      "step": 5752
    },
    {
      "epoch": 0.08159647748261963,
      "grad_norm": 2.8125,
      "learning_rate": 4.923730710900158e-05,
      "loss": 0.8876,
      "step": 5754
    },
    {
      "epoch": 0.08162483913624584,
      "grad_norm": 2.96875,
      "learning_rate": 4.923675934637232e-05,
      "loss": 0.8558,
      "step": 5756
    },
    {
      "epoch": 0.08165320078987205,
      "grad_norm": 3.28125,
      "learning_rate": 4.923621139016245e-05,
      "loss": 0.8878,
      "step": 5758
    },
    {
      "epoch": 0.08168156244349827,
      "grad_norm": 4.4375,
      "learning_rate": 4.923566324037633e-05,
      "loss": 0.9007,
      "step": 5760
    },
    {
      "epoch": 0.08170992409712448,
      "grad_norm": 3.71875,
      "learning_rate": 4.923511489701835e-05,
      "loss": 0.8933,
      "step": 5762
    },
    {
      "epoch": 0.08173828575075069,
      "grad_norm": 3.15625,
      "learning_rate": 4.92345663600929e-05,
      "loss": 0.8637,
      "step": 5764
    },
    {
      "epoch": 0.08176664740437692,
      "grad_norm": 3.265625,
      "learning_rate": 4.9234017629604357e-05,
      "loss": 0.8815,
      "step": 5766
    },
    {
      "epoch": 0.08179500905800313,
      "grad_norm": 3.015625,
      "learning_rate": 4.923346870555709e-05,
      "loss": 0.9399,
      "step": 5768
    },
    {
      "epoch": 0.08182337071162935,
      "grad_norm": 3.21875,
      "learning_rate": 4.923291958795549e-05,
      "loss": 0.9007,
      "step": 5770
    },
    {
      "epoch": 0.08185173236525556,
      "grad_norm": 2.984375,
      "learning_rate": 4.9232370276803955e-05,
      "loss": 0.8199,
      "step": 5772
    },
    {
      "epoch": 0.08188009401888177,
      "grad_norm": 3.046875,
      "learning_rate": 4.923182077210685e-05,
      "loss": 0.9075,
      "step": 5774
    },
    {
      "epoch": 0.08190845567250798,
      "grad_norm": 3.421875,
      "learning_rate": 4.923127107386858e-05,
      "loss": 0.8536,
      "step": 5776
    },
    {
      "epoch": 0.0819368173261342,
      "grad_norm": 3.125,
      "learning_rate": 4.923072118209354e-05,
      "loss": 0.9445,
      "step": 5778
    },
    {
      "epoch": 0.08196517897976041,
      "grad_norm": 3.21875,
      "learning_rate": 4.923017109678611e-05,
      "loss": 0.8769,
      "step": 5780
    },
    {
      "epoch": 0.08199354063338662,
      "grad_norm": 3.65625,
      "learning_rate": 4.922962081795068e-05,
      "loss": 0.9123,
      "step": 5782
    },
    {
      "epoch": 0.08202190228701284,
      "grad_norm": 3.5625,
      "learning_rate": 4.9229070345591656e-05,
      "loss": 0.8722,
      "step": 5784
    },
    {
      "epoch": 0.08205026394063906,
      "grad_norm": 3.109375,
      "learning_rate": 4.922851967971344e-05,
      "loss": 0.8751,
      "step": 5786
    },
    {
      "epoch": 0.08207862559426528,
      "grad_norm": 3.0,
      "learning_rate": 4.9227968820320404e-05,
      "loss": 0.9033,
      "step": 5788
    },
    {
      "epoch": 0.08210698724789149,
      "grad_norm": 2.90625,
      "learning_rate": 4.9227417767416975e-05,
      "loss": 0.9412,
      "step": 5790
    },
    {
      "epoch": 0.0821353489015177,
      "grad_norm": 3.1875,
      "learning_rate": 4.922686652100754e-05,
      "loss": 0.8889,
      "step": 5792
    },
    {
      "epoch": 0.08216371055514392,
      "grad_norm": 2.796875,
      "learning_rate": 4.922631508109652e-05,
      "loss": 0.92,
      "step": 5794
    },
    {
      "epoch": 0.08219207220877013,
      "grad_norm": 3.1875,
      "learning_rate": 4.922576344768829e-05,
      "loss": 0.903,
      "step": 5796
    },
    {
      "epoch": 0.08222043386239634,
      "grad_norm": 3.375,
      "learning_rate": 4.9225211620787284e-05,
      "loss": 0.9077,
      "step": 5798
    },
    {
      "epoch": 0.08224879551602256,
      "grad_norm": 2.984375,
      "learning_rate": 4.922465960039788e-05,
      "loss": 0.8957,
      "step": 5800
    },
    {
      "epoch": 0.08227715716964877,
      "grad_norm": 3.015625,
      "learning_rate": 4.922410738652452e-05,
      "loss": 0.8679,
      "step": 5802
    },
    {
      "epoch": 0.082305518823275,
      "grad_norm": 2.828125,
      "learning_rate": 4.922355497917159e-05,
      "loss": 0.8547,
      "step": 5804
    },
    {
      "epoch": 0.08233388047690121,
      "grad_norm": 3.21875,
      "learning_rate": 4.9223002378343514e-05,
      "loss": 0.9045,
      "step": 5806
    },
    {
      "epoch": 0.08236224213052742,
      "grad_norm": 3.1875,
      "learning_rate": 4.9222449584044707e-05,
      "loss": 0.9052,
      "step": 5808
    },
    {
      "epoch": 0.08239060378415364,
      "grad_norm": 3.0625,
      "learning_rate": 4.922189659627957e-05,
      "loss": 0.8865,
      "step": 5810
    },
    {
      "epoch": 0.08241896543777985,
      "grad_norm": 3.34375,
      "learning_rate": 4.9221343415052534e-05,
      "loss": 0.8903,
      "step": 5812
    },
    {
      "epoch": 0.08244732709140606,
      "grad_norm": 2.90625,
      "learning_rate": 4.922079004036801e-05,
      "loss": 0.8591,
      "step": 5814
    },
    {
      "epoch": 0.08247568874503228,
      "grad_norm": 3.078125,
      "learning_rate": 4.922023647223042e-05,
      "loss": 0.8843,
      "step": 5816
    },
    {
      "epoch": 0.08250405039865849,
      "grad_norm": 2.859375,
      "learning_rate": 4.9219682710644186e-05,
      "loss": 0.9006,
      "step": 5818
    },
    {
      "epoch": 0.0825324120522847,
      "grad_norm": 3.5,
      "learning_rate": 4.921912875561373e-05,
      "loss": 0.8692,
      "step": 5820
    },
    {
      "epoch": 0.08256077370591092,
      "grad_norm": 3.1875,
      "learning_rate": 4.9218574607143485e-05,
      "loss": 0.8704,
      "step": 5822
    },
    {
      "epoch": 0.08258913535953714,
      "grad_norm": 3.59375,
      "learning_rate": 4.921802026523786e-05,
      "loss": 0.8662,
      "step": 5824
    },
    {
      "epoch": 0.08261749701316336,
      "grad_norm": 2.828125,
      "learning_rate": 4.92174657299013e-05,
      "loss": 0.8701,
      "step": 5826
    },
    {
      "epoch": 0.08264585866678957,
      "grad_norm": 3.046875,
      "learning_rate": 4.921691100113822e-05,
      "loss": 0.8527,
      "step": 5828
    },
    {
      "epoch": 0.08267422032041578,
      "grad_norm": 3.46875,
      "learning_rate": 4.9216356078953064e-05,
      "loss": 0.9265,
      "step": 5830
    },
    {
      "epoch": 0.082702581974042,
      "grad_norm": 3.03125,
      "learning_rate": 4.921580096335025e-05,
      "loss": 0.8574,
      "step": 5832
    },
    {
      "epoch": 0.08273094362766821,
      "grad_norm": 3.453125,
      "learning_rate": 4.921524565433422e-05,
      "loss": 0.9335,
      "step": 5834
    },
    {
      "epoch": 0.08275930528129442,
      "grad_norm": 3.078125,
      "learning_rate": 4.9214690151909404e-05,
      "loss": 0.8836,
      "step": 5836
    },
    {
      "epoch": 0.08278766693492064,
      "grad_norm": 3.171875,
      "learning_rate": 4.921413445608025e-05,
      "loss": 0.9238,
      "step": 5838
    },
    {
      "epoch": 0.08281602858854685,
      "grad_norm": 3.09375,
      "learning_rate": 4.9213578566851185e-05,
      "loss": 0.8502,
      "step": 5840
    },
    {
      "epoch": 0.08284439024217306,
      "grad_norm": 3.03125,
      "learning_rate": 4.921302248422665e-05,
      "loss": 0.9062,
      "step": 5842
    },
    {
      "epoch": 0.08287275189579929,
      "grad_norm": 3.09375,
      "learning_rate": 4.92124662082111e-05,
      "loss": 0.8693,
      "step": 5844
    },
    {
      "epoch": 0.0829011135494255,
      "grad_norm": 3.0625,
      "learning_rate": 4.921190973880895e-05,
      "loss": 0.9055,
      "step": 5846
    },
    {
      "epoch": 0.08292947520305172,
      "grad_norm": 2.90625,
      "learning_rate": 4.9211353076024676e-05,
      "loss": 0.8249,
      "step": 5848
    },
    {
      "epoch": 0.08295783685667793,
      "grad_norm": 2.984375,
      "learning_rate": 4.9210796219862705e-05,
      "loss": 0.9075,
      "step": 5850
    },
    {
      "epoch": 0.08298619851030414,
      "grad_norm": 2.875,
      "learning_rate": 4.921023917032749e-05,
      "loss": 0.8871,
      "step": 5852
    },
    {
      "epoch": 0.08301456016393036,
      "grad_norm": 2.9375,
      "learning_rate": 4.920968192742348e-05,
      "loss": 0.8636,
      "step": 5854
    },
    {
      "epoch": 0.08304292181755657,
      "grad_norm": 2.984375,
      "learning_rate": 4.9209124491155126e-05,
      "loss": 0.8656,
      "step": 5856
    },
    {
      "epoch": 0.08307128347118278,
      "grad_norm": 3.140625,
      "learning_rate": 4.920856686152688e-05,
      "loss": 0.8762,
      "step": 5858
    },
    {
      "epoch": 0.083099645124809,
      "grad_norm": 2.9375,
      "learning_rate": 4.9208009038543196e-05,
      "loss": 0.863,
      "step": 5860
    },
    {
      "epoch": 0.08312800677843522,
      "grad_norm": 2.875,
      "learning_rate": 4.9207451022208525e-05,
      "loss": 0.8807,
      "step": 5862
    },
    {
      "epoch": 0.08315636843206144,
      "grad_norm": 2.765625,
      "learning_rate": 4.920689281252734e-05,
      "loss": 0.8575,
      "step": 5864
    },
    {
      "epoch": 0.08318473008568765,
      "grad_norm": 2.734375,
      "learning_rate": 4.920633440950407e-05,
      "loss": 0.8613,
      "step": 5866
    },
    {
      "epoch": 0.08321309173931386,
      "grad_norm": 2.84375,
      "learning_rate": 4.920577581314321e-05,
      "loss": 0.8734,
      "step": 5868
    },
    {
      "epoch": 0.08324145339294008,
      "grad_norm": 3.625,
      "learning_rate": 4.920521702344919e-05,
      "loss": 0.8866,
      "step": 5870
    },
    {
      "epoch": 0.08326981504656629,
      "grad_norm": 3.390625,
      "learning_rate": 4.9204658040426496e-05,
      "loss": 0.9465,
      "step": 5872
    },
    {
      "epoch": 0.0832981767001925,
      "grad_norm": 2.828125,
      "learning_rate": 4.920409886407959e-05,
      "loss": 0.879,
      "step": 5874
    },
    {
      "epoch": 0.08332653835381872,
      "grad_norm": 3.671875,
      "learning_rate": 4.920353949441291e-05,
      "loss": 0.8683,
      "step": 5876
    },
    {
      "epoch": 0.08335490000744493,
      "grad_norm": 3.09375,
      "learning_rate": 4.920297993143096e-05,
      "loss": 0.888,
      "step": 5878
    },
    {
      "epoch": 0.08338326166107114,
      "grad_norm": 2.84375,
      "learning_rate": 4.92024201751382e-05,
      "loss": 0.8611,
      "step": 5880
    },
    {
      "epoch": 0.08341162331469737,
      "grad_norm": 2.828125,
      "learning_rate": 4.920186022553909e-05,
      "loss": 0.8931,
      "step": 5882
    },
    {
      "epoch": 0.08343998496832358,
      "grad_norm": 3.25,
      "learning_rate": 4.92013000826381e-05,
      "loss": 0.8709,
      "step": 5884
    },
    {
      "epoch": 0.0834683466219498,
      "grad_norm": 3.28125,
      "learning_rate": 4.920073974643972e-05,
      "loss": 0.9245,
      "step": 5886
    },
    {
      "epoch": 0.08349670827557601,
      "grad_norm": 3.1875,
      "learning_rate": 4.9200179216948414e-05,
      "loss": 0.9039,
      "step": 5888
    },
    {
      "epoch": 0.08352506992920222,
      "grad_norm": 2.921875,
      "learning_rate": 4.919961849416867e-05,
      "loss": 0.8616,
      "step": 5890
    },
    {
      "epoch": 0.08355343158282844,
      "grad_norm": 3.046875,
      "learning_rate": 4.919905757810496e-05,
      "loss": 0.8229,
      "step": 5892
    },
    {
      "epoch": 0.08358179323645465,
      "grad_norm": 3.640625,
      "learning_rate": 4.919849646876175e-05,
      "loss": 0.8999,
      "step": 5894
    },
    {
      "epoch": 0.08361015489008086,
      "grad_norm": 3.59375,
      "learning_rate": 4.919793516614354e-05,
      "loss": 0.9154,
      "step": 5896
    },
    {
      "epoch": 0.08363851654370708,
      "grad_norm": 2.859375,
      "learning_rate": 4.919737367025481e-05,
      "loss": 0.8571,
      "step": 5898
    },
    {
      "epoch": 0.08366687819733329,
      "grad_norm": 3.3125,
      "learning_rate": 4.919681198110004e-05,
      "loss": 0.8856,
      "step": 5900
    },
    {
      "epoch": 0.08369523985095952,
      "grad_norm": 3.453125,
      "learning_rate": 4.919625009868373e-05,
      "loss": 0.933,
      "step": 5902
    },
    {
      "epoch": 0.08372360150458573,
      "grad_norm": 2.6875,
      "learning_rate": 4.919568802301034e-05,
      "loss": 0.816,
      "step": 5904
    },
    {
      "epoch": 0.08375196315821194,
      "grad_norm": 2.921875,
      "learning_rate": 4.9195125754084384e-05,
      "loss": 0.8829,
      "step": 5906
    },
    {
      "epoch": 0.08378032481183816,
      "grad_norm": 3.53125,
      "learning_rate": 4.9194563291910346e-05,
      "loss": 0.9458,
      "step": 5908
    },
    {
      "epoch": 0.08380868646546437,
      "grad_norm": 3.671875,
      "learning_rate": 4.9194000636492714e-05,
      "loss": 0.9059,
      "step": 5910
    },
    {
      "epoch": 0.08383704811909058,
      "grad_norm": 3.34375,
      "learning_rate": 4.9193437787835987e-05,
      "loss": 0.934,
      "step": 5912
    },
    {
      "epoch": 0.0838654097727168,
      "grad_norm": 3.390625,
      "learning_rate": 4.9192874745944664e-05,
      "loss": 0.9071,
      "step": 5914
    },
    {
      "epoch": 0.08389377142634301,
      "grad_norm": 3.03125,
      "learning_rate": 4.9192311510823226e-05,
      "loss": 0.8792,
      "step": 5916
    },
    {
      "epoch": 0.08392213307996922,
      "grad_norm": 3.0625,
      "learning_rate": 4.9191748082476184e-05,
      "loss": 0.8679,
      "step": 5918
    },
    {
      "epoch": 0.08395049473359545,
      "grad_norm": 2.984375,
      "learning_rate": 4.919118446090804e-05,
      "loss": 0.8767,
      "step": 5920
    },
    {
      "epoch": 0.08397885638722166,
      "grad_norm": 2.890625,
      "learning_rate": 4.919062064612329e-05,
      "loss": 0.9052,
      "step": 5922
    },
    {
      "epoch": 0.08400721804084788,
      "grad_norm": 3.265625,
      "learning_rate": 4.919005663812645e-05,
      "loss": 0.9091,
      "step": 5924
    },
    {
      "epoch": 0.08403557969447409,
      "grad_norm": 3.03125,
      "learning_rate": 4.9189492436922e-05,
      "loss": 0.8364,
      "step": 5926
    },
    {
      "epoch": 0.0840639413481003,
      "grad_norm": 2.71875,
      "learning_rate": 4.918892804251446e-05,
      "loss": 0.8281,
      "step": 5928
    },
    {
      "epoch": 0.08409230300172651,
      "grad_norm": 3.15625,
      "learning_rate": 4.9188363454908345e-05,
      "loss": 0.8739,
      "step": 5930
    },
    {
      "epoch": 0.08412066465535273,
      "grad_norm": 2.875,
      "learning_rate": 4.918779867410817e-05,
      "loss": 0.9088,
      "step": 5932
    },
    {
      "epoch": 0.08414902630897894,
      "grad_norm": 3.109375,
      "learning_rate": 4.9187233700118416e-05,
      "loss": 0.875,
      "step": 5934
    },
    {
      "epoch": 0.08417738796260515,
      "grad_norm": 3.09375,
      "learning_rate": 4.918666853294362e-05,
      "loss": 0.8548,
      "step": 5936
    },
    {
      "epoch": 0.08420574961623137,
      "grad_norm": 3.078125,
      "learning_rate": 4.9186103172588284e-05,
      "loss": 0.8903,
      "step": 5938
    },
    {
      "epoch": 0.0842341112698576,
      "grad_norm": 3.734375,
      "learning_rate": 4.918553761905692e-05,
      "loss": 0.9125,
      "step": 5940
    },
    {
      "epoch": 0.08426247292348381,
      "grad_norm": 2.9375,
      "learning_rate": 4.918497187235407e-05,
      "loss": 0.8749,
      "step": 5942
    },
    {
      "epoch": 0.08429083457711002,
      "grad_norm": 2.953125,
      "learning_rate": 4.9184405932484236e-05,
      "loss": 0.9243,
      "step": 5944
    },
    {
      "epoch": 0.08431919623073623,
      "grad_norm": 3.140625,
      "learning_rate": 4.918383979945194e-05,
      "loss": 0.9738,
      "step": 5946
    },
    {
      "epoch": 0.08434755788436245,
      "grad_norm": 3.296875,
      "learning_rate": 4.9183273473261695e-05,
      "loss": 0.9077,
      "step": 5948
    },
    {
      "epoch": 0.08437591953798866,
      "grad_norm": 2.90625,
      "learning_rate": 4.918270695391803e-05,
      "loss": 0.8804,
      "step": 5950
    },
    {
      "epoch": 0.08440428119161487,
      "grad_norm": 2.90625,
      "learning_rate": 4.918214024142547e-05,
      "loss": 0.9062,
      "step": 5952
    },
    {
      "epoch": 0.08443264284524109,
      "grad_norm": 3.125,
      "learning_rate": 4.918157333578855e-05,
      "loss": 0.9333,
      "step": 5954
    },
    {
      "epoch": 0.0844610044988673,
      "grad_norm": 3.28125,
      "learning_rate": 4.918100623701179e-05,
      "loss": 0.9088,
      "step": 5956
    },
    {
      "epoch": 0.08448936615249353,
      "grad_norm": 3.28125,
      "learning_rate": 4.9180438945099716e-05,
      "loss": 0.9038,
      "step": 5958
    },
    {
      "epoch": 0.08451772780611974,
      "grad_norm": 3.15625,
      "learning_rate": 4.917987146005687e-05,
      "loss": 0.8887,
      "step": 5960
    },
    {
      "epoch": 0.08454608945974595,
      "grad_norm": 3.328125,
      "learning_rate": 4.917930378188778e-05,
      "loss": 0.9105,
      "step": 5962
    },
    {
      "epoch": 0.08457445111337217,
      "grad_norm": 3.046875,
      "learning_rate": 4.917873591059696e-05,
      "loss": 0.8854,
      "step": 5964
    },
    {
      "epoch": 0.08460281276699838,
      "grad_norm": 3.046875,
      "learning_rate": 4.917816784618898e-05,
      "loss": 0.9006,
      "step": 5966
    },
    {
      "epoch": 0.0846311744206246,
      "grad_norm": 2.703125,
      "learning_rate": 4.9177599588668354e-05,
      "loss": 0.8863,
      "step": 5968
    },
    {
      "epoch": 0.08465953607425081,
      "grad_norm": 2.71875,
      "learning_rate": 4.917703113803963e-05,
      "loss": 0.8521,
      "step": 5970
    },
    {
      "epoch": 0.08468789772787702,
      "grad_norm": 3.4375,
      "learning_rate": 4.917646249430735e-05,
      "loss": 0.9113,
      "step": 5972
    },
    {
      "epoch": 0.08471625938150323,
      "grad_norm": 3.140625,
      "learning_rate": 4.9175893657476046e-05,
      "loss": 0.9184,
      "step": 5974
    },
    {
      "epoch": 0.08474462103512945,
      "grad_norm": 2.953125,
      "learning_rate": 4.917532462755027e-05,
      "loss": 0.8483,
      "step": 5976
    },
    {
      "epoch": 0.08477298268875567,
      "grad_norm": 3.109375,
      "learning_rate": 4.917475540453456e-05,
      "loss": 0.9417,
      "step": 5978
    },
    {
      "epoch": 0.08480134434238189,
      "grad_norm": 2.984375,
      "learning_rate": 4.917418598843347e-05,
      "loss": 0.9145,
      "step": 5980
    },
    {
      "epoch": 0.0848297059960081,
      "grad_norm": 3.09375,
      "learning_rate": 4.917361637925154e-05,
      "loss": 0.865,
      "step": 5982
    },
    {
      "epoch": 0.08485806764963431,
      "grad_norm": 2.828125,
      "learning_rate": 4.9173046576993325e-05,
      "loss": 0.8642,
      "step": 5984
    },
    {
      "epoch": 0.08488642930326053,
      "grad_norm": 3.03125,
      "learning_rate": 4.917247658166337e-05,
      "loss": 0.9066,
      "step": 5986
    },
    {
      "epoch": 0.08491479095688674,
      "grad_norm": 2.96875,
      "learning_rate": 4.917190639326624e-05,
      "loss": 0.9135,
      "step": 5988
    },
    {
      "epoch": 0.08494315261051295,
      "grad_norm": 3.28125,
      "learning_rate": 4.917133601180648e-05,
      "loss": 0.8589,
      "step": 5990
    },
    {
      "epoch": 0.08497151426413917,
      "grad_norm": 2.96875,
      "learning_rate": 4.9170765437288644e-05,
      "loss": 0.8566,
      "step": 5992
    },
    {
      "epoch": 0.08499987591776538,
      "grad_norm": 3.328125,
      "learning_rate": 4.9170194669717295e-05,
      "loss": 0.8724,
      "step": 5994
    },
    {
      "epoch": 0.0850282375713916,
      "grad_norm": 3.28125,
      "learning_rate": 4.9169623709096993e-05,
      "loss": 0.89,
      "step": 5996
    },
    {
      "epoch": 0.08505659922501782,
      "grad_norm": 3.421875,
      "learning_rate": 4.9169052555432286e-05,
      "loss": 0.9137,
      "step": 5998
    },
    {
      "epoch": 0.08508496087864403,
      "grad_norm": 3.28125,
      "learning_rate": 4.9168481208727746e-05,
      "loss": 0.8921,
      "step": 6000
    },
    {
      "epoch": 0.08511332253227025,
      "grad_norm": 2.921875,
      "learning_rate": 4.9167909668987935e-05,
      "loss": 0.8855,
      "step": 6002
    },
    {
      "epoch": 0.08514168418589646,
      "grad_norm": 3.421875,
      "learning_rate": 4.916733793621742e-05,
      "loss": 0.9018,
      "step": 6004
    },
    {
      "epoch": 0.08517004583952267,
      "grad_norm": 2.953125,
      "learning_rate": 4.916676601042076e-05,
      "loss": 0.8416,
      "step": 6006
    },
    {
      "epoch": 0.08519840749314889,
      "grad_norm": 3.34375,
      "learning_rate": 4.916619389160254e-05,
      "loss": 0.8922,
      "step": 6008
    },
    {
      "epoch": 0.0852267691467751,
      "grad_norm": 3.15625,
      "learning_rate": 4.9165621579767306e-05,
      "loss": 0.9117,
      "step": 6010
    },
    {
      "epoch": 0.08525513080040131,
      "grad_norm": 2.984375,
      "learning_rate": 4.9165049074919646e-05,
      "loss": 0.8551,
      "step": 6012
    },
    {
      "epoch": 0.08528349245402753,
      "grad_norm": 3.375,
      "learning_rate": 4.916447637706412e-05,
      "loss": 0.9438,
      "step": 6014
    },
    {
      "epoch": 0.08531185410765375,
      "grad_norm": 2.796875,
      "learning_rate": 4.916390348620531e-05,
      "loss": 0.8533,
      "step": 6016
    },
    {
      "epoch": 0.08534021576127997,
      "grad_norm": 3.046875,
      "learning_rate": 4.91633304023478e-05,
      "loss": 0.8668,
      "step": 6018
    },
    {
      "epoch": 0.08536857741490618,
      "grad_norm": 3.375,
      "learning_rate": 4.916275712549615e-05,
      "loss": 0.8782,
      "step": 6020
    },
    {
      "epoch": 0.0853969390685324,
      "grad_norm": 2.828125,
      "learning_rate": 4.916218365565495e-05,
      "loss": 0.9206,
      "step": 6022
    },
    {
      "epoch": 0.0854253007221586,
      "grad_norm": 2.84375,
      "learning_rate": 4.916160999282878e-05,
      "loss": 0.9033,
      "step": 6024
    },
    {
      "epoch": 0.08545366237578482,
      "grad_norm": 3.40625,
      "learning_rate": 4.916103613702222e-05,
      "loss": 0.8546,
      "step": 6026
    },
    {
      "epoch": 0.08548202402941103,
      "grad_norm": 3.03125,
      "learning_rate": 4.916046208823985e-05,
      "loss": 0.8447,
      "step": 6028
    },
    {
      "epoch": 0.08551038568303725,
      "grad_norm": 2.921875,
      "learning_rate": 4.915988784648625e-05,
      "loss": 0.8674,
      "step": 6030
    },
    {
      "epoch": 0.08553874733666346,
      "grad_norm": 3.125,
      "learning_rate": 4.915931341176603e-05,
      "loss": 0.9199,
      "step": 6032
    },
    {
      "epoch": 0.08556710899028967,
      "grad_norm": 3.09375,
      "learning_rate": 4.9158738784083754e-05,
      "loss": 0.8927,
      "step": 6034
    },
    {
      "epoch": 0.0855954706439159,
      "grad_norm": 3.0,
      "learning_rate": 4.915816396344402e-05,
      "loss": 0.8932,
      "step": 6036
    },
    {
      "epoch": 0.08562383229754211,
      "grad_norm": 3.171875,
      "learning_rate": 4.9157588949851416e-05,
      "loss": 0.8885,
      "step": 6038
    },
    {
      "epoch": 0.08565219395116833,
      "grad_norm": 3.25,
      "learning_rate": 4.9157013743310543e-05,
      "loss": 0.8951,
      "step": 6040
    },
    {
      "epoch": 0.08568055560479454,
      "grad_norm": 2.875,
      "learning_rate": 4.915643834382599e-05,
      "loss": 0.8843,
      "step": 6042
    },
    {
      "epoch": 0.08570891725842075,
      "grad_norm": 3.015625,
      "learning_rate": 4.915586275140235e-05,
      "loss": 0.8564,
      "step": 6044
    },
    {
      "epoch": 0.08573727891204697,
      "grad_norm": 2.921875,
      "learning_rate": 4.915528696604422e-05,
      "loss": 0.892,
      "step": 6046
    },
    {
      "epoch": 0.08576564056567318,
      "grad_norm": 3.0625,
      "learning_rate": 4.915471098775621e-05,
      "loss": 0.8733,
      "step": 6048
    },
    {
      "epoch": 0.08579400221929939,
      "grad_norm": 2.9375,
      "learning_rate": 4.915413481654291e-05,
      "loss": 0.874,
      "step": 6050
    },
    {
      "epoch": 0.0858223638729256,
      "grad_norm": 3.1875,
      "learning_rate": 4.915355845240892e-05,
      "loss": 0.8983,
      "step": 6052
    },
    {
      "epoch": 0.08585072552655182,
      "grad_norm": 3.0625,
      "learning_rate": 4.915298189535885e-05,
      "loss": 0.8738,
      "step": 6054
    },
    {
      "epoch": 0.08587908718017805,
      "grad_norm": 2.90625,
      "learning_rate": 4.91524051453973e-05,
      "loss": 0.8658,
      "step": 6056
    },
    {
      "epoch": 0.08590744883380426,
      "grad_norm": 3.0625,
      "learning_rate": 4.915182820252888e-05,
      "loss": 0.9114,
      "step": 6058
    },
    {
      "epoch": 0.08593581048743047,
      "grad_norm": 2.96875,
      "learning_rate": 4.91512510667582e-05,
      "loss": 0.8587,
      "step": 6060
    },
    {
      "epoch": 0.08596417214105669,
      "grad_norm": 3.015625,
      "learning_rate": 4.915067373808987e-05,
      "loss": 0.8865,
      "step": 6062
    },
    {
      "epoch": 0.0859925337946829,
      "grad_norm": 3.015625,
      "learning_rate": 4.9150096216528494e-05,
      "loss": 0.9233,
      "step": 6064
    },
    {
      "epoch": 0.08602089544830911,
      "grad_norm": 3.375,
      "learning_rate": 4.914951850207868e-05,
      "loss": 0.9379,
      "step": 6066
    },
    {
      "epoch": 0.08604925710193533,
      "grad_norm": 3.1875,
      "learning_rate": 4.914894059474506e-05,
      "loss": 0.9065,
      "step": 6068
    },
    {
      "epoch": 0.08607761875556154,
      "grad_norm": 3.390625,
      "learning_rate": 4.914836249453224e-05,
      "loss": 0.8954,
      "step": 6070
    },
    {
      "epoch": 0.08610598040918775,
      "grad_norm": 3.25,
      "learning_rate": 4.914778420144484e-05,
      "loss": 0.9241,
      "step": 6072
    },
    {
      "epoch": 0.08613434206281398,
      "grad_norm": 2.84375,
      "learning_rate": 4.914720571548747e-05,
      "loss": 0.8352,
      "step": 6074
    },
    {
      "epoch": 0.08616270371644019,
      "grad_norm": 2.96875,
      "learning_rate": 4.914662703666477e-05,
      "loss": 0.883,
      "step": 6076
    },
    {
      "epoch": 0.0861910653700664,
      "grad_norm": 2.9375,
      "learning_rate": 4.914604816498134e-05,
      "loss": 0.8734,
      "step": 6078
    },
    {
      "epoch": 0.08621942702369262,
      "grad_norm": 3.234375,
      "learning_rate": 4.914546910044181e-05,
      "loss": 0.9095,
      "step": 6080
    },
    {
      "epoch": 0.08624778867731883,
      "grad_norm": 3.46875,
      "learning_rate": 4.914488984305081e-05,
      "loss": 0.874,
      "step": 6082
    },
    {
      "epoch": 0.08627615033094505,
      "grad_norm": 2.984375,
      "learning_rate": 4.9144310392812976e-05,
      "loss": 0.871,
      "step": 6084
    },
    {
      "epoch": 0.08630451198457126,
      "grad_norm": 2.984375,
      "learning_rate": 4.914373074973291e-05,
      "loss": 0.8627,
      "step": 6086
    },
    {
      "epoch": 0.08633287363819747,
      "grad_norm": 2.890625,
      "learning_rate": 4.914315091381526e-05,
      "loss": 0.8534,
      "step": 6088
    },
    {
      "epoch": 0.08636123529182368,
      "grad_norm": 2.796875,
      "learning_rate": 4.9142570885064656e-05,
      "loss": 0.8676,
      "step": 6090
    },
    {
      "epoch": 0.0863895969454499,
      "grad_norm": 3.03125,
      "learning_rate": 4.914199066348573e-05,
      "loss": 0.9228,
      "step": 6092
    },
    {
      "epoch": 0.08641795859907613,
      "grad_norm": 3.078125,
      "learning_rate": 4.914141024908311e-05,
      "loss": 0.8517,
      "step": 6094
    },
    {
      "epoch": 0.08644632025270234,
      "grad_norm": 3.203125,
      "learning_rate": 4.914082964186143e-05,
      "loss": 0.9163,
      "step": 6096
    },
    {
      "epoch": 0.08647468190632855,
      "grad_norm": 3.046875,
      "learning_rate": 4.914024884182534e-05,
      "loss": 0.9037,
      "step": 6098
    },
    {
      "epoch": 0.08650304355995476,
      "grad_norm": 2.734375,
      "learning_rate": 4.913966784897948e-05,
      "loss": 0.9012,
      "step": 6100
    },
    {
      "epoch": 0.08653140521358098,
      "grad_norm": 2.84375,
      "learning_rate": 4.913908666332847e-05,
      "loss": 0.8996,
      "step": 6102
    },
    {
      "epoch": 0.08655976686720719,
      "grad_norm": 3.078125,
      "learning_rate": 4.913850528487697e-05,
      "loss": 0.8699,
      "step": 6104
    },
    {
      "epoch": 0.0865881285208334,
      "grad_norm": 3.015625,
      "learning_rate": 4.913792371362962e-05,
      "loss": 0.8741,
      "step": 6106
    },
    {
      "epoch": 0.08661649017445962,
      "grad_norm": 2.921875,
      "learning_rate": 4.913734194959105e-05,
      "loss": 0.8788,
      "step": 6108
    },
    {
      "epoch": 0.08664485182808583,
      "grad_norm": 2.96875,
      "learning_rate": 4.9136759992765936e-05,
      "loss": 0.905,
      "step": 6110
    },
    {
      "epoch": 0.08667321348171206,
      "grad_norm": 3.125,
      "learning_rate": 4.9136177843158906e-05,
      "loss": 0.9122,
      "step": 6112
    },
    {
      "epoch": 0.08670157513533827,
      "grad_norm": 3.0,
      "learning_rate": 4.913559550077461e-05,
      "loss": 0.8507,
      "step": 6114
    },
    {
      "epoch": 0.08672993678896448,
      "grad_norm": 3.640625,
      "learning_rate": 4.913501296561771e-05,
      "loss": 0.8961,
      "step": 6116
    },
    {
      "epoch": 0.0867582984425907,
      "grad_norm": 3.421875,
      "learning_rate": 4.9134430237692845e-05,
      "loss": 0.8672,
      "step": 6118
    },
    {
      "epoch": 0.08678666009621691,
      "grad_norm": 3.0625,
      "learning_rate": 4.913384731700468e-05,
      "loss": 0.8761,
      "step": 6120
    },
    {
      "epoch": 0.08681502174984312,
      "grad_norm": 3.046875,
      "learning_rate": 4.913326420355787e-05,
      "loss": 0.8454,
      "step": 6122
    },
    {
      "epoch": 0.08684338340346934,
      "grad_norm": 3.0625,
      "learning_rate": 4.913268089735706e-05,
      "loss": 0.9223,
      "step": 6124
    },
    {
      "epoch": 0.08687174505709555,
      "grad_norm": 3.0,
      "learning_rate": 4.9132097398406916e-05,
      "loss": 0.851,
      "step": 6126
    },
    {
      "epoch": 0.08690010671072176,
      "grad_norm": 3.15625,
      "learning_rate": 4.913151370671211e-05,
      "loss": 0.8899,
      "step": 6128
    },
    {
      "epoch": 0.08692846836434798,
      "grad_norm": 3.03125,
      "learning_rate": 4.913092982227729e-05,
      "loss": 0.8859,
      "step": 6130
    },
    {
      "epoch": 0.0869568300179742,
      "grad_norm": 3.140625,
      "learning_rate": 4.913034574510713e-05,
      "loss": 0.8453,
      "step": 6132
    },
    {
      "epoch": 0.08698519167160042,
      "grad_norm": 3.359375,
      "learning_rate": 4.9129761475206285e-05,
      "loss": 0.8815,
      "step": 6134
    },
    {
      "epoch": 0.08701355332522663,
      "grad_norm": 3.359375,
      "learning_rate": 4.9129177012579416e-05,
      "loss": 0.9054,
      "step": 6136
    },
    {
      "epoch": 0.08704191497885284,
      "grad_norm": 3.0625,
      "learning_rate": 4.912859235723122e-05,
      "loss": 0.8965,
      "step": 6138
    },
    {
      "epoch": 0.08707027663247906,
      "grad_norm": 3.171875,
      "learning_rate": 4.912800750916633e-05,
      "loss": 0.8775,
      "step": 6140
    },
    {
      "epoch": 0.08709863828610527,
      "grad_norm": 3.015625,
      "learning_rate": 4.912742246838944e-05,
      "loss": 0.8736,
      "step": 6142
    },
    {
      "epoch": 0.08712699993973148,
      "grad_norm": 3.125,
      "learning_rate": 4.912683723490522e-05,
      "loss": 0.918,
      "step": 6144
    },
    {
      "epoch": 0.0871553615933577,
      "grad_norm": 2.9375,
      "learning_rate": 4.912625180871834e-05,
      "loss": 0.8984,
      "step": 6146
    },
    {
      "epoch": 0.08718372324698391,
      "grad_norm": 2.84375,
      "learning_rate": 4.9125666189833483e-05,
      "loss": 0.9164,
      "step": 6148
    },
    {
      "epoch": 0.08721208490061012,
      "grad_norm": 2.953125,
      "learning_rate": 4.912508037825531e-05,
      "loss": 0.9069,
      "step": 6150
    },
    {
      "epoch": 0.08724044655423635,
      "grad_norm": 3.421875,
      "learning_rate": 4.9124494373988516e-05,
      "loss": 0.8824,
      "step": 6152
    },
    {
      "epoch": 0.08726880820786256,
      "grad_norm": 3.359375,
      "learning_rate": 4.912390817703778e-05,
      "loss": 0.8687,
      "step": 6154
    },
    {
      "epoch": 0.08729716986148878,
      "grad_norm": 3.453125,
      "learning_rate": 4.912332178740778e-05,
      "loss": 0.9036,
      "step": 6156
    },
    {
      "epoch": 0.08732553151511499,
      "grad_norm": 3.34375,
      "learning_rate": 4.91227352051032e-05,
      "loss": 0.8007,
      "step": 6158
    },
    {
      "epoch": 0.0873538931687412,
      "grad_norm": 3.015625,
      "learning_rate": 4.912214843012872e-05,
      "loss": 0.8963,
      "step": 6160
    },
    {
      "epoch": 0.08738225482236742,
      "grad_norm": 3.28125,
      "learning_rate": 4.912156146248903e-05,
      "loss": 0.8479,
      "step": 6162
    },
    {
      "epoch": 0.08741061647599363,
      "grad_norm": 3.03125,
      "learning_rate": 4.9120974302188824e-05,
      "loss": 0.9364,
      "step": 6164
    },
    {
      "epoch": 0.08743897812961984,
      "grad_norm": 3.109375,
      "learning_rate": 4.912038694923279e-05,
      "loss": 0.8637,
      "step": 6166
    },
    {
      "epoch": 0.08746733978324606,
      "grad_norm": 3.265625,
      "learning_rate": 4.911979940362561e-05,
      "loss": 0.9153,
      "step": 6168
    },
    {
      "epoch": 0.08749570143687228,
      "grad_norm": 3.078125,
      "learning_rate": 4.911921166537199e-05,
      "loss": 0.8486,
      "step": 6170
    },
    {
      "epoch": 0.0875240630904985,
      "grad_norm": 3.078125,
      "learning_rate": 4.911862373447661e-05,
      "loss": 0.8495,
      "step": 6172
    },
    {
      "epoch": 0.08755242474412471,
      "grad_norm": 3.078125,
      "learning_rate": 4.911803561094418e-05,
      "loss": 0.8741,
      "step": 6174
    },
    {
      "epoch": 0.08758078639775092,
      "grad_norm": 2.953125,
      "learning_rate": 4.911744729477939e-05,
      "loss": 0.8693,
      "step": 6176
    },
    {
      "epoch": 0.08760914805137714,
      "grad_norm": 2.984375,
      "learning_rate": 4.911685878598694e-05,
      "loss": 0.8658,
      "step": 6178
    },
    {
      "epoch": 0.08763750970500335,
      "grad_norm": 3.140625,
      "learning_rate": 4.9116270084571524e-05,
      "loss": 0.9404,
      "step": 6180
    },
    {
      "epoch": 0.08766587135862956,
      "grad_norm": 3.3125,
      "learning_rate": 4.911568119053786e-05,
      "loss": 0.9266,
      "step": 6182
    },
    {
      "epoch": 0.08769423301225578,
      "grad_norm": 3.828125,
      "learning_rate": 4.9115092103890636e-05,
      "loss": 0.9067,
      "step": 6184
    },
    {
      "epoch": 0.08772259466588199,
      "grad_norm": 2.96875,
      "learning_rate": 4.911450282463457e-05,
      "loss": 0.8358,
      "step": 6186
    },
    {
      "epoch": 0.0877509563195082,
      "grad_norm": 2.890625,
      "learning_rate": 4.911391335277435e-05,
      "loss": 0.891,
      "step": 6188
    },
    {
      "epoch": 0.08777931797313443,
      "grad_norm": 2.734375,
      "learning_rate": 4.91133236883147e-05,
      "loss": 0.8321,
      "step": 6190
    },
    {
      "epoch": 0.08780767962676064,
      "grad_norm": 3.234375,
      "learning_rate": 4.9112733831260325e-05,
      "loss": 0.9442,
      "step": 6192
    },
    {
      "epoch": 0.08783604128038686,
      "grad_norm": 2.734375,
      "learning_rate": 4.911214378161594e-05,
      "loss": 0.9014,
      "step": 6194
    },
    {
      "epoch": 0.08786440293401307,
      "grad_norm": 3.359375,
      "learning_rate": 4.911155353938625e-05,
      "loss": 0.8835,
      "step": 6196
    },
    {
      "epoch": 0.08789276458763928,
      "grad_norm": 3.1875,
      "learning_rate": 4.911096310457598e-05,
      "loss": 0.8834,
      "step": 6198
    },
    {
      "epoch": 0.0879211262412655,
      "grad_norm": 3.15625,
      "learning_rate": 4.9110372477189834e-05,
      "loss": 0.8237,
      "step": 6200
    },
    {
      "epoch": 0.08794948789489171,
      "grad_norm": 3.21875,
      "learning_rate": 4.910978165723253e-05,
      "loss": 0.9121,
      "step": 6202
    },
    {
      "epoch": 0.08797784954851792,
      "grad_norm": 3.125,
      "learning_rate": 4.91091906447088e-05,
      "loss": 0.8913,
      "step": 6204
    },
    {
      "epoch": 0.08800621120214414,
      "grad_norm": 3.078125,
      "learning_rate": 4.910859943962336e-05,
      "loss": 0.9398,
      "step": 6206
    },
    {
      "epoch": 0.08803457285577035,
      "grad_norm": 2.9375,
      "learning_rate": 4.910800804198092e-05,
      "loss": 0.8788,
      "step": 6208
    },
    {
      "epoch": 0.08806293450939658,
      "grad_norm": 3.1875,
      "learning_rate": 4.910741645178621e-05,
      "loss": 0.9172,
      "step": 6210
    },
    {
      "epoch": 0.08809129616302279,
      "grad_norm": 2.9375,
      "learning_rate": 4.9106824669043964e-05,
      "loss": 0.8654,
      "step": 6212
    },
    {
      "epoch": 0.088119657816649,
      "grad_norm": 2.984375,
      "learning_rate": 4.91062326937589e-05,
      "loss": 0.8587,
      "step": 6214
    },
    {
      "epoch": 0.08814801947027522,
      "grad_norm": 3.5,
      "learning_rate": 4.910564052593575e-05,
      "loss": 0.904,
      "step": 6216
    },
    {
      "epoch": 0.08817638112390143,
      "grad_norm": 2.859375,
      "learning_rate": 4.910504816557923e-05,
      "loss": 0.8675,
      "step": 6218
    },
    {
      "epoch": 0.08820474277752764,
      "grad_norm": 3.15625,
      "learning_rate": 4.910445561269409e-05,
      "loss": 0.8958,
      "step": 6220
    },
    {
      "epoch": 0.08823310443115386,
      "grad_norm": 3.359375,
      "learning_rate": 4.9103862867285056e-05,
      "loss": 0.8552,
      "step": 6222
    },
    {
      "epoch": 0.08826146608478007,
      "grad_norm": 2.90625,
      "learning_rate": 4.910326992935686e-05,
      "loss": 0.9202,
      "step": 6224
    },
    {
      "epoch": 0.08828982773840628,
      "grad_norm": 2.9375,
      "learning_rate": 4.910267679891424e-05,
      "loss": 0.8459,
      "step": 6226
    },
    {
      "epoch": 0.08831818939203251,
      "grad_norm": 3.109375,
      "learning_rate": 4.9102083475961934e-05,
      "loss": 0.8979,
      "step": 6228
    },
    {
      "epoch": 0.08834655104565872,
      "grad_norm": 2.96875,
      "learning_rate": 4.910148996050467e-05,
      "loss": 0.9162,
      "step": 6230
    },
    {
      "epoch": 0.08837491269928494,
      "grad_norm": 3.21875,
      "learning_rate": 4.9100896252547215e-05,
      "loss": 0.8816,
      "step": 6232
    },
    {
      "epoch": 0.08840327435291115,
      "grad_norm": 3.484375,
      "learning_rate": 4.910030235209428e-05,
      "loss": 0.8194,
      "step": 6234
    },
    {
      "epoch": 0.08843163600653736,
      "grad_norm": 3.21875,
      "learning_rate": 4.9099708259150634e-05,
      "loss": 0.9006,
      "step": 6236
    },
    {
      "epoch": 0.08845999766016358,
      "grad_norm": 2.84375,
      "learning_rate": 4.9099113973721e-05,
      "loss": 0.8937,
      "step": 6238
    },
    {
      "epoch": 0.08848835931378979,
      "grad_norm": 3.171875,
      "learning_rate": 4.909851949581015e-05,
      "loss": 0.8965,
      "step": 6240
    },
    {
      "epoch": 0.088516720967416,
      "grad_norm": 3.015625,
      "learning_rate": 4.90979248254228e-05,
      "loss": 0.8777,
      "step": 6242
    },
    {
      "epoch": 0.08854508262104221,
      "grad_norm": 3.015625,
      "learning_rate": 4.909732996256373e-05,
      "loss": 0.8811,
      "step": 6244
    },
    {
      "epoch": 0.08857344427466843,
      "grad_norm": 3.0,
      "learning_rate": 4.909673490723767e-05,
      "loss": 0.914,
      "step": 6246
    },
    {
      "epoch": 0.08860180592829466,
      "grad_norm": 2.75,
      "learning_rate": 4.909613965944938e-05,
      "loss": 0.873,
      "step": 6248
    },
    {
      "epoch": 0.08863016758192087,
      "grad_norm": 3.078125,
      "learning_rate": 4.909554421920362e-05,
      "loss": 0.8939,
      "step": 6250
    },
    {
      "epoch": 0.08865852923554708,
      "grad_norm": 3.328125,
      "learning_rate": 4.909494858650514e-05,
      "loss": 0.9186,
      "step": 6252
    },
    {
      "epoch": 0.0886868908891733,
      "grad_norm": 3.328125,
      "learning_rate": 4.90943527613587e-05,
      "loss": 0.8537,
      "step": 6254
    },
    {
      "epoch": 0.08871525254279951,
      "grad_norm": 3.21875,
      "learning_rate": 4.909375674376906e-05,
      "loss": 0.8592,
      "step": 6256
    },
    {
      "epoch": 0.08874361419642572,
      "grad_norm": 3.0,
      "learning_rate": 4.909316053374097e-05,
      "loss": 0.8785,
      "step": 6258
    },
    {
      "epoch": 0.08877197585005193,
      "grad_norm": 2.875,
      "learning_rate": 4.9092564131279204e-05,
      "loss": 0.8761,
      "step": 6260
    },
    {
      "epoch": 0.08880033750367815,
      "grad_norm": 3.1875,
      "learning_rate": 4.909196753638852e-05,
      "loss": 0.8668,
      "step": 6262
    },
    {
      "epoch": 0.08882869915730436,
      "grad_norm": 3.015625,
      "learning_rate": 4.909137074907368e-05,
      "loss": 0.884,
      "step": 6264
    },
    {
      "epoch": 0.08885706081093059,
      "grad_norm": 3.484375,
      "learning_rate": 4.909077376933947e-05,
      "loss": 0.9353,
      "step": 6266
    },
    {
      "epoch": 0.0888854224645568,
      "grad_norm": 2.953125,
      "learning_rate": 4.9090176597190626e-05,
      "loss": 0.9077,
      "step": 6268
    },
    {
      "epoch": 0.08891378411818301,
      "grad_norm": 2.875,
      "learning_rate": 4.908957923263194e-05,
      "loss": 0.8874,
      "step": 6270
    },
    {
      "epoch": 0.08894214577180923,
      "grad_norm": 2.75,
      "learning_rate": 4.9088981675668185e-05,
      "loss": 0.9148,
      "step": 6272
    },
    {
      "epoch": 0.08897050742543544,
      "grad_norm": 3.1875,
      "learning_rate": 4.908838392630412e-05,
      "loss": 0.8923,
      "step": 6274
    },
    {
      "epoch": 0.08899886907906165,
      "grad_norm": 3.203125,
      "learning_rate": 4.908778598454453e-05,
      "loss": 0.904,
      "step": 6276
    },
    {
      "epoch": 0.08902723073268787,
      "grad_norm": 2.78125,
      "learning_rate": 4.908718785039418e-05,
      "loss": 0.863,
      "step": 6278
    },
    {
      "epoch": 0.08905559238631408,
      "grad_norm": 3.296875,
      "learning_rate": 4.908658952385786e-05,
      "loss": 0.8702,
      "step": 6280
    },
    {
      "epoch": 0.0890839540399403,
      "grad_norm": 3.6875,
      "learning_rate": 4.908599100494034e-05,
      "loss": 0.874,
      "step": 6282
    },
    {
      "epoch": 0.08911231569356651,
      "grad_norm": 3.21875,
      "learning_rate": 4.90853922936464e-05,
      "loss": 0.8578,
      "step": 6284
    },
    {
      "epoch": 0.08914067734719273,
      "grad_norm": 2.78125,
      "learning_rate": 4.908479338998083e-05,
      "loss": 0.8572,
      "step": 6286
    },
    {
      "epoch": 0.08916903900081895,
      "grad_norm": 2.859375,
      "learning_rate": 4.9084194293948416e-05,
      "loss": 0.8532,
      "step": 6288
    },
    {
      "epoch": 0.08919740065444516,
      "grad_norm": 3.109375,
      "learning_rate": 4.9083595005553926e-05,
      "loss": 0.9366,
      "step": 6290
    },
    {
      "epoch": 0.08922576230807137,
      "grad_norm": 3.15625,
      "learning_rate": 4.9082995524802157e-05,
      "loss": 0.9382,
      "step": 6292
    },
    {
      "epoch": 0.08925412396169759,
      "grad_norm": 3.71875,
      "learning_rate": 4.908239585169789e-05,
      "loss": 0.9359,
      "step": 6294
    },
    {
      "epoch": 0.0892824856153238,
      "grad_norm": 3.078125,
      "learning_rate": 4.908179598624594e-05,
      "loss": 0.925,
      "step": 6296
    },
    {
      "epoch": 0.08931084726895001,
      "grad_norm": 3.015625,
      "learning_rate": 4.908119592845106e-05,
      "loss": 0.8669,
      "step": 6298
    },
    {
      "epoch": 0.08933920892257623,
      "grad_norm": 3.3125,
      "learning_rate": 4.908059567831808e-05,
      "loss": 0.8603,
      "step": 6300
    },
    {
      "epoch": 0.08936757057620244,
      "grad_norm": 3.609375,
      "learning_rate": 4.907999523585176e-05,
      "loss": 0.8601,
      "step": 6302
    },
    {
      "epoch": 0.08939593222982865,
      "grad_norm": 2.734375,
      "learning_rate": 4.907939460105692e-05,
      "loss": 0.8594,
      "step": 6304
    },
    {
      "epoch": 0.08942429388345488,
      "grad_norm": 3.203125,
      "learning_rate": 4.907879377393835e-05,
      "loss": 0.8719,
      "step": 6306
    },
    {
      "epoch": 0.0894526555370811,
      "grad_norm": 2.78125,
      "learning_rate": 4.907819275450084e-05,
      "loss": 0.9258,
      "step": 6308
    },
    {
      "epoch": 0.08948101719070731,
      "grad_norm": 3.390625,
      "learning_rate": 4.90775915427492e-05,
      "loss": 0.8811,
      "step": 6310
    },
    {
      "epoch": 0.08950937884433352,
      "grad_norm": 3.3125,
      "learning_rate": 4.907699013868824e-05,
      "loss": 0.9276,
      "step": 6312
    },
    {
      "epoch": 0.08953774049795973,
      "grad_norm": 3.578125,
      "learning_rate": 4.907638854232274e-05,
      "loss": 0.8658,
      "step": 6314
    },
    {
      "epoch": 0.08956610215158595,
      "grad_norm": 3.078125,
      "learning_rate": 4.907578675365753e-05,
      "loss": 0.8448,
      "step": 6316
    },
    {
      "epoch": 0.08959446380521216,
      "grad_norm": 2.9375,
      "learning_rate": 4.90751847726974e-05,
      "loss": 0.8545,
      "step": 6318
    },
    {
      "epoch": 0.08962282545883837,
      "grad_norm": 3.0,
      "learning_rate": 4.9074582599447165e-05,
      "loss": 0.8594,
      "step": 6320
    },
    {
      "epoch": 0.08965118711246459,
      "grad_norm": 3.140625,
      "learning_rate": 4.9073980233911635e-05,
      "loss": 0.864,
      "step": 6322
    },
    {
      "epoch": 0.08967954876609081,
      "grad_norm": 3.453125,
      "learning_rate": 4.9073377676095614e-05,
      "loss": 0.9123,
      "step": 6324
    },
    {
      "epoch": 0.08970791041971703,
      "grad_norm": 3.328125,
      "learning_rate": 4.907277492600393e-05,
      "loss": 0.8668,
      "step": 6326
    },
    {
      "epoch": 0.08973627207334324,
      "grad_norm": 2.984375,
      "learning_rate": 4.907217198364138e-05,
      "loss": 0.8822,
      "step": 6328
    },
    {
      "epoch": 0.08976463372696945,
      "grad_norm": 3.390625,
      "learning_rate": 4.907156884901278e-05,
      "loss": 0.829,
      "step": 6330
    },
    {
      "epoch": 0.08979299538059567,
      "grad_norm": 2.875,
      "learning_rate": 4.907096552212296e-05,
      "loss": 0.8687,
      "step": 6332
    },
    {
      "epoch": 0.08982135703422188,
      "grad_norm": 3.6875,
      "learning_rate": 4.907036200297673e-05,
      "loss": 0.8842,
      "step": 6334
    },
    {
      "epoch": 0.08984971868784809,
      "grad_norm": 3.25,
      "learning_rate": 4.906975829157893e-05,
      "loss": 0.9013,
      "step": 6336
    },
    {
      "epoch": 0.0898780803414743,
      "grad_norm": 3.296875,
      "learning_rate": 4.906915438793435e-05,
      "loss": 0.8717,
      "step": 6338
    },
    {
      "epoch": 0.08990644199510052,
      "grad_norm": 3.25,
      "learning_rate": 4.906855029204782e-05,
      "loss": 0.8495,
      "step": 6340
    },
    {
      "epoch": 0.08993480364872673,
      "grad_norm": 3.4375,
      "learning_rate": 4.906794600392419e-05,
      "loss": 0.8963,
      "step": 6342
    },
    {
      "epoch": 0.08996316530235296,
      "grad_norm": 3.25,
      "learning_rate": 4.906734152356826e-05,
      "loss": 0.9054,
      "step": 6344
    },
    {
      "epoch": 0.08999152695597917,
      "grad_norm": 3.0625,
      "learning_rate": 4.906673685098487e-05,
      "loss": 0.8951,
      "step": 6346
    },
    {
      "epoch": 0.09001988860960539,
      "grad_norm": 3.015625,
      "learning_rate": 4.906613198617886e-05,
      "loss": 0.8859,
      "step": 6348
    },
    {
      "epoch": 0.0900482502632316,
      "grad_norm": 3.015625,
      "learning_rate": 4.906552692915503e-05,
      "loss": 0.8411,
      "step": 6350
    },
    {
      "epoch": 0.09007661191685781,
      "grad_norm": 2.953125,
      "learning_rate": 4.906492167991824e-05,
      "loss": 0.8694,
      "step": 6352
    },
    {
      "epoch": 0.09010497357048403,
      "grad_norm": 2.96875,
      "learning_rate": 4.906431623847332e-05,
      "loss": 0.8775,
      "step": 6354
    },
    {
      "epoch": 0.09013333522411024,
      "grad_norm": 2.796875,
      "learning_rate": 4.9063710604825085e-05,
      "loss": 0.8967,
      "step": 6356
    },
    {
      "epoch": 0.09016169687773645,
      "grad_norm": 2.953125,
      "learning_rate": 4.90631047789784e-05,
      "loss": 0.8717,
      "step": 6358
    },
    {
      "epoch": 0.09019005853136267,
      "grad_norm": 2.84375,
      "learning_rate": 4.9062498760938094e-05,
      "loss": 0.925,
      "step": 6360
    },
    {
      "epoch": 0.09021842018498888,
      "grad_norm": 3.296875,
      "learning_rate": 4.9061892550709e-05,
      "loss": 0.9312,
      "step": 6362
    },
    {
      "epoch": 0.0902467818386151,
      "grad_norm": 2.890625,
      "learning_rate": 4.906128614829597e-05,
      "loss": 0.8783,
      "step": 6364
    },
    {
      "epoch": 0.09027514349224132,
      "grad_norm": 3.609375,
      "learning_rate": 4.906067955370384e-05,
      "loss": 0.8773,
      "step": 6366
    },
    {
      "epoch": 0.09030350514586753,
      "grad_norm": 3.078125,
      "learning_rate": 4.906007276693746e-05,
      "loss": 0.8603,
      "step": 6368
    },
    {
      "epoch": 0.09033186679949375,
      "grad_norm": 3.03125,
      "learning_rate": 4.905946578800167e-05,
      "loss": 0.8712,
      "step": 6370
    },
    {
      "epoch": 0.09036022845311996,
      "grad_norm": 2.953125,
      "learning_rate": 4.905885861690133e-05,
      "loss": 0.8405,
      "step": 6372
    },
    {
      "epoch": 0.09038859010674617,
      "grad_norm": 3.015625,
      "learning_rate": 4.905825125364127e-05,
      "loss": 0.8595,
      "step": 6374
    },
    {
      "epoch": 0.09041695176037239,
      "grad_norm": 3.1875,
      "learning_rate": 4.9057643698226355e-05,
      "loss": 0.8645,
      "step": 6376
    },
    {
      "epoch": 0.0904453134139986,
      "grad_norm": 2.84375,
      "learning_rate": 4.9057035950661435e-05,
      "loss": 0.8645,
      "step": 6378
    },
    {
      "epoch": 0.09047367506762481,
      "grad_norm": 3.15625,
      "learning_rate": 4.9056428010951374e-05,
      "loss": 0.9341,
      "step": 6380
    },
    {
      "epoch": 0.09050203672125104,
      "grad_norm": 3.234375,
      "learning_rate": 4.905581987910101e-05,
      "loss": 0.8564,
      "step": 6382
    },
    {
      "epoch": 0.09053039837487725,
      "grad_norm": 3.40625,
      "learning_rate": 4.9055211555115206e-05,
      "loss": 0.8833,
      "step": 6384
    },
    {
      "epoch": 0.09055876002850347,
      "grad_norm": 3.25,
      "learning_rate": 4.9054603038998824e-05,
      "loss": 0.8759,
      "step": 6386
    },
    {
      "epoch": 0.09058712168212968,
      "grad_norm": 3.015625,
      "learning_rate": 4.905399433075672e-05,
      "loss": 0.8734,
      "step": 6388
    },
    {
      "epoch": 0.09061548333575589,
      "grad_norm": 3.046875,
      "learning_rate": 4.905338543039377e-05,
      "loss": 0.9035,
      "step": 6390
    },
    {
      "epoch": 0.0906438449893821,
      "grad_norm": 3.015625,
      "learning_rate": 4.9052776337914816e-05,
      "loss": 0.9069,
      "step": 6392
    },
    {
      "epoch": 0.09067220664300832,
      "grad_norm": 3.4375,
      "learning_rate": 4.9052167053324736e-05,
      "loss": 0.8845,
      "step": 6394
    },
    {
      "epoch": 0.09070056829663453,
      "grad_norm": 3.765625,
      "learning_rate": 4.90515575766284e-05,
      "loss": 0.8457,
      "step": 6396
    },
    {
      "epoch": 0.09072892995026074,
      "grad_norm": 3.046875,
      "learning_rate": 4.9050947907830665e-05,
      "loss": 0.8829,
      "step": 6398
    },
    {
      "epoch": 0.09075729160388696,
      "grad_norm": 3.390625,
      "learning_rate": 4.9050338046936404e-05,
      "loss": 0.8961,
      "step": 6400
    },
    {
      "epoch": 0.09078565325751319,
      "grad_norm": 3.421875,
      "learning_rate": 4.9049727993950486e-05,
      "loss": 0.9306,
      "step": 6402
    },
    {
      "epoch": 0.0908140149111394,
      "grad_norm": 3.078125,
      "learning_rate": 4.904911774887779e-05,
      "loss": 0.8468,
      "step": 6404
    },
    {
      "epoch": 0.09084237656476561,
      "grad_norm": 2.984375,
      "learning_rate": 4.9048507311723184e-05,
      "loss": 0.9169,
      "step": 6406
    },
    {
      "epoch": 0.09087073821839182,
      "grad_norm": 3.53125,
      "learning_rate": 4.904789668249154e-05,
      "loss": 0.8817,
      "step": 6408
    },
    {
      "epoch": 0.09089909987201804,
      "grad_norm": 3.015625,
      "learning_rate": 4.9047285861187755e-05,
      "loss": 0.8795,
      "step": 6410
    },
    {
      "epoch": 0.09092746152564425,
      "grad_norm": 3.125,
      "learning_rate": 4.9046674847816685e-05,
      "loss": 0.889,
      "step": 6412
    },
    {
      "epoch": 0.09095582317927046,
      "grad_norm": 3.0625,
      "learning_rate": 4.9046063642383224e-05,
      "loss": 0.8819,
      "step": 6414
    },
    {
      "epoch": 0.09098418483289668,
      "grad_norm": 2.953125,
      "learning_rate": 4.904545224489225e-05,
      "loss": 0.8743,
      "step": 6416
    },
    {
      "epoch": 0.09101254648652289,
      "grad_norm": 2.8125,
      "learning_rate": 4.904484065534864e-05,
      "loss": 0.8897,
      "step": 6418
    },
    {
      "epoch": 0.09104090814014912,
      "grad_norm": 2.765625,
      "learning_rate": 4.904422887375728e-05,
      "loss": 0.8486,
      "step": 6420
    },
    {
      "epoch": 0.09106926979377533,
      "grad_norm": 2.6875,
      "learning_rate": 4.904361690012307e-05,
      "loss": 0.8739,
      "step": 6422
    },
    {
      "epoch": 0.09109763144740154,
      "grad_norm": 3.0,
      "learning_rate": 4.904300473445088e-05,
      "loss": 0.8787,
      "step": 6424
    },
    {
      "epoch": 0.09112599310102776,
      "grad_norm": 3.671875,
      "learning_rate": 4.904239237674561e-05,
      "loss": 0.9144,
      "step": 6426
    },
    {
      "epoch": 0.09115435475465397,
      "grad_norm": 2.96875,
      "learning_rate": 4.9041779827012156e-05,
      "loss": 0.9224,
      "step": 6428
    },
    {
      "epoch": 0.09118271640828018,
      "grad_norm": 3.0,
      "learning_rate": 4.904116708525539e-05,
      "loss": 0.8553,
      "step": 6430
    },
    {
      "epoch": 0.0912110780619064,
      "grad_norm": 2.515625,
      "learning_rate": 4.9040554151480235e-05,
      "loss": 0.8409,
      "step": 6432
    },
    {
      "epoch": 0.09123943971553261,
      "grad_norm": 2.921875,
      "learning_rate": 4.9039941025691556e-05,
      "loss": 0.938,
      "step": 6434
    },
    {
      "epoch": 0.09126780136915882,
      "grad_norm": 3.09375,
      "learning_rate": 4.9039327707894264e-05,
      "loss": 0.8864,
      "step": 6436
    },
    {
      "epoch": 0.09129616302278504,
      "grad_norm": 3.1875,
      "learning_rate": 4.903871419809327e-05,
      "loss": 0.8494,
      "step": 6438
    },
    {
      "epoch": 0.09132452467641126,
      "grad_norm": 3.0625,
      "learning_rate": 4.903810049629346e-05,
      "loss": 0.9219,
      "step": 6440
    },
    {
      "epoch": 0.09135288633003748,
      "grad_norm": 3.015625,
      "learning_rate": 4.903748660249973e-05,
      "loss": 0.9118,
      "step": 6442
    },
    {
      "epoch": 0.09138124798366369,
      "grad_norm": 3.171875,
      "learning_rate": 4.9036872516716994e-05,
      "loss": 0.9096,
      "step": 6444
    },
    {
      "epoch": 0.0914096096372899,
      "grad_norm": 2.921875,
      "learning_rate": 4.903625823895015e-05,
      "loss": 0.8269,
      "step": 6446
    },
    {
      "epoch": 0.09143797129091612,
      "grad_norm": 3.125,
      "learning_rate": 4.903564376920412e-05,
      "loss": 0.8904,
      "step": 6448
    },
    {
      "epoch": 0.09146633294454233,
      "grad_norm": 2.875,
      "learning_rate": 4.9035029107483785e-05,
      "loss": 0.8602,
      "step": 6450
    },
    {
      "epoch": 0.09149469459816854,
      "grad_norm": 3.140625,
      "learning_rate": 4.903441425379408e-05,
      "loss": 0.9092,
      "step": 6452
    },
    {
      "epoch": 0.09152305625179476,
      "grad_norm": 3.0,
      "learning_rate": 4.903379920813991e-05,
      "loss": 0.937,
      "step": 6454
    },
    {
      "epoch": 0.09155141790542097,
      "grad_norm": 2.796875,
      "learning_rate": 4.9033183970526174e-05,
      "loss": 0.8677,
      "step": 6456
    },
    {
      "epoch": 0.09157977955904718,
      "grad_norm": 2.828125,
      "learning_rate": 4.903256854095779e-05,
      "loss": 0.8993,
      "step": 6458
    },
    {
      "epoch": 0.09160814121267341,
      "grad_norm": 3.171875,
      "learning_rate": 4.903195291943969e-05,
      "loss": 0.8779,
      "step": 6460
    },
    {
      "epoch": 0.09163650286629962,
      "grad_norm": 2.78125,
      "learning_rate": 4.903133710597677e-05,
      "loss": 0.896,
      "step": 6462
    },
    {
      "epoch": 0.09166486451992584,
      "grad_norm": 3.765625,
      "learning_rate": 4.9030721100573963e-05,
      "loss": 0.9207,
      "step": 6464
    },
    {
      "epoch": 0.09169322617355205,
      "grad_norm": 3.59375,
      "learning_rate": 4.903010490323618e-05,
      "loss": 0.9265,
      "step": 6466
    },
    {
      "epoch": 0.09172158782717826,
      "grad_norm": 3.171875,
      "learning_rate": 4.9029488513968346e-05,
      "loss": 0.8462,
      "step": 6468
    },
    {
      "epoch": 0.09174994948080448,
      "grad_norm": 3.15625,
      "learning_rate": 4.902887193277539e-05,
      "loss": 0.8527,
      "step": 6470
    },
    {
      "epoch": 0.09177831113443069,
      "grad_norm": 3.21875,
      "learning_rate": 4.902825515966223e-05,
      "loss": 0.9096,
      "step": 6472
    },
    {
      "epoch": 0.0918066727880569,
      "grad_norm": 2.828125,
      "learning_rate": 4.902763819463379e-05,
      "loss": 0.8442,
      "step": 6474
    },
    {
      "epoch": 0.09183503444168312,
      "grad_norm": 3.125,
      "learning_rate": 4.9027021037695e-05,
      "loss": 0.899,
      "step": 6476
    },
    {
      "epoch": 0.09186339609530934,
      "grad_norm": 3.125,
      "learning_rate": 4.9026403688850795e-05,
      "loss": 0.8729,
      "step": 6478
    },
    {
      "epoch": 0.09189175774893556,
      "grad_norm": 3.34375,
      "learning_rate": 4.902578614810609e-05,
      "loss": 0.9065,
      "step": 6480
    },
    {
      "epoch": 0.09192011940256177,
      "grad_norm": 3.234375,
      "learning_rate": 4.902516841546584e-05,
      "loss": 0.8838,
      "step": 6482
    },
    {
      "epoch": 0.09194848105618798,
      "grad_norm": 3.265625,
      "learning_rate": 4.9024550490934965e-05,
      "loss": 0.8958,
      "step": 6484
    },
    {
      "epoch": 0.0919768427098142,
      "grad_norm": 2.984375,
      "learning_rate": 4.90239323745184e-05,
      "loss": 0.8858,
      "step": 6486
    },
    {
      "epoch": 0.09200520436344041,
      "grad_norm": 2.921875,
      "learning_rate": 4.902331406622109e-05,
      "loss": 0.828,
      "step": 6488
    },
    {
      "epoch": 0.09203356601706662,
      "grad_norm": 2.859375,
      "learning_rate": 4.902269556604796e-05,
      "loss": 0.8768,
      "step": 6490
    },
    {
      "epoch": 0.09206192767069284,
      "grad_norm": 3.25,
      "learning_rate": 4.902207687400396e-05,
      "loss": 0.8645,
      "step": 6492
    },
    {
      "epoch": 0.09209028932431905,
      "grad_norm": 3.171875,
      "learning_rate": 4.902145799009404e-05,
      "loss": 0.8542,
      "step": 6494
    },
    {
      "epoch": 0.09211865097794526,
      "grad_norm": 3.0,
      "learning_rate": 4.902083891432312e-05,
      "loss": 0.9246,
      "step": 6496
    },
    {
      "epoch": 0.09214701263157149,
      "grad_norm": 3.484375,
      "learning_rate": 4.902021964669617e-05,
      "loss": 0.8607,
      "step": 6498
    },
    {
      "epoch": 0.0921753742851977,
      "grad_norm": 3.015625,
      "learning_rate": 4.901960018721812e-05,
      "loss": 0.8936,
      "step": 6500
    },
    {
      "epoch": 0.09220373593882392,
      "grad_norm": 3.46875,
      "learning_rate": 4.901898053589391e-05,
      "loss": 0.8963,
      "step": 6502
    },
    {
      "epoch": 0.09223209759245013,
      "grad_norm": 3.03125,
      "learning_rate": 4.901836069272851e-05,
      "loss": 0.8849,
      "step": 6504
    },
    {
      "epoch": 0.09226045924607634,
      "grad_norm": 3.5,
      "learning_rate": 4.901774065772686e-05,
      "loss": 0.9654,
      "step": 6506
    },
    {
      "epoch": 0.09228882089970256,
      "grad_norm": 2.890625,
      "learning_rate": 4.901712043089392e-05,
      "loss": 0.924,
      "step": 6508
    },
    {
      "epoch": 0.09231718255332877,
      "grad_norm": 3.296875,
      "learning_rate": 4.9016500012234626e-05,
      "loss": 0.8835,
      "step": 6510
    },
    {
      "epoch": 0.09234554420695498,
      "grad_norm": 3.265625,
      "learning_rate": 4.901587940175394e-05,
      "loss": 0.8731,
      "step": 6512
    },
    {
      "epoch": 0.0923739058605812,
      "grad_norm": 3.109375,
      "learning_rate": 4.9015258599456835e-05,
      "loss": 0.8667,
      "step": 6514
    },
    {
      "epoch": 0.09240226751420741,
      "grad_norm": 2.90625,
      "learning_rate": 4.9014637605348255e-05,
      "loss": 0.8724,
      "step": 6516
    },
    {
      "epoch": 0.09243062916783364,
      "grad_norm": 3.109375,
      "learning_rate": 4.9014016419433164e-05,
      "loss": 0.894,
      "step": 6518
    },
    {
      "epoch": 0.09245899082145985,
      "grad_norm": 2.9375,
      "learning_rate": 4.9013395041716515e-05,
      "loss": 0.8424,
      "step": 6520
    },
    {
      "epoch": 0.09248735247508606,
      "grad_norm": 2.828125,
      "learning_rate": 4.901277347220329e-05,
      "loss": 0.8635,
      "step": 6522
    },
    {
      "epoch": 0.09251571412871228,
      "grad_norm": 2.984375,
      "learning_rate": 4.901215171089843e-05,
      "loss": 0.8755,
      "step": 6524
    },
    {
      "epoch": 0.09254407578233849,
      "grad_norm": 3.28125,
      "learning_rate": 4.901152975780692e-05,
      "loss": 0.8763,
      "step": 6526
    },
    {
      "epoch": 0.0925724374359647,
      "grad_norm": 3.59375,
      "learning_rate": 4.9010907612933716e-05,
      "loss": 0.9336,
      "step": 6528
    },
    {
      "epoch": 0.09260079908959092,
      "grad_norm": 3.25,
      "learning_rate": 4.90102852762838e-05,
      "loss": 0.8731,
      "step": 6530
    },
    {
      "epoch": 0.09262916074321713,
      "grad_norm": 3.140625,
      "learning_rate": 4.9009662747862115e-05,
      "loss": 0.863,
      "step": 6532
    },
    {
      "epoch": 0.09265752239684334,
      "grad_norm": 3.40625,
      "learning_rate": 4.900904002767367e-05,
      "loss": 0.8919,
      "step": 6534
    },
    {
      "epoch": 0.09268588405046957,
      "grad_norm": 3.796875,
      "learning_rate": 4.900841711572341e-05,
      "loss": 0.8807,
      "step": 6536
    },
    {
      "epoch": 0.09271424570409578,
      "grad_norm": 2.953125,
      "learning_rate": 4.900779401201633e-05,
      "loss": 0.8986,
      "step": 6538
    },
    {
      "epoch": 0.092742607357722,
      "grad_norm": 2.6875,
      "learning_rate": 4.900717071655739e-05,
      "loss": 0.8573,
      "step": 6540
    },
    {
      "epoch": 0.09277096901134821,
      "grad_norm": 3.03125,
      "learning_rate": 4.9006547229351587e-05,
      "loss": 0.8298,
      "step": 6542
    },
    {
      "epoch": 0.09279933066497442,
      "grad_norm": 2.890625,
      "learning_rate": 4.900592355040388e-05,
      "loss": 0.865,
      "step": 6544
    },
    {
      "epoch": 0.09282769231860064,
      "grad_norm": 2.953125,
      "learning_rate": 4.900529967971926e-05,
      "loss": 0.9298,
      "step": 6546
    },
    {
      "epoch": 0.09285605397222685,
      "grad_norm": 3.09375,
      "learning_rate": 4.900467561730272e-05,
      "loss": 0.8454,
      "step": 6548
    },
    {
      "epoch": 0.09288441562585306,
      "grad_norm": 3.109375,
      "learning_rate": 4.900405136315923e-05,
      "loss": 0.9067,
      "step": 6550
    },
    {
      "epoch": 0.09291277727947928,
      "grad_norm": 2.796875,
      "learning_rate": 4.900342691729378e-05,
      "loss": 0.8655,
      "step": 6552
    },
    {
      "epoch": 0.09294113893310549,
      "grad_norm": 3.234375,
      "learning_rate": 4.900280227971136e-05,
      "loss": 0.8891,
      "step": 6554
    },
    {
      "epoch": 0.09296950058673172,
      "grad_norm": 3.046875,
      "learning_rate": 4.9002177450416946e-05,
      "loss": 0.8318,
      "step": 6556
    },
    {
      "epoch": 0.09299786224035793,
      "grad_norm": 3.09375,
      "learning_rate": 4.9001552429415554e-05,
      "loss": 0.867,
      "step": 6558
    },
    {
      "epoch": 0.09302622389398414,
      "grad_norm": 3.71875,
      "learning_rate": 4.9000927216712154e-05,
      "loss": 0.945,
      "step": 6560
    },
    {
      "epoch": 0.09305458554761036,
      "grad_norm": 3.09375,
      "learning_rate": 4.900030181231175e-05,
      "loss": 0.8675,
      "step": 6562
    },
    {
      "epoch": 0.09308294720123657,
      "grad_norm": 3.234375,
      "learning_rate": 4.8999676216219335e-05,
      "loss": 0.8982,
      "step": 6564
    },
    {
      "epoch": 0.09311130885486278,
      "grad_norm": 3.109375,
      "learning_rate": 4.89990504284399e-05,
      "loss": 0.8988,
      "step": 6566
    },
    {
      "epoch": 0.093139670508489,
      "grad_norm": 3.078125,
      "learning_rate": 4.8998424448978454e-05,
      "loss": 0.9393,
      "step": 6568
    },
    {
      "epoch": 0.09316803216211521,
      "grad_norm": 3.390625,
      "learning_rate": 4.899779827783999e-05,
      "loss": 0.9388,
      "step": 6570
    },
    {
      "epoch": 0.09319639381574142,
      "grad_norm": 3.046875,
      "learning_rate": 4.899717191502951e-05,
      "loss": 0.8086,
      "step": 6572
    },
    {
      "epoch": 0.09322475546936765,
      "grad_norm": 3.171875,
      "learning_rate": 4.899654536055202e-05,
      "loss": 0.89,
      "step": 6574
    },
    {
      "epoch": 0.09325311712299386,
      "grad_norm": 3.109375,
      "learning_rate": 4.899591861441252e-05,
      "loss": 0.8898,
      "step": 6576
    },
    {
      "epoch": 0.09328147877662007,
      "grad_norm": 2.734375,
      "learning_rate": 4.899529167661602e-05,
      "loss": 0.8941,
      "step": 6578
    },
    {
      "epoch": 0.09330984043024629,
      "grad_norm": 2.9375,
      "learning_rate": 4.8994664547167524e-05,
      "loss": 0.8683,
      "step": 6580
    },
    {
      "epoch": 0.0933382020838725,
      "grad_norm": 2.875,
      "learning_rate": 4.8994037226072037e-05,
      "loss": 0.7914,
      "step": 6582
    },
    {
      "epoch": 0.09336656373749871,
      "grad_norm": 2.84375,
      "learning_rate": 4.899340971333458e-05,
      "loss": 0.8767,
      "step": 6584
    },
    {
      "epoch": 0.09339492539112493,
      "grad_norm": 3.0625,
      "learning_rate": 4.899278200896015e-05,
      "loss": 0.8491,
      "step": 6586
    },
    {
      "epoch": 0.09342328704475114,
      "grad_norm": 3.03125,
      "learning_rate": 4.8992154112953784e-05,
      "loss": 0.9286,
      "step": 6588
    },
    {
      "epoch": 0.09345164869837735,
      "grad_norm": 3.125,
      "learning_rate": 4.899152602532047e-05,
      "loss": 0.8441,
      "step": 6590
    },
    {
      "epoch": 0.09348001035200357,
      "grad_norm": 3.03125,
      "learning_rate": 4.899089774606525e-05,
      "loss": 0.8873,
      "step": 6592
    },
    {
      "epoch": 0.0935083720056298,
      "grad_norm": 3.4375,
      "learning_rate": 4.899026927519311e-05,
      "loss": 0.8617,
      "step": 6594
    },
    {
      "epoch": 0.09353673365925601,
      "grad_norm": 2.859375,
      "learning_rate": 4.89896406127091e-05,
      "loss": 0.8339,
      "step": 6596
    },
    {
      "epoch": 0.09356509531288222,
      "grad_norm": 3.09375,
      "learning_rate": 4.898901175861824e-05,
      "loss": 0.8763,
      "step": 6598
    },
    {
      "epoch": 0.09359345696650843,
      "grad_norm": 2.9375,
      "learning_rate": 4.898838271292553e-05,
      "loss": 0.8788,
      "step": 6600
    },
    {
      "epoch": 0.09362181862013465,
      "grad_norm": 3.203125,
      "learning_rate": 4.8987753475636014e-05,
      "loss": 0.9189,
      "step": 6602
    },
    {
      "epoch": 0.09365018027376086,
      "grad_norm": 3.390625,
      "learning_rate": 4.8987124046754704e-05,
      "loss": 0.8549,
      "step": 6604
    },
    {
      "epoch": 0.09367854192738707,
      "grad_norm": 3.0,
      "learning_rate": 4.8986494426286644e-05,
      "loss": 0.8558,
      "step": 6606
    },
    {
      "epoch": 0.09370690358101329,
      "grad_norm": 2.84375,
      "learning_rate": 4.898586461423684e-05,
      "loss": 0.838,
      "step": 6608
    },
    {
      "epoch": 0.0937352652346395,
      "grad_norm": 3.0,
      "learning_rate": 4.898523461061035e-05,
      "loss": 0.8473,
      "step": 6610
    },
    {
      "epoch": 0.09376362688826571,
      "grad_norm": 2.96875,
      "learning_rate": 4.898460441541218e-05,
      "loss": 0.8665,
      "step": 6612
    },
    {
      "epoch": 0.09379198854189194,
      "grad_norm": 3.296875,
      "learning_rate": 4.8983974028647386e-05,
      "loss": 0.9024,
      "step": 6614
    },
    {
      "epoch": 0.09382035019551815,
      "grad_norm": 3.15625,
      "learning_rate": 4.898334345032099e-05,
      "loss": 0.9029,
      "step": 6616
    },
    {
      "epoch": 0.09384871184914437,
      "grad_norm": 3.265625,
      "learning_rate": 4.898271268043801e-05,
      "loss": 0.9052,
      "step": 6618
    },
    {
      "epoch": 0.09387707350277058,
      "grad_norm": 3.109375,
      "learning_rate": 4.8982081719003525e-05,
      "loss": 0.8748,
      "step": 6620
    },
    {
      "epoch": 0.0939054351563968,
      "grad_norm": 3.140625,
      "learning_rate": 4.8981450566022545e-05,
      "loss": 0.8654,
      "step": 6622
    },
    {
      "epoch": 0.09393379681002301,
      "grad_norm": 3.09375,
      "learning_rate": 4.898081922150013e-05,
      "loss": 0.8366,
      "step": 6624
    },
    {
      "epoch": 0.09396215846364922,
      "grad_norm": 3.359375,
      "learning_rate": 4.89801876854413e-05,
      "loss": 0.913,
      "step": 6626
    },
    {
      "epoch": 0.09399052011727543,
      "grad_norm": 3.0,
      "learning_rate": 4.897955595785112e-05,
      "loss": 0.8899,
      "step": 6628
    },
    {
      "epoch": 0.09401888177090165,
      "grad_norm": 3.25,
      "learning_rate": 4.897892403873462e-05,
      "loss": 0.9034,
      "step": 6630
    },
    {
      "epoch": 0.09404724342452787,
      "grad_norm": 2.546875,
      "learning_rate": 4.897829192809686e-05,
      "loss": 0.8241,
      "step": 6632
    },
    {
      "epoch": 0.09407560507815409,
      "grad_norm": 3.234375,
      "learning_rate": 4.8977659625942886e-05,
      "loss": 0.8724,
      "step": 6634
    },
    {
      "epoch": 0.0941039667317803,
      "grad_norm": 3.21875,
      "learning_rate": 4.8977027132277736e-05,
      "loss": 0.8877,
      "step": 6636
    },
    {
      "epoch": 0.09413232838540651,
      "grad_norm": 2.90625,
      "learning_rate": 4.8976394447106474e-05,
      "loss": 0.8993,
      "step": 6638
    },
    {
      "epoch": 0.09416069003903273,
      "grad_norm": 2.984375,
      "learning_rate": 4.8975761570434155e-05,
      "loss": 0.8916,
      "step": 6640
    },
    {
      "epoch": 0.09418905169265894,
      "grad_norm": 2.953125,
      "learning_rate": 4.897512850226582e-05,
      "loss": 0.8337,
      "step": 6642
    },
    {
      "epoch": 0.09421741334628515,
      "grad_norm": 3.4375,
      "learning_rate": 4.897449524260654e-05,
      "loss": 0.879,
      "step": 6644
    },
    {
      "epoch": 0.09424577499991137,
      "grad_norm": 3.484375,
      "learning_rate": 4.8973861791461374e-05,
      "loss": 0.8903,
      "step": 6646
    },
    {
      "epoch": 0.09427413665353758,
      "grad_norm": 3.09375,
      "learning_rate": 4.897322814883537e-05,
      "loss": 0.8776,
      "step": 6648
    },
    {
      "epoch": 0.09430249830716379,
      "grad_norm": 3.5625,
      "learning_rate": 4.897259431473359e-05,
      "loss": 0.9234,
      "step": 6650
    },
    {
      "epoch": 0.09433085996079002,
      "grad_norm": 3.0625,
      "learning_rate": 4.89719602891611e-05,
      "loss": 0.9039,
      "step": 6652
    },
    {
      "epoch": 0.09435922161441623,
      "grad_norm": 2.671875,
      "learning_rate": 4.897132607212297e-05,
      "loss": 0.8646,
      "step": 6654
    },
    {
      "epoch": 0.09438758326804245,
      "grad_norm": 2.984375,
      "learning_rate": 4.897069166362425e-05,
      "loss": 0.8429,
      "step": 6656
    },
    {
      "epoch": 0.09441594492166866,
      "grad_norm": 3.171875,
      "learning_rate": 4.8970057063670026e-05,
      "loss": 0.877,
      "step": 6658
    },
    {
      "epoch": 0.09444430657529487,
      "grad_norm": 3.234375,
      "learning_rate": 4.896942227226535e-05,
      "loss": 0.8628,
      "step": 6660
    },
    {
      "epoch": 0.09447266822892109,
      "grad_norm": 3.140625,
      "learning_rate": 4.896878728941531e-05,
      "loss": 0.8393,
      "step": 6662
    },
    {
      "epoch": 0.0945010298825473,
      "grad_norm": 3.140625,
      "learning_rate": 4.896815211512496e-05,
      "loss": 0.8779,
      "step": 6664
    },
    {
      "epoch": 0.09452939153617351,
      "grad_norm": 2.9375,
      "learning_rate": 4.8967516749399375e-05,
      "loss": 0.8602,
      "step": 6666
    },
    {
      "epoch": 0.09455775318979973,
      "grad_norm": 3.015625,
      "learning_rate": 4.896688119224365e-05,
      "loss": 0.8643,
      "step": 6668
    },
    {
      "epoch": 0.09458611484342594,
      "grad_norm": 3.40625,
      "learning_rate": 4.896624544366284e-05,
      "loss": 0.9394,
      "step": 6670
    },
    {
      "epoch": 0.09461447649705217,
      "grad_norm": 2.890625,
      "learning_rate": 4.896560950366202e-05,
      "loss": 0.8657,
      "step": 6672
    },
    {
      "epoch": 0.09464283815067838,
      "grad_norm": 3.125,
      "learning_rate": 4.896497337224629e-05,
      "loss": 0.886,
      "step": 6674
    },
    {
      "epoch": 0.09467119980430459,
      "grad_norm": 2.8125,
      "learning_rate": 4.8964337049420716e-05,
      "loss": 0.8796,
      "step": 6676
    },
    {
      "epoch": 0.0946995614579308,
      "grad_norm": 3.46875,
      "learning_rate": 4.896370053519038e-05,
      "loss": 0.8908,
      "step": 6678
    },
    {
      "epoch": 0.09472792311155702,
      "grad_norm": 3.21875,
      "learning_rate": 4.8963063829560376e-05,
      "loss": 0.8701,
      "step": 6680
    },
    {
      "epoch": 0.09475628476518323,
      "grad_norm": 3.125,
      "learning_rate": 4.8962426932535775e-05,
      "loss": 0.8794,
      "step": 6682
    },
    {
      "epoch": 0.09478464641880945,
      "grad_norm": 3.34375,
      "learning_rate": 4.8961789844121674e-05,
      "loss": 0.8332,
      "step": 6684
    },
    {
      "epoch": 0.09481300807243566,
      "grad_norm": 3.015625,
      "learning_rate": 4.896115256432316e-05,
      "loss": 0.9035,
      "step": 6686
    },
    {
      "epoch": 0.09484136972606187,
      "grad_norm": 3.1875,
      "learning_rate": 4.896051509314532e-05,
      "loss": 0.9023,
      "step": 6688
    },
    {
      "epoch": 0.0948697313796881,
      "grad_norm": 2.78125,
      "learning_rate": 4.895987743059325e-05,
      "loss": 0.8909,
      "step": 6690
    },
    {
      "epoch": 0.09489809303331431,
      "grad_norm": 3.5625,
      "learning_rate": 4.895923957667204e-05,
      "loss": 0.8785,
      "step": 6692
    },
    {
      "epoch": 0.09492645468694053,
      "grad_norm": 2.859375,
      "learning_rate": 4.8958601531386796e-05,
      "loss": 0.8752,
      "step": 6694
    },
    {
      "epoch": 0.09495481634056674,
      "grad_norm": 2.984375,
      "learning_rate": 4.895796329474259e-05,
      "loss": 0.882,
      "step": 6696
    },
    {
      "epoch": 0.09498317799419295,
      "grad_norm": 3.046875,
      "learning_rate": 4.8957324866744534e-05,
      "loss": 0.8596,
      "step": 6698
    },
    {
      "epoch": 0.09501153964781917,
      "grad_norm": 3.40625,
      "learning_rate": 4.895668624739773e-05,
      "loss": 0.8922,
      "step": 6700
    },
    {
      "epoch": 0.09503990130144538,
      "grad_norm": 3.203125,
      "learning_rate": 4.8956047436707276e-05,
      "loss": 0.8653,
      "step": 6702
    },
    {
      "epoch": 0.09506826295507159,
      "grad_norm": 2.984375,
      "learning_rate": 4.895540843467827e-05,
      "loss": 0.9129,
      "step": 6704
    },
    {
      "epoch": 0.0950966246086978,
      "grad_norm": 3.265625,
      "learning_rate": 4.895476924131582e-05,
      "loss": 0.9321,
      "step": 6706
    },
    {
      "epoch": 0.09512498626232402,
      "grad_norm": 3.09375,
      "learning_rate": 4.895412985662504e-05,
      "loss": 0.8858,
      "step": 6708
    },
    {
      "epoch": 0.09515334791595025,
      "grad_norm": 3.28125,
      "learning_rate": 4.8953490280611015e-05,
      "loss": 0.8923,
      "step": 6710
    },
    {
      "epoch": 0.09518170956957646,
      "grad_norm": 3.546875,
      "learning_rate": 4.895285051327887e-05,
      "loss": 0.89,
      "step": 6712
    },
    {
      "epoch": 0.09521007122320267,
      "grad_norm": 3.140625,
      "learning_rate": 4.895221055463371e-05,
      "loss": 0.8784,
      "step": 6714
    },
    {
      "epoch": 0.09523843287682889,
      "grad_norm": 3.125,
      "learning_rate": 4.895157040468065e-05,
      "loss": 0.8888,
      "step": 6716
    },
    {
      "epoch": 0.0952667945304551,
      "grad_norm": 3.0,
      "learning_rate": 4.895093006342479e-05,
      "loss": 0.8994,
      "step": 6718
    },
    {
      "epoch": 0.09529515618408131,
      "grad_norm": 2.9375,
      "learning_rate": 4.895028953087126e-05,
      "loss": 0.8219,
      "step": 6720
    },
    {
      "epoch": 0.09532351783770752,
      "grad_norm": 2.96875,
      "learning_rate": 4.894964880702517e-05,
      "loss": 0.8751,
      "step": 6722
    },
    {
      "epoch": 0.09535187949133374,
      "grad_norm": 2.8125,
      "learning_rate": 4.894900789189164e-05,
      "loss": 0.8539,
      "step": 6724
    },
    {
      "epoch": 0.09538024114495995,
      "grad_norm": 3.234375,
      "learning_rate": 4.894836678547578e-05,
      "loss": 0.8832,
      "step": 6726
    },
    {
      "epoch": 0.09540860279858618,
      "grad_norm": 3.03125,
      "learning_rate": 4.894772548778272e-05,
      "loss": 0.8679,
      "step": 6728
    },
    {
      "epoch": 0.09543696445221239,
      "grad_norm": 2.953125,
      "learning_rate": 4.8947083998817576e-05,
      "loss": 0.8917,
      "step": 6730
    },
    {
      "epoch": 0.0954653261058386,
      "grad_norm": 3.25,
      "learning_rate": 4.894644231858548e-05,
      "loss": 0.8467,
      "step": 6732
    },
    {
      "epoch": 0.09549368775946482,
      "grad_norm": 3.0625,
      "learning_rate": 4.894580044709155e-05,
      "loss": 0.8977,
      "step": 6734
    },
    {
      "epoch": 0.09552204941309103,
      "grad_norm": 3.4375,
      "learning_rate": 4.8945158384340914e-05,
      "loss": 0.9222,
      "step": 6736
    },
    {
      "epoch": 0.09555041106671724,
      "grad_norm": 3.40625,
      "learning_rate": 4.89445161303387e-05,
      "loss": 0.8785,
      "step": 6738
    },
    {
      "epoch": 0.09557877272034346,
      "grad_norm": 3.109375,
      "learning_rate": 4.894387368509005e-05,
      "loss": 0.8621,
      "step": 6740
    },
    {
      "epoch": 0.09560713437396967,
      "grad_norm": 3.15625,
      "learning_rate": 4.894323104860007e-05,
      "loss": 0.8665,
      "step": 6742
    },
    {
      "epoch": 0.09563549602759588,
      "grad_norm": 3.078125,
      "learning_rate": 4.894258822087391e-05,
      "loss": 0.8774,
      "step": 6744
    },
    {
      "epoch": 0.0956638576812221,
      "grad_norm": 3.359375,
      "learning_rate": 4.8941945201916706e-05,
      "loss": 0.848,
      "step": 6746
    },
    {
      "epoch": 0.09569221933484832,
      "grad_norm": 3.171875,
      "learning_rate": 4.894130199173358e-05,
      "loss": 0.9119,
      "step": 6748
    },
    {
      "epoch": 0.09572058098847454,
      "grad_norm": 2.6875,
      "learning_rate": 4.894065859032969e-05,
      "loss": 0.8922,
      "step": 6750
    },
    {
      "epoch": 0.09574894264210075,
      "grad_norm": 3.796875,
      "learning_rate": 4.894001499771015e-05,
      "loss": 0.8957,
      "step": 6752
    },
    {
      "epoch": 0.09577730429572696,
      "grad_norm": 2.890625,
      "learning_rate": 4.893937121388012e-05,
      "loss": 0.8455,
      "step": 6754
    },
    {
      "epoch": 0.09580566594935318,
      "grad_norm": 3.0625,
      "learning_rate": 4.893872723884473e-05,
      "loss": 0.8426,
      "step": 6756
    },
    {
      "epoch": 0.09583402760297939,
      "grad_norm": 3.53125,
      "learning_rate": 4.893808307260914e-05,
      "loss": 0.8797,
      "step": 6758
    },
    {
      "epoch": 0.0958623892566056,
      "grad_norm": 3.0625,
      "learning_rate": 4.893743871517847e-05,
      "loss": 0.8905,
      "step": 6760
    },
    {
      "epoch": 0.09589075091023182,
      "grad_norm": 3.0,
      "learning_rate": 4.8936794166557895e-05,
      "loss": 0.8852,
      "step": 6762
    },
    {
      "epoch": 0.09591911256385803,
      "grad_norm": 3.265625,
      "learning_rate": 4.893614942675254e-05,
      "loss": 0.8762,
      "step": 6764
    },
    {
      "epoch": 0.09594747421748424,
      "grad_norm": 2.984375,
      "learning_rate": 4.893550449576756e-05,
      "loss": 0.8425,
      "step": 6766
    },
    {
      "epoch": 0.09597583587111047,
      "grad_norm": 3.078125,
      "learning_rate": 4.893485937360812e-05,
      "loss": 0.8618,
      "step": 6768
    },
    {
      "epoch": 0.09600419752473668,
      "grad_norm": 3.28125,
      "learning_rate": 4.893421406027935e-05,
      "loss": 0.9292,
      "step": 6770
    },
    {
      "epoch": 0.0960325591783629,
      "grad_norm": 3.359375,
      "learning_rate": 4.8933568555786416e-05,
      "loss": 0.8818,
      "step": 6772
    },
    {
      "epoch": 0.09606092083198911,
      "grad_norm": 2.921875,
      "learning_rate": 4.893292286013448e-05,
      "loss": 0.8698,
      "step": 6774
    },
    {
      "epoch": 0.09608928248561532,
      "grad_norm": 3.21875,
      "learning_rate": 4.89322769733287e-05,
      "loss": 0.9213,
      "step": 6776
    },
    {
      "epoch": 0.09611764413924154,
      "grad_norm": 2.921875,
      "learning_rate": 4.8931630895374215e-05,
      "loss": 0.8597,
      "step": 6778
    },
    {
      "epoch": 0.09614600579286775,
      "grad_norm": 3.125,
      "learning_rate": 4.89309846262762e-05,
      "loss": 0.8499,
      "step": 6780
    },
    {
      "epoch": 0.09617436744649396,
      "grad_norm": 3.609375,
      "learning_rate": 4.8930338166039815e-05,
      "loss": 0.8768,
      "step": 6782
    },
    {
      "epoch": 0.09620272910012018,
      "grad_norm": 2.921875,
      "learning_rate": 4.892969151467022e-05,
      "loss": 0.8947,
      "step": 6784
    },
    {
      "epoch": 0.0962310907537464,
      "grad_norm": 2.65625,
      "learning_rate": 4.892904467217258e-05,
      "loss": 0.8554,
      "step": 6786
    },
    {
      "epoch": 0.09625945240737262,
      "grad_norm": 3.375,
      "learning_rate": 4.892839763855208e-05,
      "loss": 0.8592,
      "step": 6788
    },
    {
      "epoch": 0.09628781406099883,
      "grad_norm": 3.078125,
      "learning_rate": 4.892775041381386e-05,
      "loss": 0.8741,
      "step": 6790
    },
    {
      "epoch": 0.09631617571462504,
      "grad_norm": 2.984375,
      "learning_rate": 4.8927102997963105e-05,
      "loss": 0.8321,
      "step": 6792
    },
    {
      "epoch": 0.09634453736825126,
      "grad_norm": 3.046875,
      "learning_rate": 4.8926455391004974e-05,
      "loss": 0.8295,
      "step": 6794
    },
    {
      "epoch": 0.09637289902187747,
      "grad_norm": 3.359375,
      "learning_rate": 4.892580759294466e-05,
      "loss": 0.9103,
      "step": 6796
    },
    {
      "epoch": 0.09640126067550368,
      "grad_norm": 3.296875,
      "learning_rate": 4.8925159603787316e-05,
      "loss": 0.8493,
      "step": 6798
    },
    {
      "epoch": 0.0964296223291299,
      "grad_norm": 3.3125,
      "learning_rate": 4.8924511423538136e-05,
      "loss": 0.8576,
      "step": 6800
    },
    {
      "epoch": 0.09645798398275611,
      "grad_norm": 2.71875,
      "learning_rate": 4.892386305220228e-05,
      "loss": 0.9026,
      "step": 6802
    },
    {
      "epoch": 0.09648634563638232,
      "grad_norm": 3.109375,
      "learning_rate": 4.892321448978494e-05,
      "loss": 0.8667,
      "step": 6804
    },
    {
      "epoch": 0.09651470729000855,
      "grad_norm": 3.078125,
      "learning_rate": 4.892256573629128e-05,
      "loss": 0.8576,
      "step": 6806
    },
    {
      "epoch": 0.09654306894363476,
      "grad_norm": 3.234375,
      "learning_rate": 4.89219167917265e-05,
      "loss": 0.8883,
      "step": 6808
    },
    {
      "epoch": 0.09657143059726098,
      "grad_norm": 2.921875,
      "learning_rate": 4.8921267656095775e-05,
      "loss": 0.8876,
      "step": 6810
    },
    {
      "epoch": 0.09659979225088719,
      "grad_norm": 3.3125,
      "learning_rate": 4.8920618329404286e-05,
      "loss": 0.9091,
      "step": 6812
    },
    {
      "epoch": 0.0966281539045134,
      "grad_norm": 3.1875,
      "learning_rate": 4.8919968811657224e-05,
      "loss": 0.893,
      "step": 6814
    },
    {
      "epoch": 0.09665651555813962,
      "grad_norm": 2.921875,
      "learning_rate": 4.891931910285978e-05,
      "loss": 0.8439,
      "step": 6816
    },
    {
      "epoch": 0.09668487721176583,
      "grad_norm": 2.984375,
      "learning_rate": 4.891866920301713e-05,
      "loss": 0.929,
      "step": 6818
    },
    {
      "epoch": 0.09671323886539204,
      "grad_norm": 2.625,
      "learning_rate": 4.8918019112134484e-05,
      "loss": 0.8322,
      "step": 6820
    },
    {
      "epoch": 0.09674160051901826,
      "grad_norm": 3.359375,
      "learning_rate": 4.891736883021701e-05,
      "loss": 0.8926,
      "step": 6822
    },
    {
      "epoch": 0.09676996217264447,
      "grad_norm": 2.9375,
      "learning_rate": 4.891671835726993e-05,
      "loss": 0.8846,
      "step": 6824
    },
    {
      "epoch": 0.0967983238262707,
      "grad_norm": 2.953125,
      "learning_rate": 4.8916067693298426e-05,
      "loss": 0.8818,
      "step": 6826
    },
    {
      "epoch": 0.09682668547989691,
      "grad_norm": 3.5,
      "learning_rate": 4.891541683830768e-05,
      "loss": 0.8972,
      "step": 6828
    },
    {
      "epoch": 0.09685504713352312,
      "grad_norm": 2.8125,
      "learning_rate": 4.891476579230292e-05,
      "loss": 0.8713,
      "step": 6830
    },
    {
      "epoch": 0.09688340878714934,
      "grad_norm": 3.484375,
      "learning_rate": 4.891411455528932e-05,
      "loss": 0.8244,
      "step": 6832
    },
    {
      "epoch": 0.09691177044077555,
      "grad_norm": 3.0,
      "learning_rate": 4.89134631272721e-05,
      "loss": 0.8848,
      "step": 6834
    },
    {
      "epoch": 0.09694013209440176,
      "grad_norm": 3.46875,
      "learning_rate": 4.891281150825644e-05,
      "loss": 0.871,
      "step": 6836
    },
    {
      "epoch": 0.09696849374802798,
      "grad_norm": 3.21875,
      "learning_rate": 4.8912159698247575e-05,
      "loss": 0.9193,
      "step": 6838
    },
    {
      "epoch": 0.09699685540165419,
      "grad_norm": 3.125,
      "learning_rate": 4.891150769725069e-05,
      "loss": 0.8347,
      "step": 6840
    },
    {
      "epoch": 0.0970252170552804,
      "grad_norm": 3.421875,
      "learning_rate": 4.8910855505271e-05,
      "loss": 0.8634,
      "step": 6842
    },
    {
      "epoch": 0.09705357870890663,
      "grad_norm": 2.84375,
      "learning_rate": 4.8910203122313705e-05,
      "loss": 0.8941,
      "step": 6844
    },
    {
      "epoch": 0.09708194036253284,
      "grad_norm": 3.109375,
      "learning_rate": 4.890955054838403e-05,
      "loss": 0.9007,
      "step": 6846
    },
    {
      "epoch": 0.09711030201615906,
      "grad_norm": 2.640625,
      "learning_rate": 4.890889778348718e-05,
      "loss": 0.8651,
      "step": 6848
    },
    {
      "epoch": 0.09713866366978527,
      "grad_norm": 3.078125,
      "learning_rate": 4.8908244827628365e-05,
      "loss": 0.7985,
      "step": 6850
    },
    {
      "epoch": 0.09716702532341148,
      "grad_norm": 2.875,
      "learning_rate": 4.89075916808128e-05,
      "loss": 0.8701,
      "step": 6852
    },
    {
      "epoch": 0.0971953869770377,
      "grad_norm": 3.046875,
      "learning_rate": 4.8906938343045716e-05,
      "loss": 0.859,
      "step": 6854
    },
    {
      "epoch": 0.09722374863066391,
      "grad_norm": 3.0625,
      "learning_rate": 4.890628481433232e-05,
      "loss": 0.8579,
      "step": 6856
    },
    {
      "epoch": 0.09725211028429012,
      "grad_norm": 3.421875,
      "learning_rate": 4.8905631094677826e-05,
      "loss": 0.8686,
      "step": 6858
    },
    {
      "epoch": 0.09728047193791634,
      "grad_norm": 3.21875,
      "learning_rate": 4.890497718408746e-05,
      "loss": 0.8887,
      "step": 6860
    },
    {
      "epoch": 0.09730883359154255,
      "grad_norm": 3.125,
      "learning_rate": 4.8904323082566456e-05,
      "loss": 0.8644,
      "step": 6862
    },
    {
      "epoch": 0.09733719524516878,
      "grad_norm": 3.296875,
      "learning_rate": 4.890366879012002e-05,
      "loss": 0.888,
      "step": 6864
    },
    {
      "epoch": 0.09736555689879499,
      "grad_norm": 3.453125,
      "learning_rate": 4.890301430675339e-05,
      "loss": 0.9147,
      "step": 6866
    },
    {
      "epoch": 0.0973939185524212,
      "grad_norm": 2.953125,
      "learning_rate": 4.890235963247179e-05,
      "loss": 0.8721,
      "step": 6868
    },
    {
      "epoch": 0.09742228020604742,
      "grad_norm": 2.859375,
      "learning_rate": 4.8901704767280456e-05,
      "loss": 0.8306,
      "step": 6870
    },
    {
      "epoch": 0.09745064185967363,
      "grad_norm": 3.4375,
      "learning_rate": 4.890104971118461e-05,
      "loss": 0.8914,
      "step": 6872
    },
    {
      "epoch": 0.09747900351329984,
      "grad_norm": 3.21875,
      "learning_rate": 4.890039446418948e-05,
      "loss": 0.9188,
      "step": 6874
    },
    {
      "epoch": 0.09750736516692605,
      "grad_norm": 3.09375,
      "learning_rate": 4.889973902630031e-05,
      "loss": 0.8805,
      "step": 6876
    },
    {
      "epoch": 0.09753572682055227,
      "grad_norm": 2.859375,
      "learning_rate": 4.889908339752233e-05,
      "loss": 0.8722,
      "step": 6878
    },
    {
      "epoch": 0.09756408847417848,
      "grad_norm": 2.921875,
      "learning_rate": 4.889842757786077e-05,
      "loss": 0.8739,
      "step": 6880
    },
    {
      "epoch": 0.09759245012780471,
      "grad_norm": 3.4375,
      "learning_rate": 4.889777156732088e-05,
      "loss": 0.8725,
      "step": 6882
    },
    {
      "epoch": 0.09762081178143092,
      "grad_norm": 3.1875,
      "learning_rate": 4.88971153659079e-05,
      "loss": 0.8994,
      "step": 6884
    },
    {
      "epoch": 0.09764917343505713,
      "grad_norm": 2.765625,
      "learning_rate": 4.8896458973627056e-05,
      "loss": 0.8805,
      "step": 6886
    },
    {
      "epoch": 0.09767753508868335,
      "grad_norm": 3.3125,
      "learning_rate": 4.8895802390483604e-05,
      "loss": 0.9145,
      "step": 6888
    },
    {
      "epoch": 0.09770589674230956,
      "grad_norm": 2.84375,
      "learning_rate": 4.8895145616482785e-05,
      "loss": 0.857,
      "step": 6890
    },
    {
      "epoch": 0.09773425839593577,
      "grad_norm": 3.046875,
      "learning_rate": 4.8894488651629844e-05,
      "loss": 0.8997,
      "step": 6892
    },
    {
      "epoch": 0.09776262004956199,
      "grad_norm": 3.328125,
      "learning_rate": 4.889383149593003e-05,
      "loss": 0.9074,
      "step": 6894
    },
    {
      "epoch": 0.0977909817031882,
      "grad_norm": 2.828125,
      "learning_rate": 4.889317414938859e-05,
      "loss": 0.8476,
      "step": 6896
    },
    {
      "epoch": 0.09781934335681441,
      "grad_norm": 2.890625,
      "learning_rate": 4.889251661201077e-05,
      "loss": 0.8836,
      "step": 6898
    },
    {
      "epoch": 0.09784770501044063,
      "grad_norm": 3.125,
      "learning_rate": 4.8891858883801825e-05,
      "loss": 0.8658,
      "step": 6900
    },
    {
      "epoch": 0.09787606666406685,
      "grad_norm": 3.125,
      "learning_rate": 4.8891200964767014e-05,
      "loss": 0.838,
      "step": 6902
    },
    {
      "epoch": 0.09790442831769307,
      "grad_norm": 3.34375,
      "learning_rate": 4.889054285491159e-05,
      "loss": 0.869,
      "step": 6904
    },
    {
      "epoch": 0.09793278997131928,
      "grad_norm": 3.390625,
      "learning_rate": 4.888988455424081e-05,
      "loss": 0.9054,
      "step": 6906
    },
    {
      "epoch": 0.0979611516249455,
      "grad_norm": 2.984375,
      "learning_rate": 4.8889226062759914e-05,
      "loss": 0.8587,
      "step": 6908
    },
    {
      "epoch": 0.09798951327857171,
      "grad_norm": 3.125,
      "learning_rate": 4.8888567380474184e-05,
      "loss": 0.8719,
      "step": 6910
    },
    {
      "epoch": 0.09801787493219792,
      "grad_norm": 3.3125,
      "learning_rate": 4.888790850738887e-05,
      "loss": 0.8956,
      "step": 6912
    },
    {
      "epoch": 0.09804623658582413,
      "grad_norm": 3.15625,
      "learning_rate": 4.888724944350924e-05,
      "loss": 0.8679,
      "step": 6914
    },
    {
      "epoch": 0.09807459823945035,
      "grad_norm": 3.125,
      "learning_rate": 4.888659018884056e-05,
      "loss": 0.8953,
      "step": 6916
    },
    {
      "epoch": 0.09810295989307656,
      "grad_norm": 3.078125,
      "learning_rate": 4.8885930743388086e-05,
      "loss": 0.8775,
      "step": 6918
    },
    {
      "epoch": 0.09813132154670277,
      "grad_norm": 3.046875,
      "learning_rate": 4.888527110715709e-05,
      "loss": 0.9056,
      "step": 6920
    },
    {
      "epoch": 0.098159683200329,
      "grad_norm": 3.015625,
      "learning_rate": 4.888461128015283e-05,
      "loss": 0.8887,
      "step": 6922
    },
    {
      "epoch": 0.09818804485395521,
      "grad_norm": 2.90625,
      "learning_rate": 4.888395126238061e-05,
      "loss": 0.8972,
      "step": 6924
    },
    {
      "epoch": 0.09821640650758143,
      "grad_norm": 2.78125,
      "learning_rate": 4.888329105384566e-05,
      "loss": 0.8811,
      "step": 6926
    },
    {
      "epoch": 0.09824476816120764,
      "grad_norm": 3.546875,
      "learning_rate": 4.888263065455328e-05,
      "loss": 0.8866,
      "step": 6928
    },
    {
      "epoch": 0.09827312981483385,
      "grad_norm": 3.109375,
      "learning_rate": 4.888197006450873e-05,
      "loss": 0.8863,
      "step": 6930
    },
    {
      "epoch": 0.09830149146846007,
      "grad_norm": 3.0,
      "learning_rate": 4.8881309283717305e-05,
      "loss": 0.8646,
      "step": 6932
    },
    {
      "epoch": 0.09832985312208628,
      "grad_norm": 3.0,
      "learning_rate": 4.888064831218425e-05,
      "loss": 0.872,
      "step": 6934
    },
    {
      "epoch": 0.0983582147757125,
      "grad_norm": 2.90625,
      "learning_rate": 4.887998714991488e-05,
      "loss": 0.8798,
      "step": 6936
    },
    {
      "epoch": 0.0983865764293387,
      "grad_norm": 2.921875,
      "learning_rate": 4.8879325796914455e-05,
      "loss": 0.9518,
      "step": 6938
    },
    {
      "epoch": 0.09841493808296493,
      "grad_norm": 2.984375,
      "learning_rate": 4.887866425318826e-05,
      "loss": 0.89,
      "step": 6940
    },
    {
      "epoch": 0.09844329973659115,
      "grad_norm": 2.703125,
      "learning_rate": 4.8878002518741585e-05,
      "loss": 0.8011,
      "step": 6942
    },
    {
      "epoch": 0.09847166139021736,
      "grad_norm": 3.484375,
      "learning_rate": 4.88773405935797e-05,
      "loss": 0.8383,
      "step": 6944
    },
    {
      "epoch": 0.09850002304384357,
      "grad_norm": 2.984375,
      "learning_rate": 4.887667847770791e-05,
      "loss": 0.8918,
      "step": 6946
    },
    {
      "epoch": 0.09852838469746979,
      "grad_norm": 3.21875,
      "learning_rate": 4.88760161711315e-05,
      "loss": 0.8785,
      "step": 6948
    },
    {
      "epoch": 0.098556746351096,
      "grad_norm": 3.28125,
      "learning_rate": 4.887535367385576e-05,
      "loss": 0.8639,
      "step": 6950
    },
    {
      "epoch": 0.09858510800472221,
      "grad_norm": 3.15625,
      "learning_rate": 4.8874690985885975e-05,
      "loss": 0.8672,
      "step": 6952
    },
    {
      "epoch": 0.09861346965834843,
      "grad_norm": 2.90625,
      "learning_rate": 4.887402810722743e-05,
      "loss": 0.8679,
      "step": 6954
    },
    {
      "epoch": 0.09864183131197464,
      "grad_norm": 2.890625,
      "learning_rate": 4.887336503788544e-05,
      "loss": 0.9038,
      "step": 6956
    },
    {
      "epoch": 0.09867019296560085,
      "grad_norm": 3.09375,
      "learning_rate": 4.887270177786529e-05,
      "loss": 0.8563,
      "step": 6958
    },
    {
      "epoch": 0.09869855461922708,
      "grad_norm": 3.234375,
      "learning_rate": 4.8872038327172276e-05,
      "loss": 0.9246,
      "step": 6960
    },
    {
      "epoch": 0.0987269162728533,
      "grad_norm": 3.109375,
      "learning_rate": 4.88713746858117e-05,
      "loss": 0.93,
      "step": 6962
    },
    {
      "epoch": 0.0987552779264795,
      "grad_norm": 3.0,
      "learning_rate": 4.887071085378886e-05,
      "loss": 0.8827,
      "step": 6964
    },
    {
      "epoch": 0.09878363958010572,
      "grad_norm": 3.046875,
      "learning_rate": 4.8870046831109064e-05,
      "loss": 0.8984,
      "step": 6966
    },
    {
      "epoch": 0.09881200123373193,
      "grad_norm": 3.28125,
      "learning_rate": 4.886938261777761e-05,
      "loss": 0.8696,
      "step": 6968
    },
    {
      "epoch": 0.09884036288735815,
      "grad_norm": 3.109375,
      "learning_rate": 4.886871821379981e-05,
      "loss": 0.8766,
      "step": 6970
    },
    {
      "epoch": 0.09886872454098436,
      "grad_norm": 3.15625,
      "learning_rate": 4.886805361918096e-05,
      "loss": 0.8598,
      "step": 6972
    },
    {
      "epoch": 0.09889708619461057,
      "grad_norm": 2.875,
      "learning_rate": 4.886738883392637e-05,
      "loss": 0.8779,
      "step": 6974
    },
    {
      "epoch": 0.09892544784823679,
      "grad_norm": 2.984375,
      "learning_rate": 4.886672385804136e-05,
      "loss": 0.892,
      "step": 6976
    },
    {
      "epoch": 0.098953809501863,
      "grad_norm": 3.296875,
      "learning_rate": 4.886605869153124e-05,
      "loss": 0.8992,
      "step": 6978
    },
    {
      "epoch": 0.09898217115548923,
      "grad_norm": 3.265625,
      "learning_rate": 4.886539333440131e-05,
      "loss": 0.9366,
      "step": 6980
    },
    {
      "epoch": 0.09901053280911544,
      "grad_norm": 3.4375,
      "learning_rate": 4.8864727786656886e-05,
      "loss": 0.8638,
      "step": 6982
    },
    {
      "epoch": 0.09903889446274165,
      "grad_norm": 3.421875,
      "learning_rate": 4.88640620483033e-05,
      "loss": 0.8521,
      "step": 6984
    },
    {
      "epoch": 0.09906725611636787,
      "grad_norm": 3.078125,
      "learning_rate": 4.886339611934585e-05,
      "loss": 0.8615,
      "step": 6986
    },
    {
      "epoch": 0.09909561776999408,
      "grad_norm": 2.984375,
      "learning_rate": 4.886272999978987e-05,
      "loss": 0.9117,
      "step": 6988
    },
    {
      "epoch": 0.09912397942362029,
      "grad_norm": 2.71875,
      "learning_rate": 4.8862063689640666e-05,
      "loss": 0.8302,
      "step": 6990
    },
    {
      "epoch": 0.0991523410772465,
      "grad_norm": 3.0625,
      "learning_rate": 4.886139718890358e-05,
      "loss": 0.8786,
      "step": 6992
    },
    {
      "epoch": 0.09918070273087272,
      "grad_norm": 3.140625,
      "learning_rate": 4.8860730497583914e-05,
      "loss": 0.8645,
      "step": 6994
    },
    {
      "epoch": 0.09920906438449893,
      "grad_norm": 3.015625,
      "learning_rate": 4.8860063615687e-05,
      "loss": 0.9151,
      "step": 6996
    },
    {
      "epoch": 0.09923742603812516,
      "grad_norm": 3.328125,
      "learning_rate": 4.885939654321817e-05,
      "loss": 0.8901,
      "step": 6998
    },
    {
      "epoch": 0.09926578769175137,
      "grad_norm": 3.21875,
      "learning_rate": 4.885872928018276e-05,
      "loss": 0.8513,
      "step": 7000
    },
    {
      "epoch": 0.09929414934537759,
      "grad_norm": 3.171875,
      "learning_rate": 4.885806182658607e-05,
      "loss": 0.9091,
      "step": 7002
    },
    {
      "epoch": 0.0993225109990038,
      "grad_norm": 2.96875,
      "learning_rate": 4.8857394182433456e-05,
      "loss": 0.8872,
      "step": 7004
    },
    {
      "epoch": 0.09935087265263001,
      "grad_norm": 2.984375,
      "learning_rate": 4.885672634773024e-05,
      "loss": 0.8998,
      "step": 7006
    },
    {
      "epoch": 0.09937923430625623,
      "grad_norm": 3.328125,
      "learning_rate": 4.885605832248177e-05,
      "loss": 0.8582,
      "step": 7008
    },
    {
      "epoch": 0.09940759595988244,
      "grad_norm": 2.71875,
      "learning_rate": 4.885539010669336e-05,
      "loss": 0.8353,
      "step": 7010
    },
    {
      "epoch": 0.09943595761350865,
      "grad_norm": 3.0625,
      "learning_rate": 4.885472170037037e-05,
      "loss": 0.9026,
      "step": 7012
    },
    {
      "epoch": 0.09946431926713487,
      "grad_norm": 3.046875,
      "learning_rate": 4.885405310351811e-05,
      "loss": 0.8705,
      "step": 7014
    },
    {
      "epoch": 0.09949268092076108,
      "grad_norm": 3.21875,
      "learning_rate": 4.885338431614195e-05,
      "loss": 0.8947,
      "step": 7016
    },
    {
      "epoch": 0.0995210425743873,
      "grad_norm": 2.953125,
      "learning_rate": 4.8852715338247214e-05,
      "loss": 0.8672,
      "step": 7018
    },
    {
      "epoch": 0.09954940422801352,
      "grad_norm": 3.328125,
      "learning_rate": 4.8852046169839245e-05,
      "loss": 0.9253,
      "step": 7020
    },
    {
      "epoch": 0.09957776588163973,
      "grad_norm": 3.296875,
      "learning_rate": 4.8851376810923396e-05,
      "loss": 0.8568,
      "step": 7022
    },
    {
      "epoch": 0.09960612753526595,
      "grad_norm": 3.0,
      "learning_rate": 4.885070726150501e-05,
      "loss": 0.8611,
      "step": 7024
    },
    {
      "epoch": 0.09963448918889216,
      "grad_norm": 3.1875,
      "learning_rate": 4.8850037521589434e-05,
      "loss": 0.8688,
      "step": 7026
    },
    {
      "epoch": 0.09966285084251837,
      "grad_norm": 2.65625,
      "learning_rate": 4.8849367591182014e-05,
      "loss": 0.8865,
      "step": 7028
    },
    {
      "epoch": 0.09969121249614459,
      "grad_norm": 3.125,
      "learning_rate": 4.8848697470288105e-05,
      "loss": 0.8672,
      "step": 7030
    },
    {
      "epoch": 0.0997195741497708,
      "grad_norm": 3.421875,
      "learning_rate": 4.884802715891307e-05,
      "loss": 0.874,
      "step": 7032
    },
    {
      "epoch": 0.09974793580339701,
      "grad_norm": 2.921875,
      "learning_rate": 4.8847356657062236e-05,
      "loss": 0.9212,
      "step": 7034
    },
    {
      "epoch": 0.09977629745702324,
      "grad_norm": 3.15625,
      "learning_rate": 4.884668596474098e-05,
      "loss": 0.8985,
      "step": 7036
    },
    {
      "epoch": 0.09980465911064945,
      "grad_norm": 3.46875,
      "learning_rate": 4.884601508195465e-05,
      "loss": 0.8831,
      "step": 7038
    },
    {
      "epoch": 0.09983302076427567,
      "grad_norm": 3.359375,
      "learning_rate": 4.8845344008708615e-05,
      "loss": 0.9153,
      "step": 7040
    },
    {
      "epoch": 0.09986138241790188,
      "grad_norm": 3.03125,
      "learning_rate": 4.884467274500822e-05,
      "loss": 0.9009,
      "step": 7042
    },
    {
      "epoch": 0.09988974407152809,
      "grad_norm": 3.28125,
      "learning_rate": 4.884400129085884e-05,
      "loss": 0.8834,
      "step": 7044
    },
    {
      "epoch": 0.0999181057251543,
      "grad_norm": 2.859375,
      "learning_rate": 4.884332964626582e-05,
      "loss": 0.8377,
      "step": 7046
    },
    {
      "epoch": 0.09994646737878052,
      "grad_norm": 2.984375,
      "learning_rate": 4.884265781123455e-05,
      "loss": 0.8779,
      "step": 7048
    },
    {
      "epoch": 0.09997482903240673,
      "grad_norm": 2.890625,
      "learning_rate": 4.884198578577037e-05,
      "loss": 0.9041,
      "step": 7050
    },
    {
      "epoch": 0.10000319068603294,
      "grad_norm": 3.046875,
      "learning_rate": 4.884131356987867e-05,
      "loss": 0.8658,
      "step": 7052
    },
    {
      "epoch": 0.10003155233965916,
      "grad_norm": 3.15625,
      "learning_rate": 4.88406411635648e-05,
      "loss": 0.9014,
      "step": 7054
    },
    {
      "epoch": 0.10005991399328538,
      "grad_norm": 2.78125,
      "learning_rate": 4.8839968566834136e-05,
      "loss": 0.8374,
      "step": 7056
    },
    {
      "epoch": 0.1000882756469116,
      "grad_norm": 3.015625,
      "learning_rate": 4.8839295779692064e-05,
      "loss": 0.8786,
      "step": 7058
    },
    {
      "epoch": 0.10011663730053781,
      "grad_norm": 3.453125,
      "learning_rate": 4.883862280214393e-05,
      "loss": 0.9005,
      "step": 7060
    },
    {
      "epoch": 0.10014499895416402,
      "grad_norm": 2.90625,
      "learning_rate": 4.883794963419514e-05,
      "loss": 0.8782,
      "step": 7062
    },
    {
      "epoch": 0.10017336060779024,
      "grad_norm": 3.125,
      "learning_rate": 4.883727627585105e-05,
      "loss": 0.8752,
      "step": 7064
    },
    {
      "epoch": 0.10020172226141645,
      "grad_norm": 3.140625,
      "learning_rate": 4.8836602727117054e-05,
      "loss": 0.8277,
      "step": 7066
    },
    {
      "epoch": 0.10023008391504266,
      "grad_norm": 3.0,
      "learning_rate": 4.8835928987998516e-05,
      "loss": 0.8267,
      "step": 7068
    },
    {
      "epoch": 0.10025844556866888,
      "grad_norm": 2.9375,
      "learning_rate": 4.883525505850082e-05,
      "loss": 0.8731,
      "step": 7070
    },
    {
      "epoch": 0.10028680722229509,
      "grad_norm": 3.15625,
      "learning_rate": 4.883458093862935e-05,
      "loss": 0.8688,
      "step": 7072
    },
    {
      "epoch": 0.1003151688759213,
      "grad_norm": 3.078125,
      "learning_rate": 4.8833906628389495e-05,
      "loss": 0.8729,
      "step": 7074
    },
    {
      "epoch": 0.10034353052954753,
      "grad_norm": 2.796875,
      "learning_rate": 4.8833232127786636e-05,
      "loss": 0.8916,
      "step": 7076
    },
    {
      "epoch": 0.10037189218317374,
      "grad_norm": 3.34375,
      "learning_rate": 4.8832557436826166e-05,
      "loss": 0.8974,
      "step": 7078
    },
    {
      "epoch": 0.10040025383679996,
      "grad_norm": 2.9375,
      "learning_rate": 4.8831882555513473e-05,
      "loss": 0.8785,
      "step": 7080
    },
    {
      "epoch": 0.10042861549042617,
      "grad_norm": 3.078125,
      "learning_rate": 4.883120748385394e-05,
      "loss": 0.8784,
      "step": 7082
    },
    {
      "epoch": 0.10045697714405238,
      "grad_norm": 3.15625,
      "learning_rate": 4.8830532221852966e-05,
      "loss": 0.88,
      "step": 7084
    },
    {
      "epoch": 0.1004853387976786,
      "grad_norm": 2.859375,
      "learning_rate": 4.882985676951594e-05,
      "loss": 0.8852,
      "step": 7086
    },
    {
      "epoch": 0.10051370045130481,
      "grad_norm": 3.34375,
      "learning_rate": 4.882918112684826e-05,
      "loss": 0.8429,
      "step": 7088
    },
    {
      "epoch": 0.10054206210493102,
      "grad_norm": 3.328125,
      "learning_rate": 4.8828505293855315e-05,
      "loss": 0.9146,
      "step": 7090
    },
    {
      "epoch": 0.10057042375855724,
      "grad_norm": 2.890625,
      "learning_rate": 4.882782927054251e-05,
      "loss": 0.8439,
      "step": 7092
    },
    {
      "epoch": 0.10059878541218346,
      "grad_norm": 2.578125,
      "learning_rate": 4.882715305691525e-05,
      "loss": 0.8956,
      "step": 7094
    },
    {
      "epoch": 0.10062714706580968,
      "grad_norm": 2.890625,
      "learning_rate": 4.8826476652978924e-05,
      "loss": 0.8756,
      "step": 7096
    },
    {
      "epoch": 0.10065550871943589,
      "grad_norm": 3.34375,
      "learning_rate": 4.882580005873894e-05,
      "loss": 0.8739,
      "step": 7098
    },
    {
      "epoch": 0.1006838703730621,
      "grad_norm": 3.046875,
      "learning_rate": 4.8825123274200703e-05,
      "loss": 0.8709,
      "step": 7100
    },
    {
      "epoch": 0.10071223202668832,
      "grad_norm": 2.921875,
      "learning_rate": 4.882444629936962e-05,
      "loss": 0.886,
      "step": 7102
    },
    {
      "epoch": 0.10074059368031453,
      "grad_norm": 3.453125,
      "learning_rate": 4.882376913425109e-05,
      "loss": 0.9081,
      "step": 7104
    },
    {
      "epoch": 0.10076895533394074,
      "grad_norm": 3.4375,
      "learning_rate": 4.882309177885053e-05,
      "loss": 0.9085,
      "step": 7106
    },
    {
      "epoch": 0.10079731698756696,
      "grad_norm": 3.328125,
      "learning_rate": 4.8822414233173344e-05,
      "loss": 0.8605,
      "step": 7108
    },
    {
      "epoch": 0.10082567864119317,
      "grad_norm": 2.734375,
      "learning_rate": 4.8821736497224954e-05,
      "loss": 0.895,
      "step": 7110
    },
    {
      "epoch": 0.10085404029481938,
      "grad_norm": 3.015625,
      "learning_rate": 4.882105857101076e-05,
      "loss": 0.8375,
      "step": 7112
    },
    {
      "epoch": 0.10088240194844561,
      "grad_norm": 3.0,
      "learning_rate": 4.8820380454536185e-05,
      "loss": 0.8829,
      "step": 7114
    },
    {
      "epoch": 0.10091076360207182,
      "grad_norm": 2.953125,
      "learning_rate": 4.8819702147806634e-05,
      "loss": 0.884,
      "step": 7116
    },
    {
      "epoch": 0.10093912525569804,
      "grad_norm": 3.078125,
      "learning_rate": 4.881902365082755e-05,
      "loss": 0.886,
      "step": 7118
    },
    {
      "epoch": 0.10096748690932425,
      "grad_norm": 2.984375,
      "learning_rate": 4.881834496360432e-05,
      "loss": 0.9413,
      "step": 7120
    },
    {
      "epoch": 0.10099584856295046,
      "grad_norm": 3.15625,
      "learning_rate": 4.881766608614238e-05,
      "loss": 0.8656,
      "step": 7122
    },
    {
      "epoch": 0.10102421021657668,
      "grad_norm": 3.28125,
      "learning_rate": 4.881698701844716e-05,
      "loss": 0.7929,
      "step": 7124
    },
    {
      "epoch": 0.10105257187020289,
      "grad_norm": 3.015625,
      "learning_rate": 4.881630776052407e-05,
      "loss": 0.8803,
      "step": 7126
    },
    {
      "epoch": 0.1010809335238291,
      "grad_norm": 3.09375,
      "learning_rate": 4.8815628312378545e-05,
      "loss": 0.8243,
      "step": 7128
    },
    {
      "epoch": 0.10110929517745532,
      "grad_norm": 2.875,
      "learning_rate": 4.8814948674016016e-05,
      "loss": 0.8313,
      "step": 7130
    },
    {
      "epoch": 0.10113765683108153,
      "grad_norm": 3.53125,
      "learning_rate": 4.881426884544189e-05,
      "loss": 0.8632,
      "step": 7132
    },
    {
      "epoch": 0.10116601848470776,
      "grad_norm": 3.0,
      "learning_rate": 4.881358882666162e-05,
      "loss": 0.8474,
      "step": 7134
    },
    {
      "epoch": 0.10119438013833397,
      "grad_norm": 3.484375,
      "learning_rate": 4.881290861768062e-05,
      "loss": 0.8663,
      "step": 7136
    },
    {
      "epoch": 0.10122274179196018,
      "grad_norm": 2.875,
      "learning_rate": 4.881222821850433e-05,
      "loss": 0.8217,
      "step": 7138
    },
    {
      "epoch": 0.1012511034455864,
      "grad_norm": 3.328125,
      "learning_rate": 4.881154762913819e-05,
      "loss": 0.8989,
      "step": 7140
    },
    {
      "epoch": 0.10127946509921261,
      "grad_norm": 2.8125,
      "learning_rate": 4.881086684958763e-05,
      "loss": 0.8452,
      "step": 7142
    },
    {
      "epoch": 0.10130782675283882,
      "grad_norm": 3.46875,
      "learning_rate": 4.881018587985809e-05,
      "loss": 0.8775,
      "step": 7144
    },
    {
      "epoch": 0.10133618840646504,
      "grad_norm": 3.3125,
      "learning_rate": 4.8809504719955e-05,
      "loss": 0.9285,
      "step": 7146
    },
    {
      "epoch": 0.10136455006009125,
      "grad_norm": 3.515625,
      "learning_rate": 4.880882336988381e-05,
      "loss": 0.9253,
      "step": 7148
    },
    {
      "epoch": 0.10139291171371746,
      "grad_norm": 3.46875,
      "learning_rate": 4.8808141829649964e-05,
      "loss": 0.9214,
      "step": 7150
    },
    {
      "epoch": 0.10142127336734369,
      "grad_norm": 2.90625,
      "learning_rate": 4.8807460099258906e-05,
      "loss": 0.8635,
      "step": 7152
    },
    {
      "epoch": 0.1014496350209699,
      "grad_norm": 2.78125,
      "learning_rate": 4.8806778178716066e-05,
      "loss": 0.8674,
      "step": 7154
    },
    {
      "epoch": 0.10147799667459612,
      "grad_norm": 3.21875,
      "learning_rate": 4.8806096068026906e-05,
      "loss": 0.8859,
      "step": 7156
    },
    {
      "epoch": 0.10150635832822233,
      "grad_norm": 2.90625,
      "learning_rate": 4.8805413767196874e-05,
      "loss": 0.8697,
      "step": 7158
    },
    {
      "epoch": 0.10153471998184854,
      "grad_norm": 2.96875,
      "learning_rate": 4.8804731276231405e-05,
      "loss": 0.8871,
      "step": 7160
    },
    {
      "epoch": 0.10156308163547476,
      "grad_norm": 3.046875,
      "learning_rate": 4.880404859513596e-05,
      "loss": 0.8387,
      "step": 7162
    },
    {
      "epoch": 0.10159144328910097,
      "grad_norm": 2.765625,
      "learning_rate": 4.8803365723915995e-05,
      "loss": 0.8323,
      "step": 7164
    },
    {
      "epoch": 0.10161980494272718,
      "grad_norm": 3.234375,
      "learning_rate": 4.880268266257696e-05,
      "loss": 0.8585,
      "step": 7166
    },
    {
      "epoch": 0.1016481665963534,
      "grad_norm": 3.109375,
      "learning_rate": 4.880199941112431e-05,
      "loss": 0.9004,
      "step": 7168
    },
    {
      "epoch": 0.10167652824997961,
      "grad_norm": 3.40625,
      "learning_rate": 4.88013159695635e-05,
      "loss": 0.9089,
      "step": 7170
    },
    {
      "epoch": 0.10170488990360584,
      "grad_norm": 3.21875,
      "learning_rate": 4.880063233789999e-05,
      "loss": 0.8965,
      "step": 7172
    },
    {
      "epoch": 0.10173325155723205,
      "grad_norm": 2.671875,
      "learning_rate": 4.879994851613925e-05,
      "loss": 0.8207,
      "step": 7174
    },
    {
      "epoch": 0.10176161321085826,
      "grad_norm": 3.03125,
      "learning_rate": 4.879926450428674e-05,
      "loss": 0.8805,
      "step": 7176
    },
    {
      "epoch": 0.10178997486448448,
      "grad_norm": 3.09375,
      "learning_rate": 4.87985803023479e-05,
      "loss": 0.8897,
      "step": 7178
    },
    {
      "epoch": 0.10181833651811069,
      "grad_norm": 3.15625,
      "learning_rate": 4.879789591032822e-05,
      "loss": 0.8809,
      "step": 7180
    },
    {
      "epoch": 0.1018466981717369,
      "grad_norm": 3.21875,
      "learning_rate": 4.879721132823315e-05,
      "loss": 0.865,
      "step": 7182
    },
    {
      "epoch": 0.10187505982536312,
      "grad_norm": 3.125,
      "learning_rate": 4.8796526556068176e-05,
      "loss": 0.8321,
      "step": 7184
    },
    {
      "epoch": 0.10190342147898933,
      "grad_norm": 2.8125,
      "learning_rate": 4.8795841593838757e-05,
      "loss": 0.8595,
      "step": 7186
    },
    {
      "epoch": 0.10193178313261554,
      "grad_norm": 2.9375,
      "learning_rate": 4.8795156441550364e-05,
      "loss": 0.8677,
      "step": 7188
    },
    {
      "epoch": 0.10196014478624177,
      "grad_norm": 3.0,
      "learning_rate": 4.879447109920846e-05,
      "loss": 0.8886,
      "step": 7190
    },
    {
      "epoch": 0.10198850643986798,
      "grad_norm": 3.125,
      "learning_rate": 4.879378556681854e-05,
      "loss": 0.883,
      "step": 7192
    },
    {
      "epoch": 0.1020168680934942,
      "grad_norm": 3.5625,
      "learning_rate": 4.879309984438606e-05,
      "loss": 0.9499,
      "step": 7194
    },
    {
      "epoch": 0.10204522974712041,
      "grad_norm": 3.328125,
      "learning_rate": 4.8792413931916504e-05,
      "loss": 0.8033,
      "step": 7196
    },
    {
      "epoch": 0.10207359140074662,
      "grad_norm": 2.84375,
      "learning_rate": 4.8791727829415356e-05,
      "loss": 0.8815,
      "step": 7198
    },
    {
      "epoch": 0.10210195305437283,
      "grad_norm": 3.109375,
      "learning_rate": 4.8791041536888086e-05,
      "loss": 0.9328,
      "step": 7200
    },
    {
      "epoch": 0.10213031470799905,
      "grad_norm": 3.09375,
      "learning_rate": 4.879035505434019e-05,
      "loss": 0.9427,
      "step": 7202
    },
    {
      "epoch": 0.10215867636162526,
      "grad_norm": 3.140625,
      "learning_rate": 4.8789668381777133e-05,
      "loss": 0.9012,
      "step": 7204
    },
    {
      "epoch": 0.10218703801525147,
      "grad_norm": 3.390625,
      "learning_rate": 4.87889815192044e-05,
      "loss": 0.847,
      "step": 7206
    },
    {
      "epoch": 0.10221539966887769,
      "grad_norm": 3.203125,
      "learning_rate": 4.87882944666275e-05,
      "loss": 0.9231,
      "step": 7208
    },
    {
      "epoch": 0.10224376132250391,
      "grad_norm": 2.828125,
      "learning_rate": 4.87876072240519e-05,
      "loss": 0.8774,
      "step": 7210
    },
    {
      "epoch": 0.10227212297613013,
      "grad_norm": 2.59375,
      "learning_rate": 4.8786919791483094e-05,
      "loss": 0.8415,
      "step": 7212
    },
    {
      "epoch": 0.10230048462975634,
      "grad_norm": 3.3125,
      "learning_rate": 4.878623216892657e-05,
      "loss": 0.8605,
      "step": 7214
    },
    {
      "epoch": 0.10232884628338255,
      "grad_norm": 3.546875,
      "learning_rate": 4.878554435638783e-05,
      "loss": 0.9108,
      "step": 7216
    },
    {
      "epoch": 0.10235720793700877,
      "grad_norm": 3.25,
      "learning_rate": 4.8784856353872355e-05,
      "loss": 0.9039,
      "step": 7218
    },
    {
      "epoch": 0.10238556959063498,
      "grad_norm": 2.96875,
      "learning_rate": 4.878416816138565e-05,
      "loss": 0.8559,
      "step": 7220
    },
    {
      "epoch": 0.1024139312442612,
      "grad_norm": 2.546875,
      "learning_rate": 4.8783479778933207e-05,
      "loss": 0.8971,
      "step": 7222
    },
    {
      "epoch": 0.10244229289788741,
      "grad_norm": 2.890625,
      "learning_rate": 4.8782791206520516e-05,
      "loss": 0.8577,
      "step": 7224
    },
    {
      "epoch": 0.10247065455151362,
      "grad_norm": 3.34375,
      "learning_rate": 4.87821024441531e-05,
      "loss": 0.8918,
      "step": 7226
    },
    {
      "epoch": 0.10249901620513983,
      "grad_norm": 3.34375,
      "learning_rate": 4.878141349183643e-05,
      "loss": 0.9011,
      "step": 7228
    },
    {
      "epoch": 0.10252737785876606,
      "grad_norm": 3.171875,
      "learning_rate": 4.878072434957604e-05,
      "loss": 0.8378,
      "step": 7230
    },
    {
      "epoch": 0.10255573951239227,
      "grad_norm": 2.890625,
      "learning_rate": 4.87800350173774e-05,
      "loss": 0.885,
      "step": 7232
    },
    {
      "epoch": 0.10258410116601849,
      "grad_norm": 3.28125,
      "learning_rate": 4.877934549524604e-05,
      "loss": 0.863,
      "step": 7234
    },
    {
      "epoch": 0.1026124628196447,
      "grad_norm": 3.5,
      "learning_rate": 4.877865578318747e-05,
      "loss": 0.8909,
      "step": 7236
    },
    {
      "epoch": 0.10264082447327091,
      "grad_norm": 2.96875,
      "learning_rate": 4.8777965881207187e-05,
      "loss": 0.8651,
      "step": 7238
    },
    {
      "epoch": 0.10266918612689713,
      "grad_norm": 2.90625,
      "learning_rate": 4.8777275789310704e-05,
      "loss": 0.8706,
      "step": 7240
    },
    {
      "epoch": 0.10269754778052334,
      "grad_norm": 2.90625,
      "learning_rate": 4.877658550750353e-05,
      "loss": 0.8697,
      "step": 7242
    },
    {
      "epoch": 0.10272590943414955,
      "grad_norm": 2.921875,
      "learning_rate": 4.877589503579119e-05,
      "loss": 0.8967,
      "step": 7244
    },
    {
      "epoch": 0.10275427108777577,
      "grad_norm": 3.1875,
      "learning_rate": 4.877520437417919e-05,
      "loss": 0.8963,
      "step": 7246
    },
    {
      "epoch": 0.102782632741402,
      "grad_norm": 3.328125,
      "learning_rate": 4.877451352267304e-05,
      "loss": 0.8984,
      "step": 7248
    },
    {
      "epoch": 0.10281099439502821,
      "grad_norm": 3.21875,
      "learning_rate": 4.877382248127826e-05,
      "loss": 0.8345,
      "step": 7250
    },
    {
      "epoch": 0.10283935604865442,
      "grad_norm": 2.640625,
      "learning_rate": 4.877313125000038e-05,
      "loss": 0.8885,
      "step": 7252
    },
    {
      "epoch": 0.10286771770228063,
      "grad_norm": 3.296875,
      "learning_rate": 4.877243982884492e-05,
      "loss": 0.875,
      "step": 7254
    },
    {
      "epoch": 0.10289607935590685,
      "grad_norm": 3.015625,
      "learning_rate": 4.877174821781739e-05,
      "loss": 0.8792,
      "step": 7256
    },
    {
      "epoch": 0.10292444100953306,
      "grad_norm": 3.015625,
      "learning_rate": 4.877105641692332e-05,
      "loss": 0.9297,
      "step": 7258
    },
    {
      "epoch": 0.10295280266315927,
      "grad_norm": 2.9375,
      "learning_rate": 4.877036442616825e-05,
      "loss": 0.889,
      "step": 7260
    },
    {
      "epoch": 0.10298116431678549,
      "grad_norm": 3.21875,
      "learning_rate": 4.876967224555768e-05,
      "loss": 0.8604,
      "step": 7262
    },
    {
      "epoch": 0.1030095259704117,
      "grad_norm": 3.125,
      "learning_rate": 4.876897987509716e-05,
      "loss": 0.8782,
      "step": 7264
    },
    {
      "epoch": 0.10303788762403791,
      "grad_norm": 3.203125,
      "learning_rate": 4.87682873147922e-05,
      "loss": 0.8677,
      "step": 7266
    },
    {
      "epoch": 0.10306624927766414,
      "grad_norm": 2.9375,
      "learning_rate": 4.876759456464836e-05,
      "loss": 0.845,
      "step": 7268
    },
    {
      "epoch": 0.10309461093129035,
      "grad_norm": 2.921875,
      "learning_rate": 4.876690162467115e-05,
      "loss": 0.8587,
      "step": 7270
    },
    {
      "epoch": 0.10312297258491657,
      "grad_norm": 3.34375,
      "learning_rate": 4.8766208494866114e-05,
      "loss": 0.8591,
      "step": 7272
    },
    {
      "epoch": 0.10315133423854278,
      "grad_norm": 3.390625,
      "learning_rate": 4.8765515175238784e-05,
      "loss": 0.8937,
      "step": 7274
    },
    {
      "epoch": 0.103179695892169,
      "grad_norm": 3.3125,
      "learning_rate": 4.8764821665794705e-05,
      "loss": 0.9215,
      "step": 7276
    },
    {
      "epoch": 0.1032080575457952,
      "grad_norm": 2.84375,
      "learning_rate": 4.87641279665394e-05,
      "loss": 0.9414,
      "step": 7278
    },
    {
      "epoch": 0.10323641919942142,
      "grad_norm": 3.140625,
      "learning_rate": 4.876343407747842e-05,
      "loss": 0.8638,
      "step": 7280
    },
    {
      "epoch": 0.10326478085304763,
      "grad_norm": 3.078125,
      "learning_rate": 4.876273999861731e-05,
      "loss": 0.8491,
      "step": 7282
    },
    {
      "epoch": 0.10329314250667385,
      "grad_norm": 2.9375,
      "learning_rate": 4.876204572996161e-05,
      "loss": 0.905,
      "step": 7284
    },
    {
      "epoch": 0.10332150416030006,
      "grad_norm": 3.0,
      "learning_rate": 4.876135127151687e-05,
      "loss": 0.8716,
      "step": 7286
    },
    {
      "epoch": 0.10334986581392629,
      "grad_norm": 3.09375,
      "learning_rate": 4.876065662328863e-05,
      "loss": 0.9147,
      "step": 7288
    },
    {
      "epoch": 0.1033782274675525,
      "grad_norm": 3.03125,
      "learning_rate": 4.875996178528244e-05,
      "loss": 0.8869,
      "step": 7290
    },
    {
      "epoch": 0.10340658912117871,
      "grad_norm": 2.90625,
      "learning_rate": 4.8759266757503855e-05,
      "loss": 0.899,
      "step": 7292
    },
    {
      "epoch": 0.10343495077480493,
      "grad_norm": 3.171875,
      "learning_rate": 4.875857153995841e-05,
      "loss": 0.8916,
      "step": 7294
    },
    {
      "epoch": 0.10346331242843114,
      "grad_norm": 3.390625,
      "learning_rate": 4.875787613265168e-05,
      "loss": 0.8721,
      "step": 7296
    },
    {
      "epoch": 0.10349167408205735,
      "grad_norm": 3.09375,
      "learning_rate": 4.8757180535589205e-05,
      "loss": 0.8522,
      "step": 7298
    },
    {
      "epoch": 0.10352003573568357,
      "grad_norm": 3.125,
      "learning_rate": 4.8756484748776545e-05,
      "loss": 0.8732,
      "step": 7300
    },
    {
      "epoch": 0.10354839738930978,
      "grad_norm": 2.828125,
      "learning_rate": 4.8755788772219256e-05,
      "loss": 0.8753,
      "step": 7302
    },
    {
      "epoch": 0.10357675904293599,
      "grad_norm": 3.046875,
      "learning_rate": 4.87550926059229e-05,
      "loss": 0.9476,
      "step": 7304
    },
    {
      "epoch": 0.10360512069656222,
      "grad_norm": 3.375,
      "learning_rate": 4.875439624989303e-05,
      "loss": 0.9234,
      "step": 7306
    },
    {
      "epoch": 0.10363348235018843,
      "grad_norm": 3.359375,
      "learning_rate": 4.8753699704135214e-05,
      "loss": 0.9191,
      "step": 7308
    },
    {
      "epoch": 0.10366184400381465,
      "grad_norm": 3.140625,
      "learning_rate": 4.875300296865502e-05,
      "loss": 0.9309,
      "step": 7310
    },
    {
      "epoch": 0.10369020565744086,
      "grad_norm": 2.8125,
      "learning_rate": 4.8752306043458e-05,
      "loss": 0.8689,
      "step": 7312
    },
    {
      "epoch": 0.10371856731106707,
      "grad_norm": 3.125,
      "learning_rate": 4.875160892854973e-05,
      "loss": 0.8908,
      "step": 7314
    },
    {
      "epoch": 0.10374692896469329,
      "grad_norm": 3.328125,
      "learning_rate": 4.875091162393578e-05,
      "loss": 0.9103,
      "step": 7316
    },
    {
      "epoch": 0.1037752906183195,
      "grad_norm": 2.953125,
      "learning_rate": 4.875021412962171e-05,
      "loss": 0.8483,
      "step": 7318
    },
    {
      "epoch": 0.10380365227194571,
      "grad_norm": 3.296875,
      "learning_rate": 4.87495164456131e-05,
      "loss": 0.8228,
      "step": 7320
    },
    {
      "epoch": 0.10383201392557193,
      "grad_norm": 3.25,
      "learning_rate": 4.874881857191551e-05,
      "loss": 0.9272,
      "step": 7322
    },
    {
      "epoch": 0.10386037557919814,
      "grad_norm": 3.4375,
      "learning_rate": 4.874812050853452e-05,
      "loss": 0.861,
      "step": 7324
    },
    {
      "epoch": 0.10388873723282437,
      "grad_norm": 2.734375,
      "learning_rate": 4.8747422255475715e-05,
      "loss": 0.8491,
      "step": 7326
    },
    {
      "epoch": 0.10391709888645058,
      "grad_norm": 3.65625,
      "learning_rate": 4.874672381274467e-05,
      "loss": 0.8729,
      "step": 7328
    },
    {
      "epoch": 0.10394546054007679,
      "grad_norm": 3.0625,
      "learning_rate": 4.874602518034694e-05,
      "loss": 0.8043,
      "step": 7330
    },
    {
      "epoch": 0.103973822193703,
      "grad_norm": 3.109375,
      "learning_rate": 4.8745326358288133e-05,
      "loss": 0.8663,
      "step": 7332
    },
    {
      "epoch": 0.10400218384732922,
      "grad_norm": 3.3125,
      "learning_rate": 4.8744627346573825e-05,
      "loss": 0.8564,
      "step": 7334
    },
    {
      "epoch": 0.10403054550095543,
      "grad_norm": 3.21875,
      "learning_rate": 4.874392814520959e-05,
      "loss": 0.8679,
      "step": 7336
    },
    {
      "epoch": 0.10405890715458165,
      "grad_norm": 2.84375,
      "learning_rate": 4.874322875420102e-05,
      "loss": 0.8373,
      "step": 7338
    },
    {
      "epoch": 0.10408726880820786,
      "grad_norm": 2.921875,
      "learning_rate": 4.87425291735537e-05,
      "loss": 0.8581,
      "step": 7340
    },
    {
      "epoch": 0.10411563046183407,
      "grad_norm": 3.484375,
      "learning_rate": 4.874182940327321e-05,
      "loss": 0.9144,
      "step": 7342
    },
    {
      "epoch": 0.1041439921154603,
      "grad_norm": 3.609375,
      "learning_rate": 4.874112944336514e-05,
      "loss": 0.8847,
      "step": 7344
    },
    {
      "epoch": 0.10417235376908651,
      "grad_norm": 3.109375,
      "learning_rate": 4.87404292938351e-05,
      "loss": 0.926,
      "step": 7346
    },
    {
      "epoch": 0.10420071542271273,
      "grad_norm": 3.328125,
      "learning_rate": 4.873972895468866e-05,
      "loss": 0.9011,
      "step": 7348
    },
    {
      "epoch": 0.10422907707633894,
      "grad_norm": 2.90625,
      "learning_rate": 4.8739028425931424e-05,
      "loss": 0.8559,
      "step": 7350
    },
    {
      "epoch": 0.10425743872996515,
      "grad_norm": 3.03125,
      "learning_rate": 4.8738327707568974e-05,
      "loss": 0.9049,
      "step": 7352
    },
    {
      "epoch": 0.10428580038359136,
      "grad_norm": 2.8125,
      "learning_rate": 4.8737626799606926e-05,
      "loss": 0.8314,
      "step": 7354
    },
    {
      "epoch": 0.10431416203721758,
      "grad_norm": 3.265625,
      "learning_rate": 4.873692570205087e-05,
      "loss": 0.9264,
      "step": 7356
    },
    {
      "epoch": 0.10434252369084379,
      "grad_norm": 3.03125,
      "learning_rate": 4.8736224414906406e-05,
      "loss": 0.845,
      "step": 7358
    },
    {
      "epoch": 0.10437088534447,
      "grad_norm": 3.0625,
      "learning_rate": 4.8735522938179135e-05,
      "loss": 0.8742,
      "step": 7360
    },
    {
      "epoch": 0.10439924699809622,
      "grad_norm": 3.546875,
      "learning_rate": 4.8734821271874656e-05,
      "loss": 0.927,
      "step": 7362
    },
    {
      "epoch": 0.10442760865172244,
      "grad_norm": 2.90625,
      "learning_rate": 4.873411941599857e-05,
      "loss": 0.8471,
      "step": 7364
    },
    {
      "epoch": 0.10445597030534866,
      "grad_norm": 3.359375,
      "learning_rate": 4.87334173705565e-05,
      "loss": 0.8937,
      "step": 7366
    },
    {
      "epoch": 0.10448433195897487,
      "grad_norm": 3.03125,
      "learning_rate": 4.873271513555404e-05,
      "loss": 0.9161,
      "step": 7368
    },
    {
      "epoch": 0.10451269361260108,
      "grad_norm": 3.1875,
      "learning_rate": 4.87320127109968e-05,
      "loss": 0.8386,
      "step": 7370
    },
    {
      "epoch": 0.1045410552662273,
      "grad_norm": 3.1875,
      "learning_rate": 4.873131009689039e-05,
      "loss": 0.8389,
      "step": 7372
    },
    {
      "epoch": 0.10456941691985351,
      "grad_norm": 3.21875,
      "learning_rate": 4.8730607293240424e-05,
      "loss": 0.9325,
      "step": 7374
    },
    {
      "epoch": 0.10459777857347972,
      "grad_norm": 3.046875,
      "learning_rate": 4.872990430005252e-05,
      "loss": 0.905,
      "step": 7376
    },
    {
      "epoch": 0.10462614022710594,
      "grad_norm": 3.0,
      "learning_rate": 4.872920111733228e-05,
      "loss": 0.895,
      "step": 7378
    },
    {
      "epoch": 0.10465450188073215,
      "grad_norm": 3.140625,
      "learning_rate": 4.8728497745085334e-05,
      "loss": 0.9069,
      "step": 7380
    },
    {
      "epoch": 0.10468286353435836,
      "grad_norm": 3.03125,
      "learning_rate": 4.872779418331729e-05,
      "loss": 0.8949,
      "step": 7382
    },
    {
      "epoch": 0.10471122518798459,
      "grad_norm": 3.0,
      "learning_rate": 4.872709043203377e-05,
      "loss": 0.9123,
      "step": 7384
    },
    {
      "epoch": 0.1047395868416108,
      "grad_norm": 3.359375,
      "learning_rate": 4.872638649124039e-05,
      "loss": 0.8278,
      "step": 7386
    },
    {
      "epoch": 0.10476794849523702,
      "grad_norm": 3.046875,
      "learning_rate": 4.872568236094279e-05,
      "loss": 0.8543,
      "step": 7388
    },
    {
      "epoch": 0.10479631014886323,
      "grad_norm": 3.171875,
      "learning_rate": 4.872497804114658e-05,
      "loss": 0.8938,
      "step": 7390
    },
    {
      "epoch": 0.10482467180248944,
      "grad_norm": 3.28125,
      "learning_rate": 4.872427353185739e-05,
      "loss": 0.9027,
      "step": 7392
    },
    {
      "epoch": 0.10485303345611566,
      "grad_norm": 2.96875,
      "learning_rate": 4.872356883308084e-05,
      "loss": 0.8652,
      "step": 7394
    },
    {
      "epoch": 0.10488139510974187,
      "grad_norm": 2.984375,
      "learning_rate": 4.872286394482257e-05,
      "loss": 0.8498,
      "step": 7396
    },
    {
      "epoch": 0.10490975676336808,
      "grad_norm": 3.265625,
      "learning_rate": 4.87221588670882e-05,
      "loss": 0.9081,
      "step": 7398
    },
    {
      "epoch": 0.1049381184169943,
      "grad_norm": 3.453125,
      "learning_rate": 4.872145359988336e-05,
      "loss": 0.8792,
      "step": 7400
    },
    {
      "epoch": 0.10496648007062052,
      "grad_norm": 2.984375,
      "learning_rate": 4.872074814321369e-05,
      "loss": 0.871,
      "step": 7402
    },
    {
      "epoch": 0.10499484172424674,
      "grad_norm": 3.03125,
      "learning_rate": 4.8720042497084825e-05,
      "loss": 0.8327,
      "step": 7404
    },
    {
      "epoch": 0.10502320337787295,
      "grad_norm": 3.015625,
      "learning_rate": 4.871933666150239e-05,
      "loss": 0.8459,
      "step": 7406
    },
    {
      "epoch": 0.10505156503149916,
      "grad_norm": 2.796875,
      "learning_rate": 4.871863063647204e-05,
      "loss": 0.8134,
      "step": 7408
    },
    {
      "epoch": 0.10507992668512538,
      "grad_norm": 3.1875,
      "learning_rate": 4.8717924421999406e-05,
      "loss": 0.8849,
      "step": 7410
    },
    {
      "epoch": 0.10510828833875159,
      "grad_norm": 3.4375,
      "learning_rate": 4.871721801809013e-05,
      "loss": 0.8259,
      "step": 7412
    },
    {
      "epoch": 0.1051366499923778,
      "grad_norm": 3.296875,
      "learning_rate": 4.8716511424749845e-05,
      "loss": 0.9026,
      "step": 7414
    },
    {
      "epoch": 0.10516501164600402,
      "grad_norm": 3.09375,
      "learning_rate": 4.87158046419842e-05,
      "loss": 0.8755,
      "step": 7416
    },
    {
      "epoch": 0.10519337329963023,
      "grad_norm": 2.859375,
      "learning_rate": 4.871509766979884e-05,
      "loss": 0.8278,
      "step": 7418
    },
    {
      "epoch": 0.10522173495325644,
      "grad_norm": 3.390625,
      "learning_rate": 4.871439050819942e-05,
      "loss": 0.8896,
      "step": 7420
    },
    {
      "epoch": 0.10525009660688267,
      "grad_norm": 2.921875,
      "learning_rate": 4.871368315719158e-05,
      "loss": 0.8555,
      "step": 7422
    },
    {
      "epoch": 0.10527845826050888,
      "grad_norm": 2.859375,
      "learning_rate": 4.871297561678096e-05,
      "loss": 0.8728,
      "step": 7424
    },
    {
      "epoch": 0.1053068199141351,
      "grad_norm": 3.03125,
      "learning_rate": 4.871226788697324e-05,
      "loss": 0.881,
      "step": 7426
    },
    {
      "epoch": 0.10533518156776131,
      "grad_norm": 3.421875,
      "learning_rate": 4.871155996777404e-05,
      "loss": 0.8798,
      "step": 7428
    },
    {
      "epoch": 0.10536354322138752,
      "grad_norm": 2.875,
      "learning_rate": 4.871085185918904e-05,
      "loss": 0.7989,
      "step": 7430
    },
    {
      "epoch": 0.10539190487501374,
      "grad_norm": 2.71875,
      "learning_rate": 4.871014356122388e-05,
      "loss": 0.8545,
      "step": 7432
    },
    {
      "epoch": 0.10542026652863995,
      "grad_norm": 3.40625,
      "learning_rate": 4.8709435073884225e-05,
      "loss": 0.8969,
      "step": 7434
    },
    {
      "epoch": 0.10544862818226616,
      "grad_norm": 2.921875,
      "learning_rate": 4.870872639717572e-05,
      "loss": 0.8759,
      "step": 7436
    },
    {
      "epoch": 0.10547698983589238,
      "grad_norm": 3.03125,
      "learning_rate": 4.8708017531104046e-05,
      "loss": 0.8897,
      "step": 7438
    },
    {
      "epoch": 0.10550535148951859,
      "grad_norm": 2.75,
      "learning_rate": 4.8707308475674854e-05,
      "loss": 0.8757,
      "step": 7440
    },
    {
      "epoch": 0.10553371314314482,
      "grad_norm": 3.125,
      "learning_rate": 4.8706599230893805e-05,
      "loss": 0.9025,
      "step": 7442
    },
    {
      "epoch": 0.10556207479677103,
      "grad_norm": 3.359375,
      "learning_rate": 4.870588979676657e-05,
      "loss": 0.8409,
      "step": 7444
    },
    {
      "epoch": 0.10559043645039724,
      "grad_norm": 3.015625,
      "learning_rate": 4.870518017329881e-05,
      "loss": 0.8367,
      "step": 7446
    },
    {
      "epoch": 0.10561879810402346,
      "grad_norm": 3.34375,
      "learning_rate": 4.870447036049619e-05,
      "loss": 0.9288,
      "step": 7448
    },
    {
      "epoch": 0.10564715975764967,
      "grad_norm": 3.140625,
      "learning_rate": 4.870376035836439e-05,
      "loss": 0.8653,
      "step": 7450
    },
    {
      "epoch": 0.10567552141127588,
      "grad_norm": 2.96875,
      "learning_rate": 4.870305016690908e-05,
      "loss": 0.896,
      "step": 7452
    },
    {
      "epoch": 0.1057038830649021,
      "grad_norm": 2.859375,
      "learning_rate": 4.8702339786135916e-05,
      "loss": 0.8739,
      "step": 7454
    },
    {
      "epoch": 0.10573224471852831,
      "grad_norm": 3.234375,
      "learning_rate": 4.870162921605059e-05,
      "loss": 0.9137,
      "step": 7456
    },
    {
      "epoch": 0.10576060637215452,
      "grad_norm": 3.125,
      "learning_rate": 4.870091845665876e-05,
      "loss": 0.8627,
      "step": 7458
    },
    {
      "epoch": 0.10578896802578075,
      "grad_norm": 3.078125,
      "learning_rate": 4.870020750796613e-05,
      "loss": 0.861,
      "step": 7460
    },
    {
      "epoch": 0.10581732967940696,
      "grad_norm": 2.859375,
      "learning_rate": 4.8699496369978346e-05,
      "loss": 0.8924,
      "step": 7462
    },
    {
      "epoch": 0.10584569133303318,
      "grad_norm": 3.203125,
      "learning_rate": 4.8698785042701114e-05,
      "loss": 0.9369,
      "step": 7464
    },
    {
      "epoch": 0.10587405298665939,
      "grad_norm": 2.8125,
      "learning_rate": 4.8698073526140106e-05,
      "loss": 0.8787,
      "step": 7466
    },
    {
      "epoch": 0.1059024146402856,
      "grad_norm": 2.828125,
      "learning_rate": 4.8697361820301e-05,
      "loss": 0.8653,
      "step": 7468
    },
    {
      "epoch": 0.10593077629391182,
      "grad_norm": 3.234375,
      "learning_rate": 4.869664992518949e-05,
      "loss": 0.8889,
      "step": 7470
    },
    {
      "epoch": 0.10595913794753803,
      "grad_norm": 3.03125,
      "learning_rate": 4.869593784081124e-05,
      "loss": 0.8551,
      "step": 7472
    },
    {
      "epoch": 0.10598749960116424,
      "grad_norm": 2.953125,
      "learning_rate": 4.869522556717198e-05,
      "loss": 0.8701,
      "step": 7474
    },
    {
      "epoch": 0.10601586125479046,
      "grad_norm": 3.15625,
      "learning_rate": 4.8694513104277354e-05,
      "loss": 0.8334,
      "step": 7476
    },
    {
      "epoch": 0.10604422290841667,
      "grad_norm": 3.078125,
      "learning_rate": 4.869380045213307e-05,
      "loss": 0.8744,
      "step": 7478
    },
    {
      "epoch": 0.1060725845620429,
      "grad_norm": 2.796875,
      "learning_rate": 4.8693087610744834e-05,
      "loss": 0.8219,
      "step": 7480
    },
    {
      "epoch": 0.10610094621566911,
      "grad_norm": 3.53125,
      "learning_rate": 4.8692374580118314e-05,
      "loss": 0.8934,
      "step": 7482
    },
    {
      "epoch": 0.10612930786929532,
      "grad_norm": 3.296875,
      "learning_rate": 4.8691661360259226e-05,
      "loss": 0.8704,
      "step": 7484
    },
    {
      "epoch": 0.10615766952292154,
      "grad_norm": 2.90625,
      "learning_rate": 4.869094795117326e-05,
      "loss": 0.8408,
      "step": 7486
    },
    {
      "epoch": 0.10618603117654775,
      "grad_norm": 3.046875,
      "learning_rate": 4.86902343528661e-05,
      "loss": 0.8397,
      "step": 7488
    },
    {
      "epoch": 0.10621439283017396,
      "grad_norm": 3.1875,
      "learning_rate": 4.868952056534347e-05,
      "loss": 0.8923,
      "step": 7490
    },
    {
      "epoch": 0.10624275448380018,
      "grad_norm": 3.1875,
      "learning_rate": 4.868880658861106e-05,
      "loss": 0.8336,
      "step": 7492
    },
    {
      "epoch": 0.10627111613742639,
      "grad_norm": 3.203125,
      "learning_rate": 4.868809242267457e-05,
      "loss": 0.9178,
      "step": 7494
    },
    {
      "epoch": 0.1062994777910526,
      "grad_norm": 3.21875,
      "learning_rate": 4.86873780675397e-05,
      "loss": 0.8791,
      "step": 7496
    },
    {
      "epoch": 0.10632783944467883,
      "grad_norm": 3.4375,
      "learning_rate": 4.8686663523212165e-05,
      "loss": 0.9042,
      "step": 7498
    },
    {
      "epoch": 0.10635620109830504,
      "grad_norm": 3.078125,
      "learning_rate": 4.8685948789697665e-05,
      "loss": 0.9165,
      "step": 7500
    },
    {
      "epoch": 0.10638456275193126,
      "grad_norm": 2.75,
      "learning_rate": 4.868523386700192e-05,
      "loss": 0.867,
      "step": 7502
    },
    {
      "epoch": 0.10641292440555747,
      "grad_norm": 3.484375,
      "learning_rate": 4.8684518755130625e-05,
      "loss": 0.8907,
      "step": 7504
    },
    {
      "epoch": 0.10644128605918368,
      "grad_norm": 3.453125,
      "learning_rate": 4.86838034540895e-05,
      "loss": 0.819,
      "step": 7506
    },
    {
      "epoch": 0.1064696477128099,
      "grad_norm": 3.21875,
      "learning_rate": 4.8683087963884265e-05,
      "loss": 0.9127,
      "step": 7508
    },
    {
      "epoch": 0.10649800936643611,
      "grad_norm": 2.859375,
      "learning_rate": 4.868237228452062e-05,
      "loss": 0.8602,
      "step": 7510
    },
    {
      "epoch": 0.10652637102006232,
      "grad_norm": 2.90625,
      "learning_rate": 4.868165641600429e-05,
      "loss": 0.8292,
      "step": 7512
    },
    {
      "epoch": 0.10655473267368853,
      "grad_norm": 2.828125,
      "learning_rate": 4.868094035834099e-05,
      "loss": 0.814,
      "step": 7514
    },
    {
      "epoch": 0.10658309432731475,
      "grad_norm": 2.703125,
      "learning_rate": 4.8680224111536446e-05,
      "loss": 0.8579,
      "step": 7516
    },
    {
      "epoch": 0.10661145598094098,
      "grad_norm": 2.890625,
      "learning_rate": 4.867950767559636e-05,
      "loss": 0.8953,
      "step": 7518
    },
    {
      "epoch": 0.10663981763456719,
      "grad_norm": 3.03125,
      "learning_rate": 4.867879105052647e-05,
      "loss": 0.8894,
      "step": 7520
    },
    {
      "epoch": 0.1066681792881934,
      "grad_norm": 2.9375,
      "learning_rate": 4.8678074236332505e-05,
      "loss": 0.9014,
      "step": 7522
    },
    {
      "epoch": 0.10669654094181961,
      "grad_norm": 2.9375,
      "learning_rate": 4.867735723302017e-05,
      "loss": 0.8258,
      "step": 7524
    },
    {
      "epoch": 0.10672490259544583,
      "grad_norm": 3.203125,
      "learning_rate": 4.867664004059521e-05,
      "loss": 0.8979,
      "step": 7526
    },
    {
      "epoch": 0.10675326424907204,
      "grad_norm": 3.53125,
      "learning_rate": 4.8675922659063346e-05,
      "loss": 0.8396,
      "step": 7528
    },
    {
      "epoch": 0.10678162590269825,
      "grad_norm": 3.359375,
      "learning_rate": 4.867520508843031e-05,
      "loss": 0.8861,
      "step": 7530
    },
    {
      "epoch": 0.10680998755632447,
      "grad_norm": 2.90625,
      "learning_rate": 4.8674487328701836e-05,
      "loss": 0.8807,
      "step": 7532
    },
    {
      "epoch": 0.10683834920995068,
      "grad_norm": 3.203125,
      "learning_rate": 4.8673769379883635e-05,
      "loss": 0.8782,
      "step": 7534
    },
    {
      "epoch": 0.1068667108635769,
      "grad_norm": 2.953125,
      "learning_rate": 4.8673051241981476e-05,
      "loss": 0.9107,
      "step": 7536
    },
    {
      "epoch": 0.10689507251720312,
      "grad_norm": 2.875,
      "learning_rate": 4.867233291500107e-05,
      "loss": 0.82,
      "step": 7538
    },
    {
      "epoch": 0.10692343417082933,
      "grad_norm": 3.109375,
      "learning_rate": 4.867161439894817e-05,
      "loss": 0.8931,
      "step": 7540
    },
    {
      "epoch": 0.10695179582445555,
      "grad_norm": 3.21875,
      "learning_rate": 4.86708956938285e-05,
      "loss": 0.9229,
      "step": 7542
    },
    {
      "epoch": 0.10698015747808176,
      "grad_norm": 2.78125,
      "learning_rate": 4.867017679964781e-05,
      "loss": 0.8501,
      "step": 7544
    },
    {
      "epoch": 0.10700851913170797,
      "grad_norm": 3.28125,
      "learning_rate": 4.8669457716411835e-05,
      "loss": 0.8319,
      "step": 7546
    },
    {
      "epoch": 0.10703688078533419,
      "grad_norm": 3.109375,
      "learning_rate": 4.8668738444126324e-05,
      "loss": 0.9763,
      "step": 7548
    },
    {
      "epoch": 0.1070652424389604,
      "grad_norm": 2.953125,
      "learning_rate": 4.8668018982797026e-05,
      "loss": 0.9042,
      "step": 7550
    },
    {
      "epoch": 0.10709360409258661,
      "grad_norm": 2.71875,
      "learning_rate": 4.866729933242968e-05,
      "loss": 0.8283,
      "step": 7552
    },
    {
      "epoch": 0.10712196574621283,
      "grad_norm": 3.4375,
      "learning_rate": 4.866657949303003e-05,
      "loss": 0.8516,
      "step": 7554
    },
    {
      "epoch": 0.10715032739983905,
      "grad_norm": 3.078125,
      "learning_rate": 4.866585946460384e-05,
      "loss": 0.862,
      "step": 7556
    },
    {
      "epoch": 0.10717868905346527,
      "grad_norm": 2.90625,
      "learning_rate": 4.8665139247156844e-05,
      "loss": 0.8642,
      "step": 7558
    },
    {
      "epoch": 0.10720705070709148,
      "grad_norm": 2.953125,
      "learning_rate": 4.86644188406948e-05,
      "loss": 0.8646,
      "step": 7560
    },
    {
      "epoch": 0.1072354123607177,
      "grad_norm": 3.046875,
      "learning_rate": 4.8663698245223466e-05,
      "loss": 0.819,
      "step": 7562
    },
    {
      "epoch": 0.10726377401434391,
      "grad_norm": 3.109375,
      "learning_rate": 4.86629774607486e-05,
      "loss": 0.8754,
      "step": 7564
    },
    {
      "epoch": 0.10729213566797012,
      "grad_norm": 3.0625,
      "learning_rate": 4.866225648727595e-05,
      "loss": 0.9102,
      "step": 7566
    },
    {
      "epoch": 0.10732049732159633,
      "grad_norm": 3.15625,
      "learning_rate": 4.866153532481129e-05,
      "loss": 0.8569,
      "step": 7568
    },
    {
      "epoch": 0.10734885897522255,
      "grad_norm": 2.984375,
      "learning_rate": 4.866081397336035e-05,
      "loss": 0.8626,
      "step": 7570
    },
    {
      "epoch": 0.10737722062884876,
      "grad_norm": 3.140625,
      "learning_rate": 4.866009243292893e-05,
      "loss": 0.873,
      "step": 7572
    },
    {
      "epoch": 0.10740558228247497,
      "grad_norm": 2.921875,
      "learning_rate": 4.8659370703522766e-05,
      "loss": 0.8573,
      "step": 7574
    },
    {
      "epoch": 0.1074339439361012,
      "grad_norm": 3.359375,
      "learning_rate": 4.865864878514763e-05,
      "loss": 0.9001,
      "step": 7576
    },
    {
      "epoch": 0.10746230558972741,
      "grad_norm": 3.0625,
      "learning_rate": 4.865792667780928e-05,
      "loss": 0.8549,
      "step": 7578
    },
    {
      "epoch": 0.10749066724335363,
      "grad_norm": 2.6875,
      "learning_rate": 4.86572043815135e-05,
      "loss": 0.8648,
      "step": 7580
    },
    {
      "epoch": 0.10751902889697984,
      "grad_norm": 2.921875,
      "learning_rate": 4.865648189626605e-05,
      "loss": 0.9075,
      "step": 7582
    },
    {
      "epoch": 0.10754739055060605,
      "grad_norm": 2.609375,
      "learning_rate": 4.86557592220727e-05,
      "loss": 0.8659,
      "step": 7584
    },
    {
      "epoch": 0.10757575220423227,
      "grad_norm": 2.8125,
      "learning_rate": 4.865503635893923e-05,
      "loss": 0.8087,
      "step": 7586
    },
    {
      "epoch": 0.10760411385785848,
      "grad_norm": 3.109375,
      "learning_rate": 4.86543133068714e-05,
      "loss": 0.899,
      "step": 7588
    },
    {
      "epoch": 0.1076324755114847,
      "grad_norm": 3.453125,
      "learning_rate": 4.8653590065874996e-05,
      "loss": 0.8587,
      "step": 7590
    },
    {
      "epoch": 0.1076608371651109,
      "grad_norm": 2.828125,
      "learning_rate": 4.865286663595578e-05,
      "loss": 0.8757,
      "step": 7592
    },
    {
      "epoch": 0.10768919881873712,
      "grad_norm": 3.203125,
      "learning_rate": 4.865214301711956e-05,
      "loss": 0.9114,
      "step": 7594
    },
    {
      "epoch": 0.10771756047236335,
      "grad_norm": 3.25,
      "learning_rate": 4.8651419209372076e-05,
      "loss": 0.8567,
      "step": 7596
    },
    {
      "epoch": 0.10774592212598956,
      "grad_norm": 3.015625,
      "learning_rate": 4.8650695212719136e-05,
      "loss": 0.8875,
      "step": 7598
    },
    {
      "epoch": 0.10777428377961577,
      "grad_norm": 3.109375,
      "learning_rate": 4.864997102716652e-05,
      "loss": 0.8102,
      "step": 7600
    },
    {
      "epoch": 0.10780264543324199,
      "grad_norm": 3.21875,
      "learning_rate": 4.8649246652720005e-05,
      "loss": 0.8942,
      "step": 7602
    },
    {
      "epoch": 0.1078310070868682,
      "grad_norm": 3.90625,
      "learning_rate": 4.864852208938538e-05,
      "loss": 0.8921,
      "step": 7604
    },
    {
      "epoch": 0.10785936874049441,
      "grad_norm": 2.890625,
      "learning_rate": 4.8647797337168435e-05,
      "loss": 0.8501,
      "step": 7606
    },
    {
      "epoch": 0.10788773039412063,
      "grad_norm": 2.765625,
      "learning_rate": 4.864707239607495e-05,
      "loss": 0.8539,
      "step": 7608
    },
    {
      "epoch": 0.10791609204774684,
      "grad_norm": 3.0625,
      "learning_rate": 4.864634726611072e-05,
      "loss": 0.867,
      "step": 7610
    },
    {
      "epoch": 0.10794445370137305,
      "grad_norm": 3.125,
      "learning_rate": 4.864562194728154e-05,
      "loss": 0.861,
      "step": 7612
    },
    {
      "epoch": 0.10797281535499928,
      "grad_norm": 2.921875,
      "learning_rate": 4.86448964395932e-05,
      "loss": 0.8554,
      "step": 7614
    },
    {
      "epoch": 0.10800117700862549,
      "grad_norm": 3.21875,
      "learning_rate": 4.8644170743051497e-05,
      "loss": 0.8579,
      "step": 7616
    },
    {
      "epoch": 0.1080295386622517,
      "grad_norm": 3.890625,
      "learning_rate": 4.8643444857662224e-05,
      "loss": 0.859,
      "step": 7618
    },
    {
      "epoch": 0.10805790031587792,
      "grad_norm": 3.4375,
      "learning_rate": 4.864271878343117e-05,
      "loss": 0.9224,
      "step": 7620
    },
    {
      "epoch": 0.10808626196950413,
      "grad_norm": 3.0625,
      "learning_rate": 4.864199252036415e-05,
      "loss": 0.9292,
      "step": 7622
    },
    {
      "epoch": 0.10811462362313035,
      "grad_norm": 3.046875,
      "learning_rate": 4.864126606846697e-05,
      "loss": 0.9004,
      "step": 7624
    },
    {
      "epoch": 0.10814298527675656,
      "grad_norm": 3.078125,
      "learning_rate": 4.8640539427745414e-05,
      "loss": 0.8514,
      "step": 7626
    },
    {
      "epoch": 0.10817134693038277,
      "grad_norm": 3.265625,
      "learning_rate": 4.863981259820529e-05,
      "loss": 0.8948,
      "step": 7628
    },
    {
      "epoch": 0.10819970858400899,
      "grad_norm": 3.03125,
      "learning_rate": 4.863908557985241e-05,
      "loss": 0.8709,
      "step": 7630
    },
    {
      "epoch": 0.1082280702376352,
      "grad_norm": 2.921875,
      "learning_rate": 4.863835837269257e-05,
      "loss": 0.8388,
      "step": 7632
    },
    {
      "epoch": 0.10825643189126143,
      "grad_norm": 2.796875,
      "learning_rate": 4.863763097673159e-05,
      "loss": 0.8828,
      "step": 7634
    },
    {
      "epoch": 0.10828479354488764,
      "grad_norm": 2.875,
      "learning_rate": 4.8636903391975274e-05,
      "loss": 0.8438,
      "step": 7636
    },
    {
      "epoch": 0.10831315519851385,
      "grad_norm": 3.046875,
      "learning_rate": 4.863617561842943e-05,
      "loss": 0.8806,
      "step": 7638
    },
    {
      "epoch": 0.10834151685214007,
      "grad_norm": 2.890625,
      "learning_rate": 4.863544765609988e-05,
      "loss": 0.8787,
      "step": 7640
    },
    {
      "epoch": 0.10836987850576628,
      "grad_norm": 2.984375,
      "learning_rate": 4.863471950499243e-05,
      "loss": 0.8523,
      "step": 7642
    },
    {
      "epoch": 0.10839824015939249,
      "grad_norm": 2.71875,
      "learning_rate": 4.86339911651129e-05,
      "loss": 0.8835,
      "step": 7644
    },
    {
      "epoch": 0.1084266018130187,
      "grad_norm": 3.40625,
      "learning_rate": 4.863326263646711e-05,
      "loss": 0.8293,
      "step": 7646
    },
    {
      "epoch": 0.10845496346664492,
      "grad_norm": 3.328125,
      "learning_rate": 4.863253391906087e-05,
      "loss": 0.8707,
      "step": 7648
    },
    {
      "epoch": 0.10848332512027113,
      "grad_norm": 3.25,
      "learning_rate": 4.863180501290001e-05,
      "loss": 0.8456,
      "step": 7650
    },
    {
      "epoch": 0.10851168677389736,
      "grad_norm": 3.140625,
      "learning_rate": 4.863107591799034e-05,
      "loss": 0.8222,
      "step": 7652
    },
    {
      "epoch": 0.10854004842752357,
      "grad_norm": 2.828125,
      "learning_rate": 4.8630346634337696e-05,
      "loss": 0.8754,
      "step": 7654
    },
    {
      "epoch": 0.10856841008114979,
      "grad_norm": 2.765625,
      "learning_rate": 4.86296171619479e-05,
      "loss": 0.808,
      "step": 7656
    },
    {
      "epoch": 0.108596771734776,
      "grad_norm": 3.140625,
      "learning_rate": 4.862888750082677e-05,
      "loss": 0.8544,
      "step": 7658
    },
    {
      "epoch": 0.10862513338840221,
      "grad_norm": 3.21875,
      "learning_rate": 4.862815765098014e-05,
      "loss": 0.9172,
      "step": 7660
    },
    {
      "epoch": 0.10865349504202843,
      "grad_norm": 3.421875,
      "learning_rate": 4.862742761241384e-05,
      "loss": 0.8594,
      "step": 7662
    },
    {
      "epoch": 0.10868185669565464,
      "grad_norm": 2.96875,
      "learning_rate": 4.86266973851337e-05,
      "loss": 0.8896,
      "step": 7664
    },
    {
      "epoch": 0.10871021834928085,
      "grad_norm": 3.078125,
      "learning_rate": 4.862596696914555e-05,
      "loss": 0.8482,
      "step": 7666
    },
    {
      "epoch": 0.10873858000290706,
      "grad_norm": 2.90625,
      "learning_rate": 4.8625236364455226e-05,
      "loss": 0.823,
      "step": 7668
    },
    {
      "epoch": 0.10876694165653328,
      "grad_norm": 3.1875,
      "learning_rate": 4.8624505571068565e-05,
      "loss": 0.8739,
      "step": 7670
    },
    {
      "epoch": 0.1087953033101595,
      "grad_norm": 3.078125,
      "learning_rate": 4.86237745889914e-05,
      "loss": 0.8752,
      "step": 7672
    },
    {
      "epoch": 0.10882366496378572,
      "grad_norm": 3.203125,
      "learning_rate": 4.862304341822958e-05,
      "loss": 0.8367,
      "step": 7674
    },
    {
      "epoch": 0.10885202661741193,
      "grad_norm": 3.046875,
      "learning_rate": 4.8622312058788935e-05,
      "loss": 0.841,
      "step": 7676
    },
    {
      "epoch": 0.10888038827103814,
      "grad_norm": 2.921875,
      "learning_rate": 4.8621580510675304e-05,
      "loss": 0.8829,
      "step": 7678
    },
    {
      "epoch": 0.10890874992466436,
      "grad_norm": 2.828125,
      "learning_rate": 4.862084877389453e-05,
      "loss": 0.8568,
      "step": 7680
    },
    {
      "epoch": 0.10893711157829057,
      "grad_norm": 3.0625,
      "learning_rate": 4.862011684845246e-05,
      "loss": 0.9124,
      "step": 7682
    },
    {
      "epoch": 0.10896547323191678,
      "grad_norm": 3.171875,
      "learning_rate": 4.8619384734354946e-05,
      "loss": 0.8851,
      "step": 7684
    },
    {
      "epoch": 0.108993834885543,
      "grad_norm": 3.28125,
      "learning_rate": 4.861865243160783e-05,
      "loss": 0.8675,
      "step": 7686
    },
    {
      "epoch": 0.10902219653916921,
      "grad_norm": 2.953125,
      "learning_rate": 4.861791994021696e-05,
      "loss": 0.8901,
      "step": 7688
    },
    {
      "epoch": 0.10905055819279542,
      "grad_norm": 3.0625,
      "learning_rate": 4.861718726018819e-05,
      "loss": 0.8517,
      "step": 7690
    },
    {
      "epoch": 0.10907891984642165,
      "grad_norm": 3.15625,
      "learning_rate": 4.861645439152738e-05,
      "loss": 0.8966,
      "step": 7692
    },
    {
      "epoch": 0.10910728150004786,
      "grad_norm": 2.828125,
      "learning_rate": 4.861572133424036e-05,
      "loss": 0.8769,
      "step": 7694
    },
    {
      "epoch": 0.10913564315367408,
      "grad_norm": 3.0,
      "learning_rate": 4.8614988088333e-05,
      "loss": 0.8444,
      "step": 7696
    },
    {
      "epoch": 0.10916400480730029,
      "grad_norm": 3.140625,
      "learning_rate": 4.861425465381115e-05,
      "loss": 0.8887,
      "step": 7698
    },
    {
      "epoch": 0.1091923664609265,
      "grad_norm": 2.859375,
      "learning_rate": 4.861352103068069e-05,
      "loss": 0.8668,
      "step": 7700
    },
    {
      "epoch": 0.10922072811455272,
      "grad_norm": 3.265625,
      "learning_rate": 4.8612787218947454e-05,
      "loss": 0.8787,
      "step": 7702
    },
    {
      "epoch": 0.10924908976817893,
      "grad_norm": 3.28125,
      "learning_rate": 4.8612053218617306e-05,
      "loss": 0.9003,
      "step": 7704
    },
    {
      "epoch": 0.10927745142180514,
      "grad_norm": 3.109375,
      "learning_rate": 4.861131902969612e-05,
      "loss": 0.8872,
      "step": 7706
    },
    {
      "epoch": 0.10930581307543136,
      "grad_norm": 2.9375,
      "learning_rate": 4.861058465218975e-05,
      "loss": 0.8587,
      "step": 7708
    },
    {
      "epoch": 0.10933417472905758,
      "grad_norm": 2.765625,
      "learning_rate": 4.860985008610407e-05,
      "loss": 0.8539,
      "step": 7710
    },
    {
      "epoch": 0.1093625363826838,
      "grad_norm": 3.03125,
      "learning_rate": 4.860911533144494e-05,
      "loss": 0.8328,
      "step": 7712
    },
    {
      "epoch": 0.10939089803631001,
      "grad_norm": 3.421875,
      "learning_rate": 4.860838038821823e-05,
      "loss": 0.8593,
      "step": 7714
    },
    {
      "epoch": 0.10941925968993622,
      "grad_norm": 2.921875,
      "learning_rate": 4.8607645256429814e-05,
      "loss": 0.8849,
      "step": 7716
    },
    {
      "epoch": 0.10944762134356244,
      "grad_norm": 3.25,
      "learning_rate": 4.860690993608556e-05,
      "loss": 0.8658,
      "step": 7718
    },
    {
      "epoch": 0.10947598299718865,
      "grad_norm": 3.265625,
      "learning_rate": 4.860617442719134e-05,
      "loss": 0.8538,
      "step": 7720
    },
    {
      "epoch": 0.10950434465081486,
      "grad_norm": 2.890625,
      "learning_rate": 4.860543872975303e-05,
      "loss": 0.8457,
      "step": 7722
    },
    {
      "epoch": 0.10953270630444108,
      "grad_norm": 2.875,
      "learning_rate": 4.86047028437765e-05,
      "loss": 0.8489,
      "step": 7724
    },
    {
      "epoch": 0.10956106795806729,
      "grad_norm": 2.90625,
      "learning_rate": 4.860396676926765e-05,
      "loss": 0.9111,
      "step": 7726
    },
    {
      "epoch": 0.1095894296116935,
      "grad_norm": 3.21875,
      "learning_rate": 4.8603230506232335e-05,
      "loss": 0.8974,
      "step": 7728
    },
    {
      "epoch": 0.10961779126531973,
      "grad_norm": 2.9375,
      "learning_rate": 4.860249405467644e-05,
      "loss": 0.8547,
      "step": 7730
    },
    {
      "epoch": 0.10964615291894594,
      "grad_norm": 2.96875,
      "learning_rate": 4.860175741460585e-05,
      "loss": 0.9357,
      "step": 7732
    },
    {
      "epoch": 0.10967451457257216,
      "grad_norm": 3.34375,
      "learning_rate": 4.860102058602646e-05,
      "loss": 0.8819,
      "step": 7734
    },
    {
      "epoch": 0.10970287622619837,
      "grad_norm": 3.65625,
      "learning_rate": 4.860028356894414e-05,
      "loss": 0.875,
      "step": 7736
    },
    {
      "epoch": 0.10973123787982458,
      "grad_norm": 3.609375,
      "learning_rate": 4.859954636336478e-05,
      "loss": 0.8494,
      "step": 7738
    },
    {
      "epoch": 0.1097595995334508,
      "grad_norm": 3.203125,
      "learning_rate": 4.859880896929426e-05,
      "loss": 0.9084,
      "step": 7740
    },
    {
      "epoch": 0.10978796118707701,
      "grad_norm": 3.296875,
      "learning_rate": 4.8598071386738485e-05,
      "loss": 0.8661,
      "step": 7742
    },
    {
      "epoch": 0.10981632284070322,
      "grad_norm": 2.90625,
      "learning_rate": 4.859733361570334e-05,
      "loss": 0.8116,
      "step": 7744
    },
    {
      "epoch": 0.10984468449432944,
      "grad_norm": 3.171875,
      "learning_rate": 4.8596595656194725e-05,
      "loss": 0.8557,
      "step": 7746
    },
    {
      "epoch": 0.10987304614795565,
      "grad_norm": 2.78125,
      "learning_rate": 4.8595857508218524e-05,
      "loss": 0.821,
      "step": 7748
    },
    {
      "epoch": 0.10990140780158188,
      "grad_norm": 3.15625,
      "learning_rate": 4.859511917178063e-05,
      "loss": 0.8575,
      "step": 7750
    },
    {
      "epoch": 0.10992976945520809,
      "grad_norm": 3.09375,
      "learning_rate": 4.8594380646886945e-05,
      "loss": 0.8532,
      "step": 7752
    },
    {
      "epoch": 0.1099581311088343,
      "grad_norm": 3.09375,
      "learning_rate": 4.8593641933543364e-05,
      "loss": 0.9046,
      "step": 7754
    },
    {
      "epoch": 0.10998649276246052,
      "grad_norm": 2.90625,
      "learning_rate": 4.85929030317558e-05,
      "loss": 0.8357,
      "step": 7756
    },
    {
      "epoch": 0.11001485441608673,
      "grad_norm": 2.84375,
      "learning_rate": 4.859216394153014e-05,
      "loss": 0.8932,
      "step": 7758
    },
    {
      "epoch": 0.11004321606971294,
      "grad_norm": 3.09375,
      "learning_rate": 4.85914246628723e-05,
      "loss": 0.8889,
      "step": 7760
    },
    {
      "epoch": 0.11007157772333916,
      "grad_norm": 3.3125,
      "learning_rate": 4.859068519578818e-05,
      "loss": 0.8687,
      "step": 7762
    },
    {
      "epoch": 0.11009993937696537,
      "grad_norm": 3.03125,
      "learning_rate": 4.858994554028367e-05,
      "loss": 0.8622,
      "step": 7764
    },
    {
      "epoch": 0.11012830103059158,
      "grad_norm": 3.359375,
      "learning_rate": 4.8589205696364705e-05,
      "loss": 0.8766,
      "step": 7766
    },
    {
      "epoch": 0.11015666268421781,
      "grad_norm": 3.03125,
      "learning_rate": 4.858846566403717e-05,
      "loss": 0.8323,
      "step": 7768
    },
    {
      "epoch": 0.11018502433784402,
      "grad_norm": 3.359375,
      "learning_rate": 4.858772544330699e-05,
      "loss": 0.8889,
      "step": 7770
    },
    {
      "epoch": 0.11021338599147024,
      "grad_norm": 3.0625,
      "learning_rate": 4.8586985034180076e-05,
      "loss": 0.8716,
      "step": 7772
    },
    {
      "epoch": 0.11024174764509645,
      "grad_norm": 3.28125,
      "learning_rate": 4.8586244436662336e-05,
      "loss": 0.8842,
      "step": 7774
    },
    {
      "epoch": 0.11027010929872266,
      "grad_norm": 2.859375,
      "learning_rate": 4.858550365075969e-05,
      "loss": 0.8635,
      "step": 7776
    },
    {
      "epoch": 0.11029847095234888,
      "grad_norm": 3.046875,
      "learning_rate": 4.858476267647806e-05,
      "loss": 0.8807,
      "step": 7778
    },
    {
      "epoch": 0.11032683260597509,
      "grad_norm": 2.828125,
      "learning_rate": 4.8584021513823344e-05,
      "loss": 0.8731,
      "step": 7780
    },
    {
      "epoch": 0.1103551942596013,
      "grad_norm": 3.578125,
      "learning_rate": 4.858328016280148e-05,
      "loss": 0.9016,
      "step": 7782
    },
    {
      "epoch": 0.11038355591322752,
      "grad_norm": 3.125,
      "learning_rate": 4.8582538623418385e-05,
      "loss": 0.8853,
      "step": 7784
    },
    {
      "epoch": 0.11041191756685373,
      "grad_norm": 2.75,
      "learning_rate": 4.8581796895679986e-05,
      "loss": 0.7872,
      "step": 7786
    },
    {
      "epoch": 0.11044027922047996,
      "grad_norm": 3.0625,
      "learning_rate": 4.85810549795922e-05,
      "loss": 0.8786,
      "step": 7788
    },
    {
      "epoch": 0.11046864087410617,
      "grad_norm": 2.78125,
      "learning_rate": 4.858031287516095e-05,
      "loss": 0.8093,
      "step": 7790
    },
    {
      "epoch": 0.11049700252773238,
      "grad_norm": 3.21875,
      "learning_rate": 4.8579570582392176e-05,
      "loss": 0.8321,
      "step": 7792
    },
    {
      "epoch": 0.1105253641813586,
      "grad_norm": 2.609375,
      "learning_rate": 4.85788281012918e-05,
      "loss": 0.812,
      "step": 7794
    },
    {
      "epoch": 0.11055372583498481,
      "grad_norm": 3.109375,
      "learning_rate": 4.857808543186574e-05,
      "loss": 0.8467,
      "step": 7796
    },
    {
      "epoch": 0.11058208748861102,
      "grad_norm": 2.921875,
      "learning_rate": 4.857734257411994e-05,
      "loss": 0.8737,
      "step": 7798
    },
    {
      "epoch": 0.11061044914223724,
      "grad_norm": 2.9375,
      "learning_rate": 4.857659952806034e-05,
      "loss": 0.8156,
      "step": 7800
    },
    {
      "epoch": 0.11063881079586345,
      "grad_norm": 3.140625,
      "learning_rate": 4.857585629369287e-05,
      "loss": 0.8812,
      "step": 7802
    },
    {
      "epoch": 0.11066717244948966,
      "grad_norm": 3.25,
      "learning_rate": 4.8575112871023456e-05,
      "loss": 0.8817,
      "step": 7804
    },
    {
      "epoch": 0.11069553410311589,
      "grad_norm": 3.125,
      "learning_rate": 4.8574369260058036e-05,
      "loss": 0.8363,
      "step": 7806
    },
    {
      "epoch": 0.1107238957567421,
      "grad_norm": 2.984375,
      "learning_rate": 4.857362546080257e-05,
      "loss": 0.8146,
      "step": 7808
    },
    {
      "epoch": 0.11075225741036832,
      "grad_norm": 3.03125,
      "learning_rate": 4.8572881473262975e-05,
      "loss": 0.8528,
      "step": 7810
    },
    {
      "epoch": 0.11078061906399453,
      "grad_norm": 3.109375,
      "learning_rate": 4.857213729744521e-05,
      "loss": 0.8651,
      "step": 7812
    },
    {
      "epoch": 0.11080898071762074,
      "grad_norm": 3.03125,
      "learning_rate": 4.857139293335521e-05,
      "loss": 0.8672,
      "step": 7814
    },
    {
      "epoch": 0.11083734237124696,
      "grad_norm": 2.9375,
      "learning_rate": 4.8570648380998926e-05,
      "loss": 0.8589,
      "step": 7816
    },
    {
      "epoch": 0.11086570402487317,
      "grad_norm": 3.203125,
      "learning_rate": 4.85699036403823e-05,
      "loss": 0.8626,
      "step": 7818
    },
    {
      "epoch": 0.11089406567849938,
      "grad_norm": 3.3125,
      "learning_rate": 4.8569158711511275e-05,
      "loss": 0.9068,
      "step": 7820
    },
    {
      "epoch": 0.1109224273321256,
      "grad_norm": 3.21875,
      "learning_rate": 4.8568413594391814e-05,
      "loss": 0.8506,
      "step": 7822
    },
    {
      "epoch": 0.11095078898575181,
      "grad_norm": 3.453125,
      "learning_rate": 4.856766828902986e-05,
      "loss": 0.8823,
      "step": 7824
    },
    {
      "epoch": 0.11097915063937804,
      "grad_norm": 3.328125,
      "learning_rate": 4.8566922795431366e-05,
      "loss": 0.8587,
      "step": 7826
    },
    {
      "epoch": 0.11100751229300425,
      "grad_norm": 3.015625,
      "learning_rate": 4.856617711360229e-05,
      "loss": 0.8912,
      "step": 7828
    },
    {
      "epoch": 0.11103587394663046,
      "grad_norm": 2.859375,
      "learning_rate": 4.856543124354858e-05,
      "loss": 0.8665,
      "step": 7830
    },
    {
      "epoch": 0.11106423560025667,
      "grad_norm": 2.890625,
      "learning_rate": 4.8564685185276204e-05,
      "loss": 0.8964,
      "step": 7832
    },
    {
      "epoch": 0.11109259725388289,
      "grad_norm": 3.1875,
      "learning_rate": 4.856393893879111e-05,
      "loss": 0.8733,
      "step": 7834
    },
    {
      "epoch": 0.1111209589075091,
      "grad_norm": 3.03125,
      "learning_rate": 4.856319250409927e-05,
      "loss": 0.8829,
      "step": 7836
    },
    {
      "epoch": 0.11114932056113531,
      "grad_norm": 3.03125,
      "learning_rate": 4.8562445881206645e-05,
      "loss": 0.8341,
      "step": 7838
    },
    {
      "epoch": 0.11117768221476153,
      "grad_norm": 3.1875,
      "learning_rate": 4.856169907011918e-05,
      "loss": 0.8954,
      "step": 7840
    },
    {
      "epoch": 0.11120604386838774,
      "grad_norm": 3.625,
      "learning_rate": 4.856095207084286e-05,
      "loss": 0.9248,
      "step": 7842
    },
    {
      "epoch": 0.11123440552201395,
      "grad_norm": 3.46875,
      "learning_rate": 4.8560204883383645e-05,
      "loss": 0.8557,
      "step": 7844
    },
    {
      "epoch": 0.11126276717564018,
      "grad_norm": 2.984375,
      "learning_rate": 4.85594575077475e-05,
      "loss": 0.8953,
      "step": 7846
    },
    {
      "epoch": 0.1112911288292664,
      "grad_norm": 2.8125,
      "learning_rate": 4.85587099439404e-05,
      "loss": 0.8676,
      "step": 7848
    },
    {
      "epoch": 0.11131949048289261,
      "grad_norm": 2.859375,
      "learning_rate": 4.855796219196831e-05,
      "loss": 0.8728,
      "step": 7850
    },
    {
      "epoch": 0.11134785213651882,
      "grad_norm": 2.640625,
      "learning_rate": 4.85572142518372e-05,
      "loss": 0.8341,
      "step": 7852
    },
    {
      "epoch": 0.11137621379014503,
      "grad_norm": 2.90625,
      "learning_rate": 4.855646612355306e-05,
      "loss": 0.7628,
      "step": 7854
    },
    {
      "epoch": 0.11140457544377125,
      "grad_norm": 3.171875,
      "learning_rate": 4.855571780712184e-05,
      "loss": 0.8515,
      "step": 7856
    },
    {
      "epoch": 0.11143293709739746,
      "grad_norm": 3.140625,
      "learning_rate": 4.855496930254954e-05,
      "loss": 0.8772,
      "step": 7858
    },
    {
      "epoch": 0.11146129875102367,
      "grad_norm": 3.171875,
      "learning_rate": 4.8554220609842124e-05,
      "loss": 0.8694,
      "step": 7860
    },
    {
      "epoch": 0.11148966040464989,
      "grad_norm": 2.984375,
      "learning_rate": 4.8553471729005577e-05,
      "loss": 0.8692,
      "step": 7862
    },
    {
      "epoch": 0.11151802205827611,
      "grad_norm": 3.15625,
      "learning_rate": 4.8552722660045884e-05,
      "loss": 0.8998,
      "step": 7864
    },
    {
      "epoch": 0.11154638371190233,
      "grad_norm": 3.1875,
      "learning_rate": 4.855197340296902e-05,
      "loss": 0.8899,
      "step": 7866
    },
    {
      "epoch": 0.11157474536552854,
      "grad_norm": 3.0,
      "learning_rate": 4.855122395778098e-05,
      "loss": 0.8794,
      "step": 7868
    },
    {
      "epoch": 0.11160310701915475,
      "grad_norm": 2.96875,
      "learning_rate": 4.8550474324487735e-05,
      "loss": 0.8914,
      "step": 7870
    },
    {
      "epoch": 0.11163146867278097,
      "grad_norm": 3.3125,
      "learning_rate": 4.8549724503095286e-05,
      "loss": 0.861,
      "step": 7872
    },
    {
      "epoch": 0.11165983032640718,
      "grad_norm": 3.0625,
      "learning_rate": 4.854897449360962e-05,
      "loss": 0.847,
      "step": 7874
    },
    {
      "epoch": 0.1116881919800334,
      "grad_norm": 2.890625,
      "learning_rate": 4.854822429603672e-05,
      "loss": 0.8836,
      "step": 7876
    },
    {
      "epoch": 0.11171655363365961,
      "grad_norm": 2.875,
      "learning_rate": 4.854747391038258e-05,
      "loss": 0.8499,
      "step": 7878
    },
    {
      "epoch": 0.11174491528728582,
      "grad_norm": 2.9375,
      "learning_rate": 4.85467233366532e-05,
      "loss": 0.9359,
      "step": 7880
    },
    {
      "epoch": 0.11177327694091203,
      "grad_norm": 3.359375,
      "learning_rate": 4.854597257485456e-05,
      "loss": 0.9127,
      "step": 7882
    },
    {
      "epoch": 0.11180163859453826,
      "grad_norm": 3.171875,
      "learning_rate": 4.8545221624992674e-05,
      "loss": 0.9026,
      "step": 7884
    },
    {
      "epoch": 0.11183000024816447,
      "grad_norm": 3.046875,
      "learning_rate": 4.854447048707354e-05,
      "loss": 0.8246,
      "step": 7886
    },
    {
      "epoch": 0.11185836190179069,
      "grad_norm": 3.28125,
      "learning_rate": 4.854371916110313e-05,
      "loss": 0.8877,
      "step": 7888
    },
    {
      "epoch": 0.1118867235554169,
      "grad_norm": 2.9375,
      "learning_rate": 4.854296764708748e-05,
      "loss": 0.8906,
      "step": 7890
    },
    {
      "epoch": 0.11191508520904311,
      "grad_norm": 2.96875,
      "learning_rate": 4.854221594503258e-05,
      "loss": 0.874,
      "step": 7892
    },
    {
      "epoch": 0.11194344686266933,
      "grad_norm": 3.015625,
      "learning_rate": 4.8541464054944416e-05,
      "loss": 0.876,
      "step": 7894
    },
    {
      "epoch": 0.11197180851629554,
      "grad_norm": 3.3125,
      "learning_rate": 4.854071197682902e-05,
      "loss": 0.8959,
      "step": 7896
    },
    {
      "epoch": 0.11200017016992175,
      "grad_norm": 3.203125,
      "learning_rate": 4.8539959710692383e-05,
      "loss": 0.8503,
      "step": 7898
    },
    {
      "epoch": 0.11202853182354797,
      "grad_norm": 3.0,
      "learning_rate": 4.853920725654052e-05,
      "loss": 0.8716,
      "step": 7900
    },
    {
      "epoch": 0.11205689347717418,
      "grad_norm": 2.828125,
      "learning_rate": 4.8538454614379435e-05,
      "loss": 0.8459,
      "step": 7902
    },
    {
      "epoch": 0.1120852551308004,
      "grad_norm": 2.84375,
      "learning_rate": 4.8537701784215145e-05,
      "loss": 0.8725,
      "step": 7904
    },
    {
      "epoch": 0.11211361678442662,
      "grad_norm": 3.578125,
      "learning_rate": 4.853694876605366e-05,
      "loss": 0.8952,
      "step": 7906
    },
    {
      "epoch": 0.11214197843805283,
      "grad_norm": 2.9375,
      "learning_rate": 4.8536195559901e-05,
      "loss": 0.8656,
      "step": 7908
    },
    {
      "epoch": 0.11217034009167905,
      "grad_norm": 3.359375,
      "learning_rate": 4.8535442165763166e-05,
      "loss": 0.9008,
      "step": 7910
    },
    {
      "epoch": 0.11219870174530526,
      "grad_norm": 2.71875,
      "learning_rate": 4.853468858364619e-05,
      "loss": 0.8839,
      "step": 7912
    },
    {
      "epoch": 0.11222706339893147,
      "grad_norm": 3.15625,
      "learning_rate": 4.853393481355609e-05,
      "loss": 0.9129,
      "step": 7914
    },
    {
      "epoch": 0.11225542505255769,
      "grad_norm": 2.671875,
      "learning_rate": 4.853318085549888e-05,
      "loss": 0.868,
      "step": 7916
    },
    {
      "epoch": 0.1122837867061839,
      "grad_norm": 2.875,
      "learning_rate": 4.853242670948059e-05,
      "loss": 0.8683,
      "step": 7918
    },
    {
      "epoch": 0.11231214835981011,
      "grad_norm": 3.234375,
      "learning_rate": 4.8531672375507234e-05,
      "loss": 0.8505,
      "step": 7920
    },
    {
      "epoch": 0.11234051001343634,
      "grad_norm": 3.421875,
      "learning_rate": 4.853091785358485e-05,
      "loss": 0.9313,
      "step": 7922
    },
    {
      "epoch": 0.11236887166706255,
      "grad_norm": 3.0625,
      "learning_rate": 4.8530163143719445e-05,
      "loss": 0.8849,
      "step": 7924
    },
    {
      "epoch": 0.11239723332068877,
      "grad_norm": 3.109375,
      "learning_rate": 4.8529408245917055e-05,
      "loss": 0.8613,
      "step": 7926
    },
    {
      "epoch": 0.11242559497431498,
      "grad_norm": 3.125,
      "learning_rate": 4.8528653160183715e-05,
      "loss": 0.9259,
      "step": 7928
    },
    {
      "epoch": 0.11245395662794119,
      "grad_norm": 3.484375,
      "learning_rate": 4.852789788652546e-05,
      "loss": 0.8758,
      "step": 7930
    },
    {
      "epoch": 0.1124823182815674,
      "grad_norm": 3.53125,
      "learning_rate": 4.8527142424948305e-05,
      "loss": 0.8767,
      "step": 7932
    },
    {
      "epoch": 0.11251067993519362,
      "grad_norm": 2.90625,
      "learning_rate": 4.8526386775458296e-05,
      "loss": 0.8126,
      "step": 7934
    },
    {
      "epoch": 0.11253904158881983,
      "grad_norm": 2.859375,
      "learning_rate": 4.8525630938061475e-05,
      "loss": 0.8656,
      "step": 7936
    },
    {
      "epoch": 0.11256740324244605,
      "grad_norm": 3.203125,
      "learning_rate": 4.8524874912763865e-05,
      "loss": 0.9178,
      "step": 7938
    },
    {
      "epoch": 0.11259576489607226,
      "grad_norm": 3.171875,
      "learning_rate": 4.8524118699571505e-05,
      "loss": 0.929,
      "step": 7940
    },
    {
      "epoch": 0.11262412654969849,
      "grad_norm": 2.890625,
      "learning_rate": 4.8523362298490446e-05,
      "loss": 0.8987,
      "step": 7942
    },
    {
      "epoch": 0.1126524882033247,
      "grad_norm": 3.125,
      "learning_rate": 4.8522605709526725e-05,
      "loss": 0.8599,
      "step": 7944
    },
    {
      "epoch": 0.11268084985695091,
      "grad_norm": 2.859375,
      "learning_rate": 4.852184893268638e-05,
      "loss": 0.8894,
      "step": 7946
    },
    {
      "epoch": 0.11270921151057713,
      "grad_norm": 3.078125,
      "learning_rate": 4.852109196797546e-05,
      "loss": 0.8124,
      "step": 7948
    },
    {
      "epoch": 0.11273757316420334,
      "grad_norm": 3.234375,
      "learning_rate": 4.852033481540001e-05,
      "loss": 0.8303,
      "step": 7950
    },
    {
      "epoch": 0.11276593481782955,
      "grad_norm": 3.203125,
      "learning_rate": 4.8519577474966074e-05,
      "loss": 0.8817,
      "step": 7952
    },
    {
      "epoch": 0.11279429647145577,
      "grad_norm": 3.234375,
      "learning_rate": 4.8518819946679705e-05,
      "loss": 0.8241,
      "step": 7954
    },
    {
      "epoch": 0.11282265812508198,
      "grad_norm": 3.34375,
      "learning_rate": 4.8518062230546954e-05,
      "loss": 0.9189,
      "step": 7956
    },
    {
      "epoch": 0.11285101977870819,
      "grad_norm": 3.015625,
      "learning_rate": 4.851730432657387e-05,
      "loss": 0.8897,
      "step": 7958
    },
    {
      "epoch": 0.11287938143233442,
      "grad_norm": 2.984375,
      "learning_rate": 4.8516546234766506e-05,
      "loss": 0.8995,
      "step": 7960
    },
    {
      "epoch": 0.11290774308596063,
      "grad_norm": 2.96875,
      "learning_rate": 4.851578795513092e-05,
      "loss": 0.8992,
      "step": 7962
    },
    {
      "epoch": 0.11293610473958685,
      "grad_norm": 3.015625,
      "learning_rate": 4.8515029487673166e-05,
      "loss": 0.9026,
      "step": 7964
    },
    {
      "epoch": 0.11296446639321306,
      "grad_norm": 2.9375,
      "learning_rate": 4.851427083239931e-05,
      "loss": 0.9006,
      "step": 7966
    },
    {
      "epoch": 0.11299282804683927,
      "grad_norm": 3.0,
      "learning_rate": 4.85135119893154e-05,
      "loss": 0.9256,
      "step": 7968
    },
    {
      "epoch": 0.11302118970046549,
      "grad_norm": 3.265625,
      "learning_rate": 4.85127529584275e-05,
      "loss": 0.9173,
      "step": 7970
    },
    {
      "epoch": 0.1130495513540917,
      "grad_norm": 3.109375,
      "learning_rate": 4.8511993739741676e-05,
      "loss": 0.8615,
      "step": 7972
    },
    {
      "epoch": 0.11307791300771791,
      "grad_norm": 3.125,
      "learning_rate": 4.8511234333263985e-05,
      "loss": 0.8467,
      "step": 7974
    },
    {
      "epoch": 0.11310627466134412,
      "grad_norm": 3.078125,
      "learning_rate": 4.85104747390005e-05,
      "loss": 0.8834,
      "step": 7976
    },
    {
      "epoch": 0.11313463631497034,
      "grad_norm": 3.15625,
      "learning_rate": 4.8509714956957284e-05,
      "loss": 0.9022,
      "step": 7978
    },
    {
      "epoch": 0.11316299796859657,
      "grad_norm": 3.21875,
      "learning_rate": 4.850895498714041e-05,
      "loss": 0.8876,
      "step": 7980
    },
    {
      "epoch": 0.11319135962222278,
      "grad_norm": 3.21875,
      "learning_rate": 4.850819482955594e-05,
      "loss": 0.8673,
      "step": 7982
    },
    {
      "epoch": 0.11321972127584899,
      "grad_norm": 3.25,
      "learning_rate": 4.850743448420995e-05,
      "loss": 0.8536,
      "step": 7984
    },
    {
      "epoch": 0.1132480829294752,
      "grad_norm": 3.34375,
      "learning_rate": 4.8506673951108516e-05,
      "loss": 0.8754,
      "step": 7986
    },
    {
      "epoch": 0.11327644458310142,
      "grad_norm": 2.84375,
      "learning_rate": 4.850591323025771e-05,
      "loss": 0.8892,
      "step": 7988
    },
    {
      "epoch": 0.11330480623672763,
      "grad_norm": 3.265625,
      "learning_rate": 4.850515232166361e-05,
      "loss": 0.8894,
      "step": 7990
    },
    {
      "epoch": 0.11333316789035384,
      "grad_norm": 3.375,
      "learning_rate": 4.8504391225332277e-05,
      "loss": 0.9214,
      "step": 7992
    },
    {
      "epoch": 0.11336152954398006,
      "grad_norm": 3.75,
      "learning_rate": 4.8503629941269815e-05,
      "loss": 0.8862,
      "step": 7994
    },
    {
      "epoch": 0.11338989119760627,
      "grad_norm": 3.65625,
      "learning_rate": 4.8502868469482294e-05,
      "loss": 0.8707,
      "step": 7996
    },
    {
      "epoch": 0.11341825285123248,
      "grad_norm": 2.96875,
      "learning_rate": 4.850210680997579e-05,
      "loss": 0.9019,
      "step": 7998
    },
    {
      "epoch": 0.11344661450485871,
      "grad_norm": 3.109375,
      "learning_rate": 4.850134496275639e-05,
      "loss": 0.917,
      "step": 8000
    },
    {
      "epoch": 0.11347497615848492,
      "grad_norm": 3.0,
      "learning_rate": 4.8500582927830185e-05,
      "loss": 0.8535,
      "step": 8002
    },
    {
      "epoch": 0.11350333781211114,
      "grad_norm": 2.828125,
      "learning_rate": 4.849982070520326e-05,
      "loss": 0.9012,
      "step": 8004
    },
    {
      "epoch": 0.11353169946573735,
      "grad_norm": 2.984375,
      "learning_rate": 4.849905829488169e-05,
      "loss": 0.8476,
      "step": 8006
    },
    {
      "epoch": 0.11356006111936356,
      "grad_norm": 3.25,
      "learning_rate": 4.849829569687158e-05,
      "loss": 0.8604,
      "step": 8008
    },
    {
      "epoch": 0.11358842277298978,
      "grad_norm": 3.171875,
      "learning_rate": 4.849753291117901e-05,
      "loss": 0.8776,
      "step": 8010
    },
    {
      "epoch": 0.11361678442661599,
      "grad_norm": 2.953125,
      "learning_rate": 4.849676993781008e-05,
      "loss": 0.8435,
      "step": 8012
    },
    {
      "epoch": 0.1136451460802422,
      "grad_norm": 3.171875,
      "learning_rate": 4.849600677677089e-05,
      "loss": 0.7998,
      "step": 8014
    },
    {
      "epoch": 0.11367350773386842,
      "grad_norm": 2.96875,
      "learning_rate": 4.849524342806751e-05,
      "loss": 0.9253,
      "step": 8016
    },
    {
      "epoch": 0.11370186938749464,
      "grad_norm": 3.5625,
      "learning_rate": 4.8494479891706065e-05,
      "loss": 0.8453,
      "step": 8018
    },
    {
      "epoch": 0.11373023104112086,
      "grad_norm": 3.03125,
      "learning_rate": 4.849371616769264e-05,
      "loss": 0.8574,
      "step": 8020
    },
    {
      "epoch": 0.11375859269474707,
      "grad_norm": 3.0,
      "learning_rate": 4.8492952256033333e-05,
      "loss": 0.9206,
      "step": 8022
    },
    {
      "epoch": 0.11378695434837328,
      "grad_norm": 3.0,
      "learning_rate": 4.8492188156734255e-05,
      "loss": 0.8401,
      "step": 8024
    },
    {
      "epoch": 0.1138153160019995,
      "grad_norm": 3.234375,
      "learning_rate": 4.84914238698015e-05,
      "loss": 0.8721,
      "step": 8026
    },
    {
      "epoch": 0.11384367765562571,
      "grad_norm": 3.265625,
      "learning_rate": 4.849065939524117e-05,
      "loss": 0.8884,
      "step": 8028
    },
    {
      "epoch": 0.11387203930925192,
      "grad_norm": 2.84375,
      "learning_rate": 4.848989473305938e-05,
      "loss": 0.8851,
      "step": 8030
    },
    {
      "epoch": 0.11390040096287814,
      "grad_norm": 2.984375,
      "learning_rate": 4.848912988326224e-05,
      "loss": 0.8398,
      "step": 8032
    },
    {
      "epoch": 0.11392876261650435,
      "grad_norm": 3.609375,
      "learning_rate": 4.848836484585585e-05,
      "loss": 0.895,
      "step": 8034
    },
    {
      "epoch": 0.11395712427013056,
      "grad_norm": 3.359375,
      "learning_rate": 4.848759962084633e-05,
      "loss": 0.9215,
      "step": 8036
    },
    {
      "epoch": 0.11398548592375679,
      "grad_norm": 3.328125,
      "learning_rate": 4.8486834208239775e-05,
      "loss": 0.8578,
      "step": 8038
    },
    {
      "epoch": 0.114013847577383,
      "grad_norm": 3.046875,
      "learning_rate": 4.848606860804231e-05,
      "loss": 0.9114,
      "step": 8040
    },
    {
      "epoch": 0.11404220923100922,
      "grad_norm": 3.046875,
      "learning_rate": 4.8485302820260045e-05,
      "loss": 0.9222,
      "step": 8042
    },
    {
      "epoch": 0.11407057088463543,
      "grad_norm": 3.015625,
      "learning_rate": 4.848453684489911e-05,
      "loss": 0.8245,
      "step": 8044
    },
    {
      "epoch": 0.11409893253826164,
      "grad_norm": 3.171875,
      "learning_rate": 4.848377068196561e-05,
      "loss": 0.8483,
      "step": 8046
    },
    {
      "epoch": 0.11412729419188786,
      "grad_norm": 2.828125,
      "learning_rate": 4.8483004331465664e-05,
      "loss": 0.8632,
      "step": 8048
    },
    {
      "epoch": 0.11415565584551407,
      "grad_norm": 2.6875,
      "learning_rate": 4.84822377934054e-05,
      "loss": 0.8302,
      "step": 8050
    },
    {
      "epoch": 0.11418401749914028,
      "grad_norm": 3.09375,
      "learning_rate": 4.848147106779093e-05,
      "loss": 0.8575,
      "step": 8052
    },
    {
      "epoch": 0.1142123791527665,
      "grad_norm": 2.921875,
      "learning_rate": 4.848070415462839e-05,
      "loss": 0.8622,
      "step": 8054
    },
    {
      "epoch": 0.11424074080639271,
      "grad_norm": 3.265625,
      "learning_rate": 4.84799370539239e-05,
      "loss": 0.8864,
      "step": 8056
    },
    {
      "epoch": 0.11426910246001894,
      "grad_norm": 3.203125,
      "learning_rate": 4.8479169765683575e-05,
      "loss": 0.8852,
      "step": 8058
    },
    {
      "epoch": 0.11429746411364515,
      "grad_norm": 3.25,
      "learning_rate": 4.8478402289913566e-05,
      "loss": 0.8212,
      "step": 8060
    },
    {
      "epoch": 0.11432582576727136,
      "grad_norm": 3.0,
      "learning_rate": 4.847763462661999e-05,
      "loss": 0.8793,
      "step": 8062
    },
    {
      "epoch": 0.11435418742089758,
      "grad_norm": 3.390625,
      "learning_rate": 4.847686677580898e-05,
      "loss": 0.8992,
      "step": 8064
    },
    {
      "epoch": 0.11438254907452379,
      "grad_norm": 3.25,
      "learning_rate": 4.847609873748667e-05,
      "loss": 0.9071,
      "step": 8066
    },
    {
      "epoch": 0.11441091072815,
      "grad_norm": 3.65625,
      "learning_rate": 4.847533051165919e-05,
      "loss": 0.886,
      "step": 8068
    },
    {
      "epoch": 0.11443927238177622,
      "grad_norm": 3.34375,
      "learning_rate": 4.8474562098332686e-05,
      "loss": 0.8981,
      "step": 8070
    },
    {
      "epoch": 0.11446763403540243,
      "grad_norm": 2.953125,
      "learning_rate": 4.847379349751329e-05,
      "loss": 0.8392,
      "step": 8072
    },
    {
      "epoch": 0.11449599568902864,
      "grad_norm": 3.046875,
      "learning_rate": 4.847302470920713e-05,
      "loss": 0.8519,
      "step": 8074
    },
    {
      "epoch": 0.11452435734265487,
      "grad_norm": 3.453125,
      "learning_rate": 4.847225573342036e-05,
      "loss": 0.8529,
      "step": 8076
    },
    {
      "epoch": 0.11455271899628108,
      "grad_norm": 3.1875,
      "learning_rate": 4.847148657015912e-05,
      "loss": 0.9027,
      "step": 8078
    },
    {
      "epoch": 0.1145810806499073,
      "grad_norm": 3.1875,
      "learning_rate": 4.847071721942955e-05,
      "loss": 0.904,
      "step": 8080
    },
    {
      "epoch": 0.11460944230353351,
      "grad_norm": 2.8125,
      "learning_rate": 4.846994768123779e-05,
      "loss": 0.8345,
      "step": 8082
    },
    {
      "epoch": 0.11463780395715972,
      "grad_norm": 2.859375,
      "learning_rate": 4.846917795559e-05,
      "loss": 0.8978,
      "step": 8084
    },
    {
      "epoch": 0.11466616561078594,
      "grad_norm": 3.046875,
      "learning_rate": 4.8468408042492316e-05,
      "loss": 0.8859,
      "step": 8086
    },
    {
      "epoch": 0.11469452726441215,
      "grad_norm": 3.109375,
      "learning_rate": 4.8467637941950895e-05,
      "loss": 0.8554,
      "step": 8088
    },
    {
      "epoch": 0.11472288891803836,
      "grad_norm": 3.140625,
      "learning_rate": 4.846686765397189e-05,
      "loss": 0.8391,
      "step": 8090
    },
    {
      "epoch": 0.11475125057166458,
      "grad_norm": 3.078125,
      "learning_rate": 4.8466097178561435e-05,
      "loss": 0.8798,
      "step": 8092
    },
    {
      "epoch": 0.11477961222529079,
      "grad_norm": 3.140625,
      "learning_rate": 4.846532651572571e-05,
      "loss": 0.9073,
      "step": 8094
    },
    {
      "epoch": 0.11480797387891702,
      "grad_norm": 3.3125,
      "learning_rate": 4.8464555665470846e-05,
      "loss": 0.8932,
      "step": 8096
    },
    {
      "epoch": 0.11483633553254323,
      "grad_norm": 2.84375,
      "learning_rate": 4.846378462780301e-05,
      "loss": 0.838,
      "step": 8098
    },
    {
      "epoch": 0.11486469718616944,
      "grad_norm": 3.234375,
      "learning_rate": 4.8463013402728376e-05,
      "loss": 0.87,
      "step": 8100
    },
    {
      "epoch": 0.11489305883979566,
      "grad_norm": 2.734375,
      "learning_rate": 4.8462241990253077e-05,
      "loss": 0.8483,
      "step": 8102
    },
    {
      "epoch": 0.11492142049342187,
      "grad_norm": 3.25,
      "learning_rate": 4.8461470390383284e-05,
      "loss": 0.9068,
      "step": 8104
    },
    {
      "epoch": 0.11494978214704808,
      "grad_norm": 3.0,
      "learning_rate": 4.8460698603125165e-05,
      "loss": 0.8483,
      "step": 8106
    },
    {
      "epoch": 0.1149781438006743,
      "grad_norm": 2.953125,
      "learning_rate": 4.8459926628484885e-05,
      "loss": 0.9265,
      "step": 8108
    },
    {
      "epoch": 0.11500650545430051,
      "grad_norm": 2.75,
      "learning_rate": 4.8459154466468606e-05,
      "loss": 0.861,
      "step": 8110
    },
    {
      "epoch": 0.11503486710792672,
      "grad_norm": 3.265625,
      "learning_rate": 4.845838211708249e-05,
      "loss": 0.8518,
      "step": 8112
    },
    {
      "epoch": 0.11506322876155295,
      "grad_norm": 2.953125,
      "learning_rate": 4.845760958033272e-05,
      "loss": 0.8733,
      "step": 8114
    },
    {
      "epoch": 0.11509159041517916,
      "grad_norm": 3.25,
      "learning_rate": 4.845683685622545e-05,
      "loss": 0.9133,
      "step": 8116
    },
    {
      "epoch": 0.11511995206880538,
      "grad_norm": 3.296875,
      "learning_rate": 4.845606394476686e-05,
      "loss": 0.8744,
      "step": 8118
    },
    {
      "epoch": 0.11514831372243159,
      "grad_norm": 2.78125,
      "learning_rate": 4.845529084596313e-05,
      "loss": 0.8499,
      "step": 8120
    },
    {
      "epoch": 0.1151766753760578,
      "grad_norm": 3.046875,
      "learning_rate": 4.8454517559820414e-05,
      "loss": 0.8626,
      "step": 8122
    },
    {
      "epoch": 0.11520503702968402,
      "grad_norm": 3.1875,
      "learning_rate": 4.845374408634491e-05,
      "loss": 0.8838,
      "step": 8124
    },
    {
      "epoch": 0.11523339868331023,
      "grad_norm": 3.109375,
      "learning_rate": 4.845297042554279e-05,
      "loss": 0.8479,
      "step": 8126
    },
    {
      "epoch": 0.11526176033693644,
      "grad_norm": 3.4375,
      "learning_rate": 4.8452196577420225e-05,
      "loss": 0.8448,
      "step": 8128
    },
    {
      "epoch": 0.11529012199056266,
      "grad_norm": 3.0,
      "learning_rate": 4.8451422541983405e-05,
      "loss": 0.8875,
      "step": 8130
    },
    {
      "epoch": 0.11531848364418887,
      "grad_norm": 3.328125,
      "learning_rate": 4.845064831923851e-05,
      "loss": 0.9164,
      "step": 8132
    },
    {
      "epoch": 0.1153468452978151,
      "grad_norm": 2.8125,
      "learning_rate": 4.8449873909191726e-05,
      "loss": 0.8871,
      "step": 8134
    },
    {
      "epoch": 0.11537520695144131,
      "grad_norm": 3.140625,
      "learning_rate": 4.844909931184923e-05,
      "loss": 0.8491,
      "step": 8136
    },
    {
      "epoch": 0.11540356860506752,
      "grad_norm": 3.21875,
      "learning_rate": 4.8448324527217214e-05,
      "loss": 0.8617,
      "step": 8138
    },
    {
      "epoch": 0.11543193025869374,
      "grad_norm": 2.90625,
      "learning_rate": 4.8447549555301865e-05,
      "loss": 0.8755,
      "step": 8140
    },
    {
      "epoch": 0.11546029191231995,
      "grad_norm": 2.9375,
      "learning_rate": 4.8446774396109375e-05,
      "loss": 0.8763,
      "step": 8142
    },
    {
      "epoch": 0.11548865356594616,
      "grad_norm": 3.34375,
      "learning_rate": 4.844599904964593e-05,
      "loss": 0.8828,
      "step": 8144
    },
    {
      "epoch": 0.11551701521957237,
      "grad_norm": 3.046875,
      "learning_rate": 4.844522351591774e-05,
      "loss": 0.8937,
      "step": 8146
    },
    {
      "epoch": 0.11554537687319859,
      "grad_norm": 3.109375,
      "learning_rate": 4.844444779493097e-05,
      "loss": 0.8962,
      "step": 8148
    },
    {
      "epoch": 0.1155737385268248,
      "grad_norm": 3.625,
      "learning_rate": 4.8443671886691835e-05,
      "loss": 0.8446,
      "step": 8150
    },
    {
      "epoch": 0.11560210018045101,
      "grad_norm": 3.078125,
      "learning_rate": 4.844289579120653e-05,
      "loss": 0.8854,
      "step": 8152
    },
    {
      "epoch": 0.11563046183407724,
      "grad_norm": 2.8125,
      "learning_rate": 4.8442119508481265e-05,
      "loss": 0.8721,
      "step": 8154
    },
    {
      "epoch": 0.11565882348770345,
      "grad_norm": 2.96875,
      "learning_rate": 4.8441343038522216e-05,
      "loss": 0.802,
      "step": 8156
    },
    {
      "epoch": 0.11568718514132967,
      "grad_norm": 3.140625,
      "learning_rate": 4.8440566381335594e-05,
      "loss": 0.8544,
      "step": 8158
    },
    {
      "epoch": 0.11571554679495588,
      "grad_norm": 3.03125,
      "learning_rate": 4.843978953692762e-05,
      "loss": 0.7877,
      "step": 8160
    },
    {
      "epoch": 0.1157439084485821,
      "grad_norm": 2.9375,
      "learning_rate": 4.8439012505304465e-05,
      "loss": 0.8352,
      "step": 8162
    },
    {
      "epoch": 0.11577227010220831,
      "grad_norm": 2.984375,
      "learning_rate": 4.843823528647236e-05,
      "loss": 0.8102,
      "step": 8164
    },
    {
      "epoch": 0.11580063175583452,
      "grad_norm": 3.1875,
      "learning_rate": 4.843745788043751e-05,
      "loss": 0.8624,
      "step": 8166
    },
    {
      "epoch": 0.11582899340946073,
      "grad_norm": 3.0,
      "learning_rate": 4.843668028720611e-05,
      "loss": 0.8971,
      "step": 8168
    },
    {
      "epoch": 0.11585735506308695,
      "grad_norm": 3.3125,
      "learning_rate": 4.84359025067844e-05,
      "loss": 0.8684,
      "step": 8170
    },
    {
      "epoch": 0.11588571671671317,
      "grad_norm": 2.890625,
      "learning_rate": 4.8435124539178564e-05,
      "loss": 0.8245,
      "step": 8172
    },
    {
      "epoch": 0.11591407837033939,
      "grad_norm": 2.9375,
      "learning_rate": 4.8434346384394824e-05,
      "loss": 0.8286,
      "step": 8174
    },
    {
      "epoch": 0.1159424400239656,
      "grad_norm": 3.234375,
      "learning_rate": 4.8433568042439396e-05,
      "loss": 0.9082,
      "step": 8176
    },
    {
      "epoch": 0.11597080167759181,
      "grad_norm": 2.78125,
      "learning_rate": 4.84327895133185e-05,
      "loss": 0.8416,
      "step": 8178
    },
    {
      "epoch": 0.11599916333121803,
      "grad_norm": 2.828125,
      "learning_rate": 4.8432010797038346e-05,
      "loss": 0.8625,
      "step": 8180
    },
    {
      "epoch": 0.11602752498484424,
      "grad_norm": 3.578125,
      "learning_rate": 4.843123189360516e-05,
      "loss": 0.8997,
      "step": 8182
    },
    {
      "epoch": 0.11605588663847045,
      "grad_norm": 3.09375,
      "learning_rate": 4.8430452803025166e-05,
      "loss": 0.9148,
      "step": 8184
    },
    {
      "epoch": 0.11608424829209667,
      "grad_norm": 2.8125,
      "learning_rate": 4.842967352530458e-05,
      "loss": 0.8548,
      "step": 8186
    },
    {
      "epoch": 0.11611260994572288,
      "grad_norm": 2.921875,
      "learning_rate": 4.842889406044963e-05,
      "loss": 0.8563,
      "step": 8188
    },
    {
      "epoch": 0.1161409715993491,
      "grad_norm": 3.078125,
      "learning_rate": 4.842811440846654e-05,
      "loss": 0.8812,
      "step": 8190
    },
    {
      "epoch": 0.11616933325297532,
      "grad_norm": 3.046875,
      "learning_rate": 4.8427334569361537e-05,
      "loss": 0.8445,
      "step": 8192
    },
    {
      "epoch": 0.11619769490660153,
      "grad_norm": 3.25,
      "learning_rate": 4.842655454314085e-05,
      "loss": 0.8815,
      "step": 8194
    },
    {
      "epoch": 0.11622605656022775,
      "grad_norm": 3.09375,
      "learning_rate": 4.842577432981071e-05,
      "loss": 0.8299,
      "step": 8196
    },
    {
      "epoch": 0.11625441821385396,
      "grad_norm": 3.03125,
      "learning_rate": 4.842499392937735e-05,
      "loss": 0.9029,
      "step": 8198
    },
    {
      "epoch": 0.11628277986748017,
      "grad_norm": 2.796875,
      "learning_rate": 4.8424213341847e-05,
      "loss": 0.8334,
      "step": 8200
    },
    {
      "epoch": 0.11631114152110639,
      "grad_norm": 3.390625,
      "learning_rate": 4.842343256722589e-05,
      "loss": 0.8758,
      "step": 8202
    },
    {
      "epoch": 0.1163395031747326,
      "grad_norm": 3.1875,
      "learning_rate": 4.8422651605520265e-05,
      "loss": 0.8164,
      "step": 8204
    },
    {
      "epoch": 0.11636786482835881,
      "grad_norm": 3.0,
      "learning_rate": 4.842187045673636e-05,
      "loss": 0.9418,
      "step": 8206
    },
    {
      "epoch": 0.11639622648198503,
      "grad_norm": 2.796875,
      "learning_rate": 4.842108912088042e-05,
      "loss": 0.8683,
      "step": 8208
    },
    {
      "epoch": 0.11642458813561124,
      "grad_norm": 3.171875,
      "learning_rate": 4.842030759795867e-05,
      "loss": 0.9266,
      "step": 8210
    },
    {
      "epoch": 0.11645294978923747,
      "grad_norm": 2.84375,
      "learning_rate": 4.841952588797736e-05,
      "loss": 0.8221,
      "step": 8212
    },
    {
      "epoch": 0.11648131144286368,
      "grad_norm": 3.03125,
      "learning_rate": 4.8418743990942736e-05,
      "loss": 0.8566,
      "step": 8214
    },
    {
      "epoch": 0.1165096730964899,
      "grad_norm": 3.09375,
      "learning_rate": 4.8417961906861056e-05,
      "loss": 0.9089,
      "step": 8216
    },
    {
      "epoch": 0.1165380347501161,
      "grad_norm": 3.15625,
      "learning_rate": 4.841717963573853e-05,
      "loss": 0.8891,
      "step": 8218
    },
    {
      "epoch": 0.11656639640374232,
      "grad_norm": 2.859375,
      "learning_rate": 4.841639717758144e-05,
      "loss": 0.8544,
      "step": 8220
    },
    {
      "epoch": 0.11659475805736853,
      "grad_norm": 2.71875,
      "learning_rate": 4.841561453239602e-05,
      "loss": 0.8665,
      "step": 8222
    },
    {
      "epoch": 0.11662311971099475,
      "grad_norm": 3.0,
      "learning_rate": 4.841483170018853e-05,
      "loss": 0.8294,
      "step": 8224
    },
    {
      "epoch": 0.11665148136462096,
      "grad_norm": 3.53125,
      "learning_rate": 4.841404868096522e-05,
      "loss": 0.8711,
      "step": 8226
    },
    {
      "epoch": 0.11667984301824717,
      "grad_norm": 2.828125,
      "learning_rate": 4.841326547473233e-05,
      "loss": 0.8554,
      "step": 8228
    },
    {
      "epoch": 0.1167082046718734,
      "grad_norm": 3.203125,
      "learning_rate": 4.841248208149614e-05,
      "loss": 0.8348,
      "step": 8230
    },
    {
      "epoch": 0.11673656632549961,
      "grad_norm": 3.015625,
      "learning_rate": 4.841169850126288e-05,
      "loss": 0.8496,
      "step": 8232
    },
    {
      "epoch": 0.11676492797912583,
      "grad_norm": 3.625,
      "learning_rate": 4.841091473403884e-05,
      "loss": 0.8694,
      "step": 8234
    },
    {
      "epoch": 0.11679328963275204,
      "grad_norm": 3.0625,
      "learning_rate": 4.841013077983025e-05,
      "loss": 0.9311,
      "step": 8236
    },
    {
      "epoch": 0.11682165128637825,
      "grad_norm": 3.109375,
      "learning_rate": 4.8409346638643384e-05,
      "loss": 0.8777,
      "step": 8238
    },
    {
      "epoch": 0.11685001294000447,
      "grad_norm": 3.1875,
      "learning_rate": 4.8408562310484506e-05,
      "loss": 0.8664,
      "step": 8240
    },
    {
      "epoch": 0.11687837459363068,
      "grad_norm": 2.890625,
      "learning_rate": 4.840777779535988e-05,
      "loss": 0.8774,
      "step": 8242
    },
    {
      "epoch": 0.11690673624725689,
      "grad_norm": 3.328125,
      "learning_rate": 4.8406993093275776e-05,
      "loss": 0.8839,
      "step": 8244
    },
    {
      "epoch": 0.1169350979008831,
      "grad_norm": 3.171875,
      "learning_rate": 4.8406208204238454e-05,
      "loss": 0.8533,
      "step": 8246
    },
    {
      "epoch": 0.11696345955450932,
      "grad_norm": 2.8125,
      "learning_rate": 4.840542312825418e-05,
      "loss": 0.881,
      "step": 8248
    },
    {
      "epoch": 0.11699182120813555,
      "grad_norm": 3.046875,
      "learning_rate": 4.840463786532924e-05,
      "loss": 0.8626,
      "step": 8250
    },
    {
      "epoch": 0.11702018286176176,
      "grad_norm": 3.09375,
      "learning_rate": 4.8403852415469885e-05,
      "loss": 0.8745,
      "step": 8252
    },
    {
      "epoch": 0.11704854451538797,
      "grad_norm": 3.171875,
      "learning_rate": 4.840306677868241e-05,
      "loss": 0.8693,
      "step": 8254
    },
    {
      "epoch": 0.11707690616901419,
      "grad_norm": 3.125,
      "learning_rate": 4.840228095497308e-05,
      "loss": 0.9052,
      "step": 8256
    },
    {
      "epoch": 0.1171052678226404,
      "grad_norm": 2.921875,
      "learning_rate": 4.840149494434816e-05,
      "loss": 0.8575,
      "step": 8258
    },
    {
      "epoch": 0.11713362947626661,
      "grad_norm": 3.25,
      "learning_rate": 4.840070874681395e-05,
      "loss": 0.8639,
      "step": 8260
    },
    {
      "epoch": 0.11716199112989283,
      "grad_norm": 3.34375,
      "learning_rate": 4.839992236237672e-05,
      "loss": 0.8783,
      "step": 8262
    },
    {
      "epoch": 0.11719035278351904,
      "grad_norm": 3.015625,
      "learning_rate": 4.839913579104274e-05,
      "loss": 0.851,
      "step": 8264
    },
    {
      "epoch": 0.11721871443714525,
      "grad_norm": 2.9375,
      "learning_rate": 4.83983490328183e-05,
      "loss": 0.9167,
      "step": 8266
    },
    {
      "epoch": 0.11724707609077148,
      "grad_norm": 3.328125,
      "learning_rate": 4.8397562087709694e-05,
      "loss": 0.8287,
      "step": 8268
    },
    {
      "epoch": 0.11727543774439769,
      "grad_norm": 3.28125,
      "learning_rate": 4.83967749557232e-05,
      "loss": 0.8643,
      "step": 8270
    },
    {
      "epoch": 0.1173037993980239,
      "grad_norm": 3.3125,
      "learning_rate": 4.839598763686509e-05,
      "loss": 0.8385,
      "step": 8272
    },
    {
      "epoch": 0.11733216105165012,
      "grad_norm": 2.9375,
      "learning_rate": 4.839520013114168e-05,
      "loss": 0.9057,
      "step": 8274
    },
    {
      "epoch": 0.11736052270527633,
      "grad_norm": 3.234375,
      "learning_rate": 4.839441243855924e-05,
      "loss": 0.8957,
      "step": 8276
    },
    {
      "epoch": 0.11738888435890255,
      "grad_norm": 2.90625,
      "learning_rate": 4.839362455912407e-05,
      "loss": 0.8609,
      "step": 8278
    },
    {
      "epoch": 0.11741724601252876,
      "grad_norm": 3.21875,
      "learning_rate": 4.839283649284246e-05,
      "loss": 0.8467,
      "step": 8280
    },
    {
      "epoch": 0.11744560766615497,
      "grad_norm": 3.0625,
      "learning_rate": 4.8392048239720703e-05,
      "loss": 0.8524,
      "step": 8282
    },
    {
      "epoch": 0.11747396931978119,
      "grad_norm": 2.875,
      "learning_rate": 4.83912597997651e-05,
      "loss": 0.8375,
      "step": 8284
    },
    {
      "epoch": 0.1175023309734074,
      "grad_norm": 3.0,
      "learning_rate": 4.839047117298195e-05,
      "loss": 0.8778,
      "step": 8286
    },
    {
      "epoch": 0.11753069262703363,
      "grad_norm": 3.03125,
      "learning_rate": 4.838968235937753e-05,
      "loss": 0.8063,
      "step": 8288
    },
    {
      "epoch": 0.11755905428065984,
      "grad_norm": 2.9375,
      "learning_rate": 4.8388893358958165e-05,
      "loss": 0.8759,
      "step": 8290
    },
    {
      "epoch": 0.11758741593428605,
      "grad_norm": 3.015625,
      "learning_rate": 4.838810417173015e-05,
      "loss": 0.8938,
      "step": 8292
    },
    {
      "epoch": 0.11761577758791227,
      "grad_norm": 3.078125,
      "learning_rate": 4.838731479769979e-05,
      "loss": 0.8824,
      "step": 8294
    },
    {
      "epoch": 0.11764413924153848,
      "grad_norm": 3.3125,
      "learning_rate": 4.8386525236873384e-05,
      "loss": 0.9117,
      "step": 8296
    },
    {
      "epoch": 0.11767250089516469,
      "grad_norm": 2.828125,
      "learning_rate": 4.838573548925724e-05,
      "loss": 0.8755,
      "step": 8298
    },
    {
      "epoch": 0.1177008625487909,
      "grad_norm": 3.203125,
      "learning_rate": 4.838494555485767e-05,
      "loss": 0.8574,
      "step": 8300
    },
    {
      "epoch": 0.11772922420241712,
      "grad_norm": 2.96875,
      "learning_rate": 4.838415543368098e-05,
      "loss": 0.8887,
      "step": 8302
    },
    {
      "epoch": 0.11775758585604333,
      "grad_norm": 2.609375,
      "learning_rate": 4.838336512573347e-05,
      "loss": 0.8382,
      "step": 8304
    },
    {
      "epoch": 0.11778594750966954,
      "grad_norm": 3.015625,
      "learning_rate": 4.838257463102147e-05,
      "loss": 0.8615,
      "step": 8306
    },
    {
      "epoch": 0.11781430916329577,
      "grad_norm": 3.375,
      "learning_rate": 4.8381783949551295e-05,
      "loss": 0.8565,
      "step": 8308
    },
    {
      "epoch": 0.11784267081692198,
      "grad_norm": 3.015625,
      "learning_rate": 4.838099308132924e-05,
      "loss": 0.8639,
      "step": 8310
    },
    {
      "epoch": 0.1178710324705482,
      "grad_norm": 3.265625,
      "learning_rate": 4.8380202026361644e-05,
      "loss": 0.8429,
      "step": 8312
    },
    {
      "epoch": 0.11789939412417441,
      "grad_norm": 3.125,
      "learning_rate": 4.837941078465481e-05,
      "loss": 0.8475,
      "step": 8314
    },
    {
      "epoch": 0.11792775577780062,
      "grad_norm": 2.84375,
      "learning_rate": 4.837861935621506e-05,
      "loss": 0.8565,
      "step": 8316
    },
    {
      "epoch": 0.11795611743142684,
      "grad_norm": 3.015625,
      "learning_rate": 4.837782774104872e-05,
      "loss": 0.8642,
      "step": 8318
    },
    {
      "epoch": 0.11798447908505305,
      "grad_norm": 3.046875,
      "learning_rate": 4.837703593916211e-05,
      "loss": 0.8899,
      "step": 8320
    },
    {
      "epoch": 0.11801284073867926,
      "grad_norm": 3.109375,
      "learning_rate": 4.837624395056155e-05,
      "loss": 0.8407,
      "step": 8322
    },
    {
      "epoch": 0.11804120239230548,
      "grad_norm": 3.203125,
      "learning_rate": 4.837545177525338e-05,
      "loss": 0.8545,
      "step": 8324
    },
    {
      "epoch": 0.1180695640459317,
      "grad_norm": 2.765625,
      "learning_rate": 4.837465941324392e-05,
      "loss": 0.8881,
      "step": 8326
    },
    {
      "epoch": 0.11809792569955792,
      "grad_norm": 3.484375,
      "learning_rate": 4.837386686453949e-05,
      "loss": 0.9075,
      "step": 8328
    },
    {
      "epoch": 0.11812628735318413,
      "grad_norm": 3.140625,
      "learning_rate": 4.8373074129146424e-05,
      "loss": 0.8665,
      "step": 8330
    },
    {
      "epoch": 0.11815464900681034,
      "grad_norm": 3.921875,
      "learning_rate": 4.837228120707106e-05,
      "loss": 0.8682,
      "step": 8332
    },
    {
      "epoch": 0.11818301066043656,
      "grad_norm": 2.765625,
      "learning_rate": 4.837148809831972e-05,
      "loss": 0.8887,
      "step": 8334
    },
    {
      "epoch": 0.11821137231406277,
      "grad_norm": 2.765625,
      "learning_rate": 4.8370694802898754e-05,
      "loss": 0.8734,
      "step": 8336
    },
    {
      "epoch": 0.11823973396768898,
      "grad_norm": 3.140625,
      "learning_rate": 4.836990132081448e-05,
      "loss": 0.9076,
      "step": 8338
    },
    {
      "epoch": 0.1182680956213152,
      "grad_norm": 2.953125,
      "learning_rate": 4.8369107652073254e-05,
      "loss": 0.9139,
      "step": 8340
    },
    {
      "epoch": 0.11829645727494141,
      "grad_norm": 3.015625,
      "learning_rate": 4.8368313796681404e-05,
      "loss": 0.9144,
      "step": 8342
    },
    {
      "epoch": 0.11832481892856762,
      "grad_norm": 3.171875,
      "learning_rate": 4.8367519754645274e-05,
      "loss": 0.8735,
      "step": 8344
    },
    {
      "epoch": 0.11835318058219385,
      "grad_norm": 3.1875,
      "learning_rate": 4.83667255259712e-05,
      "loss": 0.8377,
      "step": 8346
    },
    {
      "epoch": 0.11838154223582006,
      "grad_norm": 2.8125,
      "learning_rate": 4.8365931110665534e-05,
      "loss": 0.8461,
      "step": 8348
    },
    {
      "epoch": 0.11840990388944628,
      "grad_norm": 2.921875,
      "learning_rate": 4.836513650873462e-05,
      "loss": 0.8697,
      "step": 8350
    },
    {
      "epoch": 0.11843826554307249,
      "grad_norm": 3.28125,
      "learning_rate": 4.83643417201848e-05,
      "loss": 0.8493,
      "step": 8352
    },
    {
      "epoch": 0.1184666271966987,
      "grad_norm": 3.125,
      "learning_rate": 4.836354674502242e-05,
      "loss": 0.8517,
      "step": 8354
    },
    {
      "epoch": 0.11849498885032492,
      "grad_norm": 2.765625,
      "learning_rate": 4.8362751583253846e-05,
      "loss": 0.8306,
      "step": 8356
    },
    {
      "epoch": 0.11852335050395113,
      "grad_norm": 3.515625,
      "learning_rate": 4.836195623488541e-05,
      "loss": 0.864,
      "step": 8358
    },
    {
      "epoch": 0.11855171215757734,
      "grad_norm": 2.953125,
      "learning_rate": 4.836116069992347e-05,
      "loss": 0.8596,
      "step": 8360
    },
    {
      "epoch": 0.11858007381120356,
      "grad_norm": 3.453125,
      "learning_rate": 4.8360364978374384e-05,
      "loss": 0.9108,
      "step": 8362
    },
    {
      "epoch": 0.11860843546482977,
      "grad_norm": 3.203125,
      "learning_rate": 4.83595690702445e-05,
      "loss": 0.8731,
      "step": 8364
    },
    {
      "epoch": 0.118636797118456,
      "grad_norm": 3.25,
      "learning_rate": 4.8358772975540184e-05,
      "loss": 0.8666,
      "step": 8366
    },
    {
      "epoch": 0.11866515877208221,
      "grad_norm": 2.78125,
      "learning_rate": 4.8357976694267785e-05,
      "loss": 0.8623,
      "step": 8368
    },
    {
      "epoch": 0.11869352042570842,
      "grad_norm": 2.890625,
      "learning_rate": 4.8357180226433676e-05,
      "loss": 0.8372,
      "step": 8370
    },
    {
      "epoch": 0.11872188207933464,
      "grad_norm": 3.109375,
      "learning_rate": 4.8356383572044206e-05,
      "loss": 0.8686,
      "step": 8372
    },
    {
      "epoch": 0.11875024373296085,
      "grad_norm": 2.96875,
      "learning_rate": 4.835558673110574e-05,
      "loss": 0.8271,
      "step": 8374
    },
    {
      "epoch": 0.11877860538658706,
      "grad_norm": 2.796875,
      "learning_rate": 4.835478970362465e-05,
      "loss": 0.8645,
      "step": 8376
    },
    {
      "epoch": 0.11880696704021328,
      "grad_norm": 3.25,
      "learning_rate": 4.83539924896073e-05,
      "loss": 0.8684,
      "step": 8378
    },
    {
      "epoch": 0.11883532869383949,
      "grad_norm": 2.84375,
      "learning_rate": 4.835319508906004e-05,
      "loss": 0.8739,
      "step": 8380
    },
    {
      "epoch": 0.1188636903474657,
      "grad_norm": 3.515625,
      "learning_rate": 4.8352397501989265e-05,
      "loss": 0.8647,
      "step": 8382
    },
    {
      "epoch": 0.11889205200109193,
      "grad_norm": 2.875,
      "learning_rate": 4.835159972840133e-05,
      "loss": 0.8605,
      "step": 8384
    },
    {
      "epoch": 0.11892041365471814,
      "grad_norm": 2.734375,
      "learning_rate": 4.835080176830261e-05,
      "loss": 0.8833,
      "step": 8386
    },
    {
      "epoch": 0.11894877530834436,
      "grad_norm": 3.078125,
      "learning_rate": 4.835000362169949e-05,
      "loss": 0.8472,
      "step": 8388
    },
    {
      "epoch": 0.11897713696197057,
      "grad_norm": 2.96875,
      "learning_rate": 4.834920528859832e-05,
      "loss": 0.8665,
      "step": 8390
    },
    {
      "epoch": 0.11900549861559678,
      "grad_norm": 2.875,
      "learning_rate": 4.8348406769005494e-05,
      "loss": 0.8529,
      "step": 8392
    },
    {
      "epoch": 0.119033860269223,
      "grad_norm": 3.203125,
      "learning_rate": 4.8347608062927385e-05,
      "loss": 0.8668,
      "step": 8394
    },
    {
      "epoch": 0.11906222192284921,
      "grad_norm": 3.375,
      "learning_rate": 4.834680917037037e-05,
      "loss": 0.8712,
      "step": 8396
    },
    {
      "epoch": 0.11909058357647542,
      "grad_norm": 3.25,
      "learning_rate": 4.8346010091340836e-05,
      "loss": 0.8391,
      "step": 8398
    },
    {
      "epoch": 0.11911894523010164,
      "grad_norm": 2.96875,
      "learning_rate": 4.834521082584516e-05,
      "loss": 0.8695,
      "step": 8400
    },
    {
      "epoch": 0.11914730688372785,
      "grad_norm": 2.890625,
      "learning_rate": 4.834441137388973e-05,
      "loss": 0.8445,
      "step": 8402
    },
    {
      "epoch": 0.11917566853735408,
      "grad_norm": 3.03125,
      "learning_rate": 4.834361173548093e-05,
      "loss": 0.8103,
      "step": 8404
    },
    {
      "epoch": 0.11920403019098029,
      "grad_norm": 3.03125,
      "learning_rate": 4.834281191062515e-05,
      "loss": 0.8743,
      "step": 8406
    },
    {
      "epoch": 0.1192323918446065,
      "grad_norm": 2.71875,
      "learning_rate": 4.834201189932877e-05,
      "loss": 0.8514,
      "step": 8408
    },
    {
      "epoch": 0.11926075349823272,
      "grad_norm": 3.125,
      "learning_rate": 4.834121170159819e-05,
      "loss": 0.826,
      "step": 8410
    },
    {
      "epoch": 0.11928911515185893,
      "grad_norm": 3.046875,
      "learning_rate": 4.8340411317439785e-05,
      "loss": 0.864,
      "step": 8412
    },
    {
      "epoch": 0.11931747680548514,
      "grad_norm": 3.34375,
      "learning_rate": 4.8339610746859966e-05,
      "loss": 0.8547,
      "step": 8414
    },
    {
      "epoch": 0.11934583845911136,
      "grad_norm": 3.5625,
      "learning_rate": 4.833880998986511e-05,
      "loss": 0.9125,
      "step": 8416
    },
    {
      "epoch": 0.11937420011273757,
      "grad_norm": 2.78125,
      "learning_rate": 4.833800904646163e-05,
      "loss": 0.8258,
      "step": 8418
    },
    {
      "epoch": 0.11940256176636378,
      "grad_norm": 2.875,
      "learning_rate": 4.833720791665591e-05,
      "loss": 0.8578,
      "step": 8420
    },
    {
      "epoch": 0.11943092341999001,
      "grad_norm": 2.921875,
      "learning_rate": 4.833640660045436e-05,
      "loss": 0.8859,
      "step": 8422
    },
    {
      "epoch": 0.11945928507361622,
      "grad_norm": 3.078125,
      "learning_rate": 4.833560509786337e-05,
      "loss": 0.8563,
      "step": 8424
    },
    {
      "epoch": 0.11948764672724244,
      "grad_norm": 2.96875,
      "learning_rate": 4.8334803408889345e-05,
      "loss": 0.8291,
      "step": 8426
    },
    {
      "epoch": 0.11951600838086865,
      "grad_norm": 3.125,
      "learning_rate": 4.8334001533538685e-05,
      "loss": 0.9,
      "step": 8428
    },
    {
      "epoch": 0.11954437003449486,
      "grad_norm": 3.484375,
      "learning_rate": 4.83331994718178e-05,
      "loss": 0.8944,
      "step": 8430
    },
    {
      "epoch": 0.11957273168812108,
      "grad_norm": 3.421875,
      "learning_rate": 4.8332397223733104e-05,
      "loss": 0.9166,
      "step": 8432
    },
    {
      "epoch": 0.11960109334174729,
      "grad_norm": 2.890625,
      "learning_rate": 4.833159478929099e-05,
      "loss": 0.8563,
      "step": 8434
    },
    {
      "epoch": 0.1196294549953735,
      "grad_norm": 3.109375,
      "learning_rate": 4.833079216849787e-05,
      "loss": 0.8793,
      "step": 8436
    },
    {
      "epoch": 0.11965781664899972,
      "grad_norm": 3.15625,
      "learning_rate": 4.8329989361360155e-05,
      "loss": 0.8793,
      "step": 8438
    },
    {
      "epoch": 0.11968617830262593,
      "grad_norm": 3.015625,
      "learning_rate": 4.832918636788426e-05,
      "loss": 0.8855,
      "step": 8440
    },
    {
      "epoch": 0.11971453995625216,
      "grad_norm": 3.234375,
      "learning_rate": 4.8328383188076595e-05,
      "loss": 0.8766,
      "step": 8442
    },
    {
      "epoch": 0.11974290160987837,
      "grad_norm": 3.046875,
      "learning_rate": 4.832757982194359e-05,
      "loss": 0.8859,
      "step": 8444
    },
    {
      "epoch": 0.11977126326350458,
      "grad_norm": 2.96875,
      "learning_rate": 4.832677626949164e-05,
      "loss": 0.8623,
      "step": 8446
    },
    {
      "epoch": 0.1197996249171308,
      "grad_norm": 2.75,
      "learning_rate": 4.832597253072717e-05,
      "loss": 0.8329,
      "step": 8448
    },
    {
      "epoch": 0.11982798657075701,
      "grad_norm": 3.15625,
      "learning_rate": 4.83251686056566e-05,
      "loss": 0.8863,
      "step": 8450
    },
    {
      "epoch": 0.11985634822438322,
      "grad_norm": 3.046875,
      "learning_rate": 4.8324364494286364e-05,
      "loss": 0.8871,
      "step": 8452
    },
    {
      "epoch": 0.11988470987800944,
      "grad_norm": 2.984375,
      "learning_rate": 4.8323560196622866e-05,
      "loss": 0.8744,
      "step": 8454
    },
    {
      "epoch": 0.11991307153163565,
      "grad_norm": 3.3125,
      "learning_rate": 4.832275571267253e-05,
      "loss": 0.8234,
      "step": 8456
    },
    {
      "epoch": 0.11994143318526186,
      "grad_norm": 2.921875,
      "learning_rate": 4.83219510424418e-05,
      "loss": 0.8324,
      "step": 8458
    },
    {
      "epoch": 0.11996979483888807,
      "grad_norm": 3.5625,
      "learning_rate": 4.832114618593708e-05,
      "loss": 0.8861,
      "step": 8460
    },
    {
      "epoch": 0.1199981564925143,
      "grad_norm": 3.4375,
      "learning_rate": 4.8320341143164815e-05,
      "loss": 0.8909,
      "step": 8462
    },
    {
      "epoch": 0.12002651814614052,
      "grad_norm": 3.03125,
      "learning_rate": 4.8319535914131434e-05,
      "loss": 0.8721,
      "step": 8464
    },
    {
      "epoch": 0.12005487979976673,
      "grad_norm": 3.84375,
      "learning_rate": 4.831873049884336e-05,
      "loss": 0.824,
      "step": 8466
    },
    {
      "epoch": 0.12008324145339294,
      "grad_norm": 3.0625,
      "learning_rate": 4.8317924897307035e-05,
      "loss": 0.8189,
      "step": 8468
    },
    {
      "epoch": 0.12011160310701915,
      "grad_norm": 3.234375,
      "learning_rate": 4.8317119109528886e-05,
      "loss": 0.8414,
      "step": 8470
    },
    {
      "epoch": 0.12013996476064537,
      "grad_norm": 3.234375,
      "learning_rate": 4.8316313135515343e-05,
      "loss": 0.8718,
      "step": 8472
    },
    {
      "epoch": 0.12016832641427158,
      "grad_norm": 3.328125,
      "learning_rate": 4.831550697527286e-05,
      "loss": 0.8493,
      "step": 8474
    },
    {
      "epoch": 0.1201966880678978,
      "grad_norm": 3.328125,
      "learning_rate": 4.831470062880787e-05,
      "loss": 0.8701,
      "step": 8476
    },
    {
      "epoch": 0.12022504972152401,
      "grad_norm": 3.296875,
      "learning_rate": 4.831389409612681e-05,
      "loss": 0.8936,
      "step": 8478
    },
    {
      "epoch": 0.12025341137515023,
      "grad_norm": 3.515625,
      "learning_rate": 4.831308737723612e-05,
      "loss": 0.8974,
      "step": 8480
    },
    {
      "epoch": 0.12028177302877645,
      "grad_norm": 3.125,
      "learning_rate": 4.831228047214224e-05,
      "loss": 0.8711,
      "step": 8482
    },
    {
      "epoch": 0.12031013468240266,
      "grad_norm": 3.15625,
      "learning_rate": 4.8311473380851633e-05,
      "loss": 0.9041,
      "step": 8484
    },
    {
      "epoch": 0.12033849633602887,
      "grad_norm": 2.921875,
      "learning_rate": 4.831066610337073e-05,
      "loss": 0.8245,
      "step": 8486
    },
    {
      "epoch": 0.12036685798965509,
      "grad_norm": 2.84375,
      "learning_rate": 4.8309858639705975e-05,
      "loss": 0.8002,
      "step": 8488
    },
    {
      "epoch": 0.1203952196432813,
      "grad_norm": 3.078125,
      "learning_rate": 4.8309050989863835e-05,
      "loss": 0.8904,
      "step": 8490
    },
    {
      "epoch": 0.12042358129690751,
      "grad_norm": 3.1875,
      "learning_rate": 4.830824315385074e-05,
      "loss": 0.8781,
      "step": 8492
    },
    {
      "epoch": 0.12045194295053373,
      "grad_norm": 2.90625,
      "learning_rate": 4.8307435131673156e-05,
      "loss": 0.8535,
      "step": 8494
    },
    {
      "epoch": 0.12048030460415994,
      "grad_norm": 3.296875,
      "learning_rate": 4.830662692333753e-05,
      "loss": 0.8917,
      "step": 8496
    },
    {
      "epoch": 0.12050866625778615,
      "grad_norm": 3.125,
      "learning_rate": 4.830581852885033e-05,
      "loss": 0.9012,
      "step": 8498
    },
    {
      "epoch": 0.12053702791141238,
      "grad_norm": 3.40625,
      "learning_rate": 4.8305009948217996e-05,
      "loss": 0.868,
      "step": 8500
    },
    {
      "epoch": 0.1205653895650386,
      "grad_norm": 3.0,
      "learning_rate": 4.8304201181446994e-05,
      "loss": 0.8877,
      "step": 8502
    },
    {
      "epoch": 0.12059375121866481,
      "grad_norm": 3.0625,
      "learning_rate": 4.8303392228543774e-05,
      "loss": 0.8987,
      "step": 8504
    },
    {
      "epoch": 0.12062211287229102,
      "grad_norm": 3.078125,
      "learning_rate": 4.830258308951482e-05,
      "loss": 0.8723,
      "step": 8506
    },
    {
      "epoch": 0.12065047452591723,
      "grad_norm": 2.671875,
      "learning_rate": 4.8301773764366575e-05,
      "loss": 0.8401,
      "step": 8508
    },
    {
      "epoch": 0.12067883617954345,
      "grad_norm": 3.0,
      "learning_rate": 4.8300964253105505e-05,
      "loss": 0.8818,
      "step": 8510
    },
    {
      "epoch": 0.12070719783316966,
      "grad_norm": 2.84375,
      "learning_rate": 4.830015455573809e-05,
      "loss": 0.8766,
      "step": 8512
    },
    {
      "epoch": 0.12073555948679587,
      "grad_norm": 3.203125,
      "learning_rate": 4.829934467227077e-05,
      "loss": 0.8667,
      "step": 8514
    },
    {
      "epoch": 0.12076392114042209,
      "grad_norm": 2.984375,
      "learning_rate": 4.8298534602710046e-05,
      "loss": 0.9016,
      "step": 8516
    },
    {
      "epoch": 0.1207922827940483,
      "grad_norm": 3.203125,
      "learning_rate": 4.829772434706236e-05,
      "loss": 0.8896,
      "step": 8518
    },
    {
      "epoch": 0.12082064444767453,
      "grad_norm": 2.96875,
      "learning_rate": 4.829691390533421e-05,
      "loss": 0.8687,
      "step": 8520
    },
    {
      "epoch": 0.12084900610130074,
      "grad_norm": 2.8125,
      "learning_rate": 4.829610327753204e-05,
      "loss": 0.8608,
      "step": 8522
    },
    {
      "epoch": 0.12087736775492695,
      "grad_norm": 2.96875,
      "learning_rate": 4.829529246366234e-05,
      "loss": 0.8652,
      "step": 8524
    },
    {
      "epoch": 0.12090572940855317,
      "grad_norm": 3.203125,
      "learning_rate": 4.82944814637316e-05,
      "loss": 0.8385,
      "step": 8526
    },
    {
      "epoch": 0.12093409106217938,
      "grad_norm": 2.984375,
      "learning_rate": 4.829367027774627e-05,
      "loss": 0.8317,
      "step": 8528
    },
    {
      "epoch": 0.1209624527158056,
      "grad_norm": 3.140625,
      "learning_rate": 4.829285890571285e-05,
      "loss": 0.842,
      "step": 8530
    },
    {
      "epoch": 0.1209908143694318,
      "grad_norm": 3.046875,
      "learning_rate": 4.82920473476378e-05,
      "loss": 0.7847,
      "step": 8532
    },
    {
      "epoch": 0.12101917602305802,
      "grad_norm": 3.25,
      "learning_rate": 4.8291235603527616e-05,
      "loss": 0.8997,
      "step": 8534
    },
    {
      "epoch": 0.12104753767668423,
      "grad_norm": 3.078125,
      "learning_rate": 4.8290423673388785e-05,
      "loss": 0.8754,
      "step": 8536
    },
    {
      "epoch": 0.12107589933031046,
      "grad_norm": 3.140625,
      "learning_rate": 4.828961155722779e-05,
      "loss": 0.8678,
      "step": 8538
    },
    {
      "epoch": 0.12110426098393667,
      "grad_norm": 3.0625,
      "learning_rate": 4.828879925505111e-05,
      "loss": 0.8077,
      "step": 8540
    },
    {
      "epoch": 0.12113262263756289,
      "grad_norm": 2.890625,
      "learning_rate": 4.828798676686524e-05,
      "loss": 0.8912,
      "step": 8542
    },
    {
      "epoch": 0.1211609842911891,
      "grad_norm": 2.90625,
      "learning_rate": 4.828717409267666e-05,
      "loss": 0.9141,
      "step": 8544
    },
    {
      "epoch": 0.12118934594481531,
      "grad_norm": 3.28125,
      "learning_rate": 4.8286361232491874e-05,
      "loss": 0.8847,
      "step": 8546
    },
    {
      "epoch": 0.12121770759844153,
      "grad_norm": 2.921875,
      "learning_rate": 4.828554818631737e-05,
      "loss": 0.8409,
      "step": 8548
    },
    {
      "epoch": 0.12124606925206774,
      "grad_norm": 3.28125,
      "learning_rate": 4.8284734954159625e-05,
      "loss": 0.8453,
      "step": 8550
    },
    {
      "epoch": 0.12127443090569395,
      "grad_norm": 3.15625,
      "learning_rate": 4.828392153602516e-05,
      "loss": 0.909,
      "step": 8552
    },
    {
      "epoch": 0.12130279255932017,
      "grad_norm": 3.453125,
      "learning_rate": 4.828310793192046e-05,
      "loss": 0.8885,
      "step": 8554
    },
    {
      "epoch": 0.12133115421294638,
      "grad_norm": 2.9375,
      "learning_rate": 4.8282294141852025e-05,
      "loss": 0.8568,
      "step": 8556
    },
    {
      "epoch": 0.1213595158665726,
      "grad_norm": 2.96875,
      "learning_rate": 4.828148016582635e-05,
      "loss": 0.8679,
      "step": 8558
    },
    {
      "epoch": 0.12138787752019882,
      "grad_norm": 3.0625,
      "learning_rate": 4.828066600384994e-05,
      "loss": 0.8932,
      "step": 8560
    },
    {
      "epoch": 0.12141623917382503,
      "grad_norm": 3.21875,
      "learning_rate": 4.8279851655929295e-05,
      "loss": 0.8862,
      "step": 8562
    },
    {
      "epoch": 0.12144460082745125,
      "grad_norm": 3.6875,
      "learning_rate": 4.827903712207094e-05,
      "loss": 0.9029,
      "step": 8564
    },
    {
      "epoch": 0.12147296248107746,
      "grad_norm": 3.203125,
      "learning_rate": 4.8278222402281346e-05,
      "loss": 0.8556,
      "step": 8566
    },
    {
      "epoch": 0.12150132413470367,
      "grad_norm": 2.890625,
      "learning_rate": 4.827740749656704e-05,
      "loss": 0.8117,
      "step": 8568
    },
    {
      "epoch": 0.12152968578832989,
      "grad_norm": 2.9375,
      "learning_rate": 4.8276592404934525e-05,
      "loss": 0.8179,
      "step": 8570
    },
    {
      "epoch": 0.1215580474419561,
      "grad_norm": 2.921875,
      "learning_rate": 4.827577712739031e-05,
      "loss": 0.8246,
      "step": 8572
    },
    {
      "epoch": 0.12158640909558231,
      "grad_norm": 3.09375,
      "learning_rate": 4.827496166394092e-05,
      "loss": 0.8809,
      "step": 8574
    },
    {
      "epoch": 0.12161477074920854,
      "grad_norm": 3.1875,
      "learning_rate": 4.827414601459286e-05,
      "loss": 0.8279,
      "step": 8576
    },
    {
      "epoch": 0.12164313240283475,
      "grad_norm": 3.53125,
      "learning_rate": 4.827333017935264e-05,
      "loss": 0.8232,
      "step": 8578
    },
    {
      "epoch": 0.12167149405646097,
      "grad_norm": 3.09375,
      "learning_rate": 4.827251415822678e-05,
      "loss": 0.8568,
      "step": 8580
    },
    {
      "epoch": 0.12169985571008718,
      "grad_norm": 2.96875,
      "learning_rate": 4.8271697951221794e-05,
      "loss": 0.8042,
      "step": 8582
    },
    {
      "epoch": 0.12172821736371339,
      "grad_norm": 3.125,
      "learning_rate": 4.8270881558344204e-05,
      "loss": 0.8624,
      "step": 8584
    },
    {
      "epoch": 0.1217565790173396,
      "grad_norm": 2.8125,
      "learning_rate": 4.827006497960054e-05,
      "loss": 0.8578,
      "step": 8586
    },
    {
      "epoch": 0.12178494067096582,
      "grad_norm": 2.953125,
      "learning_rate": 4.8269248214997304e-05,
      "loss": 0.837,
      "step": 8588
    },
    {
      "epoch": 0.12181330232459203,
      "grad_norm": 3.0625,
      "learning_rate": 4.826843126454104e-05,
      "loss": 0.8317,
      "step": 8590
    },
    {
      "epoch": 0.12184166397821825,
      "grad_norm": 3.1875,
      "learning_rate": 4.826761412823825e-05,
      "loss": 0.8624,
      "step": 8592
    },
    {
      "epoch": 0.12187002563184446,
      "grad_norm": 2.96875,
      "learning_rate": 4.8266796806095485e-05,
      "loss": 0.8794,
      "step": 8594
    },
    {
      "epoch": 0.12189838728547069,
      "grad_norm": 3.25,
      "learning_rate": 4.8265979298119254e-05,
      "loss": 0.8948,
      "step": 8596
    },
    {
      "epoch": 0.1219267489390969,
      "grad_norm": 3.265625,
      "learning_rate": 4.8265161604316103e-05,
      "loss": 0.8149,
      "step": 8598
    },
    {
      "epoch": 0.12195511059272311,
      "grad_norm": 3.65625,
      "learning_rate": 4.826434372469255e-05,
      "loss": 0.8483,
      "step": 8600
    },
    {
      "epoch": 0.12198347224634933,
      "grad_norm": 3.078125,
      "learning_rate": 4.826352565925513e-05,
      "loss": 0.8608,
      "step": 8602
    },
    {
      "epoch": 0.12201183389997554,
      "grad_norm": 3.140625,
      "learning_rate": 4.8262707408010384e-05,
      "loss": 0.8456,
      "step": 8604
    },
    {
      "epoch": 0.12204019555360175,
      "grad_norm": 3.34375,
      "learning_rate": 4.826188897096484e-05,
      "loss": 0.8634,
      "step": 8606
    },
    {
      "epoch": 0.12206855720722797,
      "grad_norm": 3.203125,
      "learning_rate": 4.8261070348125035e-05,
      "loss": 0.8311,
      "step": 8608
    },
    {
      "epoch": 0.12209691886085418,
      "grad_norm": 3.5,
      "learning_rate": 4.826025153949751e-05,
      "loss": 0.8895,
      "step": 8610
    },
    {
      "epoch": 0.12212528051448039,
      "grad_norm": 3.40625,
      "learning_rate": 4.825943254508881e-05,
      "loss": 0.9245,
      "step": 8612
    },
    {
      "epoch": 0.1221536421681066,
      "grad_norm": 3.1875,
      "learning_rate": 4.825861336490547e-05,
      "loss": 0.8913,
      "step": 8614
    },
    {
      "epoch": 0.12218200382173283,
      "grad_norm": 3.109375,
      "learning_rate": 4.8257793998954024e-05,
      "loss": 0.8572,
      "step": 8616
    },
    {
      "epoch": 0.12221036547535905,
      "grad_norm": 3.09375,
      "learning_rate": 4.825697444724103e-05,
      "loss": 0.8646,
      "step": 8618
    },
    {
      "epoch": 0.12223872712898526,
      "grad_norm": 2.734375,
      "learning_rate": 4.8256154709773035e-05,
      "loss": 0.8617,
      "step": 8620
    },
    {
      "epoch": 0.12226708878261147,
      "grad_norm": 3.015625,
      "learning_rate": 4.825533478655658e-05,
      "loss": 0.8595,
      "step": 8622
    },
    {
      "epoch": 0.12229545043623768,
      "grad_norm": 2.984375,
      "learning_rate": 4.825451467759822e-05,
      "loss": 0.8369,
      "step": 8624
    },
    {
      "epoch": 0.1223238120898639,
      "grad_norm": 3.234375,
      "learning_rate": 4.825369438290449e-05,
      "loss": 0.8611,
      "step": 8626
    },
    {
      "epoch": 0.12235217374349011,
      "grad_norm": 3.0,
      "learning_rate": 4.825287390248195e-05,
      "loss": 0.8573,
      "step": 8628
    },
    {
      "epoch": 0.12238053539711632,
      "grad_norm": 3.25,
      "learning_rate": 4.825205323633717e-05,
      "loss": 0.8253,
      "step": 8630
    },
    {
      "epoch": 0.12240889705074254,
      "grad_norm": 2.734375,
      "learning_rate": 4.8251232384476675e-05,
      "loss": 0.8948,
      "step": 8632
    },
    {
      "epoch": 0.12243725870436876,
      "grad_norm": 3.8125,
      "learning_rate": 4.8250411346907044e-05,
      "loss": 0.8541,
      "step": 8634
    },
    {
      "epoch": 0.12246562035799498,
      "grad_norm": 3.375,
      "learning_rate": 4.8249590123634825e-05,
      "loss": 0.8583,
      "step": 8636
    },
    {
      "epoch": 0.12249398201162119,
      "grad_norm": 2.953125,
      "learning_rate": 4.8248768714666584e-05,
      "loss": 0.8781,
      "step": 8638
    },
    {
      "epoch": 0.1225223436652474,
      "grad_norm": 3.328125,
      "learning_rate": 4.8247947120008866e-05,
      "loss": 0.8946,
      "step": 8640
    },
    {
      "epoch": 0.12255070531887362,
      "grad_norm": 3.59375,
      "learning_rate": 4.8247125339668244e-05,
      "loss": 0.8876,
      "step": 8642
    },
    {
      "epoch": 0.12257906697249983,
      "grad_norm": 3.09375,
      "learning_rate": 4.824630337365129e-05,
      "loss": 0.8431,
      "step": 8644
    },
    {
      "epoch": 0.12260742862612604,
      "grad_norm": 3.328125,
      "learning_rate": 4.8245481221964556e-05,
      "loss": 0.8514,
      "step": 8646
    },
    {
      "epoch": 0.12263579027975226,
      "grad_norm": 3.15625,
      "learning_rate": 4.8244658884614614e-05,
      "loss": 0.882,
      "step": 8648
    },
    {
      "epoch": 0.12266415193337847,
      "grad_norm": 2.9375,
      "learning_rate": 4.824383636160803e-05,
      "loss": 0.8749,
      "step": 8650
    },
    {
      "epoch": 0.12269251358700468,
      "grad_norm": 3.0625,
      "learning_rate": 4.8243013652951374e-05,
      "loss": 0.8617,
      "step": 8652
    },
    {
      "epoch": 0.12272087524063091,
      "grad_norm": 3.046875,
      "learning_rate": 4.8242190758651215e-05,
      "loss": 0.8744,
      "step": 8654
    },
    {
      "epoch": 0.12274923689425712,
      "grad_norm": 3.0625,
      "learning_rate": 4.8241367678714136e-05,
      "loss": 0.8644,
      "step": 8656
    },
    {
      "epoch": 0.12277759854788334,
      "grad_norm": 2.9375,
      "learning_rate": 4.8240544413146704e-05,
      "loss": 0.8118,
      "step": 8658
    },
    {
      "epoch": 0.12280596020150955,
      "grad_norm": 3.328125,
      "learning_rate": 4.8239720961955484e-05,
      "loss": 0.843,
      "step": 8660
    },
    {
      "epoch": 0.12283432185513576,
      "grad_norm": 3.4375,
      "learning_rate": 4.823889732514707e-05,
      "loss": 0.8804,
      "step": 8662
    },
    {
      "epoch": 0.12286268350876198,
      "grad_norm": 3.015625,
      "learning_rate": 4.8238073502728035e-05,
      "loss": 0.8136,
      "step": 8664
    },
    {
      "epoch": 0.12289104516238819,
      "grad_norm": 2.703125,
      "learning_rate": 4.8237249494704945e-05,
      "loss": 0.8604,
      "step": 8666
    },
    {
      "epoch": 0.1229194068160144,
      "grad_norm": 3.3125,
      "learning_rate": 4.82364253010844e-05,
      "loss": 0.8516,
      "step": 8668
    },
    {
      "epoch": 0.12294776846964062,
      "grad_norm": 3.15625,
      "learning_rate": 4.823560092187298e-05,
      "loss": 0.8706,
      "step": 8670
    },
    {
      "epoch": 0.12297613012326683,
      "grad_norm": 2.890625,
      "learning_rate": 4.823477635707726e-05,
      "loss": 0.8669,
      "step": 8672
    },
    {
      "epoch": 0.12300449177689306,
      "grad_norm": 2.96875,
      "learning_rate": 4.8233951606703834e-05,
      "loss": 0.8203,
      "step": 8674
    },
    {
      "epoch": 0.12303285343051927,
      "grad_norm": 3.296875,
      "learning_rate": 4.823312667075929e-05,
      "loss": 0.8478,
      "step": 8676
    },
    {
      "epoch": 0.12306121508414548,
      "grad_norm": 2.640625,
      "learning_rate": 4.8232301549250205e-05,
      "loss": 0.8653,
      "step": 8678
    },
    {
      "epoch": 0.1230895767377717,
      "grad_norm": 2.875,
      "learning_rate": 4.823147624218318e-05,
      "loss": 0.8418,
      "step": 8680
    },
    {
      "epoch": 0.12311793839139791,
      "grad_norm": 3.15625,
      "learning_rate": 4.823065074956481e-05,
      "loss": 0.8273,
      "step": 8682
    },
    {
      "epoch": 0.12314630004502412,
      "grad_norm": 3.1875,
      "learning_rate": 4.8229825071401684e-05,
      "loss": 0.8528,
      "step": 8684
    },
    {
      "epoch": 0.12317466169865034,
      "grad_norm": 3.203125,
      "learning_rate": 4.822899920770039e-05,
      "loss": 0.924,
      "step": 8686
    },
    {
      "epoch": 0.12320302335227655,
      "grad_norm": 2.90625,
      "learning_rate": 4.822817315846754e-05,
      "loss": 0.8865,
      "step": 8688
    },
    {
      "epoch": 0.12323138500590276,
      "grad_norm": 2.859375,
      "learning_rate": 4.822734692370971e-05,
      "loss": 0.8755,
      "step": 8690
    },
    {
      "epoch": 0.12325974665952899,
      "grad_norm": 2.796875,
      "learning_rate": 4.8226520503433515e-05,
      "loss": 0.812,
      "step": 8692
    },
    {
      "epoch": 0.1232881083131552,
      "grad_norm": 3.359375,
      "learning_rate": 4.8225693897645553e-05,
      "loss": 0.857,
      "step": 8694
    },
    {
      "epoch": 0.12331646996678142,
      "grad_norm": 3.78125,
      "learning_rate": 4.822486710635242e-05,
      "loss": 0.8343,
      "step": 8696
    },
    {
      "epoch": 0.12334483162040763,
      "grad_norm": 3.515625,
      "learning_rate": 4.822404012956073e-05,
      "loss": 0.8484,
      "step": 8698
    },
    {
      "epoch": 0.12337319327403384,
      "grad_norm": 3.1875,
      "learning_rate": 4.8223212967277076e-05,
      "loss": 0.8963,
      "step": 8700
    },
    {
      "epoch": 0.12340155492766006,
      "grad_norm": 3.328125,
      "learning_rate": 4.822238561950808e-05,
      "loss": 0.872,
      "step": 8702
    },
    {
      "epoch": 0.12342991658128627,
      "grad_norm": 2.96875,
      "learning_rate": 4.8221558086260335e-05,
      "loss": 0.8551,
      "step": 8704
    },
    {
      "epoch": 0.12345827823491248,
      "grad_norm": 2.90625,
      "learning_rate": 4.8220730367540457e-05,
      "loss": 0.8356,
      "step": 8706
    },
    {
      "epoch": 0.1234866398885387,
      "grad_norm": 2.828125,
      "learning_rate": 4.8219902463355063e-05,
      "loss": 0.8848,
      "step": 8708
    },
    {
      "epoch": 0.12351500154216491,
      "grad_norm": 3.046875,
      "learning_rate": 4.821907437371075e-05,
      "loss": 0.8333,
      "step": 8710
    },
    {
      "epoch": 0.12354336319579114,
      "grad_norm": 2.9375,
      "learning_rate": 4.821824609861414e-05,
      "loss": 0.8625,
      "step": 8712
    },
    {
      "epoch": 0.12357172484941735,
      "grad_norm": 3.71875,
      "learning_rate": 4.821741763807186e-05,
      "loss": 0.9097,
      "step": 8714
    },
    {
      "epoch": 0.12360008650304356,
      "grad_norm": 3.046875,
      "learning_rate": 4.8216588992090506e-05,
      "loss": 0.8241,
      "step": 8716
    },
    {
      "epoch": 0.12362844815666978,
      "grad_norm": 3.140625,
      "learning_rate": 4.821576016067671e-05,
      "loss": 0.902,
      "step": 8718
    },
    {
      "epoch": 0.12365680981029599,
      "grad_norm": 2.921875,
      "learning_rate": 4.8214931143837094e-05,
      "loss": 0.8378,
      "step": 8720
    },
    {
      "epoch": 0.1236851714639222,
      "grad_norm": 3.0,
      "learning_rate": 4.821410194157827e-05,
      "loss": 0.8269,
      "step": 8722
    },
    {
      "epoch": 0.12371353311754842,
      "grad_norm": 3.28125,
      "learning_rate": 4.8213272553906874e-05,
      "loss": 0.9035,
      "step": 8724
    },
    {
      "epoch": 0.12374189477117463,
      "grad_norm": 2.875,
      "learning_rate": 4.8212442980829514e-05,
      "loss": 0.8323,
      "step": 8726
    },
    {
      "epoch": 0.12377025642480084,
      "grad_norm": 3.09375,
      "learning_rate": 4.821161322235282e-05,
      "loss": 0.8546,
      "step": 8728
    },
    {
      "epoch": 0.12379861807842707,
      "grad_norm": 2.984375,
      "learning_rate": 4.8210783278483436e-05,
      "loss": 0.8694,
      "step": 8730
    },
    {
      "epoch": 0.12382697973205328,
      "grad_norm": 3.78125,
      "learning_rate": 4.8209953149227966e-05,
      "loss": 0.8477,
      "step": 8732
    },
    {
      "epoch": 0.1238553413856795,
      "grad_norm": 2.84375,
      "learning_rate": 4.8209122834593055e-05,
      "loss": 0.8352,
      "step": 8734
    },
    {
      "epoch": 0.12388370303930571,
      "grad_norm": 3.203125,
      "learning_rate": 4.8208292334585335e-05,
      "loss": 0.8596,
      "step": 8736
    },
    {
      "epoch": 0.12391206469293192,
      "grad_norm": 3.125,
      "learning_rate": 4.8207461649211434e-05,
      "loss": 0.8692,
      "step": 8738
    },
    {
      "epoch": 0.12394042634655814,
      "grad_norm": 2.984375,
      "learning_rate": 4.820663077847799e-05,
      "loss": 0.8382,
      "step": 8740
    },
    {
      "epoch": 0.12396878800018435,
      "grad_norm": 3.0625,
      "learning_rate": 4.820579972239163e-05,
      "loss": 0.878,
      "step": 8742
    },
    {
      "epoch": 0.12399714965381056,
      "grad_norm": 3.75,
      "learning_rate": 4.820496848095901e-05,
      "loss": 0.8751,
      "step": 8744
    },
    {
      "epoch": 0.12402551130743678,
      "grad_norm": 2.734375,
      "learning_rate": 4.8204137054186754e-05,
      "loss": 0.8316,
      "step": 8746
    },
    {
      "epoch": 0.12405387296106299,
      "grad_norm": 3.078125,
      "learning_rate": 4.820330544208151e-05,
      "loss": 0.8491,
      "step": 8748
    },
    {
      "epoch": 0.12408223461468922,
      "grad_norm": 3.203125,
      "learning_rate": 4.820247364464991e-05,
      "loss": 0.8777,
      "step": 8750
    },
    {
      "epoch": 0.12411059626831543,
      "grad_norm": 3.5,
      "learning_rate": 4.8201641661898625e-05,
      "loss": 0.9081,
      "step": 8752
    },
    {
      "epoch": 0.12413895792194164,
      "grad_norm": 3.078125,
      "learning_rate": 4.820080949383427e-05,
      "loss": 0.8801,
      "step": 8754
    },
    {
      "epoch": 0.12416731957556786,
      "grad_norm": 3.0625,
      "learning_rate": 4.819997714046349e-05,
      "loss": 0.8725,
      "step": 8756
    },
    {
      "epoch": 0.12419568122919407,
      "grad_norm": 3.140625,
      "learning_rate": 4.819914460179295e-05,
      "loss": 0.8306,
      "step": 8758
    },
    {
      "epoch": 0.12422404288282028,
      "grad_norm": 3.234375,
      "learning_rate": 4.81983118778293e-05,
      "loss": 0.8387,
      "step": 8760
    },
    {
      "epoch": 0.1242524045364465,
      "grad_norm": 3.15625,
      "learning_rate": 4.819747896857919e-05,
      "loss": 0.882,
      "step": 8762
    },
    {
      "epoch": 0.12428076619007271,
      "grad_norm": 2.96875,
      "learning_rate": 4.819664587404926e-05,
      "loss": 0.8897,
      "step": 8764
    },
    {
      "epoch": 0.12430912784369892,
      "grad_norm": 3.0625,
      "learning_rate": 4.819581259424618e-05,
      "loss": 0.8318,
      "step": 8766
    },
    {
      "epoch": 0.12433748949732513,
      "grad_norm": 2.984375,
      "learning_rate": 4.8194979129176595e-05,
      "loss": 0.8593,
      "step": 8768
    },
    {
      "epoch": 0.12436585115095136,
      "grad_norm": 2.59375,
      "learning_rate": 4.819414547884716e-05,
      "loss": 0.8314,
      "step": 8770
    },
    {
      "epoch": 0.12439421280457758,
      "grad_norm": 2.8125,
      "learning_rate": 4.8193311643264543e-05,
      "loss": 0.8472,
      "step": 8772
    },
    {
      "epoch": 0.12442257445820379,
      "grad_norm": 3.03125,
      "learning_rate": 4.819247762243539e-05,
      "loss": 0.8711,
      "step": 8774
    },
    {
      "epoch": 0.12445093611183,
      "grad_norm": 3.21875,
      "learning_rate": 4.819164341636638e-05,
      "loss": 0.8478,
      "step": 8776
    },
    {
      "epoch": 0.12447929776545621,
      "grad_norm": 3.109375,
      "learning_rate": 4.819080902506416e-05,
      "loss": 0.8828,
      "step": 8778
    },
    {
      "epoch": 0.12450765941908243,
      "grad_norm": 2.890625,
      "learning_rate": 4.8189974448535414e-05,
      "loss": 0.8097,
      "step": 8780
    },
    {
      "epoch": 0.12453602107270864,
      "grad_norm": 3.21875,
      "learning_rate": 4.818913968678679e-05,
      "loss": 0.9076,
      "step": 8782
    },
    {
      "epoch": 0.12456438272633485,
      "grad_norm": 2.9375,
      "learning_rate": 4.8188304739824954e-05,
      "loss": 0.8711,
      "step": 8784
    },
    {
      "epoch": 0.12459274437996107,
      "grad_norm": 3.265625,
      "learning_rate": 4.8187469607656585e-05,
      "loss": 0.8879,
      "step": 8786
    },
    {
      "epoch": 0.1246211060335873,
      "grad_norm": 2.90625,
      "learning_rate": 4.818663429028836e-05,
      "loss": 0.8392,
      "step": 8788
    },
    {
      "epoch": 0.12464946768721351,
      "grad_norm": 3.3125,
      "learning_rate": 4.818579878772693e-05,
      "loss": 0.832,
      "step": 8790
    },
    {
      "epoch": 0.12467782934083972,
      "grad_norm": 3.046875,
      "learning_rate": 4.818496309997898e-05,
      "loss": 0.8323,
      "step": 8792
    },
    {
      "epoch": 0.12470619099446593,
      "grad_norm": 2.984375,
      "learning_rate": 4.818412722705118e-05,
      "loss": 0.8139,
      "step": 8794
    },
    {
      "epoch": 0.12473455264809215,
      "grad_norm": 3.0625,
      "learning_rate": 4.818329116895021e-05,
      "loss": 0.9156,
      "step": 8796
    },
    {
      "epoch": 0.12476291430171836,
      "grad_norm": 2.90625,
      "learning_rate": 4.818245492568275e-05,
      "loss": 0.862,
      "step": 8798
    },
    {
      "epoch": 0.12479127595534457,
      "grad_norm": 3.203125,
      "learning_rate": 4.818161849725548e-05,
      "loss": 0.911,
      "step": 8800
    },
    {
      "epoch": 0.12481963760897079,
      "grad_norm": 3.28125,
      "learning_rate": 4.818078188367506e-05,
      "loss": 0.8587,
      "step": 8802
    },
    {
      "epoch": 0.124847999262597,
      "grad_norm": 3.25,
      "learning_rate": 4.817994508494821e-05,
      "loss": 0.9121,
      "step": 8804
    },
    {
      "epoch": 0.12487636091622321,
      "grad_norm": 3.875,
      "learning_rate": 4.817910810108158e-05,
      "loss": 0.915,
      "step": 8806
    },
    {
      "epoch": 0.12490472256984944,
      "grad_norm": 2.84375,
      "learning_rate": 4.817827093208187e-05,
      "loss": 0.8542,
      "step": 8808
    },
    {
      "epoch": 0.12493308422347565,
      "grad_norm": 2.96875,
      "learning_rate": 4.817743357795577e-05,
      "loss": 0.8523,
      "step": 8810
    },
    {
      "epoch": 0.12496144587710187,
      "grad_norm": 3.078125,
      "learning_rate": 4.817659603870995e-05,
      "loss": 0.8904,
      "step": 8812
    },
    {
      "epoch": 0.12498980753072808,
      "grad_norm": 2.953125,
      "learning_rate": 4.8175758314351124e-05,
      "loss": 0.8686,
      "step": 8814
    },
    {
      "epoch": 0.1250181691843543,
      "grad_norm": 2.828125,
      "learning_rate": 4.8174920404885963e-05,
      "loss": 0.8899,
      "step": 8816
    },
    {
      "epoch": 0.1250465308379805,
      "grad_norm": 3.234375,
      "learning_rate": 4.817408231032117e-05,
      "loss": 0.9097,
      "step": 8818
    },
    {
      "epoch": 0.12507489249160672,
      "grad_norm": 3.09375,
      "learning_rate": 4.8173244030663435e-05,
      "loss": 0.9061,
      "step": 8820
    },
    {
      "epoch": 0.12510325414523293,
      "grad_norm": 3.28125,
      "learning_rate": 4.8172405565919456e-05,
      "loss": 0.856,
      "step": 8822
    },
    {
      "epoch": 0.12513161579885915,
      "grad_norm": 3.484375,
      "learning_rate": 4.8171566916095926e-05,
      "loss": 0.8399,
      "step": 8824
    },
    {
      "epoch": 0.12515997745248536,
      "grad_norm": 3.3125,
      "learning_rate": 4.8170728081199546e-05,
      "loss": 0.8352,
      "step": 8826
    },
    {
      "epoch": 0.12518833910611157,
      "grad_norm": 2.859375,
      "learning_rate": 4.816988906123702e-05,
      "loss": 0.8485,
      "step": 8828
    },
    {
      "epoch": 0.1252167007597378,
      "grad_norm": 2.921875,
      "learning_rate": 4.8169049856215035e-05,
      "loss": 0.8281,
      "step": 8830
    },
    {
      "epoch": 0.125245062413364,
      "grad_norm": 2.84375,
      "learning_rate": 4.816821046614031e-05,
      "loss": 0.8731,
      "step": 8832
    },
    {
      "epoch": 0.1252734240669902,
      "grad_norm": 3.0,
      "learning_rate": 4.816737089101954e-05,
      "loss": 0.8252,
      "step": 8834
    },
    {
      "epoch": 0.12530178572061645,
      "grad_norm": 3.515625,
      "learning_rate": 4.816653113085944e-05,
      "loss": 0.9154,
      "step": 8836
    },
    {
      "epoch": 0.12533014737424267,
      "grad_norm": 3.453125,
      "learning_rate": 4.81656911856667e-05,
      "loss": 0.8918,
      "step": 8838
    },
    {
      "epoch": 0.12535850902786888,
      "grad_norm": 3.0625,
      "learning_rate": 4.816485105544805e-05,
      "loss": 0.8523,
      "step": 8840
    },
    {
      "epoch": 0.1253868706814951,
      "grad_norm": 3.015625,
      "learning_rate": 4.8164010740210176e-05,
      "loss": 0.8291,
      "step": 8842
    },
    {
      "epoch": 0.1254152323351213,
      "grad_norm": 2.9375,
      "learning_rate": 4.816317023995981e-05,
      "loss": 0.8592,
      "step": 8844
    },
    {
      "epoch": 0.12544359398874752,
      "grad_norm": 2.90625,
      "learning_rate": 4.816232955470366e-05,
      "loss": 0.8897,
      "step": 8846
    },
    {
      "epoch": 0.12547195564237373,
      "grad_norm": 3.265625,
      "learning_rate": 4.816148868444844e-05,
      "loss": 0.8004,
      "step": 8848
    },
    {
      "epoch": 0.12550031729599995,
      "grad_norm": 2.90625,
      "learning_rate": 4.8160647629200864e-05,
      "loss": 0.8337,
      "step": 8850
    },
    {
      "epoch": 0.12552867894962616,
      "grad_norm": 3.578125,
      "learning_rate": 4.815980638896765e-05,
      "loss": 0.8617,
      "step": 8852
    },
    {
      "epoch": 0.12555704060325237,
      "grad_norm": 3.0625,
      "learning_rate": 4.8158964963755516e-05,
      "loss": 0.8675,
      "step": 8854
    },
    {
      "epoch": 0.1255854022568786,
      "grad_norm": 3.03125,
      "learning_rate": 4.815812335357118e-05,
      "loss": 0.8829,
      "step": 8856
    },
    {
      "epoch": 0.1256137639105048,
      "grad_norm": 3.3125,
      "learning_rate": 4.815728155842138e-05,
      "loss": 0.8577,
      "step": 8858
    },
    {
      "epoch": 0.125642125564131,
      "grad_norm": 3.4375,
      "learning_rate": 4.815643957831282e-05,
      "loss": 0.8712,
      "step": 8860
    },
    {
      "epoch": 0.12567048721775723,
      "grad_norm": 3.359375,
      "learning_rate": 4.815559741325223e-05,
      "loss": 0.8786,
      "step": 8862
    },
    {
      "epoch": 0.12569884887138344,
      "grad_norm": 3.203125,
      "learning_rate": 4.815475506324635e-05,
      "loss": 0.8536,
      "step": 8864
    },
    {
      "epoch": 0.12572721052500965,
      "grad_norm": 2.953125,
      "learning_rate": 4.815391252830189e-05,
      "loss": 0.8682,
      "step": 8866
    },
    {
      "epoch": 0.12575557217863587,
      "grad_norm": 3.4375,
      "learning_rate": 4.8153069808425585e-05,
      "loss": 0.8796,
      "step": 8868
    },
    {
      "epoch": 0.12578393383226208,
      "grad_norm": 2.765625,
      "learning_rate": 4.815222690362417e-05,
      "loss": 0.8347,
      "step": 8870
    },
    {
      "epoch": 0.1258122954858883,
      "grad_norm": 2.890625,
      "learning_rate": 4.815138381390437e-05,
      "loss": 0.8665,
      "step": 8872
    },
    {
      "epoch": 0.1258406571395145,
      "grad_norm": 3.296875,
      "learning_rate": 4.8150540539272927e-05,
      "loss": 0.8459,
      "step": 8874
    },
    {
      "epoch": 0.12586901879314075,
      "grad_norm": 3.234375,
      "learning_rate": 4.814969707973657e-05,
      "loss": 0.8658,
      "step": 8876
    },
    {
      "epoch": 0.12589738044676696,
      "grad_norm": 2.890625,
      "learning_rate": 4.8148853435302036e-05,
      "loss": 0.8425,
      "step": 8878
    },
    {
      "epoch": 0.12592574210039317,
      "grad_norm": 3.21875,
      "learning_rate": 4.8148009605976076e-05,
      "loss": 0.8801,
      "step": 8880
    },
    {
      "epoch": 0.1259541037540194,
      "grad_norm": 3.65625,
      "learning_rate": 4.814716559176541e-05,
      "loss": 0.9048,
      "step": 8882
    },
    {
      "epoch": 0.1259824654076456,
      "grad_norm": 3.03125,
      "learning_rate": 4.8146321392676784e-05,
      "loss": 0.8931,
      "step": 8884
    },
    {
      "epoch": 0.1260108270612718,
      "grad_norm": 3.765625,
      "learning_rate": 4.814547700871696e-05,
      "loss": 0.8837,
      "step": 8886
    },
    {
      "epoch": 0.12603918871489803,
      "grad_norm": 2.984375,
      "learning_rate": 4.814463243989266e-05,
      "loss": 0.8589,
      "step": 8888
    },
    {
      "epoch": 0.12606755036852424,
      "grad_norm": 3.25,
      "learning_rate": 4.814378768621063e-05,
      "loss": 0.9094,
      "step": 8890
    },
    {
      "epoch": 0.12609591202215045,
      "grad_norm": 3.3125,
      "learning_rate": 4.8142942747677634e-05,
      "loss": 0.883,
      "step": 8892
    },
    {
      "epoch": 0.12612427367577667,
      "grad_norm": 3.109375,
      "learning_rate": 4.814209762430041e-05,
      "loss": 0.8434,
      "step": 8894
    },
    {
      "epoch": 0.12615263532940288,
      "grad_norm": 3.1875,
      "learning_rate": 4.814125231608571e-05,
      "loss": 0.8382,
      "step": 8896
    },
    {
      "epoch": 0.1261809969830291,
      "grad_norm": 3.296875,
      "learning_rate": 4.814040682304027e-05,
      "loss": 0.9266,
      "step": 8898
    },
    {
      "epoch": 0.1262093586366553,
      "grad_norm": 3.171875,
      "learning_rate": 4.813956114517087e-05,
      "loss": 0.8859,
      "step": 8900
    },
    {
      "epoch": 0.12623772029028152,
      "grad_norm": 3.046875,
      "learning_rate": 4.813871528248425e-05,
      "loss": 0.8471,
      "step": 8902
    },
    {
      "epoch": 0.12626608194390773,
      "grad_norm": 2.953125,
      "learning_rate": 4.813786923498717e-05,
      "loss": 0.7952,
      "step": 8904
    },
    {
      "epoch": 0.12629444359753395,
      "grad_norm": 3.1875,
      "learning_rate": 4.8137023002686375e-05,
      "loss": 0.8673,
      "step": 8906
    },
    {
      "epoch": 0.12632280525116016,
      "grad_norm": 3.046875,
      "learning_rate": 4.813617658558864e-05,
      "loss": 0.8803,
      "step": 8908
    },
    {
      "epoch": 0.12635116690478637,
      "grad_norm": 2.671875,
      "learning_rate": 4.813532998370072e-05,
      "loss": 0.8262,
      "step": 8910
    },
    {
      "epoch": 0.12637952855841258,
      "grad_norm": 3.0625,
      "learning_rate": 4.8134483197029376e-05,
      "loss": 0.845,
      "step": 8912
    },
    {
      "epoch": 0.12640789021203883,
      "grad_norm": 2.90625,
      "learning_rate": 4.8133636225581375e-05,
      "loss": 0.8407,
      "step": 8914
    },
    {
      "epoch": 0.12643625186566504,
      "grad_norm": 3.09375,
      "learning_rate": 4.813278906936348e-05,
      "loss": 0.8272,
      "step": 8916
    },
    {
      "epoch": 0.12646461351929125,
      "grad_norm": 3.421875,
      "learning_rate": 4.813194172838245e-05,
      "loss": 0.8921,
      "step": 8918
    },
    {
      "epoch": 0.12649297517291747,
      "grad_norm": 2.859375,
      "learning_rate": 4.813109420264506e-05,
      "loss": 0.8778,
      "step": 8920
    },
    {
      "epoch": 0.12652133682654368,
      "grad_norm": 2.984375,
      "learning_rate": 4.813024649215807e-05,
      "loss": 0.8918,
      "step": 8922
    },
    {
      "epoch": 0.1265496984801699,
      "grad_norm": 2.625,
      "learning_rate": 4.812939859692826e-05,
      "loss": 0.8372,
      "step": 8924
    },
    {
      "epoch": 0.1265780601337961,
      "grad_norm": 3.1875,
      "learning_rate": 4.812855051696241e-05,
      "loss": 0.8307,
      "step": 8926
    },
    {
      "epoch": 0.12660642178742232,
      "grad_norm": 2.953125,
      "learning_rate": 4.8127702252267285e-05,
      "loss": 0.8263,
      "step": 8928
    },
    {
      "epoch": 0.12663478344104853,
      "grad_norm": 2.96875,
      "learning_rate": 4.8126853802849655e-05,
      "loss": 0.8785,
      "step": 8930
    },
    {
      "epoch": 0.12666314509467475,
      "grad_norm": 3.5,
      "learning_rate": 4.8126005168716305e-05,
      "loss": 0.8976,
      "step": 8932
    },
    {
      "epoch": 0.12669150674830096,
      "grad_norm": 3.234375,
      "learning_rate": 4.8125156349874004e-05,
      "loss": 0.8982,
      "step": 8934
    },
    {
      "epoch": 0.12671986840192717,
      "grad_norm": 3.234375,
      "learning_rate": 4.812430734632954e-05,
      "loss": 0.8745,
      "step": 8936
    },
    {
      "epoch": 0.12674823005555338,
      "grad_norm": 3.109375,
      "learning_rate": 4.8123458158089687e-05,
      "loss": 0.8837,
      "step": 8938
    },
    {
      "epoch": 0.1267765917091796,
      "grad_norm": 3.203125,
      "learning_rate": 4.812260878516124e-05,
      "loss": 0.8361,
      "step": 8940
    },
    {
      "epoch": 0.1268049533628058,
      "grad_norm": 3.609375,
      "learning_rate": 4.812175922755096e-05,
      "loss": 0.8675,
      "step": 8942
    },
    {
      "epoch": 0.12683331501643202,
      "grad_norm": 2.734375,
      "learning_rate": 4.8120909485265657e-05,
      "loss": 0.8505,
      "step": 8944
    },
    {
      "epoch": 0.12686167667005824,
      "grad_norm": 2.890625,
      "learning_rate": 4.8120059558312106e-05,
      "loss": 0.8447,
      "step": 8946
    },
    {
      "epoch": 0.12689003832368445,
      "grad_norm": 2.9375,
      "learning_rate": 4.81192094466971e-05,
      "loss": 0.8789,
      "step": 8948
    },
    {
      "epoch": 0.12691839997731066,
      "grad_norm": 3.15625,
      "learning_rate": 4.811835915042742e-05,
      "loss": 0.8275,
      "step": 8950
    },
    {
      "epoch": 0.1269467616309369,
      "grad_norm": 3.40625,
      "learning_rate": 4.811750866950986e-05,
      "loss": 0.8483,
      "step": 8952
    },
    {
      "epoch": 0.12697512328456312,
      "grad_norm": 3.015625,
      "learning_rate": 4.811665800395123e-05,
      "loss": 0.8461,
      "step": 8954
    },
    {
      "epoch": 0.12700348493818933,
      "grad_norm": 2.9375,
      "learning_rate": 4.81158071537583e-05,
      "loss": 0.8788,
      "step": 8956
    },
    {
      "epoch": 0.12703184659181554,
      "grad_norm": 3.015625,
      "learning_rate": 4.811495611893788e-05,
      "loss": 0.8773,
      "step": 8958
    },
    {
      "epoch": 0.12706020824544176,
      "grad_norm": 2.90625,
      "learning_rate": 4.8114104899496764e-05,
      "loss": 0.853,
      "step": 8960
    },
    {
      "epoch": 0.12708856989906797,
      "grad_norm": 3.28125,
      "learning_rate": 4.8113253495441745e-05,
      "loss": 0.8581,
      "step": 8962
    },
    {
      "epoch": 0.12711693155269418,
      "grad_norm": 3.015625,
      "learning_rate": 4.8112401906779636e-05,
      "loss": 0.8587,
      "step": 8964
    },
    {
      "epoch": 0.1271452932063204,
      "grad_norm": 3.484375,
      "learning_rate": 4.811155013351723e-05,
      "loss": 0.8961,
      "step": 8966
    },
    {
      "epoch": 0.1271736548599466,
      "grad_norm": 3.15625,
      "learning_rate": 4.8110698175661326e-05,
      "loss": 0.9038,
      "step": 8968
    },
    {
      "epoch": 0.12720201651357282,
      "grad_norm": 3.234375,
      "learning_rate": 4.810984603321874e-05,
      "loss": 0.8236,
      "step": 8970
    },
    {
      "epoch": 0.12723037816719904,
      "grad_norm": 2.703125,
      "learning_rate": 4.810899370619627e-05,
      "loss": 0.8669,
      "step": 8972
    },
    {
      "epoch": 0.12725873982082525,
      "grad_norm": 3.296875,
      "learning_rate": 4.810814119460073e-05,
      "loss": 0.8748,
      "step": 8974
    },
    {
      "epoch": 0.12728710147445146,
      "grad_norm": 3.125,
      "learning_rate": 4.8107288498438916e-05,
      "loss": 0.8738,
      "step": 8976
    },
    {
      "epoch": 0.12731546312807768,
      "grad_norm": 3.203125,
      "learning_rate": 4.810643561771766e-05,
      "loss": 0.8775,
      "step": 8978
    },
    {
      "epoch": 0.1273438247817039,
      "grad_norm": 3.609375,
      "learning_rate": 4.810558255244375e-05,
      "loss": 0.8725,
      "step": 8980
    },
    {
      "epoch": 0.1273721864353301,
      "grad_norm": 2.9375,
      "learning_rate": 4.810472930262402e-05,
      "loss": 0.8497,
      "step": 8982
    },
    {
      "epoch": 0.12740054808895632,
      "grad_norm": 2.9375,
      "learning_rate": 4.810387586826527e-05,
      "loss": 0.9181,
      "step": 8984
    },
    {
      "epoch": 0.12742890974258253,
      "grad_norm": 2.8125,
      "learning_rate": 4.810302224937432e-05,
      "loss": 0.8439,
      "step": 8986
    },
    {
      "epoch": 0.12745727139620874,
      "grad_norm": 2.9375,
      "learning_rate": 4.8102168445957996e-05,
      "loss": 0.808,
      "step": 8988
    },
    {
      "epoch": 0.12748563304983498,
      "grad_norm": 3.25,
      "learning_rate": 4.810131445802312e-05,
      "loss": 0.8537,
      "step": 8990
    },
    {
      "epoch": 0.1275139947034612,
      "grad_norm": 3.21875,
      "learning_rate": 4.810046028557649e-05,
      "loss": 0.8944,
      "step": 8992
    },
    {
      "epoch": 0.1275423563570874,
      "grad_norm": 3.28125,
      "learning_rate": 4.8099605928624944e-05,
      "loss": 0.8728,
      "step": 8994
    },
    {
      "epoch": 0.12757071801071362,
      "grad_norm": 3.09375,
      "learning_rate": 4.809875138717531e-05,
      "loss": 0.9011,
      "step": 8996
    },
    {
      "epoch": 0.12759907966433984,
      "grad_norm": 3.171875,
      "learning_rate": 4.80978966612344e-05,
      "loss": 0.8736,
      "step": 8998
    },
    {
      "epoch": 0.12762744131796605,
      "grad_norm": 2.90625,
      "learning_rate": 4.809704175080906e-05,
      "loss": 0.8695,
      "step": 9000
    },
    {
      "epoch": 0.12765580297159226,
      "grad_norm": 3.296875,
      "learning_rate": 4.80961866559061e-05,
      "loss": 0.8884,
      "step": 9002
    },
    {
      "epoch": 0.12768416462521848,
      "grad_norm": 2.875,
      "learning_rate": 4.809533137653236e-05,
      "loss": 0.8581,
      "step": 9004
    },
    {
      "epoch": 0.1277125262788447,
      "grad_norm": 2.921875,
      "learning_rate": 4.809447591269467e-05,
      "loss": 0.8966,
      "step": 9006
    },
    {
      "epoch": 0.1277408879324709,
      "grad_norm": 2.96875,
      "learning_rate": 4.8093620264399855e-05,
      "loss": 0.8069,
      "step": 9008
    },
    {
      "epoch": 0.12776924958609712,
      "grad_norm": 3.3125,
      "learning_rate": 4.809276443165475e-05,
      "loss": 0.8908,
      "step": 9010
    },
    {
      "epoch": 0.12779761123972333,
      "grad_norm": 3.078125,
      "learning_rate": 4.8091908414466206e-05,
      "loss": 0.9391,
      "step": 9012
    },
    {
      "epoch": 0.12782597289334954,
      "grad_norm": 3.1875,
      "learning_rate": 4.809105221284105e-05,
      "loss": 0.8821,
      "step": 9014
    },
    {
      "epoch": 0.12785433454697576,
      "grad_norm": 2.984375,
      "learning_rate": 4.809019582678611e-05,
      "loss": 0.8484,
      "step": 9016
    },
    {
      "epoch": 0.12788269620060197,
      "grad_norm": 3.234375,
      "learning_rate": 4.808933925630824e-05,
      "loss": 0.8844,
      "step": 9018
    },
    {
      "epoch": 0.12791105785422818,
      "grad_norm": 2.6875,
      "learning_rate": 4.8088482501414275e-05,
      "loss": 0.8529,
      "step": 9020
    },
    {
      "epoch": 0.1279394195078544,
      "grad_norm": 2.71875,
      "learning_rate": 4.808762556211106e-05,
      "loss": 0.8581,
      "step": 9022
    },
    {
      "epoch": 0.1279677811614806,
      "grad_norm": 3.015625,
      "learning_rate": 4.808676843840545e-05,
      "loss": 0.8596,
      "step": 9024
    },
    {
      "epoch": 0.12799614281510682,
      "grad_norm": 3.171875,
      "learning_rate": 4.808591113030427e-05,
      "loss": 0.8472,
      "step": 9026
    },
    {
      "epoch": 0.12802450446873304,
      "grad_norm": 2.96875,
      "learning_rate": 4.808505363781438e-05,
      "loss": 0.8094,
      "step": 9028
    },
    {
      "epoch": 0.12805286612235928,
      "grad_norm": 3.5,
      "learning_rate": 4.8084195960942625e-05,
      "loss": 0.8465,
      "step": 9030
    },
    {
      "epoch": 0.1280812277759855,
      "grad_norm": 3.140625,
      "learning_rate": 4.8083338099695864e-05,
      "loss": 0.8526,
      "step": 9032
    },
    {
      "epoch": 0.1281095894296117,
      "grad_norm": 3.21875,
      "learning_rate": 4.808248005408094e-05,
      "loss": 0.8141,
      "step": 9034
    },
    {
      "epoch": 0.12813795108323792,
      "grad_norm": 3.265625,
      "learning_rate": 4.808162182410471e-05,
      "loss": 0.8798,
      "step": 9036
    },
    {
      "epoch": 0.12816631273686413,
      "grad_norm": 3.34375,
      "learning_rate": 4.8080763409774016e-05,
      "loss": 0.8588,
      "step": 9038
    },
    {
      "epoch": 0.12819467439049034,
      "grad_norm": 3.015625,
      "learning_rate": 4.8079904811095736e-05,
      "loss": 0.8075,
      "step": 9040
    },
    {
      "epoch": 0.12822303604411656,
      "grad_norm": 3.3125,
      "learning_rate": 4.807904602807671e-05,
      "loss": 0.8171,
      "step": 9042
    },
    {
      "epoch": 0.12825139769774277,
      "grad_norm": 2.984375,
      "learning_rate": 4.8078187060723807e-05,
      "loss": 0.8923,
      "step": 9044
    },
    {
      "epoch": 0.12827975935136898,
      "grad_norm": 3.140625,
      "learning_rate": 4.807732790904388e-05,
      "loss": 0.8604,
      "step": 9046
    },
    {
      "epoch": 0.1283081210049952,
      "grad_norm": 3.171875,
      "learning_rate": 4.80764685730438e-05,
      "loss": 0.8598,
      "step": 9048
    },
    {
      "epoch": 0.1283364826586214,
      "grad_norm": 3.171875,
      "learning_rate": 4.807560905273043e-05,
      "loss": 0.8373,
      "step": 9050
    },
    {
      "epoch": 0.12836484431224762,
      "grad_norm": 2.78125,
      "learning_rate": 4.807474934811063e-05,
      "loss": 0.8712,
      "step": 9052
    },
    {
      "epoch": 0.12839320596587384,
      "grad_norm": 2.875,
      "learning_rate": 4.807388945919127e-05,
      "loss": 0.8161,
      "step": 9054
    },
    {
      "epoch": 0.12842156761950005,
      "grad_norm": 3.03125,
      "learning_rate": 4.8073029385979204e-05,
      "loss": 0.8262,
      "step": 9056
    },
    {
      "epoch": 0.12844992927312626,
      "grad_norm": 3.265625,
      "learning_rate": 4.8072169128481324e-05,
      "loss": 0.9214,
      "step": 9058
    },
    {
      "epoch": 0.12847829092675248,
      "grad_norm": 3.078125,
      "learning_rate": 4.807130868670449e-05,
      "loss": 0.8353,
      "step": 9060
    },
    {
      "epoch": 0.1285066525803787,
      "grad_norm": 3.25,
      "learning_rate": 4.807044806065557e-05,
      "loss": 0.8481,
      "step": 9062
    },
    {
      "epoch": 0.1285350142340049,
      "grad_norm": 3.0625,
      "learning_rate": 4.8069587250341444e-05,
      "loss": 0.8691,
      "step": 9064
    },
    {
      "epoch": 0.12856337588763112,
      "grad_norm": 3.15625,
      "learning_rate": 4.806872625576898e-05,
      "loss": 0.8491,
      "step": 9066
    },
    {
      "epoch": 0.12859173754125736,
      "grad_norm": 3.03125,
      "learning_rate": 4.8067865076945075e-05,
      "loss": 0.8643,
      "step": 9068
    },
    {
      "epoch": 0.12862009919488357,
      "grad_norm": 3.28125,
      "learning_rate": 4.806700371387658e-05,
      "loss": 0.878,
      "step": 9070
    },
    {
      "epoch": 0.12864846084850978,
      "grad_norm": 3.125,
      "learning_rate": 4.8066142166570397e-05,
      "loss": 0.956,
      "step": 9072
    },
    {
      "epoch": 0.128676822502136,
      "grad_norm": 3.1875,
      "learning_rate": 4.806528043503339e-05,
      "loss": 0.8875,
      "step": 9074
    },
    {
      "epoch": 0.1287051841557622,
      "grad_norm": 3.3125,
      "learning_rate": 4.806441851927245e-05,
      "loss": 0.8668,
      "step": 9076
    },
    {
      "epoch": 0.12873354580938842,
      "grad_norm": 3.28125,
      "learning_rate": 4.8063556419294466e-05,
      "loss": 0.8465,
      "step": 9078
    },
    {
      "epoch": 0.12876190746301464,
      "grad_norm": 3.265625,
      "learning_rate": 4.8062694135106316e-05,
      "loss": 0.8528,
      "step": 9080
    },
    {
      "epoch": 0.12879026911664085,
      "grad_norm": 2.796875,
      "learning_rate": 4.806183166671489e-05,
      "loss": 0.8483,
      "step": 9082
    },
    {
      "epoch": 0.12881863077026706,
      "grad_norm": 3.890625,
      "learning_rate": 4.8060969014127074e-05,
      "loss": 0.8898,
      "step": 9084
    },
    {
      "epoch": 0.12884699242389328,
      "grad_norm": 3.28125,
      "learning_rate": 4.806010617734977e-05,
      "loss": 0.9074,
      "step": 9086
    },
    {
      "epoch": 0.1288753540775195,
      "grad_norm": 3.0,
      "learning_rate": 4.805924315638985e-05,
      "loss": 0.8672,
      "step": 9088
    },
    {
      "epoch": 0.1289037157311457,
      "grad_norm": 2.5625,
      "learning_rate": 4.8058379951254215e-05,
      "loss": 0.8655,
      "step": 9090
    },
    {
      "epoch": 0.12893207738477191,
      "grad_norm": 2.96875,
      "learning_rate": 4.805751656194977e-05,
      "loss": 0.8731,
      "step": 9092
    },
    {
      "epoch": 0.12896043903839813,
      "grad_norm": 3.140625,
      "learning_rate": 4.8056652988483396e-05,
      "loss": 0.816,
      "step": 9094
    },
    {
      "epoch": 0.12898880069202434,
      "grad_norm": 3.1875,
      "learning_rate": 4.8055789230862006e-05,
      "loss": 0.8694,
      "step": 9096
    },
    {
      "epoch": 0.12901716234565055,
      "grad_norm": 3.34375,
      "learning_rate": 4.8054925289092475e-05,
      "loss": 0.8725,
      "step": 9098
    },
    {
      "epoch": 0.12904552399927677,
      "grad_norm": 3.0,
      "learning_rate": 4.805406116318173e-05,
      "loss": 0.9176,
      "step": 9100
    },
    {
      "epoch": 0.12907388565290298,
      "grad_norm": 2.984375,
      "learning_rate": 4.805319685313666e-05,
      "loss": 0.8544,
      "step": 9102
    },
    {
      "epoch": 0.1291022473065292,
      "grad_norm": 3.171875,
      "learning_rate": 4.8052332358964155e-05,
      "loss": 0.8694,
      "step": 9104
    },
    {
      "epoch": 0.12913060896015544,
      "grad_norm": 2.90625,
      "learning_rate": 4.8051467680671145e-05,
      "loss": 0.8471,
      "step": 9106
    },
    {
      "epoch": 0.12915897061378165,
      "grad_norm": 3.21875,
      "learning_rate": 4.8050602818264526e-05,
      "loss": 0.8667,
      "step": 9108
    },
    {
      "epoch": 0.12918733226740786,
      "grad_norm": 3.078125,
      "learning_rate": 4.80497377717512e-05,
      "loss": 0.8992,
      "step": 9110
    },
    {
      "epoch": 0.12921569392103407,
      "grad_norm": 3.21875,
      "learning_rate": 4.804887254113809e-05,
      "loss": 0.8551,
      "step": 9112
    },
    {
      "epoch": 0.1292440555746603,
      "grad_norm": 2.96875,
      "learning_rate": 4.804800712643209e-05,
      "loss": 0.8483,
      "step": 9114
    },
    {
      "epoch": 0.1292724172282865,
      "grad_norm": 3.03125,
      "learning_rate": 4.804714152764012e-05,
      "loss": 0.8873,
      "step": 9116
    },
    {
      "epoch": 0.12930077888191271,
      "grad_norm": 3.703125,
      "learning_rate": 4.80462757447691e-05,
      "loss": 0.8553,
      "step": 9118
    },
    {
      "epoch": 0.12932914053553893,
      "grad_norm": 3.125,
      "learning_rate": 4.8045409777825934e-05,
      "loss": 0.8563,
      "step": 9120
    },
    {
      "epoch": 0.12935750218916514,
      "grad_norm": 2.8125,
      "learning_rate": 4.804454362681754e-05,
      "loss": 0.8281,
      "step": 9122
    },
    {
      "epoch": 0.12938586384279135,
      "grad_norm": 3.15625,
      "learning_rate": 4.8043677291750835e-05,
      "loss": 0.8523,
      "step": 9124
    },
    {
      "epoch": 0.12941422549641757,
      "grad_norm": 3.078125,
      "learning_rate": 4.804281077263275e-05,
      "loss": 0.8406,
      "step": 9126
    },
    {
      "epoch": 0.12944258715004378,
      "grad_norm": 3.28125,
      "learning_rate": 4.80419440694702e-05,
      "loss": 0.8729,
      "step": 9128
    },
    {
      "epoch": 0.12947094880367,
      "grad_norm": 3.09375,
      "learning_rate": 4.80410771822701e-05,
      "loss": 0.8524,
      "step": 9130
    },
    {
      "epoch": 0.1294993104572962,
      "grad_norm": 2.65625,
      "learning_rate": 4.804021011103939e-05,
      "loss": 0.8637,
      "step": 9132
    },
    {
      "epoch": 0.12952767211092242,
      "grad_norm": 3.109375,
      "learning_rate": 4.8039342855784974e-05,
      "loss": 0.849,
      "step": 9134
    },
    {
      "epoch": 0.12955603376454863,
      "grad_norm": 3.171875,
      "learning_rate": 4.80384754165138e-05,
      "loss": 0.8894,
      "step": 9136
    },
    {
      "epoch": 0.12958439541817485,
      "grad_norm": 3.21875,
      "learning_rate": 4.803760779323278e-05,
      "loss": 0.8689,
      "step": 9138
    },
    {
      "epoch": 0.12961275707180106,
      "grad_norm": 2.8125,
      "learning_rate": 4.8036739985948845e-05,
      "loss": 0.8518,
      "step": 9140
    },
    {
      "epoch": 0.12964111872542727,
      "grad_norm": 3.234375,
      "learning_rate": 4.803587199466893e-05,
      "loss": 0.9258,
      "step": 9142
    },
    {
      "epoch": 0.12966948037905351,
      "grad_norm": 2.96875,
      "learning_rate": 4.803500381939997e-05,
      "loss": 0.899,
      "step": 9144
    },
    {
      "epoch": 0.12969784203267973,
      "grad_norm": 3.25,
      "learning_rate": 4.803413546014891e-05,
      "loss": 0.8846,
      "step": 9146
    },
    {
      "epoch": 0.12972620368630594,
      "grad_norm": 3.21875,
      "learning_rate": 4.8033266916922666e-05,
      "loss": 0.8986,
      "step": 9148
    },
    {
      "epoch": 0.12975456533993215,
      "grad_norm": 2.921875,
      "learning_rate": 4.8032398189728176e-05,
      "loss": 0.8403,
      "step": 9150
    },
    {
      "epoch": 0.12978292699355837,
      "grad_norm": 3.078125,
      "learning_rate": 4.8031529278572394e-05,
      "loss": 0.8755,
      "step": 9152
    },
    {
      "epoch": 0.12981128864718458,
      "grad_norm": 3.109375,
      "learning_rate": 4.803066018346225e-05,
      "loss": 0.8886,
      "step": 9154
    },
    {
      "epoch": 0.1298396503008108,
      "grad_norm": 2.984375,
      "learning_rate": 4.802979090440469e-05,
      "loss": 0.83,
      "step": 9156
    },
    {
      "epoch": 0.129868011954437,
      "grad_norm": 3.0,
      "learning_rate": 4.802892144140665e-05,
      "loss": 0.8827,
      "step": 9158
    },
    {
      "epoch": 0.12989637360806322,
      "grad_norm": 3.03125,
      "learning_rate": 4.802805179447508e-05,
      "loss": 0.8363,
      "step": 9160
    },
    {
      "epoch": 0.12992473526168943,
      "grad_norm": 3.046875,
      "learning_rate": 4.802718196361692e-05,
      "loss": 0.9085,
      "step": 9162
    },
    {
      "epoch": 0.12995309691531565,
      "grad_norm": 2.953125,
      "learning_rate": 4.802631194883913e-05,
      "loss": 0.925,
      "step": 9164
    },
    {
      "epoch": 0.12998145856894186,
      "grad_norm": 3.28125,
      "learning_rate": 4.802544175014865e-05,
      "loss": 0.8595,
      "step": 9166
    },
    {
      "epoch": 0.13000982022256807,
      "grad_norm": 3.140625,
      "learning_rate": 4.802457136755242e-05,
      "loss": 0.8998,
      "step": 9168
    },
    {
      "epoch": 0.1300381818761943,
      "grad_norm": 3.125,
      "learning_rate": 4.802370080105741e-05,
      "loss": 0.8732,
      "step": 9170
    },
    {
      "epoch": 0.1300665435298205,
      "grad_norm": 3.1875,
      "learning_rate": 4.802283005067057e-05,
      "loss": 0.848,
      "step": 9172
    },
    {
      "epoch": 0.1300949051834467,
      "grad_norm": 3.171875,
      "learning_rate": 4.802195911639884e-05,
      "loss": 0.8926,
      "step": 9174
    },
    {
      "epoch": 0.13012326683707293,
      "grad_norm": 3.625,
      "learning_rate": 4.802108799824919e-05,
      "loss": 0.8575,
      "step": 9176
    },
    {
      "epoch": 0.13015162849069914,
      "grad_norm": 3.125,
      "learning_rate": 4.802021669622858e-05,
      "loss": 0.855,
      "step": 9178
    },
    {
      "epoch": 0.13017999014432535,
      "grad_norm": 3.515625,
      "learning_rate": 4.801934521034396e-05,
      "loss": 0.8964,
      "step": 9180
    },
    {
      "epoch": 0.13020835179795157,
      "grad_norm": 2.984375,
      "learning_rate": 4.80184735406023e-05,
      "loss": 0.8436,
      "step": 9182
    },
    {
      "epoch": 0.1302367134515778,
      "grad_norm": 3.046875,
      "learning_rate": 4.801760168701055e-05,
      "loss": 0.7832,
      "step": 9184
    },
    {
      "epoch": 0.13026507510520402,
      "grad_norm": 3.0625,
      "learning_rate": 4.8016729649575686e-05,
      "loss": 0.8683,
      "step": 9186
    },
    {
      "epoch": 0.13029343675883023,
      "grad_norm": 2.84375,
      "learning_rate": 4.801585742830466e-05,
      "loss": 0.8638,
      "step": 9188
    },
    {
      "epoch": 0.13032179841245645,
      "grad_norm": 3.1875,
      "learning_rate": 4.801498502320445e-05,
      "loss": 0.8526,
      "step": 9190
    },
    {
      "epoch": 0.13035016006608266,
      "grad_norm": 3.125,
      "learning_rate": 4.801411243428202e-05,
      "loss": 0.8005,
      "step": 9192
    },
    {
      "epoch": 0.13037852171970887,
      "grad_norm": 3.125,
      "learning_rate": 4.801323966154434e-05,
      "loss": 0.874,
      "step": 9194
    },
    {
      "epoch": 0.1304068833733351,
      "grad_norm": 3.125,
      "learning_rate": 4.801236670499837e-05,
      "loss": 0.8263,
      "step": 9196
    },
    {
      "epoch": 0.1304352450269613,
      "grad_norm": 3.125,
      "learning_rate": 4.80114935646511e-05,
      "loss": 0.8854,
      "step": 9198
    },
    {
      "epoch": 0.1304636066805875,
      "grad_norm": 2.8125,
      "learning_rate": 4.8010620240509495e-05,
      "loss": 0.8502,
      "step": 9200
    },
    {
      "epoch": 0.13049196833421373,
      "grad_norm": 3.140625,
      "learning_rate": 4.800974673258054e-05,
      "loss": 0.8639,
      "step": 9202
    },
    {
      "epoch": 0.13052032998783994,
      "grad_norm": 3.546875,
      "learning_rate": 4.8008873040871196e-05,
      "loss": 0.8744,
      "step": 9204
    },
    {
      "epoch": 0.13054869164146615,
      "grad_norm": 2.953125,
      "learning_rate": 4.800799916538844e-05,
      "loss": 0.8234,
      "step": 9206
    },
    {
      "epoch": 0.13057705329509237,
      "grad_norm": 2.75,
      "learning_rate": 4.800712510613927e-05,
      "loss": 0.7832,
      "step": 9208
    },
    {
      "epoch": 0.13060541494871858,
      "grad_norm": 3.015625,
      "learning_rate": 4.800625086313065e-05,
      "loss": 0.8322,
      "step": 9210
    },
    {
      "epoch": 0.1306337766023448,
      "grad_norm": 2.828125,
      "learning_rate": 4.8005376436369576e-05,
      "loss": 0.8051,
      "step": 9212
    },
    {
      "epoch": 0.130662138255971,
      "grad_norm": 2.984375,
      "learning_rate": 4.8004501825863026e-05,
      "loss": 0.8336,
      "step": 9214
    },
    {
      "epoch": 0.13069049990959722,
      "grad_norm": 3.34375,
      "learning_rate": 4.800362703161798e-05,
      "loss": 0.8402,
      "step": 9216
    },
    {
      "epoch": 0.13071886156322343,
      "grad_norm": 3.859375,
      "learning_rate": 4.800275205364144e-05,
      "loss": 0.8642,
      "step": 9218
    },
    {
      "epoch": 0.13074722321684965,
      "grad_norm": 2.875,
      "learning_rate": 4.800187689194038e-05,
      "loss": 0.8994,
      "step": 9220
    },
    {
      "epoch": 0.13077558487047589,
      "grad_norm": 3.109375,
      "learning_rate": 4.80010015465218e-05,
      "loss": 0.8558,
      "step": 9222
    },
    {
      "epoch": 0.1308039465241021,
      "grad_norm": 3.09375,
      "learning_rate": 4.8000126017392676e-05,
      "loss": 0.865,
      "step": 9224
    },
    {
      "epoch": 0.1308323081777283,
      "grad_norm": 3.265625,
      "learning_rate": 4.7999250304560017e-05,
      "loss": 0.8295,
      "step": 9226
    },
    {
      "epoch": 0.13086066983135453,
      "grad_norm": 3.046875,
      "learning_rate": 4.7998374408030814e-05,
      "loss": 0.8605,
      "step": 9228
    },
    {
      "epoch": 0.13088903148498074,
      "grad_norm": 2.9375,
      "learning_rate": 4.7997498327812054e-05,
      "loss": 0.877,
      "step": 9230
    },
    {
      "epoch": 0.13091739313860695,
      "grad_norm": 3.0,
      "learning_rate": 4.7996622063910744e-05,
      "loss": 0.8349,
      "step": 9232
    },
    {
      "epoch": 0.13094575479223317,
      "grad_norm": 2.84375,
      "learning_rate": 4.7995745616333876e-05,
      "loss": 0.8846,
      "step": 9234
    },
    {
      "epoch": 0.13097411644585938,
      "grad_norm": 3.484375,
      "learning_rate": 4.799486898508846e-05,
      "loss": 0.877,
      "step": 9236
    },
    {
      "epoch": 0.1310024780994856,
      "grad_norm": 2.921875,
      "learning_rate": 4.799399217018149e-05,
      "loss": 0.8521,
      "step": 9238
    },
    {
      "epoch": 0.1310308397531118,
      "grad_norm": 3.359375,
      "learning_rate": 4.7993115171619975e-05,
      "loss": 0.8627,
      "step": 9240
    },
    {
      "epoch": 0.13105920140673802,
      "grad_norm": 3.15625,
      "learning_rate": 4.7992237989410904e-05,
      "loss": 0.8441,
      "step": 9242
    },
    {
      "epoch": 0.13108756306036423,
      "grad_norm": 3.0,
      "learning_rate": 4.79913606235613e-05,
      "loss": 0.8465,
      "step": 9244
    },
    {
      "epoch": 0.13111592471399044,
      "grad_norm": 3.171875,
      "learning_rate": 4.7990483074078175e-05,
      "loss": 0.8582,
      "step": 9246
    },
    {
      "epoch": 0.13114428636761666,
      "grad_norm": 3.28125,
      "learning_rate": 4.798960534096851e-05,
      "loss": 0.916,
      "step": 9248
    },
    {
      "epoch": 0.13117264802124287,
      "grad_norm": 3.15625,
      "learning_rate": 4.798872742423935e-05,
      "loss": 0.8798,
      "step": 9250
    },
    {
      "epoch": 0.13120100967486908,
      "grad_norm": 3.328125,
      "learning_rate": 4.798784932389768e-05,
      "loss": 0.8598,
      "step": 9252
    },
    {
      "epoch": 0.1312293713284953,
      "grad_norm": 3.359375,
      "learning_rate": 4.798697103995053e-05,
      "loss": 0.8761,
      "step": 9254
    },
    {
      "epoch": 0.1312577329821215,
      "grad_norm": 3.09375,
      "learning_rate": 4.7986092572404906e-05,
      "loss": 0.8479,
      "step": 9256
    },
    {
      "epoch": 0.13128609463574772,
      "grad_norm": 2.9375,
      "learning_rate": 4.7985213921267834e-05,
      "loss": 0.8668,
      "step": 9258
    },
    {
      "epoch": 0.13131445628937397,
      "grad_norm": 2.96875,
      "learning_rate": 4.798433508654632e-05,
      "loss": 0.8423,
      "step": 9260
    },
    {
      "epoch": 0.13134281794300018,
      "grad_norm": 3.265625,
      "learning_rate": 4.798345606824739e-05,
      "loss": 0.8759,
      "step": 9262
    },
    {
      "epoch": 0.1313711795966264,
      "grad_norm": 2.921875,
      "learning_rate": 4.798257686637806e-05,
      "loss": 0.8022,
      "step": 9264
    },
    {
      "epoch": 0.1313995412502526,
      "grad_norm": 3.203125,
      "learning_rate": 4.798169748094536e-05,
      "loss": 0.8389,
      "step": 9266
    },
    {
      "epoch": 0.13142790290387882,
      "grad_norm": 2.90625,
      "learning_rate": 4.798081791195631e-05,
      "loss": 0.841,
      "step": 9268
    },
    {
      "epoch": 0.13145626455750503,
      "grad_norm": 3.03125,
      "learning_rate": 4.7979938159417925e-05,
      "loss": 0.7989,
      "step": 9270
    },
    {
      "epoch": 0.13148462621113124,
      "grad_norm": 3.46875,
      "learning_rate": 4.7979058223337246e-05,
      "loss": 0.8867,
      "step": 9272
    },
    {
      "epoch": 0.13151298786475746,
      "grad_norm": 2.9375,
      "learning_rate": 4.7978178103721306e-05,
      "loss": 0.8614,
      "step": 9274
    },
    {
      "epoch": 0.13154134951838367,
      "grad_norm": 2.859375,
      "learning_rate": 4.797729780057712e-05,
      "loss": 0.8182,
      "step": 9276
    },
    {
      "epoch": 0.13156971117200988,
      "grad_norm": 3.171875,
      "learning_rate": 4.7976417313911715e-05,
      "loss": 0.8683,
      "step": 9278
    },
    {
      "epoch": 0.1315980728256361,
      "grad_norm": 3.390625,
      "learning_rate": 4.797553664373213e-05,
      "loss": 0.8519,
      "step": 9280
    },
    {
      "epoch": 0.1316264344792623,
      "grad_norm": 3.140625,
      "learning_rate": 4.797465579004542e-05,
      "loss": 0.9274,
      "step": 9282
    },
    {
      "epoch": 0.13165479613288852,
      "grad_norm": 3.25,
      "learning_rate": 4.797377475285859e-05,
      "loss": 0.8448,
      "step": 9284
    },
    {
      "epoch": 0.13168315778651474,
      "grad_norm": 2.890625,
      "learning_rate": 4.797289353217869e-05,
      "loss": 0.8573,
      "step": 9286
    },
    {
      "epoch": 0.13171151944014095,
      "grad_norm": 3.0625,
      "learning_rate": 4.7972012128012754e-05,
      "loss": 0.8346,
      "step": 9288
    },
    {
      "epoch": 0.13173988109376716,
      "grad_norm": 3.078125,
      "learning_rate": 4.797113054036783e-05,
      "loss": 0.8266,
      "step": 9290
    },
    {
      "epoch": 0.13176824274739338,
      "grad_norm": 3.0,
      "learning_rate": 4.797024876925095e-05,
      "loss": 0.8027,
      "step": 9292
    },
    {
      "epoch": 0.1317966044010196,
      "grad_norm": 2.765625,
      "learning_rate": 4.796936681466916e-05,
      "loss": 0.8075,
      "step": 9294
    },
    {
      "epoch": 0.1318249660546458,
      "grad_norm": 3.25,
      "learning_rate": 4.796848467662951e-05,
      "loss": 0.8408,
      "step": 9296
    },
    {
      "epoch": 0.13185332770827204,
      "grad_norm": 3.296875,
      "learning_rate": 4.796760235513904e-05,
      "loss": 0.8735,
      "step": 9298
    },
    {
      "epoch": 0.13188168936189826,
      "grad_norm": 3.140625,
      "learning_rate": 4.79667198502048e-05,
      "loss": 0.8966,
      "step": 9300
    },
    {
      "epoch": 0.13191005101552447,
      "grad_norm": 2.875,
      "learning_rate": 4.796583716183383e-05,
      "loss": 0.8467,
      "step": 9302
    },
    {
      "epoch": 0.13193841266915068,
      "grad_norm": 3.265625,
      "learning_rate": 4.7964954290033195e-05,
      "loss": 0.8881,
      "step": 9304
    },
    {
      "epoch": 0.1319667743227769,
      "grad_norm": 3.21875,
      "learning_rate": 4.796407123480993e-05,
      "loss": 0.8007,
      "step": 9306
    },
    {
      "epoch": 0.1319951359764031,
      "grad_norm": 3.15625,
      "learning_rate": 4.79631879961711e-05,
      "loss": 0.8423,
      "step": 9308
    },
    {
      "epoch": 0.13202349763002932,
      "grad_norm": 2.890625,
      "learning_rate": 4.7962304574123754e-05,
      "loss": 0.8666,
      "step": 9310
    },
    {
      "epoch": 0.13205185928365554,
      "grad_norm": 3.28125,
      "learning_rate": 4.7961420968674955e-05,
      "loss": 0.8424,
      "step": 9312
    },
    {
      "epoch": 0.13208022093728175,
      "grad_norm": 2.640625,
      "learning_rate": 4.796053717983175e-05,
      "loss": 0.8447,
      "step": 9314
    },
    {
      "epoch": 0.13210858259090796,
      "grad_norm": 2.796875,
      "learning_rate": 4.795965320760121e-05,
      "loss": 0.8442,
      "step": 9316
    },
    {
      "epoch": 0.13213694424453418,
      "grad_norm": 3.0625,
      "learning_rate": 4.795876905199038e-05,
      "loss": 0.8652,
      "step": 9318
    },
    {
      "epoch": 0.1321653058981604,
      "grad_norm": 3.4375,
      "learning_rate": 4.7957884713006334e-05,
      "loss": 0.8523,
      "step": 9320
    },
    {
      "epoch": 0.1321936675517866,
      "grad_norm": 3.25,
      "learning_rate": 4.7957000190656134e-05,
      "loss": 0.7876,
      "step": 9322
    },
    {
      "epoch": 0.13222202920541282,
      "grad_norm": 3.015625,
      "learning_rate": 4.795611548494684e-05,
      "loss": 0.8601,
      "step": 9324
    },
    {
      "epoch": 0.13225039085903903,
      "grad_norm": 3.40625,
      "learning_rate": 4.795523059588552e-05,
      "loss": 0.9373,
      "step": 9326
    },
    {
      "epoch": 0.13227875251266524,
      "grad_norm": 3.125,
      "learning_rate": 4.7954345523479235e-05,
      "loss": 0.8288,
      "step": 9328
    },
    {
      "epoch": 0.13230711416629146,
      "grad_norm": 2.953125,
      "learning_rate": 4.795346026773507e-05,
      "loss": 0.8257,
      "step": 9330
    },
    {
      "epoch": 0.13233547581991767,
      "grad_norm": 3.296875,
      "learning_rate": 4.7952574828660086e-05,
      "loss": 0.8779,
      "step": 9332
    },
    {
      "epoch": 0.13236383747354388,
      "grad_norm": 3.171875,
      "learning_rate": 4.795168920626135e-05,
      "loss": 0.8499,
      "step": 9334
    },
    {
      "epoch": 0.1323921991271701,
      "grad_norm": 3.046875,
      "learning_rate": 4.795080340054595e-05,
      "loss": 0.9124,
      "step": 9336
    },
    {
      "epoch": 0.13242056078079634,
      "grad_norm": 3.171875,
      "learning_rate": 4.794991741152095e-05,
      "loss": 0.848,
      "step": 9338
    },
    {
      "epoch": 0.13244892243442255,
      "grad_norm": 3.25,
      "learning_rate": 4.794903123919343e-05,
      "loss": 0.8716,
      "step": 9340
    },
    {
      "epoch": 0.13247728408804876,
      "grad_norm": 3.125,
      "learning_rate": 4.794814488357046e-05,
      "loss": 0.8554,
      "step": 9342
    },
    {
      "epoch": 0.13250564574167498,
      "grad_norm": 2.984375,
      "learning_rate": 4.794725834465913e-05,
      "loss": 0.8326,
      "step": 9344
    },
    {
      "epoch": 0.1325340073953012,
      "grad_norm": 3.21875,
      "learning_rate": 4.794637162246652e-05,
      "loss": 0.8678,
      "step": 9346
    },
    {
      "epoch": 0.1325623690489274,
      "grad_norm": 3.21875,
      "learning_rate": 4.794548471699971e-05,
      "loss": 0.8293,
      "step": 9348
    },
    {
      "epoch": 0.13259073070255362,
      "grad_norm": 3.15625,
      "learning_rate": 4.794459762826578e-05,
      "loss": 0.8594,
      "step": 9350
    },
    {
      "epoch": 0.13261909235617983,
      "grad_norm": 3.265625,
      "learning_rate": 4.7943710356271816e-05,
      "loss": 0.9143,
      "step": 9352
    },
    {
      "epoch": 0.13264745400980604,
      "grad_norm": 3.1875,
      "learning_rate": 4.7942822901024906e-05,
      "loss": 0.8479,
      "step": 9354
    },
    {
      "epoch": 0.13267581566343226,
      "grad_norm": 2.96875,
      "learning_rate": 4.794193526253215e-05,
      "loss": 0.8305,
      "step": 9356
    },
    {
      "epoch": 0.13270417731705847,
      "grad_norm": 3.328125,
      "learning_rate": 4.794104744080061e-05,
      "loss": 0.8722,
      "step": 9358
    },
    {
      "epoch": 0.13273253897068468,
      "grad_norm": 3.140625,
      "learning_rate": 4.794015943583741e-05,
      "loss": 0.8982,
      "step": 9360
    },
    {
      "epoch": 0.1327609006243109,
      "grad_norm": 2.984375,
      "learning_rate": 4.793927124764962e-05,
      "loss": 0.837,
      "step": 9362
    },
    {
      "epoch": 0.1327892622779371,
      "grad_norm": 3.046875,
      "learning_rate": 4.7938382876244334e-05,
      "loss": 0.8604,
      "step": 9364
    },
    {
      "epoch": 0.13281762393156332,
      "grad_norm": 3.234375,
      "learning_rate": 4.793749432162866e-05,
      "loss": 0.8655,
      "step": 9366
    },
    {
      "epoch": 0.13284598558518954,
      "grad_norm": 2.90625,
      "learning_rate": 4.793660558380969e-05,
      "loss": 0.8463,
      "step": 9368
    },
    {
      "epoch": 0.13287434723881575,
      "grad_norm": 2.78125,
      "learning_rate": 4.793571666279452e-05,
      "loss": 0.835,
      "step": 9370
    },
    {
      "epoch": 0.13290270889244196,
      "grad_norm": 3.015625,
      "learning_rate": 4.793482755859026e-05,
      "loss": 0.8698,
      "step": 9372
    },
    {
      "epoch": 0.13293107054606818,
      "grad_norm": 3.0625,
      "learning_rate": 4.793393827120399e-05,
      "loss": 0.8619,
      "step": 9374
    },
    {
      "epoch": 0.13295943219969442,
      "grad_norm": 3.25,
      "learning_rate": 4.7933048800642824e-05,
      "loss": 0.8675,
      "step": 9376
    },
    {
      "epoch": 0.13298779385332063,
      "grad_norm": 2.84375,
      "learning_rate": 4.793215914691388e-05,
      "loss": 0.7908,
      "step": 9378
    },
    {
      "epoch": 0.13301615550694684,
      "grad_norm": 3.03125,
      "learning_rate": 4.793126931002424e-05,
      "loss": 0.8718,
      "step": 9380
    },
    {
      "epoch": 0.13304451716057306,
      "grad_norm": 3.09375,
      "learning_rate": 4.793037928998103e-05,
      "loss": 0.8544,
      "step": 9382
    },
    {
      "epoch": 0.13307287881419927,
      "grad_norm": 3.234375,
      "learning_rate": 4.7929489086791345e-05,
      "loss": 0.8418,
      "step": 9384
    },
    {
      "epoch": 0.13310124046782548,
      "grad_norm": 3.109375,
      "learning_rate": 4.79285987004623e-05,
      "loss": 0.8832,
      "step": 9386
    },
    {
      "epoch": 0.1331296021214517,
      "grad_norm": 3.171875,
      "learning_rate": 4.792770813100101e-05,
      "loss": 0.8773,
      "step": 9388
    },
    {
      "epoch": 0.1331579637750779,
      "grad_norm": 2.96875,
      "learning_rate": 4.792681737841459e-05,
      "loss": 0.8121,
      "step": 9390
    },
    {
      "epoch": 0.13318632542870412,
      "grad_norm": 3.03125,
      "learning_rate": 4.792592644271015e-05,
      "loss": 0.8542,
      "step": 9392
    },
    {
      "epoch": 0.13321468708233034,
      "grad_norm": 2.8125,
      "learning_rate": 4.792503532389481e-05,
      "loss": 0.8474,
      "step": 9394
    },
    {
      "epoch": 0.13324304873595655,
      "grad_norm": 3.234375,
      "learning_rate": 4.792414402197568e-05,
      "loss": 0.887,
      "step": 9396
    },
    {
      "epoch": 0.13327141038958276,
      "grad_norm": 3.421875,
      "learning_rate": 4.792325253695987e-05,
      "loss": 0.8668,
      "step": 9398
    },
    {
      "epoch": 0.13329977204320897,
      "grad_norm": 3.5625,
      "learning_rate": 4.792236086885453e-05,
      "loss": 0.8674,
      "step": 9400
    },
    {
      "epoch": 0.1333281336968352,
      "grad_norm": 2.734375,
      "learning_rate": 4.7921469017666756e-05,
      "loss": 0.8189,
      "step": 9402
    },
    {
      "epoch": 0.1333564953504614,
      "grad_norm": 3.28125,
      "learning_rate": 4.7920576983403684e-05,
      "loss": 0.8895,
      "step": 9404
    },
    {
      "epoch": 0.13338485700408761,
      "grad_norm": 3.609375,
      "learning_rate": 4.791968476607244e-05,
      "loss": 0.8661,
      "step": 9406
    },
    {
      "epoch": 0.13341321865771383,
      "grad_norm": 2.734375,
      "learning_rate": 4.791879236568013e-05,
      "loss": 0.8701,
      "step": 9408
    },
    {
      "epoch": 0.13344158031134004,
      "grad_norm": 2.875,
      "learning_rate": 4.791789978223391e-05,
      "loss": 0.8424,
      "step": 9410
    },
    {
      "epoch": 0.13346994196496625,
      "grad_norm": 3.078125,
      "learning_rate": 4.791700701574089e-05,
      "loss": 0.9,
      "step": 9412
    },
    {
      "epoch": 0.1334983036185925,
      "grad_norm": 3.171875,
      "learning_rate": 4.791611406620821e-05,
      "loss": 0.8082,
      "step": 9414
    },
    {
      "epoch": 0.1335266652722187,
      "grad_norm": 3.15625,
      "learning_rate": 4.791522093364299e-05,
      "loss": 0.8809,
      "step": 9416
    },
    {
      "epoch": 0.13355502692584492,
      "grad_norm": 3.34375,
      "learning_rate": 4.791432761805238e-05,
      "loss": 0.9061,
      "step": 9418
    },
    {
      "epoch": 0.13358338857947114,
      "grad_norm": 3.625,
      "learning_rate": 4.79134341194435e-05,
      "loss": 0.8561,
      "step": 9420
    },
    {
      "epoch": 0.13361175023309735,
      "grad_norm": 3.265625,
      "learning_rate": 4.7912540437823506e-05,
      "loss": 0.8975,
      "step": 9422
    },
    {
      "epoch": 0.13364011188672356,
      "grad_norm": 3.109375,
      "learning_rate": 4.791164657319951e-05,
      "loss": 0.8804,
      "step": 9424
    },
    {
      "epoch": 0.13366847354034977,
      "grad_norm": 2.875,
      "learning_rate": 4.7910752525578676e-05,
      "loss": 0.7955,
      "step": 9426
    },
    {
      "epoch": 0.133696835193976,
      "grad_norm": 2.796875,
      "learning_rate": 4.790985829496812e-05,
      "loss": 0.8097,
      "step": 9428
    },
    {
      "epoch": 0.1337251968476022,
      "grad_norm": 3.21875,
      "learning_rate": 4.790896388137501e-05,
      "loss": 0.7972,
      "step": 9430
    },
    {
      "epoch": 0.13375355850122841,
      "grad_norm": 3.3125,
      "learning_rate": 4.790806928480647e-05,
      "loss": 0.8619,
      "step": 9432
    },
    {
      "epoch": 0.13378192015485463,
      "grad_norm": 2.921875,
      "learning_rate": 4.790717450526966e-05,
      "loss": 0.8208,
      "step": 9434
    },
    {
      "epoch": 0.13381028180848084,
      "grad_norm": 3.03125,
      "learning_rate": 4.7906279542771714e-05,
      "loss": 0.8632,
      "step": 9436
    },
    {
      "epoch": 0.13383864346210705,
      "grad_norm": 2.71875,
      "learning_rate": 4.7905384397319784e-05,
      "loss": 0.8662,
      "step": 9438
    },
    {
      "epoch": 0.13386700511573327,
      "grad_norm": 3.125,
      "learning_rate": 4.790448906892102e-05,
      "loss": 0.8648,
      "step": 9440
    },
    {
      "epoch": 0.13389536676935948,
      "grad_norm": 3.296875,
      "learning_rate": 4.790359355758258e-05,
      "loss": 0.8868,
      "step": 9442
    },
    {
      "epoch": 0.1339237284229857,
      "grad_norm": 3.265625,
      "learning_rate": 4.790269786331161e-05,
      "loss": 0.864,
      "step": 9444
    },
    {
      "epoch": 0.1339520900766119,
      "grad_norm": 3.0625,
      "learning_rate": 4.7901801986115254e-05,
      "loss": 0.858,
      "step": 9446
    },
    {
      "epoch": 0.13398045173023812,
      "grad_norm": 2.921875,
      "learning_rate": 4.790090592600069e-05,
      "loss": 0.8854,
      "step": 9448
    },
    {
      "epoch": 0.13400881338386433,
      "grad_norm": 3.0625,
      "learning_rate": 4.7900009682975056e-05,
      "loss": 0.8537,
      "step": 9450
    },
    {
      "epoch": 0.13403717503749057,
      "grad_norm": 3.59375,
      "learning_rate": 4.789911325704552e-05,
      "loss": 0.8308,
      "step": 9452
    },
    {
      "epoch": 0.1340655366911168,
      "grad_norm": 2.96875,
      "learning_rate": 4.789821664821924e-05,
      "loss": 0.8557,
      "step": 9454
    },
    {
      "epoch": 0.134093898344743,
      "grad_norm": 3.046875,
      "learning_rate": 4.789731985650338e-05,
      "loss": 0.8707,
      "step": 9456
    },
    {
      "epoch": 0.13412225999836921,
      "grad_norm": 3.265625,
      "learning_rate": 4.789642288190509e-05,
      "loss": 0.8556,
      "step": 9458
    },
    {
      "epoch": 0.13415062165199543,
      "grad_norm": 2.71875,
      "learning_rate": 4.7895525724431546e-05,
      "loss": 0.8327,
      "step": 9460
    },
    {
      "epoch": 0.13417898330562164,
      "grad_norm": 3.28125,
      "learning_rate": 4.789462838408991e-05,
      "loss": 0.8553,
      "step": 9462
    },
    {
      "epoch": 0.13420734495924785,
      "grad_norm": 3.140625,
      "learning_rate": 4.789373086088736e-05,
      "loss": 0.9202,
      "step": 9464
    },
    {
      "epoch": 0.13423570661287407,
      "grad_norm": 2.890625,
      "learning_rate": 4.789283315483104e-05,
      "loss": 0.8443,
      "step": 9466
    },
    {
      "epoch": 0.13426406826650028,
      "grad_norm": 2.796875,
      "learning_rate": 4.7891935265928146e-05,
      "loss": 0.8217,
      "step": 9468
    },
    {
      "epoch": 0.1342924299201265,
      "grad_norm": 2.96875,
      "learning_rate": 4.7891037194185834e-05,
      "loss": 0.8868,
      "step": 9470
    },
    {
      "epoch": 0.1343207915737527,
      "grad_norm": 3.234375,
      "learning_rate": 4.7890138939611285e-05,
      "loss": 0.8625,
      "step": 9472
    },
    {
      "epoch": 0.13434915322737892,
      "grad_norm": 3.78125,
      "learning_rate": 4.788924050221166e-05,
      "loss": 0.8653,
      "step": 9474
    },
    {
      "epoch": 0.13437751488100513,
      "grad_norm": 3.21875,
      "learning_rate": 4.788834188199415e-05,
      "loss": 0.8778,
      "step": 9476
    },
    {
      "epoch": 0.13440587653463135,
      "grad_norm": 3.03125,
      "learning_rate": 4.788744307896593e-05,
      "loss": 0.8455,
      "step": 9478
    },
    {
      "epoch": 0.13443423818825756,
      "grad_norm": 3.25,
      "learning_rate": 4.788654409313417e-05,
      "loss": 0.8521,
      "step": 9480
    },
    {
      "epoch": 0.13446259984188377,
      "grad_norm": 2.828125,
      "learning_rate": 4.788564492450606e-05,
      "loss": 0.8577,
      "step": 9482
    },
    {
      "epoch": 0.13449096149551,
      "grad_norm": 3.015625,
      "learning_rate": 4.788474557308877e-05,
      "loss": 0.8495,
      "step": 9484
    },
    {
      "epoch": 0.1345193231491362,
      "grad_norm": 3.21875,
      "learning_rate": 4.78838460388895e-05,
      "loss": 0.8899,
      "step": 9486
    },
    {
      "epoch": 0.1345476848027624,
      "grad_norm": 3.203125,
      "learning_rate": 4.788294632191542e-05,
      "loss": 0.8866,
      "step": 9488
    },
    {
      "epoch": 0.13457604645638863,
      "grad_norm": 3.21875,
      "learning_rate": 4.788204642217373e-05,
      "loss": 0.9318,
      "step": 9490
    },
    {
      "epoch": 0.13460440811001487,
      "grad_norm": 3.046875,
      "learning_rate": 4.78811463396716e-05,
      "loss": 0.8219,
      "step": 9492
    },
    {
      "epoch": 0.13463276976364108,
      "grad_norm": 2.921875,
      "learning_rate": 4.7880246074416224e-05,
      "loss": 0.8227,
      "step": 9494
    },
    {
      "epoch": 0.1346611314172673,
      "grad_norm": 3.296875,
      "learning_rate": 4.7879345626414804e-05,
      "loss": 0.9071,
      "step": 9496
    },
    {
      "epoch": 0.1346894930708935,
      "grad_norm": 3.15625,
      "learning_rate": 4.787844499567452e-05,
      "loss": 0.8425,
      "step": 9498
    },
    {
      "epoch": 0.13471785472451972,
      "grad_norm": 3.046875,
      "learning_rate": 4.787754418220258e-05,
      "loss": 0.8426,
      "step": 9500
    },
    {
      "epoch": 0.13474621637814593,
      "grad_norm": 3.234375,
      "learning_rate": 4.787664318600615e-05,
      "loss": 0.8813,
      "step": 9502
    },
    {
      "epoch": 0.13477457803177215,
      "grad_norm": 3.1875,
      "learning_rate": 4.787574200709246e-05,
      "loss": 0.8562,
      "step": 9504
    },
    {
      "epoch": 0.13480293968539836,
      "grad_norm": 2.828125,
      "learning_rate": 4.787484064546869e-05,
      "loss": 0.8287,
      "step": 9506
    },
    {
      "epoch": 0.13483130133902457,
      "grad_norm": 3.5,
      "learning_rate": 4.787393910114204e-05,
      "loss": 0.8647,
      "step": 9508
    },
    {
      "epoch": 0.1348596629926508,
      "grad_norm": 3.4375,
      "learning_rate": 4.7873037374119716e-05,
      "loss": 0.8606,
      "step": 9510
    },
    {
      "epoch": 0.134888024646277,
      "grad_norm": 3.125,
      "learning_rate": 4.787213546440892e-05,
      "loss": 0.8586,
      "step": 9512
    },
    {
      "epoch": 0.1349163862999032,
      "grad_norm": 3.21875,
      "learning_rate": 4.7871233372016844e-05,
      "loss": 0.8125,
      "step": 9514
    },
    {
      "epoch": 0.13494474795352943,
      "grad_norm": 3.03125,
      "learning_rate": 4.7870331096950706e-05,
      "loss": 0.8427,
      "step": 9516
    },
    {
      "epoch": 0.13497310960715564,
      "grad_norm": 3.359375,
      "learning_rate": 4.7869428639217706e-05,
      "loss": 0.8612,
      "step": 9518
    },
    {
      "epoch": 0.13500147126078185,
      "grad_norm": 3.40625,
      "learning_rate": 4.786852599882505e-05,
      "loss": 0.891,
      "step": 9520
    },
    {
      "epoch": 0.13502983291440807,
      "grad_norm": 2.734375,
      "learning_rate": 4.7867623175779955e-05,
      "loss": 0.8563,
      "step": 9522
    },
    {
      "epoch": 0.13505819456803428,
      "grad_norm": 2.90625,
      "learning_rate": 4.786672017008963e-05,
      "loss": 0.8838,
      "step": 9524
    },
    {
      "epoch": 0.1350865562216605,
      "grad_norm": 3.09375,
      "learning_rate": 4.786581698176129e-05,
      "loss": 0.8651,
      "step": 9526
    },
    {
      "epoch": 0.1351149178752867,
      "grad_norm": 2.84375,
      "learning_rate": 4.786491361080214e-05,
      "loss": 0.8792,
      "step": 9528
    },
    {
      "epoch": 0.13514327952891295,
      "grad_norm": 2.90625,
      "learning_rate": 4.78640100572194e-05,
      "loss": 0.8631,
      "step": 9530
    },
    {
      "epoch": 0.13517164118253916,
      "grad_norm": 3.21875,
      "learning_rate": 4.7863106321020285e-05,
      "loss": 0.8525,
      "step": 9532
    },
    {
      "epoch": 0.13520000283616537,
      "grad_norm": 2.828125,
      "learning_rate": 4.786220240221202e-05,
      "loss": 0.8351,
      "step": 9534
    },
    {
      "epoch": 0.13522836448979159,
      "grad_norm": 3.140625,
      "learning_rate": 4.786129830080181e-05,
      "loss": 0.8511,
      "step": 9536
    },
    {
      "epoch": 0.1352567261434178,
      "grad_norm": 2.9375,
      "learning_rate": 4.78603940167969e-05,
      "loss": 0.8393,
      "step": 9538
    },
    {
      "epoch": 0.135285087797044,
      "grad_norm": 2.796875,
      "learning_rate": 4.785948955020448e-05,
      "loss": 0.8694,
      "step": 9540
    },
    {
      "epoch": 0.13531344945067023,
      "grad_norm": 2.90625,
      "learning_rate": 4.78585849010318e-05,
      "loss": 0.8945,
      "step": 9542
    },
    {
      "epoch": 0.13534181110429644,
      "grad_norm": 3.09375,
      "learning_rate": 4.785768006928608e-05,
      "loss": 0.9118,
      "step": 9544
    },
    {
      "epoch": 0.13537017275792265,
      "grad_norm": 3.015625,
      "learning_rate": 4.785677505497454e-05,
      "loss": 0.8287,
      "step": 9546
    },
    {
      "epoch": 0.13539853441154887,
      "grad_norm": 3.0,
      "learning_rate": 4.7855869858104426e-05,
      "loss": 0.8384,
      "step": 9548
    },
    {
      "epoch": 0.13542689606517508,
      "grad_norm": 3.015625,
      "learning_rate": 4.785496447868295e-05,
      "loss": 0.8659,
      "step": 9550
    },
    {
      "epoch": 0.1354552577188013,
      "grad_norm": 3.5625,
      "learning_rate": 4.7854058916717336e-05,
      "loss": 0.8575,
      "step": 9552
    },
    {
      "epoch": 0.1354836193724275,
      "grad_norm": 2.703125,
      "learning_rate": 4.785315317221484e-05,
      "loss": 0.8098,
      "step": 9554
    },
    {
      "epoch": 0.13551198102605372,
      "grad_norm": 3.0,
      "learning_rate": 4.785224724518268e-05,
      "loss": 0.8222,
      "step": 9556
    },
    {
      "epoch": 0.13554034267967993,
      "grad_norm": 3.1875,
      "learning_rate": 4.78513411356281e-05,
      "loss": 0.8712,
      "step": 9558
    },
    {
      "epoch": 0.13556870433330614,
      "grad_norm": 3.015625,
      "learning_rate": 4.7850434843558335e-05,
      "loss": 0.8667,
      "step": 9560
    },
    {
      "epoch": 0.13559706598693236,
      "grad_norm": 3.046875,
      "learning_rate": 4.784952836898062e-05,
      "loss": 0.8606,
      "step": 9562
    },
    {
      "epoch": 0.13562542764055857,
      "grad_norm": 3.234375,
      "learning_rate": 4.78486217119022e-05,
      "loss": 0.8305,
      "step": 9564
    },
    {
      "epoch": 0.13565378929418478,
      "grad_norm": 3.109375,
      "learning_rate": 4.7847714872330316e-05,
      "loss": 0.8512,
      "step": 9566
    },
    {
      "epoch": 0.13568215094781103,
      "grad_norm": 3.171875,
      "learning_rate": 4.784680785027221e-05,
      "loss": 0.8591,
      "step": 9568
    },
    {
      "epoch": 0.13571051260143724,
      "grad_norm": 3.0,
      "learning_rate": 4.784590064573512e-05,
      "loss": 0.7925,
      "step": 9570
    },
    {
      "epoch": 0.13573887425506345,
      "grad_norm": 3.09375,
      "learning_rate": 4.78449932587263e-05,
      "loss": 0.9063,
      "step": 9572
    },
    {
      "epoch": 0.13576723590868967,
      "grad_norm": 3.515625,
      "learning_rate": 4.7844085689253e-05,
      "loss": 0.8593,
      "step": 9574
    },
    {
      "epoch": 0.13579559756231588,
      "grad_norm": 3.09375,
      "learning_rate": 4.784317793732246e-05,
      "loss": 0.8915,
      "step": 9576
    },
    {
      "epoch": 0.1358239592159421,
      "grad_norm": 2.96875,
      "learning_rate": 4.7842270002941934e-05,
      "loss": 0.8596,
      "step": 9578
    },
    {
      "epoch": 0.1358523208695683,
      "grad_norm": 3.421875,
      "learning_rate": 4.784136188611867e-05,
      "loss": 0.8806,
      "step": 9580
    },
    {
      "epoch": 0.13588068252319452,
      "grad_norm": 3.0,
      "learning_rate": 4.784045358685993e-05,
      "loss": 0.8176,
      "step": 9582
    },
    {
      "epoch": 0.13590904417682073,
      "grad_norm": 3.015625,
      "learning_rate": 4.783954510517296e-05,
      "loss": 0.8076,
      "step": 9584
    },
    {
      "epoch": 0.13593740583044694,
      "grad_norm": 3.390625,
      "learning_rate": 4.783863644106502e-05,
      "loss": 0.8612,
      "step": 9586
    },
    {
      "epoch": 0.13596576748407316,
      "grad_norm": 3.109375,
      "learning_rate": 4.7837727594543374e-05,
      "loss": 0.9014,
      "step": 9588
    },
    {
      "epoch": 0.13599412913769937,
      "grad_norm": 3.375,
      "learning_rate": 4.783681856561527e-05,
      "loss": 0.8419,
      "step": 9590
    },
    {
      "epoch": 0.13602249079132558,
      "grad_norm": 3.09375,
      "learning_rate": 4.7835909354287975e-05,
      "loss": 0.8436,
      "step": 9592
    },
    {
      "epoch": 0.1360508524449518,
      "grad_norm": 2.9375,
      "learning_rate": 4.783499996056875e-05,
      "loss": 0.8537,
      "step": 9594
    },
    {
      "epoch": 0.136079214098578,
      "grad_norm": 3.34375,
      "learning_rate": 4.783409038446485e-05,
      "loss": 0.8933,
      "step": 9596
    },
    {
      "epoch": 0.13610757575220422,
      "grad_norm": 3.109375,
      "learning_rate": 4.783318062598355e-05,
      "loss": 0.9062,
      "step": 9598
    },
    {
      "epoch": 0.13613593740583044,
      "grad_norm": 3.4375,
      "learning_rate": 4.783227068513212e-05,
      "loss": 0.8731,
      "step": 9600
    },
    {
      "epoch": 0.13616429905945665,
      "grad_norm": 2.953125,
      "learning_rate": 4.783136056191781e-05,
      "loss": 0.8348,
      "step": 9602
    },
    {
      "epoch": 0.13619266071308286,
      "grad_norm": 3.5,
      "learning_rate": 4.7830450256347907e-05,
      "loss": 0.8867,
      "step": 9604
    },
    {
      "epoch": 0.1362210223667091,
      "grad_norm": 3.0625,
      "learning_rate": 4.782953976842968e-05,
      "loss": 0.8585,
      "step": 9606
    },
    {
      "epoch": 0.13624938402033532,
      "grad_norm": 3.296875,
      "learning_rate": 4.7828629098170384e-05,
      "loss": 0.8652,
      "step": 9608
    },
    {
      "epoch": 0.13627774567396153,
      "grad_norm": 3.359375,
      "learning_rate": 4.7827718245577313e-05,
      "loss": 0.8762,
      "step": 9610
    },
    {
      "epoch": 0.13630610732758774,
      "grad_norm": 3.046875,
      "learning_rate": 4.782680721065773e-05,
      "loss": 0.8931,
      "step": 9612
    },
    {
      "epoch": 0.13633446898121396,
      "grad_norm": 2.96875,
      "learning_rate": 4.7825895993418915e-05,
      "loss": 0.8768,
      "step": 9614
    },
    {
      "epoch": 0.13636283063484017,
      "grad_norm": 3.125,
      "learning_rate": 4.782498459386815e-05,
      "loss": 0.8769,
      "step": 9616
    },
    {
      "epoch": 0.13639119228846638,
      "grad_norm": 3.125,
      "learning_rate": 4.7824073012012704e-05,
      "loss": 0.8461,
      "step": 9618
    },
    {
      "epoch": 0.1364195539420926,
      "grad_norm": 3.046875,
      "learning_rate": 4.782316124785987e-05,
      "loss": 0.84,
      "step": 9620
    },
    {
      "epoch": 0.1364479155957188,
      "grad_norm": 2.796875,
      "learning_rate": 4.782224930141693e-05,
      "loss": 0.8248,
      "step": 9622
    },
    {
      "epoch": 0.13647627724934502,
      "grad_norm": 2.75,
      "learning_rate": 4.782133717269115e-05,
      "loss": 0.828,
      "step": 9624
    },
    {
      "epoch": 0.13650463890297124,
      "grad_norm": 3.296875,
      "learning_rate": 4.7820424861689835e-05,
      "loss": 0.8086,
      "step": 9626
    },
    {
      "epoch": 0.13653300055659745,
      "grad_norm": 2.9375,
      "learning_rate": 4.7819512368420265e-05,
      "loss": 0.8648,
      "step": 9628
    },
    {
      "epoch": 0.13656136221022366,
      "grad_norm": 2.953125,
      "learning_rate": 4.781859969288973e-05,
      "loss": 0.8266,
      "step": 9630
    },
    {
      "epoch": 0.13658972386384988,
      "grad_norm": 2.8125,
      "learning_rate": 4.7817686835105513e-05,
      "loss": 0.8522,
      "step": 9632
    },
    {
      "epoch": 0.1366180855174761,
      "grad_norm": 3.421875,
      "learning_rate": 4.781677379507491e-05,
      "loss": 0.8819,
      "step": 9634
    },
    {
      "epoch": 0.1366464471711023,
      "grad_norm": 2.796875,
      "learning_rate": 4.781586057280521e-05,
      "loss": 0.9147,
      "step": 9636
    },
    {
      "epoch": 0.13667480882472852,
      "grad_norm": 2.828125,
      "learning_rate": 4.7814947168303716e-05,
      "loss": 0.8184,
      "step": 9638
    },
    {
      "epoch": 0.13670317047835473,
      "grad_norm": 3.3125,
      "learning_rate": 4.781403358157771e-05,
      "loss": 0.9015,
      "step": 9640
    },
    {
      "epoch": 0.13673153213198094,
      "grad_norm": 3.078125,
      "learning_rate": 4.78131198126345e-05,
      "loss": 0.8835,
      "step": 9642
    },
    {
      "epoch": 0.13675989378560716,
      "grad_norm": 3.03125,
      "learning_rate": 4.781220586148138e-05,
      "loss": 0.8635,
      "step": 9644
    },
    {
      "epoch": 0.1367882554392334,
      "grad_norm": 3.25,
      "learning_rate": 4.781129172812565e-05,
      "loss": 0.9092,
      "step": 9646
    },
    {
      "epoch": 0.1368166170928596,
      "grad_norm": 2.84375,
      "learning_rate": 4.781037741257461e-05,
      "loss": 0.8278,
      "step": 9648
    },
    {
      "epoch": 0.13684497874648582,
      "grad_norm": 2.765625,
      "learning_rate": 4.780946291483557e-05,
      "loss": 0.8255,
      "step": 9650
    },
    {
      "epoch": 0.13687334040011204,
      "grad_norm": 3.078125,
      "learning_rate": 4.7808548234915826e-05,
      "loss": 0.807,
      "step": 9652
    },
    {
      "epoch": 0.13690170205373825,
      "grad_norm": 3.015625,
      "learning_rate": 4.780763337282268e-05,
      "loss": 0.8346,
      "step": 9654
    },
    {
      "epoch": 0.13693006370736446,
      "grad_norm": 3.1875,
      "learning_rate": 4.780671832856345e-05,
      "loss": 0.9045,
      "step": 9656
    },
    {
      "epoch": 0.13695842536099068,
      "grad_norm": 3.15625,
      "learning_rate": 4.780580310214544e-05,
      "loss": 0.8907,
      "step": 9658
    },
    {
      "epoch": 0.1369867870146169,
      "grad_norm": 3.078125,
      "learning_rate": 4.780488769357596e-05,
      "loss": 0.8827,
      "step": 9660
    },
    {
      "epoch": 0.1370151486682431,
      "grad_norm": 3.03125,
      "learning_rate": 4.7803972102862314e-05,
      "loss": 0.8603,
      "step": 9662
    },
    {
      "epoch": 0.13704351032186932,
      "grad_norm": 2.8125,
      "learning_rate": 4.780305633001183e-05,
      "loss": 0.8513,
      "step": 9664
    },
    {
      "epoch": 0.13707187197549553,
      "grad_norm": 3.0625,
      "learning_rate": 4.780214037503181e-05,
      "loss": 0.8648,
      "step": 9666
    },
    {
      "epoch": 0.13710023362912174,
      "grad_norm": 2.84375,
      "learning_rate": 4.7801224237929574e-05,
      "loss": 0.8421,
      "step": 9668
    },
    {
      "epoch": 0.13712859528274796,
      "grad_norm": 2.71875,
      "learning_rate": 4.780030791871244e-05,
      "loss": 0.8696,
      "step": 9670
    },
    {
      "epoch": 0.13715695693637417,
      "grad_norm": 3.078125,
      "learning_rate": 4.7799391417387727e-05,
      "loss": 0.8732,
      "step": 9672
    },
    {
      "epoch": 0.13718531859000038,
      "grad_norm": 3.015625,
      "learning_rate": 4.779847473396275e-05,
      "loss": 0.8645,
      "step": 9674
    },
    {
      "epoch": 0.1372136802436266,
      "grad_norm": 3.125,
      "learning_rate": 4.779755786844484e-05,
      "loss": 0.892,
      "step": 9676
    },
    {
      "epoch": 0.1372420418972528,
      "grad_norm": 3.125,
      "learning_rate": 4.779664082084131e-05,
      "loss": 0.8746,
      "step": 9678
    },
    {
      "epoch": 0.13727040355087902,
      "grad_norm": 3.3125,
      "learning_rate": 4.779572359115949e-05,
      "loss": 0.8194,
      "step": 9680
    },
    {
      "epoch": 0.13729876520450524,
      "grad_norm": 2.953125,
      "learning_rate": 4.77948061794067e-05,
      "loss": 0.8346,
      "step": 9682
    },
    {
      "epoch": 0.13732712685813148,
      "grad_norm": 3.296875,
      "learning_rate": 4.779388858559028e-05,
      "loss": 0.8399,
      "step": 9684
    },
    {
      "epoch": 0.1373554885117577,
      "grad_norm": 3.46875,
      "learning_rate": 4.779297080971756e-05,
      "loss": 0.8884,
      "step": 9686
    },
    {
      "epoch": 0.1373838501653839,
      "grad_norm": 2.9375,
      "learning_rate": 4.7792052851795856e-05,
      "loss": 0.8905,
      "step": 9688
    },
    {
      "epoch": 0.13741221181901012,
      "grad_norm": 2.9375,
      "learning_rate": 4.77911347118325e-05,
      "loss": 0.8121,
      "step": 9690
    },
    {
      "epoch": 0.13744057347263633,
      "grad_norm": 2.9375,
      "learning_rate": 4.779021638983483e-05,
      "loss": 0.8386,
      "step": 9692
    },
    {
      "epoch": 0.13746893512626254,
      "grad_norm": 3.484375,
      "learning_rate": 4.778929788581018e-05,
      "loss": 0.8886,
      "step": 9694
    },
    {
      "epoch": 0.13749729677988876,
      "grad_norm": 3.484375,
      "learning_rate": 4.77883791997659e-05,
      "loss": 0.8446,
      "step": 9696
    },
    {
      "epoch": 0.13752565843351497,
      "grad_norm": 2.9375,
      "learning_rate": 4.77874603317093e-05,
      "loss": 0.8943,
      "step": 9698
    },
    {
      "epoch": 0.13755402008714118,
      "grad_norm": 3.0,
      "learning_rate": 4.7786541281647745e-05,
      "loss": 0.8148,
      "step": 9700
    },
    {
      "epoch": 0.1375823817407674,
      "grad_norm": 3.375,
      "learning_rate": 4.778562204958856e-05,
      "loss": 0.9129,
      "step": 9702
    },
    {
      "epoch": 0.1376107433943936,
      "grad_norm": 3.234375,
      "learning_rate": 4.778470263553909e-05,
      "loss": 0.8649,
      "step": 9704
    },
    {
      "epoch": 0.13763910504801982,
      "grad_norm": 2.953125,
      "learning_rate": 4.778378303950669e-05,
      "loss": 0.8664,
      "step": 9706
    },
    {
      "epoch": 0.13766746670164604,
      "grad_norm": 3.84375,
      "learning_rate": 4.7782863261498686e-05,
      "loss": 0.9259,
      "step": 9708
    },
    {
      "epoch": 0.13769582835527225,
      "grad_norm": 3.015625,
      "learning_rate": 4.778194330152243e-05,
      "loss": 0.8506,
      "step": 9710
    },
    {
      "epoch": 0.13772419000889846,
      "grad_norm": 3.09375,
      "learning_rate": 4.7781023159585295e-05,
      "loss": 0.9133,
      "step": 9712
    },
    {
      "epoch": 0.13775255166252467,
      "grad_norm": 3.28125,
      "learning_rate": 4.7780102835694583e-05,
      "loss": 0.8754,
      "step": 9714
    },
    {
      "epoch": 0.1377809133161509,
      "grad_norm": 3.0625,
      "learning_rate": 4.7779182329857685e-05,
      "loss": 0.8788,
      "step": 9716
    },
    {
      "epoch": 0.1378092749697771,
      "grad_norm": 3.5,
      "learning_rate": 4.777826164208193e-05,
      "loss": 0.9233,
      "step": 9718
    },
    {
      "epoch": 0.13783763662340331,
      "grad_norm": 3.0,
      "learning_rate": 4.777734077237469e-05,
      "loss": 0.8675,
      "step": 9720
    },
    {
      "epoch": 0.13786599827702956,
      "grad_norm": 3.03125,
      "learning_rate": 4.777641972074331e-05,
      "loss": 0.8138,
      "step": 9722
    },
    {
      "epoch": 0.13789435993065577,
      "grad_norm": 3.5,
      "learning_rate": 4.7775498487195135e-05,
      "loss": 0.9171,
      "step": 9724
    },
    {
      "epoch": 0.13792272158428198,
      "grad_norm": 3.25,
      "learning_rate": 4.7774577071737536e-05,
      "loss": 0.8829,
      "step": 9726
    },
    {
      "epoch": 0.1379510832379082,
      "grad_norm": 2.78125,
      "learning_rate": 4.777365547437788e-05,
      "loss": 0.823,
      "step": 9728
    },
    {
      "epoch": 0.1379794448915344,
      "grad_norm": 3.03125,
      "learning_rate": 4.777273369512351e-05,
      "loss": 0.7923,
      "step": 9730
    },
    {
      "epoch": 0.13800780654516062,
      "grad_norm": 4.0625,
      "learning_rate": 4.7771811733981797e-05,
      "loss": 0.8381,
      "step": 9732
    },
    {
      "epoch": 0.13803616819878683,
      "grad_norm": 2.921875,
      "learning_rate": 4.777088959096011e-05,
      "loss": 0.8322,
      "step": 9734
    },
    {
      "epoch": 0.13806452985241305,
      "grad_norm": 3.1875,
      "learning_rate": 4.776996726606581e-05,
      "loss": 0.8754,
      "step": 9736
    },
    {
      "epoch": 0.13809289150603926,
      "grad_norm": 3.0,
      "learning_rate": 4.7769044759306256e-05,
      "loss": 0.8735,
      "step": 9738
    },
    {
      "epoch": 0.13812125315966547,
      "grad_norm": 3.015625,
      "learning_rate": 4.776812207068883e-05,
      "loss": 0.8242,
      "step": 9740
    },
    {
      "epoch": 0.1381496148132917,
      "grad_norm": 3.0625,
      "learning_rate": 4.776719920022089e-05,
      "loss": 0.8891,
      "step": 9742
    },
    {
      "epoch": 0.1381779764669179,
      "grad_norm": 2.875,
      "learning_rate": 4.776627614790982e-05,
      "loss": 0.8782,
      "step": 9744
    },
    {
      "epoch": 0.13820633812054411,
      "grad_norm": 3.34375,
      "learning_rate": 4.776535291376297e-05,
      "loss": 0.863,
      "step": 9746
    },
    {
      "epoch": 0.13823469977417033,
      "grad_norm": 3.0625,
      "learning_rate": 4.776442949778773e-05,
      "loss": 0.8439,
      "step": 9748
    },
    {
      "epoch": 0.13826306142779654,
      "grad_norm": 2.921875,
      "learning_rate": 4.776350589999148e-05,
      "loss": 0.8195,
      "step": 9750
    },
    {
      "epoch": 0.13829142308142275,
      "grad_norm": 2.8125,
      "learning_rate": 4.776258212038159e-05,
      "loss": 0.8392,
      "step": 9752
    },
    {
      "epoch": 0.13831978473504897,
      "grad_norm": 3.140625,
      "learning_rate": 4.776165815896544e-05,
      "loss": 0.8137,
      "step": 9754
    },
    {
      "epoch": 0.13834814638867518,
      "grad_norm": 2.953125,
      "learning_rate": 4.77607340157504e-05,
      "loss": 0.8081,
      "step": 9756
    },
    {
      "epoch": 0.1383765080423014,
      "grad_norm": 2.90625,
      "learning_rate": 4.775980969074386e-05,
      "loss": 0.8619,
      "step": 9758
    },
    {
      "epoch": 0.13840486969592763,
      "grad_norm": 3.234375,
      "learning_rate": 4.775888518395321e-05,
      "loss": 0.8601,
      "step": 9760
    },
    {
      "epoch": 0.13843323134955385,
      "grad_norm": 3.328125,
      "learning_rate": 4.775796049538582e-05,
      "loss": 0.9255,
      "step": 9762
    },
    {
      "epoch": 0.13846159300318006,
      "grad_norm": 3.3125,
      "learning_rate": 4.7757035625049084e-05,
      "loss": 0.8264,
      "step": 9764
    },
    {
      "epoch": 0.13848995465680627,
      "grad_norm": 2.90625,
      "learning_rate": 4.775611057295039e-05,
      "loss": 0.8394,
      "step": 9766
    },
    {
      "epoch": 0.1385183163104325,
      "grad_norm": 3.15625,
      "learning_rate": 4.775518533909711e-05,
      "loss": 0.8104,
      "step": 9768
    },
    {
      "epoch": 0.1385466779640587,
      "grad_norm": 3.328125,
      "learning_rate": 4.775425992349666e-05,
      "loss": 0.8605,
      "step": 9770
    },
    {
      "epoch": 0.13857503961768491,
      "grad_norm": 2.90625,
      "learning_rate": 4.775333432615641e-05,
      "loss": 0.8251,
      "step": 9772
    },
    {
      "epoch": 0.13860340127131113,
      "grad_norm": 3.21875,
      "learning_rate": 4.775240854708376e-05,
      "loss": 0.8508,
      "step": 9774
    },
    {
      "epoch": 0.13863176292493734,
      "grad_norm": 3.109375,
      "learning_rate": 4.775148258628611e-05,
      "loss": 0.7978,
      "step": 9776
    },
    {
      "epoch": 0.13866012457856355,
      "grad_norm": 3.4375,
      "learning_rate": 4.775055644377086e-05,
      "loss": 0.8389,
      "step": 9778
    },
    {
      "epoch": 0.13868848623218977,
      "grad_norm": 3.21875,
      "learning_rate": 4.774963011954538e-05,
      "loss": 0.8475,
      "step": 9780
    },
    {
      "epoch": 0.13871684788581598,
      "grad_norm": 3.21875,
      "learning_rate": 4.7748703613617095e-05,
      "loss": 0.8287,
      "step": 9782
    },
    {
      "epoch": 0.1387452095394422,
      "grad_norm": 3.34375,
      "learning_rate": 4.7747776925993404e-05,
      "loss": 0.8915,
      "step": 9784
    },
    {
      "epoch": 0.1387735711930684,
      "grad_norm": 3.546875,
      "learning_rate": 4.774685005668169e-05,
      "loss": 0.8942,
      "step": 9786
    },
    {
      "epoch": 0.13880193284669462,
      "grad_norm": 3.34375,
      "learning_rate": 4.7745923005689374e-05,
      "loss": 0.8831,
      "step": 9788
    },
    {
      "epoch": 0.13883029450032083,
      "grad_norm": 2.9375,
      "learning_rate": 4.774499577302385e-05,
      "loss": 0.9236,
      "step": 9790
    },
    {
      "epoch": 0.13885865615394705,
      "grad_norm": 2.78125,
      "learning_rate": 4.774406835869253e-05,
      "loss": 0.8147,
      "step": 9792
    },
    {
      "epoch": 0.13888701780757326,
      "grad_norm": 2.703125,
      "learning_rate": 4.774314076270282e-05,
      "loss": 0.8255,
      "step": 9794
    },
    {
      "epoch": 0.13891537946119947,
      "grad_norm": 3.5,
      "learning_rate": 4.774221298506213e-05,
      "loss": 0.8877,
      "step": 9796
    },
    {
      "epoch": 0.1389437411148257,
      "grad_norm": 3.1875,
      "learning_rate": 4.774128502577786e-05,
      "loss": 0.8505,
      "step": 9798
    },
    {
      "epoch": 0.13897210276845193,
      "grad_norm": 2.96875,
      "learning_rate": 4.774035688485744e-05,
      "loss": 0.8191,
      "step": 9800
    },
    {
      "epoch": 0.13900046442207814,
      "grad_norm": 3.203125,
      "learning_rate": 4.7739428562308266e-05,
      "loss": 0.8258,
      "step": 9802
    },
    {
      "epoch": 0.13902882607570435,
      "grad_norm": 3.0625,
      "learning_rate": 4.773850005813776e-05,
      "loss": 0.8342,
      "step": 9804
    },
    {
      "epoch": 0.13905718772933057,
      "grad_norm": 3.15625,
      "learning_rate": 4.7737571372353335e-05,
      "loss": 0.8472,
      "step": 9806
    },
    {
      "epoch": 0.13908554938295678,
      "grad_norm": 3.046875,
      "learning_rate": 4.773664250496241e-05,
      "loss": 0.8194,
      "step": 9808
    },
    {
      "epoch": 0.139113911036583,
      "grad_norm": 3.625,
      "learning_rate": 4.773571345597241e-05,
      "loss": 0.9342,
      "step": 9810
    },
    {
      "epoch": 0.1391422726902092,
      "grad_norm": 3.109375,
      "learning_rate": 4.773478422539075e-05,
      "loss": 0.8469,
      "step": 9812
    },
    {
      "epoch": 0.13917063434383542,
      "grad_norm": 3.0,
      "learning_rate": 4.7733854813224855e-05,
      "loss": 0.8441,
      "step": 9814
    },
    {
      "epoch": 0.13919899599746163,
      "grad_norm": 3.3125,
      "learning_rate": 4.773292521948214e-05,
      "loss": 0.8465,
      "step": 9816
    },
    {
      "epoch": 0.13922735765108785,
      "grad_norm": 3.234375,
      "learning_rate": 4.773199544417003e-05,
      "loss": 0.8675,
      "step": 9818
    },
    {
      "epoch": 0.13925571930471406,
      "grad_norm": 3.015625,
      "learning_rate": 4.7731065487295966e-05,
      "loss": 0.8517,
      "step": 9820
    },
    {
      "epoch": 0.13928408095834027,
      "grad_norm": 3.203125,
      "learning_rate": 4.773013534886735e-05,
      "loss": 0.8653,
      "step": 9822
    },
    {
      "epoch": 0.13931244261196649,
      "grad_norm": 3.15625,
      "learning_rate": 4.772920502889164e-05,
      "loss": 0.8876,
      "step": 9824
    },
    {
      "epoch": 0.1393408042655927,
      "grad_norm": 2.8125,
      "learning_rate": 4.772827452737626e-05,
      "loss": 0.8165,
      "step": 9826
    },
    {
      "epoch": 0.1393691659192189,
      "grad_norm": 3.0625,
      "learning_rate": 4.7727343844328626e-05,
      "loss": 0.8395,
      "step": 9828
    },
    {
      "epoch": 0.13939752757284513,
      "grad_norm": 2.828125,
      "learning_rate": 4.7726412979756183e-05,
      "loss": 0.9051,
      "step": 9830
    },
    {
      "epoch": 0.13942588922647134,
      "grad_norm": 3.140625,
      "learning_rate": 4.772548193366636e-05,
      "loss": 0.8455,
      "step": 9832
    },
    {
      "epoch": 0.13945425088009755,
      "grad_norm": 2.921875,
      "learning_rate": 4.7724550706066604e-05,
      "loss": 0.8069,
      "step": 9834
    },
    {
      "epoch": 0.13948261253372377,
      "grad_norm": 3.921875,
      "learning_rate": 4.772361929696434e-05,
      "loss": 0.8995,
      "step": 9836
    },
    {
      "epoch": 0.13951097418735,
      "grad_norm": 3.109375,
      "learning_rate": 4.772268770636701e-05,
      "loss": 0.8693,
      "step": 9838
    },
    {
      "epoch": 0.13953933584097622,
      "grad_norm": 3.3125,
      "learning_rate": 4.772175593428206e-05,
      "loss": 0.8537,
      "step": 9840
    },
    {
      "epoch": 0.13956769749460243,
      "grad_norm": 3.453125,
      "learning_rate": 4.7720823980716934e-05,
      "loss": 0.8723,
      "step": 9842
    },
    {
      "epoch": 0.13959605914822865,
      "grad_norm": 3.28125,
      "learning_rate": 4.771989184567907e-05,
      "loss": 0.8922,
      "step": 9844
    },
    {
      "epoch": 0.13962442080185486,
      "grad_norm": 2.828125,
      "learning_rate": 4.7718959529175915e-05,
      "loss": 0.8948,
      "step": 9846
    },
    {
      "epoch": 0.13965278245548107,
      "grad_norm": 3.71875,
      "learning_rate": 4.771802703121492e-05,
      "loss": 0.8548,
      "step": 9848
    },
    {
      "epoch": 0.13968114410910729,
      "grad_norm": 3.421875,
      "learning_rate": 4.7717094351803515e-05,
      "loss": 0.871,
      "step": 9850
    },
    {
      "epoch": 0.1397095057627335,
      "grad_norm": 2.71875,
      "learning_rate": 4.771616149094917e-05,
      "loss": 0.8462,
      "step": 9852
    },
    {
      "epoch": 0.1397378674163597,
      "grad_norm": 2.84375,
      "learning_rate": 4.771522844865932e-05,
      "loss": 0.8349,
      "step": 9854
    },
    {
      "epoch": 0.13976622906998593,
      "grad_norm": 3.453125,
      "learning_rate": 4.771429522494144e-05,
      "loss": 0.8386,
      "step": 9856
    },
    {
      "epoch": 0.13979459072361214,
      "grad_norm": 3.046875,
      "learning_rate": 4.771336181980296e-05,
      "loss": 0.8664,
      "step": 9858
    },
    {
      "epoch": 0.13982295237723835,
      "grad_norm": 3.453125,
      "learning_rate": 4.7712428233251336e-05,
      "loss": 0.9148,
      "step": 9860
    },
    {
      "epoch": 0.13985131403086457,
      "grad_norm": 3.125,
      "learning_rate": 4.771149446529405e-05,
      "loss": 0.8802,
      "step": 9862
    },
    {
      "epoch": 0.13987967568449078,
      "grad_norm": 2.984375,
      "learning_rate": 4.771056051593853e-05,
      "loss": 0.8807,
      "step": 9864
    },
    {
      "epoch": 0.139908037338117,
      "grad_norm": 3.0625,
      "learning_rate": 4.770962638519225e-05,
      "loss": 0.8685,
      "step": 9866
    },
    {
      "epoch": 0.1399363989917432,
      "grad_norm": 3.3125,
      "learning_rate": 4.7708692073062674e-05,
      "loss": 0.824,
      "step": 9868
    },
    {
      "epoch": 0.13996476064536942,
      "grad_norm": 3.25,
      "learning_rate": 4.770775757955726e-05,
      "loss": 0.8914,
      "step": 9870
    },
    {
      "epoch": 0.13999312229899563,
      "grad_norm": 3.09375,
      "learning_rate": 4.770682290468347e-05,
      "loss": 0.8658,
      "step": 9872
    },
    {
      "epoch": 0.14002148395262184,
      "grad_norm": 3.203125,
      "learning_rate": 4.7705888048448774e-05,
      "loss": 0.8861,
      "step": 9874
    },
    {
      "epoch": 0.14004984560624809,
      "grad_norm": 2.703125,
      "learning_rate": 4.770495301086063e-05,
      "loss": 0.8257,
      "step": 9876
    },
    {
      "epoch": 0.1400782072598743,
      "grad_norm": 3.265625,
      "learning_rate": 4.770401779192651e-05,
      "loss": 0.7909,
      "step": 9878
    },
    {
      "epoch": 0.1401065689135005,
      "grad_norm": 3.078125,
      "learning_rate": 4.770308239165389e-05,
      "loss": 0.8363,
      "step": 9880
    },
    {
      "epoch": 0.14013493056712673,
      "grad_norm": 3.171875,
      "learning_rate": 4.770214681005024e-05,
      "loss": 0.818,
      "step": 9882
    },
    {
      "epoch": 0.14016329222075294,
      "grad_norm": 3.1875,
      "learning_rate": 4.7701211047123025e-05,
      "loss": 0.8739,
      "step": 9884
    },
    {
      "epoch": 0.14019165387437915,
      "grad_norm": 3.015625,
      "learning_rate": 4.770027510287972e-05,
      "loss": 0.8339,
      "step": 9886
    },
    {
      "epoch": 0.14022001552800537,
      "grad_norm": 2.921875,
      "learning_rate": 4.769933897732781e-05,
      "loss": 0.8304,
      "step": 9888
    },
    {
      "epoch": 0.14024837718163158,
      "grad_norm": 2.890625,
      "learning_rate": 4.769840267047476e-05,
      "loss": 0.8929,
      "step": 9890
    },
    {
      "epoch": 0.1402767388352578,
      "grad_norm": 3.203125,
      "learning_rate": 4.769746618232805e-05,
      "loss": 0.8473,
      "step": 9892
    },
    {
      "epoch": 0.140305100488884,
      "grad_norm": 3.8125,
      "learning_rate": 4.769652951289518e-05,
      "loss": 0.8598,
      "step": 9894
    },
    {
      "epoch": 0.14033346214251022,
      "grad_norm": 3.65625,
      "learning_rate": 4.76955926621836e-05,
      "loss": 0.8372,
      "step": 9896
    },
    {
      "epoch": 0.14036182379613643,
      "grad_norm": 3.359375,
      "learning_rate": 4.769465563020081e-05,
      "loss": 0.8666,
      "step": 9898
    },
    {
      "epoch": 0.14039018544976264,
      "grad_norm": 3.0,
      "learning_rate": 4.7693718416954296e-05,
      "loss": 0.8561,
      "step": 9900
    },
    {
      "epoch": 0.14041854710338886,
      "grad_norm": 3.109375,
      "learning_rate": 4.7692781022451536e-05,
      "loss": 0.8811,
      "step": 9902
    },
    {
      "epoch": 0.14044690875701507,
      "grad_norm": 3.03125,
      "learning_rate": 4.7691843446700024e-05,
      "loss": 0.8468,
      "step": 9904
    },
    {
      "epoch": 0.14047527041064128,
      "grad_norm": 2.859375,
      "learning_rate": 4.7690905689707245e-05,
      "loss": 0.8298,
      "step": 9906
    },
    {
      "epoch": 0.1405036320642675,
      "grad_norm": 2.859375,
      "learning_rate": 4.7689967751480694e-05,
      "loss": 0.8544,
      "step": 9908
    },
    {
      "epoch": 0.1405319937178937,
      "grad_norm": 2.859375,
      "learning_rate": 4.768902963202784e-05,
      "loss": 0.8216,
      "step": 9910
    },
    {
      "epoch": 0.14056035537151992,
      "grad_norm": 2.84375,
      "learning_rate": 4.76880913313562e-05,
      "loss": 0.8521,
      "step": 9912
    },
    {
      "epoch": 0.14058871702514616,
      "grad_norm": 3.125,
      "learning_rate": 4.768715284947327e-05,
      "loss": 0.8552,
      "step": 9914
    },
    {
      "epoch": 0.14061707867877238,
      "grad_norm": 3.25,
      "learning_rate": 4.768621418638653e-05,
      "loss": 0.8592,
      "step": 9916
    },
    {
      "epoch": 0.1406454403323986,
      "grad_norm": 2.78125,
      "learning_rate": 4.768527534210349e-05,
      "loss": 0.8285,
      "step": 9918
    },
    {
      "epoch": 0.1406738019860248,
      "grad_norm": 3.796875,
      "learning_rate": 4.768433631663163e-05,
      "loss": 0.8686,
      "step": 9920
    },
    {
      "epoch": 0.14070216363965102,
      "grad_norm": 2.875,
      "learning_rate": 4.768339710997847e-05,
      "loss": 0.859,
      "step": 9922
    },
    {
      "epoch": 0.14073052529327723,
      "grad_norm": 3.09375,
      "learning_rate": 4.768245772215151e-05,
      "loss": 0.823,
      "step": 9924
    },
    {
      "epoch": 0.14075888694690344,
      "grad_norm": 3.21875,
      "learning_rate": 4.768151815315824e-05,
      "loss": 0.8552,
      "step": 9926
    },
    {
      "epoch": 0.14078724860052966,
      "grad_norm": 3.234375,
      "learning_rate": 4.768057840300617e-05,
      "loss": 0.8626,
      "step": 9928
    },
    {
      "epoch": 0.14081561025415587,
      "grad_norm": 3.046875,
      "learning_rate": 4.7679638471702817e-05,
      "loss": 0.8933,
      "step": 9930
    },
    {
      "epoch": 0.14084397190778208,
      "grad_norm": 2.828125,
      "learning_rate": 4.767869835925567e-05,
      "loss": 0.8456,
      "step": 9932
    },
    {
      "epoch": 0.1408723335614083,
      "grad_norm": 3.0625,
      "learning_rate": 4.767775806567225e-05,
      "loss": 0.8259,
      "step": 9934
    },
    {
      "epoch": 0.1409006952150345,
      "grad_norm": 3.65625,
      "learning_rate": 4.767681759096006e-05,
      "loss": 0.8667,
      "step": 9936
    },
    {
      "epoch": 0.14092905686866072,
      "grad_norm": 3.5625,
      "learning_rate": 4.767587693512662e-05,
      "loss": 0.8894,
      "step": 9938
    },
    {
      "epoch": 0.14095741852228694,
      "grad_norm": 3.390625,
      "learning_rate": 4.767493609817943e-05,
      "loss": 0.8851,
      "step": 9940
    },
    {
      "epoch": 0.14098578017591315,
      "grad_norm": 3.0625,
      "learning_rate": 4.767399508012603e-05,
      "loss": 0.8475,
      "step": 9942
    },
    {
      "epoch": 0.14101414182953936,
      "grad_norm": 2.734375,
      "learning_rate": 4.7673053880973904e-05,
      "loss": 0.8871,
      "step": 9944
    },
    {
      "epoch": 0.14104250348316558,
      "grad_norm": 2.984375,
      "learning_rate": 4.7672112500730594e-05,
      "loss": 0.8521,
      "step": 9946
    },
    {
      "epoch": 0.1410708651367918,
      "grad_norm": 2.859375,
      "learning_rate": 4.76711709394036e-05,
      "loss": 0.8152,
      "step": 9948
    },
    {
      "epoch": 0.141099226790418,
      "grad_norm": 3.015625,
      "learning_rate": 4.767022919700046e-05,
      "loss": 0.8853,
      "step": 9950
    },
    {
      "epoch": 0.14112758844404422,
      "grad_norm": 3.203125,
      "learning_rate": 4.7669287273528676e-05,
      "loss": 0.814,
      "step": 9952
    },
    {
      "epoch": 0.14115595009767046,
      "grad_norm": 3.25,
      "learning_rate": 4.7668345168995794e-05,
      "loss": 0.8751,
      "step": 9954
    },
    {
      "epoch": 0.14118431175129667,
      "grad_norm": 2.921875,
      "learning_rate": 4.7667402883409315e-05,
      "loss": 0.8147,
      "step": 9956
    },
    {
      "epoch": 0.14121267340492288,
      "grad_norm": 3.390625,
      "learning_rate": 4.7666460416776794e-05,
      "loss": 0.91,
      "step": 9958
    },
    {
      "epoch": 0.1412410350585491,
      "grad_norm": 3.3125,
      "learning_rate": 4.7665517769105726e-05,
      "loss": 0.8375,
      "step": 9960
    },
    {
      "epoch": 0.1412693967121753,
      "grad_norm": 3.078125,
      "learning_rate": 4.7664574940403666e-05,
      "loss": 0.8531,
      "step": 9962
    },
    {
      "epoch": 0.14129775836580152,
      "grad_norm": 3.015625,
      "learning_rate": 4.766363193067813e-05,
      "loss": 0.8683,
      "step": 9964
    },
    {
      "epoch": 0.14132612001942774,
      "grad_norm": 3.046875,
      "learning_rate": 4.766268873993666e-05,
      "loss": 0.783,
      "step": 9966
    },
    {
      "epoch": 0.14135448167305395,
      "grad_norm": 3.734375,
      "learning_rate": 4.766174536818677e-05,
      "loss": 0.8508,
      "step": 9968
    },
    {
      "epoch": 0.14138284332668016,
      "grad_norm": 3.078125,
      "learning_rate": 4.7660801815436015e-05,
      "loss": 0.9063,
      "step": 9970
    },
    {
      "epoch": 0.14141120498030638,
      "grad_norm": 3.09375,
      "learning_rate": 4.7659858081691936e-05,
      "loss": 0.8375,
      "step": 9972
    },
    {
      "epoch": 0.1414395666339326,
      "grad_norm": 3.140625,
      "learning_rate": 4.765891416696204e-05,
      "loss": 0.8516,
      "step": 9974
    },
    {
      "epoch": 0.1414679282875588,
      "grad_norm": 3.171875,
      "learning_rate": 4.76579700712539e-05,
      "loss": 0.8882,
      "step": 9976
    },
    {
      "epoch": 0.14149628994118502,
      "grad_norm": 3.234375,
      "learning_rate": 4.765702579457504e-05,
      "loss": 0.8333,
      "step": 9978
    },
    {
      "epoch": 0.14152465159481123,
      "grad_norm": 3.203125,
      "learning_rate": 4.765608133693299e-05,
      "loss": 0.8676,
      "step": 9980
    },
    {
      "epoch": 0.14155301324843744,
      "grad_norm": 2.953125,
      "learning_rate": 4.7655136698335326e-05,
      "loss": 0.8437,
      "step": 9982
    },
    {
      "epoch": 0.14158137490206366,
      "grad_norm": 3.015625,
      "learning_rate": 4.765419187878956e-05,
      "loss": 0.8221,
      "step": 9984
    },
    {
      "epoch": 0.14160973655568987,
      "grad_norm": 2.78125,
      "learning_rate": 4.765324687830326e-05,
      "loss": 0.8472,
      "step": 9986
    },
    {
      "epoch": 0.14163809820931608,
      "grad_norm": 3.3125,
      "learning_rate": 4.7652301696883964e-05,
      "loss": 0.8415,
      "step": 9988
    },
    {
      "epoch": 0.1416664598629423,
      "grad_norm": 2.921875,
      "learning_rate": 4.765135633453922e-05,
      "loss": 0.9079,
      "step": 9990
    },
    {
      "epoch": 0.14169482151656854,
      "grad_norm": 3.265625,
      "learning_rate": 4.7650410791276584e-05,
      "loss": 0.902,
      "step": 9992
    },
    {
      "epoch": 0.14172318317019475,
      "grad_norm": 2.9375,
      "learning_rate": 4.7649465067103606e-05,
      "loss": 0.836,
      "step": 9994
    },
    {
      "epoch": 0.14175154482382096,
      "grad_norm": 2.765625,
      "learning_rate": 4.7648519162027836e-05,
      "loss": 0.7711,
      "step": 9996
    },
    {
      "epoch": 0.14177990647744718,
      "grad_norm": 3.015625,
      "learning_rate": 4.764757307605684e-05,
      "loss": 0.8369,
      "step": 9998
    },
    {
      "epoch": 0.1418082681310734,
      "grad_norm": 2.9375,
      "learning_rate": 4.764662680919817e-05,
      "loss": 0.8296,
      "step": 10000
    },
    {
      "epoch": 0.1418366297846996,
      "grad_norm": 3.421875,
      "learning_rate": 4.764568036145938e-05,
      "loss": 0.8628,
      "step": 10002
    },
    {
      "epoch": 0.14186499143832582,
      "grad_norm": 3.28125,
      "learning_rate": 4.7644733732848025e-05,
      "loss": 0.889,
      "step": 10004
    },
    {
      "epoch": 0.14189335309195203,
      "grad_norm": 3.375,
      "learning_rate": 4.764378692337167e-05,
      "loss": 0.8759,
      "step": 10006
    },
    {
      "epoch": 0.14192171474557824,
      "grad_norm": 3.328125,
      "learning_rate": 4.764283993303788e-05,
      "loss": 0.8765,
      "step": 10008
    },
    {
      "epoch": 0.14195007639920446,
      "grad_norm": 2.875,
      "learning_rate": 4.7641892761854216e-05,
      "loss": 0.8435,
      "step": 10010
    },
    {
      "epoch": 0.14197843805283067,
      "grad_norm": 3.40625,
      "learning_rate": 4.7640945409828255e-05,
      "loss": 0.8862,
      "step": 10012
    },
    {
      "epoch": 0.14200679970645688,
      "grad_norm": 3.4375,
      "learning_rate": 4.763999787696754e-05,
      "loss": 0.8723,
      "step": 10014
    },
    {
      "epoch": 0.1420351613600831,
      "grad_norm": 2.921875,
      "learning_rate": 4.7639050163279654e-05,
      "loss": 0.8466,
      "step": 10016
    },
    {
      "epoch": 0.1420635230137093,
      "grad_norm": 3.21875,
      "learning_rate": 4.7638102268772166e-05,
      "loss": 0.8607,
      "step": 10018
    },
    {
      "epoch": 0.14209188466733552,
      "grad_norm": 3.078125,
      "learning_rate": 4.763715419345265e-05,
      "loss": 0.851,
      "step": 10020
    },
    {
      "epoch": 0.14212024632096174,
      "grad_norm": 3.234375,
      "learning_rate": 4.763620593732867e-05,
      "loss": 0.8344,
      "step": 10022
    },
    {
      "epoch": 0.14214860797458795,
      "grad_norm": 2.890625,
      "learning_rate": 4.7635257500407804e-05,
      "loss": 0.8483,
      "step": 10024
    },
    {
      "epoch": 0.14217696962821416,
      "grad_norm": 2.875,
      "learning_rate": 4.763430888269762e-05,
      "loss": 0.8426,
      "step": 10026
    },
    {
      "epoch": 0.14220533128184037,
      "grad_norm": 3.234375,
      "learning_rate": 4.763336008420571e-05,
      "loss": 0.8417,
      "step": 10028
    },
    {
      "epoch": 0.14223369293546662,
      "grad_norm": 3.0,
      "learning_rate": 4.763241110493964e-05,
      "loss": 0.8286,
      "step": 10030
    },
    {
      "epoch": 0.14226205458909283,
      "grad_norm": 3.296875,
      "learning_rate": 4.7631461944906994e-05,
      "loss": 0.8649,
      "step": 10032
    },
    {
      "epoch": 0.14229041624271904,
      "grad_norm": 3.75,
      "learning_rate": 4.7630512604115354e-05,
      "loss": 0.8481,
      "step": 10034
    },
    {
      "epoch": 0.14231877789634526,
      "grad_norm": 3.046875,
      "learning_rate": 4.762956308257229e-05,
      "loss": 0.8222,
      "step": 10036
    },
    {
      "epoch": 0.14234713954997147,
      "grad_norm": 3.453125,
      "learning_rate": 4.7628613380285405e-05,
      "loss": 0.8978,
      "step": 10038
    },
    {
      "epoch": 0.14237550120359768,
      "grad_norm": 3.359375,
      "learning_rate": 4.7627663497262276e-05,
      "loss": 0.8074,
      "step": 10040
    },
    {
      "epoch": 0.1424038628572239,
      "grad_norm": 3.125,
      "learning_rate": 4.7626713433510485e-05,
      "loss": 0.8174,
      "step": 10042
    },
    {
      "epoch": 0.1424322245108501,
      "grad_norm": 3.234375,
      "learning_rate": 4.762576318903763e-05,
      "loss": 0.8344,
      "step": 10044
    },
    {
      "epoch": 0.14246058616447632,
      "grad_norm": 3.140625,
      "learning_rate": 4.762481276385129e-05,
      "loss": 0.895,
      "step": 10046
    },
    {
      "epoch": 0.14248894781810253,
      "grad_norm": 3.015625,
      "learning_rate": 4.762386215795907e-05,
      "loss": 0.8598,
      "step": 10048
    },
    {
      "epoch": 0.14251730947172875,
      "grad_norm": 3.125,
      "learning_rate": 4.762291137136854e-05,
      "loss": 0.8493,
      "step": 10050
    },
    {
      "epoch": 0.14254567112535496,
      "grad_norm": 3.34375,
      "learning_rate": 4.7621960404087316e-05,
      "loss": 0.8935,
      "step": 10052
    },
    {
      "epoch": 0.14257403277898117,
      "grad_norm": 2.890625,
      "learning_rate": 4.7621009256122985e-05,
      "loss": 0.8644,
      "step": 10054
    },
    {
      "epoch": 0.1426023944326074,
      "grad_norm": 3.140625,
      "learning_rate": 4.7620057927483145e-05,
      "loss": 0.8487,
      "step": 10056
    },
    {
      "epoch": 0.1426307560862336,
      "grad_norm": 3.46875,
      "learning_rate": 4.761910641817539e-05,
      "loss": 0.9167,
      "step": 10058
    },
    {
      "epoch": 0.14265911773985981,
      "grad_norm": 3.234375,
      "learning_rate": 4.761815472820733e-05,
      "loss": 0.9637,
      "step": 10060
    },
    {
      "epoch": 0.14268747939348603,
      "grad_norm": 2.96875,
      "learning_rate": 4.761720285758655e-05,
      "loss": 0.817,
      "step": 10062
    },
    {
      "epoch": 0.14271584104711224,
      "grad_norm": 2.96875,
      "learning_rate": 4.761625080632067e-05,
      "loss": 0.8475,
      "step": 10064
    },
    {
      "epoch": 0.14274420270073845,
      "grad_norm": 3.0,
      "learning_rate": 4.761529857441728e-05,
      "loss": 0.8297,
      "step": 10066
    },
    {
      "epoch": 0.1427725643543647,
      "grad_norm": 3.09375,
      "learning_rate": 4.7614346161884e-05,
      "loss": 0.8912,
      "step": 10068
    },
    {
      "epoch": 0.1428009260079909,
      "grad_norm": 2.890625,
      "learning_rate": 4.761339356872843e-05,
      "loss": 0.8751,
      "step": 10070
    },
    {
      "epoch": 0.14282928766161712,
      "grad_norm": 3.171875,
      "learning_rate": 4.761244079495817e-05,
      "loss": 0.8644,
      "step": 10072
    },
    {
      "epoch": 0.14285764931524333,
      "grad_norm": 3.015625,
      "learning_rate": 4.761148784058084e-05,
      "loss": 0.8362,
      "step": 10074
    },
    {
      "epoch": 0.14288601096886955,
      "grad_norm": 3.21875,
      "learning_rate": 4.761053470560404e-05,
      "loss": 0.8419,
      "step": 10076
    },
    {
      "epoch": 0.14291437262249576,
      "grad_norm": 3.0625,
      "learning_rate": 4.76095813900354e-05,
      "loss": 0.9393,
      "step": 10078
    },
    {
      "epoch": 0.14294273427612197,
      "grad_norm": 2.96875,
      "learning_rate": 4.760862789388253e-05,
      "loss": 0.8559,
      "step": 10080
    },
    {
      "epoch": 0.1429710959297482,
      "grad_norm": 3.09375,
      "learning_rate": 4.7607674217153034e-05,
      "loss": 0.8438,
      "step": 10082
    },
    {
      "epoch": 0.1429994575833744,
      "grad_norm": 2.90625,
      "learning_rate": 4.760672035985453e-05,
      "loss": 0.9016,
      "step": 10084
    },
    {
      "epoch": 0.14302781923700061,
      "grad_norm": 3.015625,
      "learning_rate": 4.7605766321994646e-05,
      "loss": 0.8682,
      "step": 10086
    },
    {
      "epoch": 0.14305618089062683,
      "grad_norm": 3.3125,
      "learning_rate": 4.760481210358101e-05,
      "loss": 0.8857,
      "step": 10088
    },
    {
      "epoch": 0.14308454254425304,
      "grad_norm": 3.109375,
      "learning_rate": 4.7603857704621226e-05,
      "loss": 0.8203,
      "step": 10090
    },
    {
      "epoch": 0.14311290419787925,
      "grad_norm": 2.9375,
      "learning_rate": 4.7602903125122914e-05,
      "loss": 0.8423,
      "step": 10092
    },
    {
      "epoch": 0.14314126585150547,
      "grad_norm": 3.09375,
      "learning_rate": 4.760194836509372e-05,
      "loss": 0.8828,
      "step": 10094
    },
    {
      "epoch": 0.14316962750513168,
      "grad_norm": 3.0,
      "learning_rate": 4.760099342454125e-05,
      "loss": 0.8315,
      "step": 10096
    },
    {
      "epoch": 0.1431979891587579,
      "grad_norm": 3.0625,
      "learning_rate": 4.760003830347314e-05,
      "loss": 0.8567,
      "step": 10098
    },
    {
      "epoch": 0.1432263508123841,
      "grad_norm": 3.0625,
      "learning_rate": 4.759908300189702e-05,
      "loss": 0.8294,
      "step": 10100
    },
    {
      "epoch": 0.14325471246601032,
      "grad_norm": 3.296875,
      "learning_rate": 4.75981275198205e-05,
      "loss": 0.8795,
      "step": 10102
    },
    {
      "epoch": 0.14328307411963653,
      "grad_norm": 2.875,
      "learning_rate": 4.759717185725124e-05,
      "loss": 0.8557,
      "step": 10104
    },
    {
      "epoch": 0.14331143577326275,
      "grad_norm": 3.25,
      "learning_rate": 4.759621601419687e-05,
      "loss": 0.85,
      "step": 10106
    },
    {
      "epoch": 0.143339797426889,
      "grad_norm": 3.25,
      "learning_rate": 4.7595259990665004e-05,
      "loss": 0.8639,
      "step": 10108
    },
    {
      "epoch": 0.1433681590805152,
      "grad_norm": 3.53125,
      "learning_rate": 4.759430378666329e-05,
      "loss": 0.864,
      "step": 10110
    },
    {
      "epoch": 0.1433965207341414,
      "grad_norm": 2.75,
      "learning_rate": 4.759334740219937e-05,
      "loss": 0.8776,
      "step": 10112
    },
    {
      "epoch": 0.14342488238776763,
      "grad_norm": 2.734375,
      "learning_rate": 4.7592390837280875e-05,
      "loss": 0.8177,
      "step": 10114
    },
    {
      "epoch": 0.14345324404139384,
      "grad_norm": 3.203125,
      "learning_rate": 4.759143409191544e-05,
      "loss": 0.8178,
      "step": 10116
    },
    {
      "epoch": 0.14348160569502005,
      "grad_norm": 3.0625,
      "learning_rate": 4.7590477166110725e-05,
      "loss": 0.8379,
      "step": 10118
    },
    {
      "epoch": 0.14350996734864627,
      "grad_norm": 3.375,
      "learning_rate": 4.758952005987436e-05,
      "loss": 0.8449,
      "step": 10120
    },
    {
      "epoch": 0.14353832900227248,
      "grad_norm": 3.078125,
      "learning_rate": 4.758856277321398e-05,
      "loss": 0.8611,
      "step": 10122
    },
    {
      "epoch": 0.1435666906558987,
      "grad_norm": 3.28125,
      "learning_rate": 4.758760530613726e-05,
      "loss": 0.8688,
      "step": 10124
    },
    {
      "epoch": 0.1435950523095249,
      "grad_norm": 2.890625,
      "learning_rate": 4.758664765865182e-05,
      "loss": 0.8216,
      "step": 10126
    },
    {
      "epoch": 0.14362341396315112,
      "grad_norm": 2.875,
      "learning_rate": 4.7585689830765324e-05,
      "loss": 0.8344,
      "step": 10128
    },
    {
      "epoch": 0.14365177561677733,
      "grad_norm": 3.328125,
      "learning_rate": 4.7584731822485414e-05,
      "loss": 0.8587,
      "step": 10130
    },
    {
      "epoch": 0.14368013727040355,
      "grad_norm": 3.078125,
      "learning_rate": 4.758377363381974e-05,
      "loss": 0.8801,
      "step": 10132
    },
    {
      "epoch": 0.14370849892402976,
      "grad_norm": 3.75,
      "learning_rate": 4.758281526477596e-05,
      "loss": 0.8767,
      "step": 10134
    },
    {
      "epoch": 0.14373686057765597,
      "grad_norm": 3.125,
      "learning_rate": 4.758185671536174e-05,
      "loss": 0.8736,
      "step": 10136
    },
    {
      "epoch": 0.14376522223128219,
      "grad_norm": 3.03125,
      "learning_rate": 4.758089798558471e-05,
      "loss": 0.8429,
      "step": 10138
    },
    {
      "epoch": 0.1437935838849084,
      "grad_norm": 3.125,
      "learning_rate": 4.757993907545255e-05,
      "loss": 0.8449,
      "step": 10140
    },
    {
      "epoch": 0.1438219455385346,
      "grad_norm": 3.109375,
      "learning_rate": 4.75789799849729e-05,
      "loss": 0.8694,
      "step": 10142
    },
    {
      "epoch": 0.14385030719216083,
      "grad_norm": 2.875,
      "learning_rate": 4.7578020714153446e-05,
      "loss": 0.8685,
      "step": 10144
    },
    {
      "epoch": 0.14387866884578707,
      "grad_norm": 3.078125,
      "learning_rate": 4.757706126300183e-05,
      "loss": 0.8633,
      "step": 10146
    },
    {
      "epoch": 0.14390703049941328,
      "grad_norm": 3.1875,
      "learning_rate": 4.757610163152572e-05,
      "loss": 0.8711,
      "step": 10148
    },
    {
      "epoch": 0.1439353921530395,
      "grad_norm": 3.03125,
      "learning_rate": 4.7575141819732774e-05,
      "loss": 0.8226,
      "step": 10150
    },
    {
      "epoch": 0.1439637538066657,
      "grad_norm": 3.03125,
      "learning_rate": 4.7574181827630666e-05,
      "loss": 0.8743,
      "step": 10152
    },
    {
      "epoch": 0.14399211546029192,
      "grad_norm": 3.4375,
      "learning_rate": 4.7573221655227064e-05,
      "loss": 0.8883,
      "step": 10154
    },
    {
      "epoch": 0.14402047711391813,
      "grad_norm": 3.328125,
      "learning_rate": 4.7572261302529633e-05,
      "loss": 0.8785,
      "step": 10156
    },
    {
      "epoch": 0.14404883876754435,
      "grad_norm": 3.625,
      "learning_rate": 4.7571300769546044e-05,
      "loss": 0.8874,
      "step": 10158
    },
    {
      "epoch": 0.14407720042117056,
      "grad_norm": 3.0,
      "learning_rate": 4.757034005628398e-05,
      "loss": 0.8502,
      "step": 10160
    },
    {
      "epoch": 0.14410556207479677,
      "grad_norm": 3.578125,
      "learning_rate": 4.7569379162751094e-05,
      "loss": 0.8543,
      "step": 10162
    },
    {
      "epoch": 0.14413392372842299,
      "grad_norm": 3.3125,
      "learning_rate": 4.756841808895508e-05,
      "loss": 0.8199,
      "step": 10164
    },
    {
      "epoch": 0.1441622853820492,
      "grad_norm": 3.15625,
      "learning_rate": 4.75674568349036e-05,
      "loss": 0.8687,
      "step": 10166
    },
    {
      "epoch": 0.1441906470356754,
      "grad_norm": 2.984375,
      "learning_rate": 4.756649540060434e-05,
      "loss": 0.8429,
      "step": 10168
    },
    {
      "epoch": 0.14421900868930163,
      "grad_norm": 2.921875,
      "learning_rate": 4.756553378606497e-05,
      "loss": 0.8835,
      "step": 10170
    },
    {
      "epoch": 0.14424737034292784,
      "grad_norm": 3.125,
      "learning_rate": 4.7564571991293184e-05,
      "loss": 0.8422,
      "step": 10172
    },
    {
      "epoch": 0.14427573199655405,
      "grad_norm": 2.953125,
      "learning_rate": 4.756361001629664e-05,
      "loss": 0.8378,
      "step": 10174
    },
    {
      "epoch": 0.14430409365018027,
      "grad_norm": 3.515625,
      "learning_rate": 4.756264786108306e-05,
      "loss": 0.834,
      "step": 10176
    },
    {
      "epoch": 0.14433245530380648,
      "grad_norm": 2.828125,
      "learning_rate": 4.75616855256601e-05,
      "loss": 0.8525,
      "step": 10178
    },
    {
      "epoch": 0.1443608169574327,
      "grad_norm": 3.078125,
      "learning_rate": 4.7560723010035446e-05,
      "loss": 0.8611,
      "step": 10180
    },
    {
      "epoch": 0.1443891786110589,
      "grad_norm": 2.953125,
      "learning_rate": 4.7559760314216794e-05,
      "loss": 0.8685,
      "step": 10182
    },
    {
      "epoch": 0.14441754026468515,
      "grad_norm": 3.0,
      "learning_rate": 4.755879743821183e-05,
      "loss": 0.8823,
      "step": 10184
    },
    {
      "epoch": 0.14444590191831136,
      "grad_norm": 3.21875,
      "learning_rate": 4.755783438202825e-05,
      "loss": 0.8883,
      "step": 10186
    },
    {
      "epoch": 0.14447426357193757,
      "grad_norm": 3.296875,
      "learning_rate": 4.755687114567375e-05,
      "loss": 0.8464,
      "step": 10188
    },
    {
      "epoch": 0.14450262522556379,
      "grad_norm": 2.953125,
      "learning_rate": 4.755590772915601e-05,
      "loss": 0.8345,
      "step": 10190
    },
    {
      "epoch": 0.14453098687919,
      "grad_norm": 2.859375,
      "learning_rate": 4.7554944132482724e-05,
      "loss": 0.8596,
      "step": 10192
    },
    {
      "epoch": 0.1445593485328162,
      "grad_norm": 2.828125,
      "learning_rate": 4.75539803556616e-05,
      "loss": 0.8417,
      "step": 10194
    },
    {
      "epoch": 0.14458771018644243,
      "grad_norm": 3.609375,
      "learning_rate": 4.7553016398700333e-05,
      "loss": 0.8772,
      "step": 10196
    },
    {
      "epoch": 0.14461607184006864,
      "grad_norm": 2.671875,
      "learning_rate": 4.755205226160662e-05,
      "loss": 0.8432,
      "step": 10198
    },
    {
      "epoch": 0.14464443349369485,
      "grad_norm": 2.9375,
      "learning_rate": 4.7551087944388153e-05,
      "loss": 0.8546,
      "step": 10200
    },
    {
      "epoch": 0.14467279514732106,
      "grad_norm": 2.984375,
      "learning_rate": 4.7550123447052646e-05,
      "loss": 0.8456,
      "step": 10202
    },
    {
      "epoch": 0.14470115680094728,
      "grad_norm": 3.109375,
      "learning_rate": 4.75491587696078e-05,
      "loss": 0.8132,
      "step": 10204
    },
    {
      "epoch": 0.1447295184545735,
      "grad_norm": 3.234375,
      "learning_rate": 4.754819391206132e-05,
      "loss": 0.9357,
      "step": 10206
    },
    {
      "epoch": 0.1447578801081997,
      "grad_norm": 3.171875,
      "learning_rate": 4.754722887442091e-05,
      "loss": 0.8581,
      "step": 10208
    },
    {
      "epoch": 0.14478624176182592,
      "grad_norm": 2.90625,
      "learning_rate": 4.754626365669428e-05,
      "loss": 0.9172,
      "step": 10210
    },
    {
      "epoch": 0.14481460341545213,
      "grad_norm": 3.03125,
      "learning_rate": 4.754529825888914e-05,
      "loss": 0.8368,
      "step": 10212
    },
    {
      "epoch": 0.14484296506907834,
      "grad_norm": 3.21875,
      "learning_rate": 4.7544332681013194e-05,
      "loss": 0.7964,
      "step": 10214
    },
    {
      "epoch": 0.14487132672270456,
      "grad_norm": 2.59375,
      "learning_rate": 4.754336692307416e-05,
      "loss": 0.8434,
      "step": 10216
    },
    {
      "epoch": 0.14489968837633077,
      "grad_norm": 3.3125,
      "learning_rate": 4.7542400985079755e-05,
      "loss": 0.9134,
      "step": 10218
    },
    {
      "epoch": 0.14492805002995698,
      "grad_norm": 2.796875,
      "learning_rate": 4.754143486703768e-05,
      "loss": 0.8269,
      "step": 10220
    },
    {
      "epoch": 0.14495641168358323,
      "grad_norm": 3.59375,
      "learning_rate": 4.754046856895568e-05,
      "loss": 0.8812,
      "step": 10222
    },
    {
      "epoch": 0.14498477333720944,
      "grad_norm": 2.921875,
      "learning_rate": 4.753950209084144e-05,
      "loss": 0.8358,
      "step": 10224
    },
    {
      "epoch": 0.14501313499083565,
      "grad_norm": 3.359375,
      "learning_rate": 4.753853543270269e-05,
      "loss": 0.8704,
      "step": 10226
    },
    {
      "epoch": 0.14504149664446186,
      "grad_norm": 3.015625,
      "learning_rate": 4.753756859454717e-05,
      "loss": 0.7999,
      "step": 10228
    },
    {
      "epoch": 0.14506985829808808,
      "grad_norm": 3.359375,
      "learning_rate": 4.753660157638257e-05,
      "loss": 0.8822,
      "step": 10230
    },
    {
      "epoch": 0.1450982199517143,
      "grad_norm": 3.125,
      "learning_rate": 4.7535634378216636e-05,
      "loss": 0.8762,
      "step": 10232
    },
    {
      "epoch": 0.1451265816053405,
      "grad_norm": 2.71875,
      "learning_rate": 4.7534667000057084e-05,
      "loss": 0.8235,
      "step": 10234
    },
    {
      "epoch": 0.14515494325896672,
      "grad_norm": 3.390625,
      "learning_rate": 4.753369944191165e-05,
      "loss": 0.8641,
      "step": 10236
    },
    {
      "epoch": 0.14518330491259293,
      "grad_norm": 3.09375,
      "learning_rate": 4.753273170378805e-05,
      "loss": 0.8542,
      "step": 10238
    },
    {
      "epoch": 0.14521166656621914,
      "grad_norm": 2.859375,
      "learning_rate": 4.753176378569403e-05,
      "loss": 0.8415,
      "step": 10240
    },
    {
      "epoch": 0.14524002821984536,
      "grad_norm": 3.34375,
      "learning_rate": 4.75307956876373e-05,
      "loss": 0.8207,
      "step": 10242
    },
    {
      "epoch": 0.14526838987347157,
      "grad_norm": 2.734375,
      "learning_rate": 4.75298274096256e-05,
      "loss": 0.8583,
      "step": 10244
    },
    {
      "epoch": 0.14529675152709778,
      "grad_norm": 3.125,
      "learning_rate": 4.752885895166667e-05,
      "loss": 0.8689,
      "step": 10246
    },
    {
      "epoch": 0.145325113180724,
      "grad_norm": 2.859375,
      "learning_rate": 4.752789031376824e-05,
      "loss": 0.8414,
      "step": 10248
    },
    {
      "epoch": 0.1453534748343502,
      "grad_norm": 3.203125,
      "learning_rate": 4.7526921495938046e-05,
      "loss": 0.8676,
      "step": 10250
    },
    {
      "epoch": 0.14538183648797642,
      "grad_norm": 3.390625,
      "learning_rate": 4.752595249818383e-05,
      "loss": 0.8573,
      "step": 10252
    },
    {
      "epoch": 0.14541019814160264,
      "grad_norm": 2.671875,
      "learning_rate": 4.752498332051333e-05,
      "loss": 0.824,
      "step": 10254
    },
    {
      "epoch": 0.14543855979522885,
      "grad_norm": 2.78125,
      "learning_rate": 4.7524013962934285e-05,
      "loss": 0.8317,
      "step": 10256
    },
    {
      "epoch": 0.14546692144885506,
      "grad_norm": 3.40625,
      "learning_rate": 4.752304442545443e-05,
      "loss": 0.8743,
      "step": 10258
    },
    {
      "epoch": 0.14549528310248128,
      "grad_norm": 3.28125,
      "learning_rate": 4.752207470808153e-05,
      "loss": 0.8796,
      "step": 10260
    },
    {
      "epoch": 0.14552364475610752,
      "grad_norm": 3.453125,
      "learning_rate": 4.752110481082331e-05,
      "loss": 0.8975,
      "step": 10262
    },
    {
      "epoch": 0.14555200640973373,
      "grad_norm": 3.171875,
      "learning_rate": 4.7520134733687524e-05,
      "loss": 0.8503,
      "step": 10264
    },
    {
      "epoch": 0.14558036806335994,
      "grad_norm": 3.109375,
      "learning_rate": 4.751916447668192e-05,
      "loss": 0.857,
      "step": 10266
    },
    {
      "epoch": 0.14560872971698616,
      "grad_norm": 2.890625,
      "learning_rate": 4.751819403981425e-05,
      "loss": 0.8186,
      "step": 10268
    },
    {
      "epoch": 0.14563709137061237,
      "grad_norm": 4.1875,
      "learning_rate": 4.7517223423092255e-05,
      "loss": 0.8538,
      "step": 10270
    },
    {
      "epoch": 0.14566545302423858,
      "grad_norm": 3.265625,
      "learning_rate": 4.75162526265237e-05,
      "loss": 0.8414,
      "step": 10272
    },
    {
      "epoch": 0.1456938146778648,
      "grad_norm": 3.203125,
      "learning_rate": 4.751528165011633e-05,
      "loss": 0.8533,
      "step": 10274
    },
    {
      "epoch": 0.145722176331491,
      "grad_norm": 3.140625,
      "learning_rate": 4.751431049387791e-05,
      "loss": 0.9075,
      "step": 10276
    },
    {
      "epoch": 0.14575053798511722,
      "grad_norm": 3.03125,
      "learning_rate": 4.751333915781618e-05,
      "loss": 0.8609,
      "step": 10278
    },
    {
      "epoch": 0.14577889963874344,
      "grad_norm": 3.40625,
      "learning_rate": 4.751236764193892e-05,
      "loss": 0.9451,
      "step": 10280
    },
    {
      "epoch": 0.14580726129236965,
      "grad_norm": 3.15625,
      "learning_rate": 4.751139594625388e-05,
      "loss": 0.8609,
      "step": 10282
    },
    {
      "epoch": 0.14583562294599586,
      "grad_norm": 3.140625,
      "learning_rate": 4.7510424070768814e-05,
      "loss": 0.87,
      "step": 10284
    },
    {
      "epoch": 0.14586398459962208,
      "grad_norm": 3.328125,
      "learning_rate": 4.7509452015491486e-05,
      "loss": 0.8629,
      "step": 10286
    },
    {
      "epoch": 0.1458923462532483,
      "grad_norm": 2.96875,
      "learning_rate": 4.750847978042966e-05,
      "loss": 0.8541,
      "step": 10288
    },
    {
      "epoch": 0.1459207079068745,
      "grad_norm": 3.0,
      "learning_rate": 4.750750736559111e-05,
      "loss": 0.8653,
      "step": 10290
    },
    {
      "epoch": 0.14594906956050072,
      "grad_norm": 3.09375,
      "learning_rate": 4.7506534770983595e-05,
      "loss": 0.94,
      "step": 10292
    },
    {
      "epoch": 0.14597743121412693,
      "grad_norm": 2.890625,
      "learning_rate": 4.750556199661489e-05,
      "loss": 0.8824,
      "step": 10294
    },
    {
      "epoch": 0.14600579286775314,
      "grad_norm": 3.28125,
      "learning_rate": 4.750458904249276e-05,
      "loss": 0.8646,
      "step": 10296
    },
    {
      "epoch": 0.14603415452137936,
      "grad_norm": 3.1875,
      "learning_rate": 4.750361590862498e-05,
      "loss": 0.866,
      "step": 10298
    },
    {
      "epoch": 0.1460625161750056,
      "grad_norm": 3.09375,
      "learning_rate": 4.750264259501931e-05,
      "loss": 0.8373,
      "step": 10300
    },
    {
      "epoch": 0.1460908778286318,
      "grad_norm": 3.3125,
      "learning_rate": 4.7501669101683535e-05,
      "loss": 0.7949,
      "step": 10302
    },
    {
      "epoch": 0.14611923948225802,
      "grad_norm": 3.34375,
      "learning_rate": 4.750069542862543e-05,
      "loss": 0.8714,
      "step": 10304
    },
    {
      "epoch": 0.14614760113588424,
      "grad_norm": 2.84375,
      "learning_rate": 4.749972157585276e-05,
      "loss": 0.8706,
      "step": 10306
    },
    {
      "epoch": 0.14617596278951045,
      "grad_norm": 3.25,
      "learning_rate": 4.7498747543373325e-05,
      "loss": 0.844,
      "step": 10308
    },
    {
      "epoch": 0.14620432444313666,
      "grad_norm": 2.953125,
      "learning_rate": 4.749777333119489e-05,
      "loss": 0.8659,
      "step": 10310
    },
    {
      "epoch": 0.14623268609676288,
      "grad_norm": 3.28125,
      "learning_rate": 4.749679893932524e-05,
      "loss": 0.9178,
      "step": 10312
    },
    {
      "epoch": 0.1462610477503891,
      "grad_norm": 3.359375,
      "learning_rate": 4.749582436777215e-05,
      "loss": 0.8853,
      "step": 10314
    },
    {
      "epoch": 0.1462894094040153,
      "grad_norm": 2.71875,
      "learning_rate": 4.7494849616543416e-05,
      "loss": 0.8272,
      "step": 10316
    },
    {
      "epoch": 0.14631777105764152,
      "grad_norm": 3.109375,
      "learning_rate": 4.7493874685646816e-05,
      "loss": 0.9145,
      "step": 10318
    },
    {
      "epoch": 0.14634613271126773,
      "grad_norm": 2.921875,
      "learning_rate": 4.7492899575090144e-05,
      "loss": 0.8589,
      "step": 10320
    },
    {
      "epoch": 0.14637449436489394,
      "grad_norm": 2.890625,
      "learning_rate": 4.749192428488117e-05,
      "loss": 0.8397,
      "step": 10322
    },
    {
      "epoch": 0.14640285601852016,
      "grad_norm": 2.984375,
      "learning_rate": 4.749094881502771e-05,
      "loss": 0.9036,
      "step": 10324
    },
    {
      "epoch": 0.14643121767214637,
      "grad_norm": 2.859375,
      "learning_rate": 4.748997316553754e-05,
      "loss": 0.8489,
      "step": 10326
    },
    {
      "epoch": 0.14645957932577258,
      "grad_norm": 3.171875,
      "learning_rate": 4.7488997336418445e-05,
      "loss": 0.8667,
      "step": 10328
    },
    {
      "epoch": 0.1464879409793988,
      "grad_norm": 3.390625,
      "learning_rate": 4.7488021327678234e-05,
      "loss": 0.8245,
      "step": 10330
    },
    {
      "epoch": 0.146516302633025,
      "grad_norm": 3.265625,
      "learning_rate": 4.748704513932469e-05,
      "loss": 0.8592,
      "step": 10332
    },
    {
      "epoch": 0.14654466428665122,
      "grad_norm": 2.96875,
      "learning_rate": 4.748606877136563e-05,
      "loss": 0.8478,
      "step": 10334
    },
    {
      "epoch": 0.14657302594027743,
      "grad_norm": 2.96875,
      "learning_rate": 4.748509222380883e-05,
      "loss": 0.8704,
      "step": 10336
    },
    {
      "epoch": 0.14660138759390368,
      "grad_norm": 3.234375,
      "learning_rate": 4.7484115496662095e-05,
      "loss": 0.8525,
      "step": 10338
    },
    {
      "epoch": 0.1466297492475299,
      "grad_norm": 3.296875,
      "learning_rate": 4.7483138589933244e-05,
      "loss": 0.877,
      "step": 10340
    },
    {
      "epoch": 0.1466581109011561,
      "grad_norm": 3.03125,
      "learning_rate": 4.7482161503630053e-05,
      "loss": 0.8192,
      "step": 10342
    },
    {
      "epoch": 0.14668647255478232,
      "grad_norm": 3.21875,
      "learning_rate": 4.748118423776034e-05,
      "loss": 0.8604,
      "step": 10344
    },
    {
      "epoch": 0.14671483420840853,
      "grad_norm": 2.5625,
      "learning_rate": 4.748020679233191e-05,
      "loss": 0.8537,
      "step": 10346
    },
    {
      "epoch": 0.14674319586203474,
      "grad_norm": 3.46875,
      "learning_rate": 4.7479229167352565e-05,
      "loss": 0.8026,
      "step": 10348
    },
    {
      "epoch": 0.14677155751566096,
      "grad_norm": 3.046875,
      "learning_rate": 4.747825136283013e-05,
      "loss": 0.7927,
      "step": 10350
    },
    {
      "epoch": 0.14679991916928717,
      "grad_norm": 2.875,
      "learning_rate": 4.747727337877239e-05,
      "loss": 0.8409,
      "step": 10352
    },
    {
      "epoch": 0.14682828082291338,
      "grad_norm": 3.0,
      "learning_rate": 4.747629521518717e-05,
      "loss": 0.8509,
      "step": 10354
    },
    {
      "epoch": 0.1468566424765396,
      "grad_norm": 2.640625,
      "learning_rate": 4.7475316872082286e-05,
      "loss": 0.8428,
      "step": 10356
    },
    {
      "epoch": 0.1468850041301658,
      "grad_norm": 3.4375,
      "learning_rate": 4.7474338349465544e-05,
      "loss": 0.8995,
      "step": 10358
    },
    {
      "epoch": 0.14691336578379202,
      "grad_norm": 3.09375,
      "learning_rate": 4.747335964734476e-05,
      "loss": 0.8691,
      "step": 10360
    },
    {
      "epoch": 0.14694172743741823,
      "grad_norm": 3.078125,
      "learning_rate": 4.747238076572777e-05,
      "loss": 0.8624,
      "step": 10362
    },
    {
      "epoch": 0.14697008909104445,
      "grad_norm": 3.0625,
      "learning_rate": 4.747140170462235e-05,
      "loss": 0.8676,
      "step": 10364
    },
    {
      "epoch": 0.14699845074467066,
      "grad_norm": 2.96875,
      "learning_rate": 4.747042246403636e-05,
      "loss": 0.8325,
      "step": 10366
    },
    {
      "epoch": 0.14702681239829687,
      "grad_norm": 3.03125,
      "learning_rate": 4.74694430439776e-05,
      "loss": 0.9016,
      "step": 10368
    },
    {
      "epoch": 0.1470551740519231,
      "grad_norm": 3.28125,
      "learning_rate": 4.746846344445391e-05,
      "loss": 0.8643,
      "step": 10370
    },
    {
      "epoch": 0.1470835357055493,
      "grad_norm": 2.9375,
      "learning_rate": 4.74674836654731e-05,
      "loss": 0.893,
      "step": 10372
    },
    {
      "epoch": 0.14711189735917551,
      "grad_norm": 2.984375,
      "learning_rate": 4.7466503707043e-05,
      "loss": 0.8298,
      "step": 10374
    },
    {
      "epoch": 0.14714025901280176,
      "grad_norm": 2.890625,
      "learning_rate": 4.746552356917143e-05,
      "loss": 0.8708,
      "step": 10376
    },
    {
      "epoch": 0.14716862066642797,
      "grad_norm": 2.875,
      "learning_rate": 4.7464543251866226e-05,
      "loss": 0.9084,
      "step": 10378
    },
    {
      "epoch": 0.14719698232005418,
      "grad_norm": 2.96875,
      "learning_rate": 4.746356275513522e-05,
      "loss": 0.8751,
      "step": 10380
    },
    {
      "epoch": 0.1472253439736804,
      "grad_norm": 3.234375,
      "learning_rate": 4.746258207898624e-05,
      "loss": 0.8707,
      "step": 10382
    },
    {
      "epoch": 0.1472537056273066,
      "grad_norm": 3.140625,
      "learning_rate": 4.746160122342712e-05,
      "loss": 0.8554,
      "step": 10384
    },
    {
      "epoch": 0.14728206728093282,
      "grad_norm": 2.84375,
      "learning_rate": 4.746062018846569e-05,
      "loss": 0.8656,
      "step": 10386
    },
    {
      "epoch": 0.14731042893455903,
      "grad_norm": 3.265625,
      "learning_rate": 4.745963897410979e-05,
      "loss": 0.8141,
      "step": 10388
    },
    {
      "epoch": 0.14733879058818525,
      "grad_norm": 3.140625,
      "learning_rate": 4.745865758036725e-05,
      "loss": 0.8266,
      "step": 10390
    },
    {
      "epoch": 0.14736715224181146,
      "grad_norm": 3.234375,
      "learning_rate": 4.745767600724592e-05,
      "loss": 0.8982,
      "step": 10392
    },
    {
      "epoch": 0.14739551389543767,
      "grad_norm": 3.125,
      "learning_rate": 4.7456694254753635e-05,
      "loss": 0.8489,
      "step": 10394
    },
    {
      "epoch": 0.1474238755490639,
      "grad_norm": 2.984375,
      "learning_rate": 4.7455712322898236e-05,
      "loss": 0.8759,
      "step": 10396
    },
    {
      "epoch": 0.1474522372026901,
      "grad_norm": 2.953125,
      "learning_rate": 4.745473021168756e-05,
      "loss": 0.8554,
      "step": 10398
    },
    {
      "epoch": 0.14748059885631631,
      "grad_norm": 2.96875,
      "learning_rate": 4.7453747921129456e-05,
      "loss": 0.9012,
      "step": 10400
    },
    {
      "epoch": 0.14750896050994253,
      "grad_norm": 3.5,
      "learning_rate": 4.7452765451231776e-05,
      "loss": 0.8792,
      "step": 10402
    },
    {
      "epoch": 0.14753732216356874,
      "grad_norm": 3.546875,
      "learning_rate": 4.7451782802002354e-05,
      "loss": 0.845,
      "step": 10404
    },
    {
      "epoch": 0.14756568381719495,
      "grad_norm": 3.09375,
      "learning_rate": 4.7450799973449044e-05,
      "loss": 0.8844,
      "step": 10406
    },
    {
      "epoch": 0.14759404547082117,
      "grad_norm": 3.34375,
      "learning_rate": 4.74498169655797e-05,
      "loss": 0.8484,
      "step": 10408
    },
    {
      "epoch": 0.14762240712444738,
      "grad_norm": 3.046875,
      "learning_rate": 4.744883377840217e-05,
      "loss": 0.8403,
      "step": 10410
    },
    {
      "epoch": 0.1476507687780736,
      "grad_norm": 3.34375,
      "learning_rate": 4.744785041192431e-05,
      "loss": 0.8683,
      "step": 10412
    },
    {
      "epoch": 0.1476791304316998,
      "grad_norm": 3.25,
      "learning_rate": 4.744686686615397e-05,
      "loss": 0.8753,
      "step": 10414
    },
    {
      "epoch": 0.14770749208532605,
      "grad_norm": 2.96875,
      "learning_rate": 4.744588314109901e-05,
      "loss": 0.852,
      "step": 10416
    },
    {
      "epoch": 0.14773585373895226,
      "grad_norm": 3.125,
      "learning_rate": 4.744489923676728e-05,
      "loss": 0.8806,
      "step": 10418
    },
    {
      "epoch": 0.14776421539257847,
      "grad_norm": 2.75,
      "learning_rate": 4.744391515316664e-05,
      "loss": 0.8465,
      "step": 10420
    },
    {
      "epoch": 0.1477925770462047,
      "grad_norm": 2.796875,
      "learning_rate": 4.744293089030496e-05,
      "loss": 0.8609,
      "step": 10422
    },
    {
      "epoch": 0.1478209386998309,
      "grad_norm": 2.84375,
      "learning_rate": 4.7441946448190086e-05,
      "loss": 0.8562,
      "step": 10424
    },
    {
      "epoch": 0.1478493003534571,
      "grad_norm": 3.015625,
      "learning_rate": 4.74409618268299e-05,
      "loss": 0.8534,
      "step": 10426
    },
    {
      "epoch": 0.14787766200708333,
      "grad_norm": 3.015625,
      "learning_rate": 4.743997702623225e-05,
      "loss": 0.899,
      "step": 10428
    },
    {
      "epoch": 0.14790602366070954,
      "grad_norm": 3.21875,
      "learning_rate": 4.7438992046404994e-05,
      "loss": 0.8842,
      "step": 10430
    },
    {
      "epoch": 0.14793438531433575,
      "grad_norm": 3.234375,
      "learning_rate": 4.743800688735603e-05,
      "loss": 0.8542,
      "step": 10432
    },
    {
      "epoch": 0.14796274696796197,
      "grad_norm": 2.875,
      "learning_rate": 4.74370215490932e-05,
      "loss": 0.8614,
      "step": 10434
    },
    {
      "epoch": 0.14799110862158818,
      "grad_norm": 2.984375,
      "learning_rate": 4.7436036031624385e-05,
      "loss": 0.8652,
      "step": 10436
    },
    {
      "epoch": 0.1480194702752144,
      "grad_norm": 2.921875,
      "learning_rate": 4.743505033495746e-05,
      "loss": 0.844,
      "step": 10438
    },
    {
      "epoch": 0.1480478319288406,
      "grad_norm": 3.5625,
      "learning_rate": 4.7434064459100275e-05,
      "loss": 0.8386,
      "step": 10440
    },
    {
      "epoch": 0.14807619358246682,
      "grad_norm": 3.203125,
      "learning_rate": 4.743307840406073e-05,
      "loss": 0.8524,
      "step": 10442
    },
    {
      "epoch": 0.14810455523609303,
      "grad_norm": 3.1875,
      "learning_rate": 4.743209216984669e-05,
      "loss": 0.8453,
      "step": 10444
    },
    {
      "epoch": 0.14813291688971925,
      "grad_norm": 2.796875,
      "learning_rate": 4.743110575646603e-05,
      "loss": 0.8895,
      "step": 10446
    },
    {
      "epoch": 0.14816127854334546,
      "grad_norm": 3.0,
      "learning_rate": 4.743011916392664e-05,
      "loss": 0.8314,
      "step": 10448
    },
    {
      "epoch": 0.14818964019697167,
      "grad_norm": 3.0,
      "learning_rate": 4.7429132392236384e-05,
      "loss": 0.8272,
      "step": 10450
    },
    {
      "epoch": 0.14821800185059789,
      "grad_norm": 3.015625,
      "learning_rate": 4.742814544140316e-05,
      "loss": 0.83,
      "step": 10452
    },
    {
      "epoch": 0.14824636350422413,
      "grad_norm": 3.0625,
      "learning_rate": 4.7427158311434835e-05,
      "loss": 0.845,
      "step": 10454
    },
    {
      "epoch": 0.14827472515785034,
      "grad_norm": 2.921875,
      "learning_rate": 4.7426171002339306e-05,
      "loss": 0.8253,
      "step": 10456
    },
    {
      "epoch": 0.14830308681147655,
      "grad_norm": 3.015625,
      "learning_rate": 4.742518351412446e-05,
      "loss": 0.8757,
      "step": 10458
    },
    {
      "epoch": 0.14833144846510277,
      "grad_norm": 3.21875,
      "learning_rate": 4.742419584679816e-05,
      "loss": 0.8963,
      "step": 10460
    },
    {
      "epoch": 0.14835981011872898,
      "grad_norm": 3.28125,
      "learning_rate": 4.742320800036832e-05,
      "loss": 0.8683,
      "step": 10462
    },
    {
      "epoch": 0.1483881717723552,
      "grad_norm": 3.0,
      "learning_rate": 4.742221997484283e-05,
      "loss": 0.7973,
      "step": 10464
    },
    {
      "epoch": 0.1484165334259814,
      "grad_norm": 3.09375,
      "learning_rate": 4.742123177022957e-05,
      "loss": 0.8561,
      "step": 10466
    },
    {
      "epoch": 0.14844489507960762,
      "grad_norm": 3.140625,
      "learning_rate": 4.742024338653643e-05,
      "loss": 0.8887,
      "step": 10468
    },
    {
      "epoch": 0.14847325673323383,
      "grad_norm": 2.65625,
      "learning_rate": 4.741925482377131e-05,
      "loss": 0.8312,
      "step": 10470
    },
    {
      "epoch": 0.14850161838686005,
      "grad_norm": 2.875,
      "learning_rate": 4.7418266081942116e-05,
      "loss": 0.8632,
      "step": 10472
    },
    {
      "epoch": 0.14852998004048626,
      "grad_norm": 3.453125,
      "learning_rate": 4.741727716105673e-05,
      "loss": 0.8744,
      "step": 10474
    },
    {
      "epoch": 0.14855834169411247,
      "grad_norm": 3.65625,
      "learning_rate": 4.741628806112305e-05,
      "loss": 0.8407,
      "step": 10476
    },
    {
      "epoch": 0.14858670334773869,
      "grad_norm": 2.6875,
      "learning_rate": 4.741529878214899e-05,
      "loss": 0.8285,
      "step": 10478
    },
    {
      "epoch": 0.1486150650013649,
      "grad_norm": 3.203125,
      "learning_rate": 4.741430932414245e-05,
      "loss": 0.8436,
      "step": 10480
    },
    {
      "epoch": 0.1486434266549911,
      "grad_norm": 3.203125,
      "learning_rate": 4.741331968711131e-05,
      "loss": 0.858,
      "step": 10482
    },
    {
      "epoch": 0.14867178830861733,
      "grad_norm": 3.046875,
      "learning_rate": 4.74123298710635e-05,
      "loss": 0.8377,
      "step": 10484
    },
    {
      "epoch": 0.14870014996224354,
      "grad_norm": 3.09375,
      "learning_rate": 4.741133987600691e-05,
      "loss": 0.8519,
      "step": 10486
    },
    {
      "epoch": 0.14872851161586975,
      "grad_norm": 3.1875,
      "learning_rate": 4.741034970194946e-05,
      "loss": 0.8491,
      "step": 10488
    },
    {
      "epoch": 0.14875687326949597,
      "grad_norm": 3.828125,
      "learning_rate": 4.740935934889905e-05,
      "loss": 0.8534,
      "step": 10490
    },
    {
      "epoch": 0.1487852349231222,
      "grad_norm": 3.15625,
      "learning_rate": 4.7408368816863596e-05,
      "loss": 0.8587,
      "step": 10492
    },
    {
      "epoch": 0.14881359657674842,
      "grad_norm": 3.234375,
      "learning_rate": 4.7407378105851e-05,
      "loss": 0.7781,
      "step": 10494
    },
    {
      "epoch": 0.14884195823037463,
      "grad_norm": 3.125,
      "learning_rate": 4.7406387215869184e-05,
      "loss": 0.8945,
      "step": 10496
    },
    {
      "epoch": 0.14887031988400085,
      "grad_norm": 3.171875,
      "learning_rate": 4.7405396146926065e-05,
      "loss": 0.8522,
      "step": 10498
    },
    {
      "epoch": 0.14889868153762706,
      "grad_norm": 3.34375,
      "learning_rate": 4.740440489902954e-05,
      "loss": 0.902,
      "step": 10500
    },
    {
      "epoch": 0.14892704319125327,
      "grad_norm": 3.203125,
      "learning_rate": 4.740341347218754e-05,
      "loss": 0.8663,
      "step": 10502
    },
    {
      "epoch": 0.14895540484487949,
      "grad_norm": 3.140625,
      "learning_rate": 4.740242186640799e-05,
      "loss": 0.8985,
      "step": 10504
    },
    {
      "epoch": 0.1489837664985057,
      "grad_norm": 3.53125,
      "learning_rate": 4.740143008169879e-05,
      "loss": 0.8672,
      "step": 10506
    },
    {
      "epoch": 0.1490121281521319,
      "grad_norm": 3.734375,
      "learning_rate": 4.7400438118067894e-05,
      "loss": 0.9126,
      "step": 10508
    },
    {
      "epoch": 0.14904048980575813,
      "grad_norm": 3.625,
      "learning_rate": 4.7399445975523185e-05,
      "loss": 0.8612,
      "step": 10510
    },
    {
      "epoch": 0.14906885145938434,
      "grad_norm": 3.078125,
      "learning_rate": 4.7398453654072616e-05,
      "loss": 0.8557,
      "step": 10512
    },
    {
      "epoch": 0.14909721311301055,
      "grad_norm": 3.453125,
      "learning_rate": 4.73974611537241e-05,
      "loss": 0.8151,
      "step": 10514
    },
    {
      "epoch": 0.14912557476663676,
      "grad_norm": 2.890625,
      "learning_rate": 4.7396468474485575e-05,
      "loss": 0.823,
      "step": 10516
    },
    {
      "epoch": 0.14915393642026298,
      "grad_norm": 2.921875,
      "learning_rate": 4.739547561636496e-05,
      "loss": 0.8812,
      "step": 10518
    },
    {
      "epoch": 0.1491822980738892,
      "grad_norm": 2.828125,
      "learning_rate": 4.739448257937018e-05,
      "loss": 0.8435,
      "step": 10520
    },
    {
      "epoch": 0.1492106597275154,
      "grad_norm": 2.953125,
      "learning_rate": 4.739348936350918e-05,
      "loss": 0.8101,
      "step": 10522
    },
    {
      "epoch": 0.14923902138114162,
      "grad_norm": 2.9375,
      "learning_rate": 4.7392495968789885e-05,
      "loss": 0.8998,
      "step": 10524
    },
    {
      "epoch": 0.14926738303476783,
      "grad_norm": 3.28125,
      "learning_rate": 4.7391502395220234e-05,
      "loss": 0.8674,
      "step": 10526
    },
    {
      "epoch": 0.14929574468839404,
      "grad_norm": 3.171875,
      "learning_rate": 4.7390508642808155e-05,
      "loss": 0.9009,
      "step": 10528
    },
    {
      "epoch": 0.14932410634202029,
      "grad_norm": 3.046875,
      "learning_rate": 4.73895147115616e-05,
      "loss": 0.8576,
      "step": 10530
    },
    {
      "epoch": 0.1493524679956465,
      "grad_norm": 3.140625,
      "learning_rate": 4.738852060148849e-05,
      "loss": 0.7996,
      "step": 10532
    },
    {
      "epoch": 0.1493808296492727,
      "grad_norm": 3.21875,
      "learning_rate": 4.738752631259677e-05,
      "loss": 0.8229,
      "step": 10534
    },
    {
      "epoch": 0.14940919130289892,
      "grad_norm": 2.921875,
      "learning_rate": 4.738653184489439e-05,
      "loss": 0.8034,
      "step": 10536
    },
    {
      "epoch": 0.14943755295652514,
      "grad_norm": 3.234375,
      "learning_rate": 4.738553719838928e-05,
      "loss": 0.8418,
      "step": 10538
    },
    {
      "epoch": 0.14946591461015135,
      "grad_norm": 3.0625,
      "learning_rate": 4.738454237308939e-05,
      "loss": 0.9024,
      "step": 10540
    },
    {
      "epoch": 0.14949427626377756,
      "grad_norm": 3.734375,
      "learning_rate": 4.738354736900268e-05,
      "loss": 0.88,
      "step": 10542
    },
    {
      "epoch": 0.14952263791740378,
      "grad_norm": 2.71875,
      "learning_rate": 4.7382552186137074e-05,
      "loss": 0.8594,
      "step": 10544
    },
    {
      "epoch": 0.14955099957103,
      "grad_norm": 3.078125,
      "learning_rate": 4.7381556824500536e-05,
      "loss": 0.8171,
      "step": 10546
    },
    {
      "epoch": 0.1495793612246562,
      "grad_norm": 2.96875,
      "learning_rate": 4.7380561284101e-05,
      "loss": 0.8811,
      "step": 10548
    },
    {
      "epoch": 0.14960772287828242,
      "grad_norm": 3.25,
      "learning_rate": 4.737956556494644e-05,
      "loss": 0.852,
      "step": 10550
    },
    {
      "epoch": 0.14963608453190863,
      "grad_norm": 2.859375,
      "learning_rate": 4.737856966704479e-05,
      "loss": 0.8325,
      "step": 10552
    },
    {
      "epoch": 0.14966444618553484,
      "grad_norm": 3.390625,
      "learning_rate": 4.7377573590404026e-05,
      "loss": 0.8702,
      "step": 10554
    },
    {
      "epoch": 0.14969280783916106,
      "grad_norm": 3.25,
      "learning_rate": 4.737657733503207e-05,
      "loss": 0.8734,
      "step": 10556
    },
    {
      "epoch": 0.14972116949278727,
      "grad_norm": 3.296875,
      "learning_rate": 4.737558090093691e-05,
      "loss": 0.8427,
      "step": 10558
    },
    {
      "epoch": 0.14974953114641348,
      "grad_norm": 2.96875,
      "learning_rate": 4.737458428812649e-05,
      "loss": 0.8637,
      "step": 10560
    },
    {
      "epoch": 0.1497778928000397,
      "grad_norm": 3.046875,
      "learning_rate": 4.737358749660877e-05,
      "loss": 0.8285,
      "step": 10562
    },
    {
      "epoch": 0.1498062544536659,
      "grad_norm": 2.890625,
      "learning_rate": 4.7372590526391715e-05,
      "loss": 0.9196,
      "step": 10564
    },
    {
      "epoch": 0.14983461610729212,
      "grad_norm": 3.09375,
      "learning_rate": 4.737159337748329e-05,
      "loss": 0.8683,
      "step": 10566
    },
    {
      "epoch": 0.14986297776091834,
      "grad_norm": 2.921875,
      "learning_rate": 4.737059604989145e-05,
      "loss": 0.8668,
      "step": 10568
    },
    {
      "epoch": 0.14989133941454458,
      "grad_norm": 2.984375,
      "learning_rate": 4.736959854362417e-05,
      "loss": 0.9091,
      "step": 10570
    },
    {
      "epoch": 0.1499197010681708,
      "grad_norm": 3.078125,
      "learning_rate": 4.736860085868942e-05,
      "loss": 0.8966,
      "step": 10572
    },
    {
      "epoch": 0.149948062721797,
      "grad_norm": 3.671875,
      "learning_rate": 4.736760299509516e-05,
      "loss": 0.8502,
      "step": 10574
    },
    {
      "epoch": 0.14997642437542322,
      "grad_norm": 3.046875,
      "learning_rate": 4.7366604952849364e-05,
      "loss": 0.8359,
      "step": 10576
    },
    {
      "epoch": 0.15000478602904943,
      "grad_norm": 2.78125,
      "learning_rate": 4.736560673196e-05,
      "loss": 0.7901,
      "step": 10578
    },
    {
      "epoch": 0.15003314768267564,
      "grad_norm": 3.40625,
      "learning_rate": 4.736460833243505e-05,
      "loss": 0.8647,
      "step": 10580
    },
    {
      "epoch": 0.15006150933630186,
      "grad_norm": 2.9375,
      "learning_rate": 4.7363609754282466e-05,
      "loss": 0.8754,
      "step": 10582
    },
    {
      "epoch": 0.15008987098992807,
      "grad_norm": 2.90625,
      "learning_rate": 4.7362610997510255e-05,
      "loss": 0.8299,
      "step": 10584
    },
    {
      "epoch": 0.15011823264355428,
      "grad_norm": 3.140625,
      "learning_rate": 4.7361612062126374e-05,
      "loss": 0.8453,
      "step": 10586
    },
    {
      "epoch": 0.1501465942971805,
      "grad_norm": 3.125,
      "learning_rate": 4.73606129481388e-05,
      "loss": 0.8733,
      "step": 10588
    },
    {
      "epoch": 0.1501749559508067,
      "grad_norm": 2.875,
      "learning_rate": 4.735961365555552e-05,
      "loss": 0.8369,
      "step": 10590
    },
    {
      "epoch": 0.15020331760443292,
      "grad_norm": 3.328125,
      "learning_rate": 4.735861418438452e-05,
      "loss": 0.854,
      "step": 10592
    },
    {
      "epoch": 0.15023167925805914,
      "grad_norm": 3.40625,
      "learning_rate": 4.735761453463378e-05,
      "loss": 0.8656,
      "step": 10594
    },
    {
      "epoch": 0.15026004091168535,
      "grad_norm": 3.3125,
      "learning_rate": 4.735661470631127e-05,
      "loss": 0.8859,
      "step": 10596
    },
    {
      "epoch": 0.15028840256531156,
      "grad_norm": 3.28125,
      "learning_rate": 4.735561469942499e-05,
      "loss": 0.8423,
      "step": 10598
    },
    {
      "epoch": 0.15031676421893778,
      "grad_norm": 3.203125,
      "learning_rate": 4.7354614513982934e-05,
      "loss": 0.8685,
      "step": 10600
    },
    {
      "epoch": 0.150345125872564,
      "grad_norm": 3.078125,
      "learning_rate": 4.7353614149993074e-05,
      "loss": 0.8093,
      "step": 10602
    },
    {
      "epoch": 0.1503734875261902,
      "grad_norm": 2.6875,
      "learning_rate": 4.73526136074634e-05,
      "loss": 0.8242,
      "step": 10604
    },
    {
      "epoch": 0.15040184917981642,
      "grad_norm": 2.828125,
      "learning_rate": 4.735161288640192e-05,
      "loss": 0.8224,
      "step": 10606
    },
    {
      "epoch": 0.15043021083344266,
      "grad_norm": 2.734375,
      "learning_rate": 4.7350611986816616e-05,
      "loss": 0.8413,
      "step": 10608
    },
    {
      "epoch": 0.15045857248706887,
      "grad_norm": 2.859375,
      "learning_rate": 4.734961090871548e-05,
      "loss": 0.8802,
      "step": 10610
    },
    {
      "epoch": 0.15048693414069508,
      "grad_norm": 3.15625,
      "learning_rate": 4.734860965210651e-05,
      "loss": 0.8258,
      "step": 10612
    },
    {
      "epoch": 0.1505152957943213,
      "grad_norm": 3.078125,
      "learning_rate": 4.734760821699771e-05,
      "loss": 0.8835,
      "step": 10614
    },
    {
      "epoch": 0.1505436574479475,
      "grad_norm": 3.109375,
      "learning_rate": 4.734660660339707e-05,
      "loss": 0.853,
      "step": 10616
    },
    {
      "epoch": 0.15057201910157372,
      "grad_norm": 3.15625,
      "learning_rate": 4.734560481131259e-05,
      "loss": 0.8682,
      "step": 10618
    },
    {
      "epoch": 0.15060038075519994,
      "grad_norm": 2.875,
      "learning_rate": 4.734460284075227e-05,
      "loss": 0.867,
      "step": 10620
    },
    {
      "epoch": 0.15062874240882615,
      "grad_norm": 3.0625,
      "learning_rate": 4.734360069172413e-05,
      "loss": 0.8725,
      "step": 10622
    },
    {
      "epoch": 0.15065710406245236,
      "grad_norm": 3.15625,
      "learning_rate": 4.734259836423615e-05,
      "loss": 0.9182,
      "step": 10624
    },
    {
      "epoch": 0.15068546571607858,
      "grad_norm": 3.296875,
      "learning_rate": 4.734159585829635e-05,
      "loss": 0.8568,
      "step": 10626
    },
    {
      "epoch": 0.1507138273697048,
      "grad_norm": 3.359375,
      "learning_rate": 4.7340593173912737e-05,
      "loss": 0.8534,
      "step": 10628
    },
    {
      "epoch": 0.150742189023331,
      "grad_norm": 3.203125,
      "learning_rate": 4.733959031109331e-05,
      "loss": 0.8352,
      "step": 10630
    },
    {
      "epoch": 0.15077055067695722,
      "grad_norm": 3.34375,
      "learning_rate": 4.733858726984609e-05,
      "loss": 0.8328,
      "step": 10632
    },
    {
      "epoch": 0.15079891233058343,
      "grad_norm": 3.03125,
      "learning_rate": 4.7337584050179074e-05,
      "loss": 0.8525,
      "step": 10634
    },
    {
      "epoch": 0.15082727398420964,
      "grad_norm": 3.0625,
      "learning_rate": 4.733658065210029e-05,
      "loss": 0.8715,
      "step": 10636
    },
    {
      "epoch": 0.15085563563783586,
      "grad_norm": 3.328125,
      "learning_rate": 4.733557707561775e-05,
      "loss": 0.8584,
      "step": 10638
    },
    {
      "epoch": 0.15088399729146207,
      "grad_norm": 3.09375,
      "learning_rate": 4.7334573320739466e-05,
      "loss": 0.8288,
      "step": 10640
    },
    {
      "epoch": 0.15091235894508828,
      "grad_norm": 3.15625,
      "learning_rate": 4.733356938747345e-05,
      "loss": 0.8389,
      "step": 10642
    },
    {
      "epoch": 0.1509407205987145,
      "grad_norm": 2.90625,
      "learning_rate": 4.7332565275827733e-05,
      "loss": 0.7991,
      "step": 10644
    },
    {
      "epoch": 0.15096908225234074,
      "grad_norm": 3.125,
      "learning_rate": 4.733156098581032e-05,
      "loss": 0.8429,
      "step": 10646
    },
    {
      "epoch": 0.15099744390596695,
      "grad_norm": 3.546875,
      "learning_rate": 4.733055651742925e-05,
      "loss": 0.8911,
      "step": 10648
    },
    {
      "epoch": 0.15102580555959316,
      "grad_norm": 3.328125,
      "learning_rate": 4.732955187069252e-05,
      "loss": 0.8577,
      "step": 10650
    },
    {
      "epoch": 0.15105416721321938,
      "grad_norm": 2.875,
      "learning_rate": 4.7328547045608185e-05,
      "loss": 0.851,
      "step": 10652
    },
    {
      "epoch": 0.1510825288668456,
      "grad_norm": 2.765625,
      "learning_rate": 4.7327542042184245e-05,
      "loss": 0.836,
      "step": 10654
    },
    {
      "epoch": 0.1511108905204718,
      "grad_norm": 3.015625,
      "learning_rate": 4.7326536860428746e-05,
      "loss": 0.8914,
      "step": 10656
    },
    {
      "epoch": 0.15113925217409802,
      "grad_norm": 3.078125,
      "learning_rate": 4.73255315003497e-05,
      "loss": 0.8547,
      "step": 10658
    },
    {
      "epoch": 0.15116761382772423,
      "grad_norm": 3.171875,
      "learning_rate": 4.7324525961955144e-05,
      "loss": 0.8766,
      "step": 10660
    },
    {
      "epoch": 0.15119597548135044,
      "grad_norm": 3.046875,
      "learning_rate": 4.7323520245253114e-05,
      "loss": 0.8682,
      "step": 10662
    },
    {
      "epoch": 0.15122433713497666,
      "grad_norm": 3.40625,
      "learning_rate": 4.7322514350251645e-05,
      "loss": 0.8539,
      "step": 10664
    },
    {
      "epoch": 0.15125269878860287,
      "grad_norm": 2.984375,
      "learning_rate": 4.732150827695876e-05,
      "loss": 0.876,
      "step": 10666
    },
    {
      "epoch": 0.15128106044222908,
      "grad_norm": 3.46875,
      "learning_rate": 4.7320502025382495e-05,
      "loss": 0.8412,
      "step": 10668
    },
    {
      "epoch": 0.1513094220958553,
      "grad_norm": 2.9375,
      "learning_rate": 4.7319495595530894e-05,
      "loss": 0.8098,
      "step": 10670
    },
    {
      "epoch": 0.1513377837494815,
      "grad_norm": 3.078125,
      "learning_rate": 4.7318488987411994e-05,
      "loss": 0.8816,
      "step": 10672
    },
    {
      "epoch": 0.15136614540310772,
      "grad_norm": 3.390625,
      "learning_rate": 4.731748220103384e-05,
      "loss": 0.8814,
      "step": 10674
    },
    {
      "epoch": 0.15139450705673393,
      "grad_norm": 3.171875,
      "learning_rate": 4.7316475236404454e-05,
      "loss": 0.7992,
      "step": 10676
    },
    {
      "epoch": 0.15142286871036015,
      "grad_norm": 3.171875,
      "learning_rate": 4.73154680935319e-05,
      "loss": 0.8243,
      "step": 10678
    },
    {
      "epoch": 0.15145123036398636,
      "grad_norm": 2.953125,
      "learning_rate": 4.731446077242422e-05,
      "loss": 0.8025,
      "step": 10680
    },
    {
      "epoch": 0.15147959201761257,
      "grad_norm": 3.078125,
      "learning_rate": 4.7313453273089445e-05,
      "loss": 0.8614,
      "step": 10682
    },
    {
      "epoch": 0.15150795367123882,
      "grad_norm": 3.21875,
      "learning_rate": 4.7312445595535625e-05,
      "loss": 0.8442,
      "step": 10684
    },
    {
      "epoch": 0.15153631532486503,
      "grad_norm": 2.75,
      "learning_rate": 4.7311437739770824e-05,
      "loss": 0.8514,
      "step": 10686
    },
    {
      "epoch": 0.15156467697849124,
      "grad_norm": 3.078125,
      "learning_rate": 4.731042970580307e-05,
      "loss": 0.8575,
      "step": 10688
    },
    {
      "epoch": 0.15159303863211746,
      "grad_norm": 3.125,
      "learning_rate": 4.7309421493640436e-05,
      "loss": 0.8497,
      "step": 10690
    },
    {
      "epoch": 0.15162140028574367,
      "grad_norm": 3.234375,
      "learning_rate": 4.730841310329096e-05,
      "loss": 0.876,
      "step": 10692
    },
    {
      "epoch": 0.15164976193936988,
      "grad_norm": 3.140625,
      "learning_rate": 4.730740453476271e-05,
      "loss": 0.9269,
      "step": 10694
    },
    {
      "epoch": 0.1516781235929961,
      "grad_norm": 3.234375,
      "learning_rate": 4.7306395788063724e-05,
      "loss": 0.8301,
      "step": 10696
    },
    {
      "epoch": 0.1517064852466223,
      "grad_norm": 3.171875,
      "learning_rate": 4.7305386863202065e-05,
      "loss": 0.9016,
      "step": 10698
    },
    {
      "epoch": 0.15173484690024852,
      "grad_norm": 3.15625,
      "learning_rate": 4.7304377760185796e-05,
      "loss": 0.8385,
      "step": 10700
    },
    {
      "epoch": 0.15176320855387473,
      "grad_norm": 3.171875,
      "learning_rate": 4.7303368479022974e-05,
      "loss": 0.8781,
      "step": 10702
    },
    {
      "epoch": 0.15179157020750095,
      "grad_norm": 3.046875,
      "learning_rate": 4.730235901972166e-05,
      "loss": 0.8595,
      "step": 10704
    },
    {
      "epoch": 0.15181993186112716,
      "grad_norm": 3.078125,
      "learning_rate": 4.730134938228992e-05,
      "loss": 0.8832,
      "step": 10706
    },
    {
      "epoch": 0.15184829351475337,
      "grad_norm": 3.140625,
      "learning_rate": 4.7300339566735805e-05,
      "loss": 0.8634,
      "step": 10708
    },
    {
      "epoch": 0.1518766551683796,
      "grad_norm": 2.96875,
      "learning_rate": 4.72993295730674e-05,
      "loss": 0.8175,
      "step": 10710
    },
    {
      "epoch": 0.1519050168220058,
      "grad_norm": 3.359375,
      "learning_rate": 4.729831940129277e-05,
      "loss": 0.7978,
      "step": 10712
    },
    {
      "epoch": 0.151933378475632,
      "grad_norm": 3.28125,
      "learning_rate": 4.729730905141996e-05,
      "loss": 0.8862,
      "step": 10714
    },
    {
      "epoch": 0.15196174012925823,
      "grad_norm": 3.078125,
      "learning_rate": 4.729629852345706e-05,
      "loss": 0.8722,
      "step": 10716
    },
    {
      "epoch": 0.15199010178288444,
      "grad_norm": 3.375,
      "learning_rate": 4.7295287817412136e-05,
      "loss": 0.8525,
      "step": 10718
    },
    {
      "epoch": 0.15201846343651065,
      "grad_norm": 3.0,
      "learning_rate": 4.7294276933293265e-05,
      "loss": 0.8345,
      "step": 10720
    },
    {
      "epoch": 0.15204682509013687,
      "grad_norm": 2.890625,
      "learning_rate": 4.729326587110852e-05,
      "loss": 0.8258,
      "step": 10722
    },
    {
      "epoch": 0.1520751867437631,
      "grad_norm": 2.734375,
      "learning_rate": 4.729225463086596e-05,
      "loss": 0.8902,
      "step": 10724
    },
    {
      "epoch": 0.15210354839738932,
      "grad_norm": 3.15625,
      "learning_rate": 4.729124321257369e-05,
      "loss": 0.8481,
      "step": 10726
    },
    {
      "epoch": 0.15213191005101553,
      "grad_norm": 3.203125,
      "learning_rate": 4.729023161623977e-05,
      "loss": 0.8615,
      "step": 10728
    },
    {
      "epoch": 0.15216027170464175,
      "grad_norm": 3.4375,
      "learning_rate": 4.7289219841872273e-05,
      "loss": 0.8372,
      "step": 10730
    },
    {
      "epoch": 0.15218863335826796,
      "grad_norm": 3.375,
      "learning_rate": 4.72882078894793e-05,
      "loss": 0.8692,
      "step": 10732
    },
    {
      "epoch": 0.15221699501189417,
      "grad_norm": 3.0,
      "learning_rate": 4.728719575906892e-05,
      "loss": 0.8189,
      "step": 10734
    },
    {
      "epoch": 0.1522453566655204,
      "grad_norm": 3.0,
      "learning_rate": 4.7286183450649224e-05,
      "loss": 0.8397,
      "step": 10736
    },
    {
      "epoch": 0.1522737183191466,
      "grad_norm": 2.921875,
      "learning_rate": 4.7285170964228294e-05,
      "loss": 0.8087,
      "step": 10738
    },
    {
      "epoch": 0.1523020799727728,
      "grad_norm": 3.0,
      "learning_rate": 4.728415829981422e-05,
      "loss": 0.8147,
      "step": 10740
    },
    {
      "epoch": 0.15233044162639903,
      "grad_norm": 3.34375,
      "learning_rate": 4.728314545741508e-05,
      "loss": 0.8569,
      "step": 10742
    },
    {
      "epoch": 0.15235880328002524,
      "grad_norm": 3.21875,
      "learning_rate": 4.7282132437038975e-05,
      "loss": 0.8676,
      "step": 10744
    },
    {
      "epoch": 0.15238716493365145,
      "grad_norm": 3.015625,
      "learning_rate": 4.728111923869399e-05,
      "loss": 0.8694,
      "step": 10746
    },
    {
      "epoch": 0.15241552658727767,
      "grad_norm": 2.96875,
      "learning_rate": 4.728010586238822e-05,
      "loss": 0.859,
      "step": 10748
    },
    {
      "epoch": 0.15244388824090388,
      "grad_norm": 3.296875,
      "learning_rate": 4.727909230812976e-05,
      "loss": 0.8582,
      "step": 10750
    },
    {
      "epoch": 0.1524722498945301,
      "grad_norm": 3.09375,
      "learning_rate": 4.72780785759267e-05,
      "loss": 0.8241,
      "step": 10752
    },
    {
      "epoch": 0.1525006115481563,
      "grad_norm": 3.234375,
      "learning_rate": 4.727706466578714e-05,
      "loss": 0.8889,
      "step": 10754
    },
    {
      "epoch": 0.15252897320178252,
      "grad_norm": 3.09375,
      "learning_rate": 4.727605057771919e-05,
      "loss": 0.8372,
      "step": 10756
    },
    {
      "epoch": 0.15255733485540873,
      "grad_norm": 3.3125,
      "learning_rate": 4.727503631173092e-05,
      "loss": 0.8457,
      "step": 10758
    },
    {
      "epoch": 0.15258569650903495,
      "grad_norm": 2.78125,
      "learning_rate": 4.7274021867830465e-05,
      "loss": 0.8614,
      "step": 10760
    },
    {
      "epoch": 0.1526140581626612,
      "grad_norm": 2.90625,
      "learning_rate": 4.727300724602591e-05,
      "loss": 0.8552,
      "step": 10762
    },
    {
      "epoch": 0.1526424198162874,
      "grad_norm": 3.234375,
      "learning_rate": 4.727199244632535e-05,
      "loss": 0.8533,
      "step": 10764
    },
    {
      "epoch": 0.1526707814699136,
      "grad_norm": 2.84375,
      "learning_rate": 4.727097746873691e-05,
      "loss": 0.8328,
      "step": 10766
    },
    {
      "epoch": 0.15269914312353983,
      "grad_norm": 3.1875,
      "learning_rate": 4.726996231326869e-05,
      "loss": 0.8426,
      "step": 10768
    },
    {
      "epoch": 0.15272750477716604,
      "grad_norm": 2.859375,
      "learning_rate": 4.726894697992879e-05,
      "loss": 0.8401,
      "step": 10770
    },
    {
      "epoch": 0.15275586643079225,
      "grad_norm": 3.375,
      "learning_rate": 4.7267931468725326e-05,
      "loss": 0.8962,
      "step": 10772
    },
    {
      "epoch": 0.15278422808441847,
      "grad_norm": 3.296875,
      "learning_rate": 4.7266915779666405e-05,
      "loss": 0.8235,
      "step": 10774
    },
    {
      "epoch": 0.15281258973804468,
      "grad_norm": 2.9375,
      "learning_rate": 4.7265899912760144e-05,
      "loss": 0.8862,
      "step": 10776
    },
    {
      "epoch": 0.1528409513916709,
      "grad_norm": 2.953125,
      "learning_rate": 4.7264883868014654e-05,
      "loss": 0.8473,
      "step": 10778
    },
    {
      "epoch": 0.1528693130452971,
      "grad_norm": 3.578125,
      "learning_rate": 4.726386764543805e-05,
      "loss": 0.8784,
      "step": 10780
    },
    {
      "epoch": 0.15289767469892332,
      "grad_norm": 3.15625,
      "learning_rate": 4.7262851245038456e-05,
      "loss": 0.8825,
      "step": 10782
    },
    {
      "epoch": 0.15292603635254953,
      "grad_norm": 3.046875,
      "learning_rate": 4.726183466682399e-05,
      "loss": 0.8681,
      "step": 10784
    },
    {
      "epoch": 0.15295439800617575,
      "grad_norm": 3.0625,
      "learning_rate": 4.726081791080276e-05,
      "loss": 0.8223,
      "step": 10786
    },
    {
      "epoch": 0.15298275965980196,
      "grad_norm": 2.796875,
      "learning_rate": 4.725980097698288e-05,
      "loss": 0.797,
      "step": 10788
    },
    {
      "epoch": 0.15301112131342817,
      "grad_norm": 3.296875,
      "learning_rate": 4.72587838653725e-05,
      "loss": 0.8692,
      "step": 10790
    },
    {
      "epoch": 0.15303948296705439,
      "grad_norm": 2.921875,
      "learning_rate": 4.725776657597972e-05,
      "loss": 0.8666,
      "step": 10792
    },
    {
      "epoch": 0.1530678446206806,
      "grad_norm": 3.046875,
      "learning_rate": 4.725674910881268e-05,
      "loss": 0.9008,
      "step": 10794
    },
    {
      "epoch": 0.1530962062743068,
      "grad_norm": 3.125,
      "learning_rate": 4.725573146387949e-05,
      "loss": 0.8792,
      "step": 10796
    },
    {
      "epoch": 0.15312456792793303,
      "grad_norm": 3.125,
      "learning_rate": 4.7254713641188304e-05,
      "loss": 0.8704,
      "step": 10798
    },
    {
      "epoch": 0.15315292958155927,
      "grad_norm": 3.015625,
      "learning_rate": 4.725369564074723e-05,
      "loss": 0.8353,
      "step": 10800
    },
    {
      "epoch": 0.15318129123518548,
      "grad_norm": 3.34375,
      "learning_rate": 4.72526774625644e-05,
      "loss": 0.853,
      "step": 10802
    },
    {
      "epoch": 0.1532096528888117,
      "grad_norm": 3.40625,
      "learning_rate": 4.725165910664795e-05,
      "loss": 0.8656,
      "step": 10804
    },
    {
      "epoch": 0.1532380145424379,
      "grad_norm": 3.109375,
      "learning_rate": 4.725064057300602e-05,
      "loss": 0.8262,
      "step": 10806
    },
    {
      "epoch": 0.15326637619606412,
      "grad_norm": 2.96875,
      "learning_rate": 4.724962186164674e-05,
      "loss": 0.8688,
      "step": 10808
    },
    {
      "epoch": 0.15329473784969033,
      "grad_norm": 2.953125,
      "learning_rate": 4.724860297257824e-05,
      "loss": 0.8439,
      "step": 10810
    },
    {
      "epoch": 0.15332309950331655,
      "grad_norm": 3.171875,
      "learning_rate": 4.724758390580867e-05,
      "loss": 0.8513,
      "step": 10812
    },
    {
      "epoch": 0.15335146115694276,
      "grad_norm": 3.34375,
      "learning_rate": 4.724656466134616e-05,
      "loss": 0.8601,
      "step": 10814
    },
    {
      "epoch": 0.15337982281056897,
      "grad_norm": 2.890625,
      "learning_rate": 4.7245545239198855e-05,
      "loss": 0.8412,
      "step": 10816
    },
    {
      "epoch": 0.15340818446419519,
      "grad_norm": 4.0,
      "learning_rate": 4.724452563937489e-05,
      "loss": 0.8694,
      "step": 10818
    },
    {
      "epoch": 0.1534365461178214,
      "grad_norm": 3.484375,
      "learning_rate": 4.7243505861882434e-05,
      "loss": 0.851,
      "step": 10820
    },
    {
      "epoch": 0.1534649077714476,
      "grad_norm": 3.25,
      "learning_rate": 4.724248590672959e-05,
      "loss": 0.8515,
      "step": 10822
    },
    {
      "epoch": 0.15349326942507382,
      "grad_norm": 2.859375,
      "learning_rate": 4.724146577392454e-05,
      "loss": 0.8753,
      "step": 10824
    },
    {
      "epoch": 0.15352163107870004,
      "grad_norm": 3.09375,
      "learning_rate": 4.724044546347542e-05,
      "loss": 0.8399,
      "step": 10826
    },
    {
      "epoch": 0.15354999273232625,
      "grad_norm": 3.046875,
      "learning_rate": 4.723942497539038e-05,
      "loss": 0.8265,
      "step": 10828
    },
    {
      "epoch": 0.15357835438595246,
      "grad_norm": 2.90625,
      "learning_rate": 4.723840430967756e-05,
      "loss": 0.867,
      "step": 10830
    },
    {
      "epoch": 0.15360671603957868,
      "grad_norm": 3.34375,
      "learning_rate": 4.723738346634513e-05,
      "loss": 0.8405,
      "step": 10832
    },
    {
      "epoch": 0.1536350776932049,
      "grad_norm": 3.15625,
      "learning_rate": 4.7236362445401227e-05,
      "loss": 0.8562,
      "step": 10834
    },
    {
      "epoch": 0.1536634393468311,
      "grad_norm": 3.609375,
      "learning_rate": 4.723534124685402e-05,
      "loss": 0.8798,
      "step": 10836
    },
    {
      "epoch": 0.15369180100045735,
      "grad_norm": 3.109375,
      "learning_rate": 4.723431987071166e-05,
      "loss": 0.8262,
      "step": 10838
    },
    {
      "epoch": 0.15372016265408356,
      "grad_norm": 3.0625,
      "learning_rate": 4.72332983169823e-05,
      "loss": 0.8394,
      "step": 10840
    },
    {
      "epoch": 0.15374852430770977,
      "grad_norm": 3.109375,
      "learning_rate": 4.723227658567411e-05,
      "loss": 0.8423,
      "step": 10842
    },
    {
      "epoch": 0.15377688596133599,
      "grad_norm": 2.953125,
      "learning_rate": 4.723125467679523e-05,
      "loss": 0.8594,
      "step": 10844
    },
    {
      "epoch": 0.1538052476149622,
      "grad_norm": 2.890625,
      "learning_rate": 4.723023259035385e-05,
      "loss": 0.8316,
      "step": 10846
    },
    {
      "epoch": 0.1538336092685884,
      "grad_norm": 3.359375,
      "learning_rate": 4.722921032635812e-05,
      "loss": 0.9154,
      "step": 10848
    },
    {
      "epoch": 0.15386197092221462,
      "grad_norm": 3.25,
      "learning_rate": 4.72281878848162e-05,
      "loss": 0.8611,
      "step": 10850
    },
    {
      "epoch": 0.15389033257584084,
      "grad_norm": 2.875,
      "learning_rate": 4.722716526573626e-05,
      "loss": 0.8358,
      "step": 10852
    },
    {
      "epoch": 0.15391869422946705,
      "grad_norm": 3.421875,
      "learning_rate": 4.722614246912647e-05,
      "loss": 0.8122,
      "step": 10854
    },
    {
      "epoch": 0.15394705588309326,
      "grad_norm": 2.859375,
      "learning_rate": 4.7225119494995e-05,
      "loss": 0.8489,
      "step": 10856
    },
    {
      "epoch": 0.15397541753671948,
      "grad_norm": 3.40625,
      "learning_rate": 4.722409634335001e-05,
      "loss": 0.8791,
      "step": 10858
    },
    {
      "epoch": 0.1540037791903457,
      "grad_norm": 3.0,
      "learning_rate": 4.722307301419968e-05,
      "loss": 0.8559,
      "step": 10860
    },
    {
      "epoch": 0.1540321408439719,
      "grad_norm": 3.15625,
      "learning_rate": 4.722204950755219e-05,
      "loss": 0.8293,
      "step": 10862
    },
    {
      "epoch": 0.15406050249759812,
      "grad_norm": 3.1875,
      "learning_rate": 4.722102582341571e-05,
      "loss": 0.8668,
      "step": 10864
    },
    {
      "epoch": 0.15408886415122433,
      "grad_norm": 3.125,
      "learning_rate": 4.722000196179841e-05,
      "loss": 0.8517,
      "step": 10866
    },
    {
      "epoch": 0.15411722580485054,
      "grad_norm": 3.265625,
      "learning_rate": 4.721897792270848e-05,
      "loss": 0.8068,
      "step": 10868
    },
    {
      "epoch": 0.15414558745847676,
      "grad_norm": 2.765625,
      "learning_rate": 4.7217953706154074e-05,
      "loss": 0.8635,
      "step": 10870
    },
    {
      "epoch": 0.15417394911210297,
      "grad_norm": 2.859375,
      "learning_rate": 4.7216929312143396e-05,
      "loss": 0.8683,
      "step": 10872
    },
    {
      "epoch": 0.15420231076572918,
      "grad_norm": 3.109375,
      "learning_rate": 4.7215904740684634e-05,
      "loss": 0.8272,
      "step": 10874
    },
    {
      "epoch": 0.1542306724193554,
      "grad_norm": 3.1875,
      "learning_rate": 4.7214879991785955e-05,
      "loss": 0.8546,
      "step": 10876
    },
    {
      "epoch": 0.15425903407298164,
      "grad_norm": 3.125,
      "learning_rate": 4.721385506545554e-05,
      "loss": 0.8566,
      "step": 10878
    },
    {
      "epoch": 0.15428739572660785,
      "grad_norm": 2.625,
      "learning_rate": 4.721282996170159e-05,
      "loss": 0.8234,
      "step": 10880
    },
    {
      "epoch": 0.15431575738023406,
      "grad_norm": 3.375,
      "learning_rate": 4.7211804680532276e-05,
      "loss": 0.8506,
      "step": 10882
    },
    {
      "epoch": 0.15434411903386028,
      "grad_norm": 3.0625,
      "learning_rate": 4.72107792219558e-05,
      "loss": 0.8685,
      "step": 10884
    },
    {
      "epoch": 0.1543724806874865,
      "grad_norm": 3.140625,
      "learning_rate": 4.7209753585980355e-05,
      "loss": 0.8528,
      "step": 10886
    },
    {
      "epoch": 0.1544008423411127,
      "grad_norm": 3.0625,
      "learning_rate": 4.7208727772614125e-05,
      "loss": 0.8811,
      "step": 10888
    },
    {
      "epoch": 0.15442920399473892,
      "grad_norm": 2.734375,
      "learning_rate": 4.72077017818653e-05,
      "loss": 0.8816,
      "step": 10890
    },
    {
      "epoch": 0.15445756564836513,
      "grad_norm": 2.734375,
      "learning_rate": 4.7206675613742084e-05,
      "loss": 0.7986,
      "step": 10892
    },
    {
      "epoch": 0.15448592730199134,
      "grad_norm": 2.6875,
      "learning_rate": 4.720564926825267e-05,
      "loss": 0.8251,
      "step": 10894
    },
    {
      "epoch": 0.15451428895561756,
      "grad_norm": 3.421875,
      "learning_rate": 4.7204622745405255e-05,
      "loss": 0.8616,
      "step": 10896
    },
    {
      "epoch": 0.15454265060924377,
      "grad_norm": 3.5,
      "learning_rate": 4.720359604520803e-05,
      "loss": 0.8627,
      "step": 10898
    },
    {
      "epoch": 0.15457101226286998,
      "grad_norm": 3.3125,
      "learning_rate": 4.7202569167669206e-05,
      "loss": 0.869,
      "step": 10900
    },
    {
      "epoch": 0.1545993739164962,
      "grad_norm": 3.15625,
      "learning_rate": 4.720154211279698e-05,
      "loss": 0.8751,
      "step": 10902
    },
    {
      "epoch": 0.1546277355701224,
      "grad_norm": 3.0625,
      "learning_rate": 4.720051488059956e-05,
      "loss": 0.7982,
      "step": 10904
    },
    {
      "epoch": 0.15465609722374862,
      "grad_norm": 2.796875,
      "learning_rate": 4.7199487471085145e-05,
      "loss": 0.82,
      "step": 10906
    },
    {
      "epoch": 0.15468445887737484,
      "grad_norm": 3.40625,
      "learning_rate": 4.719845988426195e-05,
      "loss": 0.8417,
      "step": 10908
    },
    {
      "epoch": 0.15471282053100105,
      "grad_norm": 2.953125,
      "learning_rate": 4.719743212013816e-05,
      "loss": 0.8134,
      "step": 10910
    },
    {
      "epoch": 0.15474118218462726,
      "grad_norm": 2.921875,
      "learning_rate": 4.719640417872201e-05,
      "loss": 0.8642,
      "step": 10912
    },
    {
      "epoch": 0.15476954383825348,
      "grad_norm": 3.484375,
      "learning_rate": 4.719537606002169e-05,
      "loss": 0.8987,
      "step": 10914
    },
    {
      "epoch": 0.15479790549187972,
      "grad_norm": 3.703125,
      "learning_rate": 4.7194347764045435e-05,
      "loss": 0.9024,
      "step": 10916
    },
    {
      "epoch": 0.15482626714550593,
      "grad_norm": 3.40625,
      "learning_rate": 4.7193319290801427e-05,
      "loss": 0.8815,
      "step": 10918
    },
    {
      "epoch": 0.15485462879913214,
      "grad_norm": 3.09375,
      "learning_rate": 4.719229064029791e-05,
      "loss": 0.8757,
      "step": 10920
    },
    {
      "epoch": 0.15488299045275836,
      "grad_norm": 3.234375,
      "learning_rate": 4.7191261812543084e-05,
      "loss": 0.8833,
      "step": 10922
    },
    {
      "epoch": 0.15491135210638457,
      "grad_norm": 3.0625,
      "learning_rate": 4.719023280754517e-05,
      "loss": 0.9141,
      "step": 10924
    },
    {
      "epoch": 0.15493971376001078,
      "grad_norm": 2.890625,
      "learning_rate": 4.718920362531239e-05,
      "loss": 0.8534,
      "step": 10926
    },
    {
      "epoch": 0.154968075413637,
      "grad_norm": 3.0625,
      "learning_rate": 4.718817426585296e-05,
      "loss": 0.8669,
      "step": 10928
    },
    {
      "epoch": 0.1549964370672632,
      "grad_norm": 3.234375,
      "learning_rate": 4.718714472917509e-05,
      "loss": 0.8801,
      "step": 10930
    },
    {
      "epoch": 0.15502479872088942,
      "grad_norm": 2.96875,
      "learning_rate": 4.718611501528703e-05,
      "loss": 0.8831,
      "step": 10932
    },
    {
      "epoch": 0.15505316037451564,
      "grad_norm": 3.078125,
      "learning_rate": 4.718508512419698e-05,
      "loss": 0.8782,
      "step": 10934
    },
    {
      "epoch": 0.15508152202814185,
      "grad_norm": 2.953125,
      "learning_rate": 4.718405505591318e-05,
      "loss": 0.8657,
      "step": 10936
    },
    {
      "epoch": 0.15510988368176806,
      "grad_norm": 3.203125,
      "learning_rate": 4.718302481044385e-05,
      "loss": 0.8123,
      "step": 10938
    },
    {
      "epoch": 0.15513824533539428,
      "grad_norm": 2.890625,
      "learning_rate": 4.718199438779722e-05,
      "loss": 0.8473,
      "step": 10940
    },
    {
      "epoch": 0.1551666069890205,
      "grad_norm": 2.953125,
      "learning_rate": 4.718096378798153e-05,
      "loss": 0.8624,
      "step": 10942
    },
    {
      "epoch": 0.1551949686426467,
      "grad_norm": 3.0625,
      "learning_rate": 4.717993301100499e-05,
      "loss": 0.8787,
      "step": 10944
    },
    {
      "epoch": 0.15522333029627292,
      "grad_norm": 2.765625,
      "learning_rate": 4.717890205687586e-05,
      "loss": 0.8178,
      "step": 10946
    },
    {
      "epoch": 0.15525169194989913,
      "grad_norm": 3.34375,
      "learning_rate": 4.7177870925602354e-05,
      "loss": 0.8567,
      "step": 10948
    },
    {
      "epoch": 0.15528005360352534,
      "grad_norm": 2.671875,
      "learning_rate": 4.717683961719271e-05,
      "loss": 0.8356,
      "step": 10950
    },
    {
      "epoch": 0.15530841525715156,
      "grad_norm": 3.203125,
      "learning_rate": 4.717580813165517e-05,
      "loss": 0.8421,
      "step": 10952
    },
    {
      "epoch": 0.1553367769107778,
      "grad_norm": 3.015625,
      "learning_rate": 4.717477646899797e-05,
      "loss": 0.8622,
      "step": 10954
    },
    {
      "epoch": 0.155365138564404,
      "grad_norm": 3.890625,
      "learning_rate": 4.7173744629229354e-05,
      "loss": 0.8358,
      "step": 10956
    },
    {
      "epoch": 0.15539350021803022,
      "grad_norm": 3.65625,
      "learning_rate": 4.717271261235756e-05,
      "loss": 0.8596,
      "step": 10958
    },
    {
      "epoch": 0.15542186187165644,
      "grad_norm": 2.921875,
      "learning_rate": 4.717168041839083e-05,
      "loss": 0.8515,
      "step": 10960
    },
    {
      "epoch": 0.15545022352528265,
      "grad_norm": 3.34375,
      "learning_rate": 4.7170648047337415e-05,
      "loss": 0.8765,
      "step": 10962
    },
    {
      "epoch": 0.15547858517890886,
      "grad_norm": 2.9375,
      "learning_rate": 4.7169615499205546e-05,
      "loss": 0.8578,
      "step": 10964
    },
    {
      "epoch": 0.15550694683253508,
      "grad_norm": 3.28125,
      "learning_rate": 4.7168582774003486e-05,
      "loss": 0.8834,
      "step": 10966
    },
    {
      "epoch": 0.1555353084861613,
      "grad_norm": 3.1875,
      "learning_rate": 4.716754987173948e-05,
      "loss": 0.8671,
      "step": 10968
    },
    {
      "epoch": 0.1555636701397875,
      "grad_norm": 2.796875,
      "learning_rate": 4.7166516792421765e-05,
      "loss": 0.8555,
      "step": 10970
    },
    {
      "epoch": 0.15559203179341372,
      "grad_norm": 3.109375,
      "learning_rate": 4.7165483536058605e-05,
      "loss": 0.8604,
      "step": 10972
    },
    {
      "epoch": 0.15562039344703993,
      "grad_norm": 3.203125,
      "learning_rate": 4.716445010265826e-05,
      "loss": 0.8221,
      "step": 10974
    },
    {
      "epoch": 0.15564875510066614,
      "grad_norm": 3.0,
      "learning_rate": 4.7163416492228966e-05,
      "loss": 0.8499,
      "step": 10976
    },
    {
      "epoch": 0.15567711675429236,
      "grad_norm": 3.046875,
      "learning_rate": 4.716238270477899e-05,
      "loss": 0.8699,
      "step": 10978
    },
    {
      "epoch": 0.15570547840791857,
      "grad_norm": 2.984375,
      "learning_rate": 4.716134874031657e-05,
      "loss": 0.8367,
      "step": 10980
    },
    {
      "epoch": 0.15573384006154478,
      "grad_norm": 2.71875,
      "learning_rate": 4.716031459884999e-05,
      "loss": 0.8115,
      "step": 10982
    },
    {
      "epoch": 0.155762201715171,
      "grad_norm": 3.640625,
      "learning_rate": 4.7159280280387495e-05,
      "loss": 0.8136,
      "step": 10984
    },
    {
      "epoch": 0.1557905633687972,
      "grad_norm": 2.90625,
      "learning_rate": 4.715824578493736e-05,
      "loss": 0.9136,
      "step": 10986
    },
    {
      "epoch": 0.15581892502242342,
      "grad_norm": 3.09375,
      "learning_rate": 4.715721111250783e-05,
      "loss": 0.8731,
      "step": 10988
    },
    {
      "epoch": 0.15584728667604963,
      "grad_norm": 3.078125,
      "learning_rate": 4.715617626310718e-05,
      "loss": 0.8863,
      "step": 10990
    },
    {
      "epoch": 0.15587564832967588,
      "grad_norm": 3.421875,
      "learning_rate": 4.715514123674367e-05,
      "loss": 0.8613,
      "step": 10992
    },
    {
      "epoch": 0.1559040099833021,
      "grad_norm": 3.109375,
      "learning_rate": 4.715410603342557e-05,
      "loss": 0.7919,
      "step": 10994
    },
    {
      "epoch": 0.1559323716369283,
      "grad_norm": 3.046875,
      "learning_rate": 4.715307065316115e-05,
      "loss": 0.8035,
      "step": 10996
    },
    {
      "epoch": 0.15596073329055452,
      "grad_norm": 2.890625,
      "learning_rate": 4.715203509595868e-05,
      "loss": 0.8812,
      "step": 10998
    },
    {
      "epoch": 0.15598909494418073,
      "grad_norm": 3.015625,
      "learning_rate": 4.715099936182642e-05,
      "loss": 0.8406,
      "step": 11000
    },
    {
      "epoch": 0.15601745659780694,
      "grad_norm": 3.078125,
      "learning_rate": 4.714996345077265e-05,
      "loss": 0.8831,
      "step": 11002
    },
    {
      "epoch": 0.15604581825143315,
      "grad_norm": 3.4375,
      "learning_rate": 4.714892736280565e-05,
      "loss": 0.8271,
      "step": 11004
    },
    {
      "epoch": 0.15607417990505937,
      "grad_norm": 3.109375,
      "learning_rate": 4.71478910979337e-05,
      "loss": 0.8397,
      "step": 11006
    },
    {
      "epoch": 0.15610254155868558,
      "grad_norm": 3.0625,
      "learning_rate": 4.7146854656165045e-05,
      "loss": 0.8772,
      "step": 11008
    },
    {
      "epoch": 0.1561309032123118,
      "grad_norm": 3.046875,
      "learning_rate": 4.7145818037507996e-05,
      "loss": 0.849,
      "step": 11010
    },
    {
      "epoch": 0.156159264865938,
      "grad_norm": 3.359375,
      "learning_rate": 4.7144781241970815e-05,
      "loss": 0.8993,
      "step": 11012
    },
    {
      "epoch": 0.15618762651956422,
      "grad_norm": 3.203125,
      "learning_rate": 4.71437442695618e-05,
      "loss": 0.8754,
      "step": 11014
    },
    {
      "epoch": 0.15621598817319043,
      "grad_norm": 2.84375,
      "learning_rate": 4.714270712028921e-05,
      "loss": 0.8987,
      "step": 11016
    },
    {
      "epoch": 0.15624434982681665,
      "grad_norm": 3.34375,
      "learning_rate": 4.7141669794161345e-05,
      "loss": 0.8291,
      "step": 11018
    },
    {
      "epoch": 0.15627271148044286,
      "grad_norm": 2.640625,
      "learning_rate": 4.714063229118649e-05,
      "loss": 0.8518,
      "step": 11020
    },
    {
      "epoch": 0.15630107313406907,
      "grad_norm": 3.109375,
      "learning_rate": 4.713959461137293e-05,
      "loss": 0.8206,
      "step": 11022
    },
    {
      "epoch": 0.1563294347876953,
      "grad_norm": 2.78125,
      "learning_rate": 4.7138556754728944e-05,
      "loss": 0.8363,
      "step": 11024
    },
    {
      "epoch": 0.1563577964413215,
      "grad_norm": 3.21875,
      "learning_rate": 4.713751872126284e-05,
      "loss": 0.8701,
      "step": 11026
    },
    {
      "epoch": 0.1563861580949477,
      "grad_norm": 2.984375,
      "learning_rate": 4.713648051098289e-05,
      "loss": 0.8793,
      "step": 11028
    },
    {
      "epoch": 0.15641451974857393,
      "grad_norm": 3.109375,
      "learning_rate": 4.713544212389739e-05,
      "loss": 0.8601,
      "step": 11030
    },
    {
      "epoch": 0.15644288140220017,
      "grad_norm": 3.203125,
      "learning_rate": 4.713440356001464e-05,
      "loss": 0.8541,
      "step": 11032
    },
    {
      "epoch": 0.15647124305582638,
      "grad_norm": 3.0,
      "learning_rate": 4.713336481934294e-05,
      "loss": 0.8945,
      "step": 11034
    },
    {
      "epoch": 0.1564996047094526,
      "grad_norm": 2.8125,
      "learning_rate": 4.713232590189057e-05,
      "loss": 0.7985,
      "step": 11036
    },
    {
      "epoch": 0.1565279663630788,
      "grad_norm": 2.921875,
      "learning_rate": 4.713128680766584e-05,
      "loss": 0.8518,
      "step": 11038
    },
    {
      "epoch": 0.15655632801670502,
      "grad_norm": 3.03125,
      "learning_rate": 4.713024753667704e-05,
      "loss": 0.8538,
      "step": 11040
    },
    {
      "epoch": 0.15658468967033123,
      "grad_norm": 2.828125,
      "learning_rate": 4.712920808893249e-05,
      "loss": 0.8243,
      "step": 11042
    },
    {
      "epoch": 0.15661305132395745,
      "grad_norm": 3.125,
      "learning_rate": 4.712816846444047e-05,
      "loss": 0.9115,
      "step": 11044
    },
    {
      "epoch": 0.15664141297758366,
      "grad_norm": 3.3125,
      "learning_rate": 4.712712866320929e-05,
      "loss": 0.8904,
      "step": 11046
    },
    {
      "epoch": 0.15666977463120987,
      "grad_norm": 3.0,
      "learning_rate": 4.712608868524726e-05,
      "loss": 0.8497,
      "step": 11048
    },
    {
      "epoch": 0.1566981362848361,
      "grad_norm": 3.3125,
      "learning_rate": 4.7125048530562686e-05,
      "loss": 0.8311,
      "step": 11050
    },
    {
      "epoch": 0.1567264979384623,
      "grad_norm": 3.359375,
      "learning_rate": 4.712400819916387e-05,
      "loss": 0.8532,
      "step": 11052
    },
    {
      "epoch": 0.1567548595920885,
      "grad_norm": 3.171875,
      "learning_rate": 4.712296769105913e-05,
      "loss": 0.7876,
      "step": 11054
    },
    {
      "epoch": 0.15678322124571473,
      "grad_norm": 3.203125,
      "learning_rate": 4.712192700625677e-05,
      "loss": 0.8316,
      "step": 11056
    },
    {
      "epoch": 0.15681158289934094,
      "grad_norm": 3.28125,
      "learning_rate": 4.71208861447651e-05,
      "loss": 0.8606,
      "step": 11058
    },
    {
      "epoch": 0.15683994455296715,
      "grad_norm": 2.78125,
      "learning_rate": 4.711984510659244e-05,
      "loss": 0.8486,
      "step": 11060
    },
    {
      "epoch": 0.15686830620659337,
      "grad_norm": 3.09375,
      "learning_rate": 4.71188038917471e-05,
      "loss": 0.8634,
      "step": 11062
    },
    {
      "epoch": 0.15689666786021958,
      "grad_norm": 3.578125,
      "learning_rate": 4.7117762500237405e-05,
      "loss": 0.8459,
      "step": 11064
    },
    {
      "epoch": 0.1569250295138458,
      "grad_norm": 3.234375,
      "learning_rate": 4.711672093207166e-05,
      "loss": 0.8516,
      "step": 11066
    },
    {
      "epoch": 0.156953391167472,
      "grad_norm": 3.421875,
      "learning_rate": 4.711567918725819e-05,
      "loss": 0.882,
      "step": 11068
    },
    {
      "epoch": 0.15698175282109825,
      "grad_norm": 3.53125,
      "learning_rate": 4.711463726580531e-05,
      "loss": 0.8967,
      "step": 11070
    },
    {
      "epoch": 0.15701011447472446,
      "grad_norm": 3.3125,
      "learning_rate": 4.711359516772135e-05,
      "loss": 0.913,
      "step": 11072
    },
    {
      "epoch": 0.15703847612835067,
      "grad_norm": 3.140625,
      "learning_rate": 4.7112552893014634e-05,
      "loss": 0.8731,
      "step": 11074
    },
    {
      "epoch": 0.1570668377819769,
      "grad_norm": 2.921875,
      "learning_rate": 4.711151044169348e-05,
      "loss": 0.8617,
      "step": 11076
    },
    {
      "epoch": 0.1570951994356031,
      "grad_norm": 2.921875,
      "learning_rate": 4.7110467813766215e-05,
      "loss": 0.8272,
      "step": 11078
    },
    {
      "epoch": 0.1571235610892293,
      "grad_norm": 3.03125,
      "learning_rate": 4.710942500924118e-05,
      "loss": 0.7927,
      "step": 11080
    },
    {
      "epoch": 0.15715192274285553,
      "grad_norm": 3.046875,
      "learning_rate": 4.710838202812668e-05,
      "loss": 0.8717,
      "step": 11082
    },
    {
      "epoch": 0.15718028439648174,
      "grad_norm": 3.375,
      "learning_rate": 4.710733887043106e-05,
      "loss": 0.9206,
      "step": 11084
    },
    {
      "epoch": 0.15720864605010795,
      "grad_norm": 2.796875,
      "learning_rate": 4.710629553616266e-05,
      "loss": 0.843,
      "step": 11086
    },
    {
      "epoch": 0.15723700770373417,
      "grad_norm": 2.90625,
      "learning_rate": 4.7105252025329795e-05,
      "loss": 0.8475,
      "step": 11088
    },
    {
      "epoch": 0.15726536935736038,
      "grad_norm": 3.046875,
      "learning_rate": 4.71042083379408e-05,
      "loss": 0.807,
      "step": 11090
    },
    {
      "epoch": 0.1572937310109866,
      "grad_norm": 3.046875,
      "learning_rate": 4.7103164474004037e-05,
      "loss": 0.8238,
      "step": 11092
    },
    {
      "epoch": 0.1573220926646128,
      "grad_norm": 3.09375,
      "learning_rate": 4.7102120433527806e-05,
      "loss": 0.8518,
      "step": 11094
    },
    {
      "epoch": 0.15735045431823902,
      "grad_norm": 3.203125,
      "learning_rate": 4.7101076216520476e-05,
      "loss": 0.8442,
      "step": 11096
    },
    {
      "epoch": 0.15737881597186523,
      "grad_norm": 2.9375,
      "learning_rate": 4.7100031822990376e-05,
      "loss": 0.8668,
      "step": 11098
    },
    {
      "epoch": 0.15740717762549145,
      "grad_norm": 2.9375,
      "learning_rate": 4.7098987252945846e-05,
      "loss": 0.8282,
      "step": 11100
    },
    {
      "epoch": 0.15743553927911766,
      "grad_norm": 3.078125,
      "learning_rate": 4.709794250639523e-05,
      "loss": 0.7823,
      "step": 11102
    },
    {
      "epoch": 0.15746390093274387,
      "grad_norm": 3.109375,
      "learning_rate": 4.709689758334688e-05,
      "loss": 0.8433,
      "step": 11104
    },
    {
      "epoch": 0.15749226258637009,
      "grad_norm": 3.109375,
      "learning_rate": 4.709585248380912e-05,
      "loss": 0.8488,
      "step": 11106
    },
    {
      "epoch": 0.15752062423999633,
      "grad_norm": 3.015625,
      "learning_rate": 4.709480720779033e-05,
      "loss": 0.8455,
      "step": 11108
    },
    {
      "epoch": 0.15754898589362254,
      "grad_norm": 3.21875,
      "learning_rate": 4.7093761755298825e-05,
      "loss": 0.8606,
      "step": 11110
    },
    {
      "epoch": 0.15757734754724875,
      "grad_norm": 2.9375,
      "learning_rate": 4.709271612634298e-05,
      "loss": 0.8425,
      "step": 11112
    },
    {
      "epoch": 0.15760570920087497,
      "grad_norm": 3.265625,
      "learning_rate": 4.7091670320931134e-05,
      "loss": 0.8196,
      "step": 11114
    },
    {
      "epoch": 0.15763407085450118,
      "grad_norm": 3.03125,
      "learning_rate": 4.709062433907165e-05,
      "loss": 0.8583,
      "step": 11116
    },
    {
      "epoch": 0.1576624325081274,
      "grad_norm": 3.0625,
      "learning_rate": 4.7089578180772874e-05,
      "loss": 0.8326,
      "step": 11118
    },
    {
      "epoch": 0.1576907941617536,
      "grad_norm": 3.5,
      "learning_rate": 4.708853184604316e-05,
      "loss": 0.9074,
      "step": 11120
    },
    {
      "epoch": 0.15771915581537982,
      "grad_norm": 3.234375,
      "learning_rate": 4.7087485334890866e-05,
      "loss": 0.8143,
      "step": 11122
    },
    {
      "epoch": 0.15774751746900603,
      "grad_norm": 3.4375,
      "learning_rate": 4.708643864732436e-05,
      "loss": 0.8554,
      "step": 11124
    },
    {
      "epoch": 0.15777587912263225,
      "grad_norm": 3.0625,
      "learning_rate": 4.7085391783352e-05,
      "loss": 0.8634,
      "step": 11126
    },
    {
      "epoch": 0.15780424077625846,
      "grad_norm": 2.984375,
      "learning_rate": 4.708434474298213e-05,
      "loss": 0.8856,
      "step": 11128
    },
    {
      "epoch": 0.15783260242988467,
      "grad_norm": 2.828125,
      "learning_rate": 4.708329752622314e-05,
      "loss": 0.8834,
      "step": 11130
    },
    {
      "epoch": 0.15786096408351089,
      "grad_norm": 3.125,
      "learning_rate": 4.708225013308336e-05,
      "loss": 0.8756,
      "step": 11132
    },
    {
      "epoch": 0.1578893257371371,
      "grad_norm": 3.34375,
      "learning_rate": 4.7081202563571194e-05,
      "loss": 0.84,
      "step": 11134
    },
    {
      "epoch": 0.1579176873907633,
      "grad_norm": 3.21875,
      "learning_rate": 4.7080154817694976e-05,
      "loss": 0.8784,
      "step": 11136
    },
    {
      "epoch": 0.15794604904438952,
      "grad_norm": 3.328125,
      "learning_rate": 4.70791068954631e-05,
      "loss": 0.8593,
      "step": 11138
    },
    {
      "epoch": 0.15797441069801574,
      "grad_norm": 3.203125,
      "learning_rate": 4.707805879688392e-05,
      "loss": 0.854,
      "step": 11140
    },
    {
      "epoch": 0.15800277235164195,
      "grad_norm": 2.71875,
      "learning_rate": 4.7077010521965816e-05,
      "loss": 0.8378,
      "step": 11142
    },
    {
      "epoch": 0.15803113400526816,
      "grad_norm": 3.296875,
      "learning_rate": 4.707596207071715e-05,
      "loss": 0.8812,
      "step": 11144
    },
    {
      "epoch": 0.1580594956588944,
      "grad_norm": 2.890625,
      "learning_rate": 4.70749134431463e-05,
      "loss": 0.8696,
      "step": 11146
    },
    {
      "epoch": 0.15808785731252062,
      "grad_norm": 3.5,
      "learning_rate": 4.707386463926165e-05,
      "loss": 0.8111,
      "step": 11148
    },
    {
      "epoch": 0.15811621896614683,
      "grad_norm": 3.328125,
      "learning_rate": 4.707281565907157e-05,
      "loss": 0.8954,
      "step": 11150
    },
    {
      "epoch": 0.15814458061977305,
      "grad_norm": 2.953125,
      "learning_rate": 4.707176650258444e-05,
      "loss": 0.8123,
      "step": 11152
    },
    {
      "epoch": 0.15817294227339926,
      "grad_norm": 3.109375,
      "learning_rate": 4.707071716980863e-05,
      "loss": 0.9134,
      "step": 11154
    },
    {
      "epoch": 0.15820130392702547,
      "grad_norm": 3.109375,
      "learning_rate": 4.7069667660752544e-05,
      "loss": 0.8849,
      "step": 11156
    },
    {
      "epoch": 0.15822966558065168,
      "grad_norm": 3.1875,
      "learning_rate": 4.7068617975424546e-05,
      "loss": 0.8468,
      "step": 11158
    },
    {
      "epoch": 0.1582580272342779,
      "grad_norm": 2.6875,
      "learning_rate": 4.706756811383302e-05,
      "loss": 0.7827,
      "step": 11160
    },
    {
      "epoch": 0.1582863888879041,
      "grad_norm": 3.359375,
      "learning_rate": 4.706651807598635e-05,
      "loss": 0.8526,
      "step": 11162
    },
    {
      "epoch": 0.15831475054153032,
      "grad_norm": 3.546875,
      "learning_rate": 4.706546786189294e-05,
      "loss": 0.8705,
      "step": 11164
    },
    {
      "epoch": 0.15834311219515654,
      "grad_norm": 3.328125,
      "learning_rate": 4.706441747156116e-05,
      "loss": 0.8156,
      "step": 11166
    },
    {
      "epoch": 0.15837147384878275,
      "grad_norm": 3.40625,
      "learning_rate": 4.7063366904999404e-05,
      "loss": 0.829,
      "step": 11168
    },
    {
      "epoch": 0.15839983550240896,
      "grad_norm": 2.640625,
      "learning_rate": 4.706231616221607e-05,
      "loss": 0.8178,
      "step": 11170
    },
    {
      "epoch": 0.15842819715603518,
      "grad_norm": 2.90625,
      "learning_rate": 4.706126524321954e-05,
      "loss": 0.8704,
      "step": 11172
    },
    {
      "epoch": 0.1584565588096614,
      "grad_norm": 3.421875,
      "learning_rate": 4.706021414801822e-05,
      "loss": 0.8898,
      "step": 11174
    },
    {
      "epoch": 0.1584849204632876,
      "grad_norm": 2.953125,
      "learning_rate": 4.70591628766205e-05,
      "loss": 0.8189,
      "step": 11176
    },
    {
      "epoch": 0.15851328211691382,
      "grad_norm": 3.140625,
      "learning_rate": 4.705811142903477e-05,
      "loss": 0.8909,
      "step": 11178
    },
    {
      "epoch": 0.15854164377054003,
      "grad_norm": 3.3125,
      "learning_rate": 4.7057059805269434e-05,
      "loss": 0.8583,
      "step": 11180
    },
    {
      "epoch": 0.15857000542416624,
      "grad_norm": 3.03125,
      "learning_rate": 4.7056008005332886e-05,
      "loss": 0.8883,
      "step": 11182
    },
    {
      "epoch": 0.15859836707779246,
      "grad_norm": 3.203125,
      "learning_rate": 4.7054956029233535e-05,
      "loss": 0.8679,
      "step": 11184
    },
    {
      "epoch": 0.1586267287314187,
      "grad_norm": 3.171875,
      "learning_rate": 4.705390387697978e-05,
      "loss": 0.9084,
      "step": 11186
    },
    {
      "epoch": 0.1586550903850449,
      "grad_norm": 3.046875,
      "learning_rate": 4.705285154858002e-05,
      "loss": 0.8186,
      "step": 11188
    },
    {
      "epoch": 0.15868345203867112,
      "grad_norm": 3.125,
      "learning_rate": 4.7051799044042664e-05,
      "loss": 0.8662,
      "step": 11190
    },
    {
      "epoch": 0.15871181369229734,
      "grad_norm": 3.09375,
      "learning_rate": 4.705074636337612e-05,
      "loss": 0.8466,
      "step": 11192
    },
    {
      "epoch": 0.15874017534592355,
      "grad_norm": 3.109375,
      "learning_rate": 4.704969350658879e-05,
      "loss": 0.8762,
      "step": 11194
    },
    {
      "epoch": 0.15876853699954976,
      "grad_norm": 3.03125,
      "learning_rate": 4.7048640473689086e-05,
      "loss": 0.8659,
      "step": 11196
    },
    {
      "epoch": 0.15879689865317598,
      "grad_norm": 3.421875,
      "learning_rate": 4.7047587264685434e-05,
      "loss": 0.9,
      "step": 11198
    },
    {
      "epoch": 0.1588252603068022,
      "grad_norm": 2.78125,
      "learning_rate": 4.704653387958622e-05,
      "loss": 0.8421,
      "step": 11200
    },
    {
      "epoch": 0.1588536219604284,
      "grad_norm": 2.984375,
      "learning_rate": 4.704548031839987e-05,
      "loss": 0.9011,
      "step": 11202
    },
    {
      "epoch": 0.15888198361405462,
      "grad_norm": 3.0625,
      "learning_rate": 4.70444265811348e-05,
      "loss": 0.8748,
      "step": 11204
    },
    {
      "epoch": 0.15891034526768083,
      "grad_norm": 3.5625,
      "learning_rate": 4.704337266779942e-05,
      "loss": 0.855,
      "step": 11206
    },
    {
      "epoch": 0.15893870692130704,
      "grad_norm": 3.34375,
      "learning_rate": 4.704231857840216e-05,
      "loss": 0.8265,
      "step": 11208
    },
    {
      "epoch": 0.15896706857493326,
      "grad_norm": 3.375,
      "learning_rate": 4.704126431295143e-05,
      "loss": 0.8315,
      "step": 11210
    },
    {
      "epoch": 0.15899543022855947,
      "grad_norm": 3.234375,
      "learning_rate": 4.704020987145565e-05,
      "loss": 0.8484,
      "step": 11212
    },
    {
      "epoch": 0.15902379188218568,
      "grad_norm": 3.640625,
      "learning_rate": 4.703915525392324e-05,
      "loss": 0.8211,
      "step": 11214
    },
    {
      "epoch": 0.1590521535358119,
      "grad_norm": 3.625,
      "learning_rate": 4.703810046036263e-05,
      "loss": 0.8512,
      "step": 11216
    },
    {
      "epoch": 0.1590805151894381,
      "grad_norm": 2.96875,
      "learning_rate": 4.703704549078225e-05,
      "loss": 0.8499,
      "step": 11218
    },
    {
      "epoch": 0.15910887684306432,
      "grad_norm": 3.203125,
      "learning_rate": 4.7035990345190504e-05,
      "loss": 0.8608,
      "step": 11220
    },
    {
      "epoch": 0.15913723849669054,
      "grad_norm": 3.015625,
      "learning_rate": 4.703493502359584e-05,
      "loss": 0.8526,
      "step": 11222
    },
    {
      "epoch": 0.15916560015031678,
      "grad_norm": 3.21875,
      "learning_rate": 4.7033879526006674e-05,
      "loss": 0.9033,
      "step": 11224
    },
    {
      "epoch": 0.159193961803943,
      "grad_norm": 3.046875,
      "learning_rate": 4.703282385243145e-05,
      "loss": 0.8407,
      "step": 11226
    },
    {
      "epoch": 0.1592223234575692,
      "grad_norm": 3.234375,
      "learning_rate": 4.7031768002878596e-05,
      "loss": 0.8395,
      "step": 11228
    },
    {
      "epoch": 0.15925068511119542,
      "grad_norm": 3.71875,
      "learning_rate": 4.7030711977356536e-05,
      "loss": 0.9158,
      "step": 11230
    },
    {
      "epoch": 0.15927904676482163,
      "grad_norm": 3.15625,
      "learning_rate": 4.702965577587371e-05,
      "loss": 0.8865,
      "step": 11232
    },
    {
      "epoch": 0.15930740841844784,
      "grad_norm": 2.875,
      "learning_rate": 4.702859939843854e-05,
      "loss": 0.8097,
      "step": 11234
    },
    {
      "epoch": 0.15933577007207406,
      "grad_norm": 2.71875,
      "learning_rate": 4.70275428450595e-05,
      "loss": 0.8103,
      "step": 11236
    },
    {
      "epoch": 0.15936413172570027,
      "grad_norm": 2.984375,
      "learning_rate": 4.7026486115744995e-05,
      "loss": 0.8915,
      "step": 11238
    },
    {
      "epoch": 0.15939249337932648,
      "grad_norm": 3.4375,
      "learning_rate": 4.702542921050348e-05,
      "loss": 0.9178,
      "step": 11240
    },
    {
      "epoch": 0.1594208550329527,
      "grad_norm": 2.90625,
      "learning_rate": 4.702437212934339e-05,
      "loss": 0.8455,
      "step": 11242
    },
    {
      "epoch": 0.1594492166865789,
      "grad_norm": 3.0625,
      "learning_rate": 4.702331487227317e-05,
      "loss": 0.8746,
      "step": 11244
    },
    {
      "epoch": 0.15947757834020512,
      "grad_norm": 3.0,
      "learning_rate": 4.7022257439301266e-05,
      "loss": 0.8578,
      "step": 11246
    },
    {
      "epoch": 0.15950593999383134,
      "grad_norm": 3.109375,
      "learning_rate": 4.702119983043612e-05,
      "loss": 0.848,
      "step": 11248
    },
    {
      "epoch": 0.15953430164745755,
      "grad_norm": 3.09375,
      "learning_rate": 4.702014204568619e-05,
      "loss": 0.8508,
      "step": 11250
    },
    {
      "epoch": 0.15956266330108376,
      "grad_norm": 3.03125,
      "learning_rate": 4.701908408505992e-05,
      "loss": 0.8568,
      "step": 11252
    },
    {
      "epoch": 0.15959102495470998,
      "grad_norm": 3.171875,
      "learning_rate": 4.701802594856575e-05,
      "loss": 0.8186,
      "step": 11254
    },
    {
      "epoch": 0.1596193866083362,
      "grad_norm": 3.15625,
      "learning_rate": 4.701696763621213e-05,
      "loss": 0.8525,
      "step": 11256
    },
    {
      "epoch": 0.1596477482619624,
      "grad_norm": 3.125,
      "learning_rate": 4.701590914800753e-05,
      "loss": 0.8951,
      "step": 11258
    },
    {
      "epoch": 0.15967610991558862,
      "grad_norm": 3.015625,
      "learning_rate": 4.701485048396039e-05,
      "loss": 0.8287,
      "step": 11260
    },
    {
      "epoch": 0.15970447156921486,
      "grad_norm": 3.21875,
      "learning_rate": 4.701379164407917e-05,
      "loss": 0.8651,
      "step": 11262
    },
    {
      "epoch": 0.15973283322284107,
      "grad_norm": 3.640625,
      "learning_rate": 4.701273262837233e-05,
      "loss": 0.8517,
      "step": 11264
    },
    {
      "epoch": 0.15976119487646728,
      "grad_norm": 2.90625,
      "learning_rate": 4.7011673436848334e-05,
      "loss": 0.8433,
      "step": 11266
    },
    {
      "epoch": 0.1597895565300935,
      "grad_norm": 3.296875,
      "learning_rate": 4.701061406951563e-05,
      "loss": 0.9015,
      "step": 11268
    },
    {
      "epoch": 0.1598179181837197,
      "grad_norm": 2.703125,
      "learning_rate": 4.700955452638268e-05,
      "loss": 0.8102,
      "step": 11270
    },
    {
      "epoch": 0.15984627983734592,
      "grad_norm": 2.921875,
      "learning_rate": 4.7008494807457954e-05,
      "loss": 0.852,
      "step": 11272
    },
    {
      "epoch": 0.15987464149097214,
      "grad_norm": 2.953125,
      "learning_rate": 4.700743491274991e-05,
      "loss": 0.8546,
      "step": 11274
    },
    {
      "epoch": 0.15990300314459835,
      "grad_norm": 3.390625,
      "learning_rate": 4.700637484226702e-05,
      "loss": 0.8702,
      "step": 11276
    },
    {
      "epoch": 0.15993136479822456,
      "grad_norm": 3.15625,
      "learning_rate": 4.7005314596017736e-05,
      "loss": 0.8966,
      "step": 11278
    },
    {
      "epoch": 0.15995972645185078,
      "grad_norm": 2.875,
      "learning_rate": 4.700425417401054e-05,
      "loss": 0.8192,
      "step": 11280
    },
    {
      "epoch": 0.159988088105477,
      "grad_norm": 2.828125,
      "learning_rate": 4.70031935762539e-05,
      "loss": 0.869,
      "step": 11282
    },
    {
      "epoch": 0.1600164497591032,
      "grad_norm": 3.015625,
      "learning_rate": 4.700213280275629e-05,
      "loss": 0.8911,
      "step": 11284
    },
    {
      "epoch": 0.16004481141272942,
      "grad_norm": 3.296875,
      "learning_rate": 4.700107185352617e-05,
      "loss": 0.8333,
      "step": 11286
    },
    {
      "epoch": 0.16007317306635563,
      "grad_norm": 3.140625,
      "learning_rate": 4.7000010728572026e-05,
      "loss": 0.8438,
      "step": 11288
    },
    {
      "epoch": 0.16010153471998184,
      "grad_norm": 3.3125,
      "learning_rate": 4.699894942790233e-05,
      "loss": 0.8883,
      "step": 11290
    },
    {
      "epoch": 0.16012989637360805,
      "grad_norm": 2.828125,
      "learning_rate": 4.699788795152555e-05,
      "loss": 0.8369,
      "step": 11292
    },
    {
      "epoch": 0.16015825802723427,
      "grad_norm": 3.0,
      "learning_rate": 4.6996826299450184e-05,
      "loss": 0.8655,
      "step": 11294
    },
    {
      "epoch": 0.16018661968086048,
      "grad_norm": 3.140625,
      "learning_rate": 4.6995764471684686e-05,
      "loss": 0.8583,
      "step": 11296
    },
    {
      "epoch": 0.1602149813344867,
      "grad_norm": 2.84375,
      "learning_rate": 4.6994702468237555e-05,
      "loss": 0.8532,
      "step": 11298
    },
    {
      "epoch": 0.16024334298811294,
      "grad_norm": 2.90625,
      "learning_rate": 4.699364028911727e-05,
      "loss": 0.8679,
      "step": 11300
    },
    {
      "epoch": 0.16027170464173915,
      "grad_norm": 3.0,
      "learning_rate": 4.6992577934332315e-05,
      "loss": 0.828,
      "step": 11302
    },
    {
      "epoch": 0.16030006629536536,
      "grad_norm": 3.34375,
      "learning_rate": 4.6991515403891174e-05,
      "loss": 0.8946,
      "step": 11304
    },
    {
      "epoch": 0.16032842794899158,
      "grad_norm": 3.234375,
      "learning_rate": 4.699045269780232e-05,
      "loss": 0.8637,
      "step": 11306
    },
    {
      "epoch": 0.1603567896026178,
      "grad_norm": 2.828125,
      "learning_rate": 4.698938981607427e-05,
      "loss": 0.8357,
      "step": 11308
    },
    {
      "epoch": 0.160385151256244,
      "grad_norm": 3.265625,
      "learning_rate": 4.698832675871548e-05,
      "loss": 0.8588,
      "step": 11310
    },
    {
      "epoch": 0.16041351290987022,
      "grad_norm": 2.765625,
      "learning_rate": 4.6987263525734474e-05,
      "loss": 0.8119,
      "step": 11312
    },
    {
      "epoch": 0.16044187456349643,
      "grad_norm": 3.421875,
      "learning_rate": 4.6986200117139724e-05,
      "loss": 0.8713,
      "step": 11314
    },
    {
      "epoch": 0.16047023621712264,
      "grad_norm": 2.84375,
      "learning_rate": 4.6985136532939724e-05,
      "loss": 0.8768,
      "step": 11316
    },
    {
      "epoch": 0.16049859787074885,
      "grad_norm": 2.953125,
      "learning_rate": 4.698407277314297e-05,
      "loss": 0.8498,
      "step": 11318
    },
    {
      "epoch": 0.16052695952437507,
      "grad_norm": 3.265625,
      "learning_rate": 4.698300883775796e-05,
      "loss": 0.8207,
      "step": 11320
    },
    {
      "epoch": 0.16055532117800128,
      "grad_norm": 3.359375,
      "learning_rate": 4.698194472679319e-05,
      "loss": 0.8436,
      "step": 11322
    },
    {
      "epoch": 0.1605836828316275,
      "grad_norm": 2.75,
      "learning_rate": 4.6980880440257166e-05,
      "loss": 0.8522,
      "step": 11324
    },
    {
      "epoch": 0.1606120444852537,
      "grad_norm": 3.125,
      "learning_rate": 4.6979815978158386e-05,
      "loss": 0.8744,
      "step": 11326
    },
    {
      "epoch": 0.16064040613887992,
      "grad_norm": 2.75,
      "learning_rate": 4.6978751340505345e-05,
      "loss": 0.8471,
      "step": 11328
    },
    {
      "epoch": 0.16066876779250613,
      "grad_norm": 2.84375,
      "learning_rate": 4.6977686527306556e-05,
      "loss": 0.8471,
      "step": 11330
    },
    {
      "epoch": 0.16069712944613235,
      "grad_norm": 3.15625,
      "learning_rate": 4.697662153857052e-05,
      "loss": 0.83,
      "step": 11332
    },
    {
      "epoch": 0.16072549109975856,
      "grad_norm": 3.140625,
      "learning_rate": 4.6975556374305734e-05,
      "loss": 0.8638,
      "step": 11334
    },
    {
      "epoch": 0.16075385275338477,
      "grad_norm": 3.328125,
      "learning_rate": 4.697449103452072e-05,
      "loss": 0.8425,
      "step": 11336
    },
    {
      "epoch": 0.160782214407011,
      "grad_norm": 3.1875,
      "learning_rate": 4.697342551922398e-05,
      "loss": 0.9139,
      "step": 11338
    },
    {
      "epoch": 0.16081057606063723,
      "grad_norm": 3.25,
      "learning_rate": 4.697235982842402e-05,
      "loss": 0.897,
      "step": 11340
    },
    {
      "epoch": 0.16083893771426344,
      "grad_norm": 3.328125,
      "learning_rate": 4.697129396212936e-05,
      "loss": 0.8933,
      "step": 11342
    },
    {
      "epoch": 0.16086729936788965,
      "grad_norm": 2.859375,
      "learning_rate": 4.697022792034851e-05,
      "loss": 0.834,
      "step": 11344
    },
    {
      "epoch": 0.16089566102151587,
      "grad_norm": 3.140625,
      "learning_rate": 4.696916170308998e-05,
      "loss": 0.8006,
      "step": 11346
    },
    {
      "epoch": 0.16092402267514208,
      "grad_norm": 2.96875,
      "learning_rate": 4.696809531036229e-05,
      "loss": 0.8654,
      "step": 11348
    },
    {
      "epoch": 0.1609523843287683,
      "grad_norm": 2.84375,
      "learning_rate": 4.6967028742173964e-05,
      "loss": 0.8287,
      "step": 11350
    },
    {
      "epoch": 0.1609807459823945,
      "grad_norm": 3.5,
      "learning_rate": 4.696596199853351e-05,
      "loss": 0.8639,
      "step": 11352
    },
    {
      "epoch": 0.16100910763602072,
      "grad_norm": 3.234375,
      "learning_rate": 4.6964895079449456e-05,
      "loss": 0.8723,
      "step": 11354
    },
    {
      "epoch": 0.16103746928964693,
      "grad_norm": 3.25,
      "learning_rate": 4.696382798493032e-05,
      "loss": 0.8628,
      "step": 11356
    },
    {
      "epoch": 0.16106583094327315,
      "grad_norm": 2.734375,
      "learning_rate": 4.696276071498462e-05,
      "loss": 0.8354,
      "step": 11358
    },
    {
      "epoch": 0.16109419259689936,
      "grad_norm": 3.078125,
      "learning_rate": 4.696169326962088e-05,
      "loss": 0.8005,
      "step": 11360
    },
    {
      "epoch": 0.16112255425052557,
      "grad_norm": 2.75,
      "learning_rate": 4.696062564884764e-05,
      "loss": 0.8375,
      "step": 11362
    },
    {
      "epoch": 0.1611509159041518,
      "grad_norm": 3.0625,
      "learning_rate": 4.695955785267342e-05,
      "loss": 0.8701,
      "step": 11364
    },
    {
      "epoch": 0.161179277557778,
      "grad_norm": 2.734375,
      "learning_rate": 4.6958489881106736e-05,
      "loss": 0.8459,
      "step": 11366
    },
    {
      "epoch": 0.1612076392114042,
      "grad_norm": 3.328125,
      "learning_rate": 4.695742173415614e-05,
      "loss": 0.8407,
      "step": 11368
    },
    {
      "epoch": 0.16123600086503043,
      "grad_norm": 2.8125,
      "learning_rate": 4.6956353411830144e-05,
      "loss": 0.8236,
      "step": 11370
    },
    {
      "epoch": 0.16126436251865664,
      "grad_norm": 3.34375,
      "learning_rate": 4.69552849141373e-05,
      "loss": 0.9502,
      "step": 11372
    },
    {
      "epoch": 0.16129272417228285,
      "grad_norm": 2.9375,
      "learning_rate": 4.695421624108612e-05,
      "loss": 0.8808,
      "step": 11374
    },
    {
      "epoch": 0.16132108582590907,
      "grad_norm": 3.0,
      "learning_rate": 4.695314739268516e-05,
      "loss": 0.8186,
      "step": 11376
    },
    {
      "epoch": 0.1613494474795353,
      "grad_norm": 3.0625,
      "learning_rate": 4.6952078368942934e-05,
      "loss": 0.8791,
      "step": 11378
    },
    {
      "epoch": 0.16137780913316152,
      "grad_norm": 2.90625,
      "learning_rate": 4.6951009169868e-05,
      "loss": 0.79,
      "step": 11380
    },
    {
      "epoch": 0.16140617078678773,
      "grad_norm": 3.265625,
      "learning_rate": 4.69499397954689e-05,
      "loss": 0.896,
      "step": 11382
    },
    {
      "epoch": 0.16143453244041395,
      "grad_norm": 3.515625,
      "learning_rate": 4.694887024575415e-05,
      "loss": 0.8577,
      "step": 11384
    },
    {
      "epoch": 0.16146289409404016,
      "grad_norm": 2.96875,
      "learning_rate": 4.694780052073232e-05,
      "loss": 0.8402,
      "step": 11386
    },
    {
      "epoch": 0.16149125574766637,
      "grad_norm": 3.1875,
      "learning_rate": 4.694673062041195e-05,
      "loss": 0.8484,
      "step": 11388
    },
    {
      "epoch": 0.1615196174012926,
      "grad_norm": 3.359375,
      "learning_rate": 4.694566054480157e-05,
      "loss": 0.852,
      "step": 11390
    },
    {
      "epoch": 0.1615479790549188,
      "grad_norm": 3.125,
      "learning_rate": 4.694459029390973e-05,
      "loss": 0.8122,
      "step": 11392
    },
    {
      "epoch": 0.161576340708545,
      "grad_norm": 3.15625,
      "learning_rate": 4.694351986774499e-05,
      "loss": 0.8588,
      "step": 11394
    },
    {
      "epoch": 0.16160470236217123,
      "grad_norm": 3.203125,
      "learning_rate": 4.6942449266315894e-05,
      "loss": 0.8571,
      "step": 11396
    },
    {
      "epoch": 0.16163306401579744,
      "grad_norm": 3.125,
      "learning_rate": 4.694137848963099e-05,
      "loss": 0.8692,
      "step": 11398
    },
    {
      "epoch": 0.16166142566942365,
      "grad_norm": 3.09375,
      "learning_rate": 4.6940307537698833e-05,
      "loss": 0.8334,
      "step": 11400
    },
    {
      "epoch": 0.16168978732304987,
      "grad_norm": 3.015625,
      "learning_rate": 4.693923641052798e-05,
      "loss": 0.8515,
      "step": 11402
    },
    {
      "epoch": 0.16171814897667608,
      "grad_norm": 3.015625,
      "learning_rate": 4.693816510812698e-05,
      "loss": 0.8818,
      "step": 11404
    },
    {
      "epoch": 0.1617465106303023,
      "grad_norm": 2.78125,
      "learning_rate": 4.693709363050439e-05,
      "loss": 0.8506,
      "step": 11406
    },
    {
      "epoch": 0.1617748722839285,
      "grad_norm": 3.53125,
      "learning_rate": 4.6936021977668775e-05,
      "loss": 0.8685,
      "step": 11408
    },
    {
      "epoch": 0.16180323393755472,
      "grad_norm": 3.046875,
      "learning_rate": 4.693495014962869e-05,
      "loss": 0.849,
      "step": 11410
    },
    {
      "epoch": 0.16183159559118093,
      "grad_norm": 3.296875,
      "learning_rate": 4.6933878146392685e-05,
      "loss": 0.8936,
      "step": 11412
    },
    {
      "epoch": 0.16185995724480715,
      "grad_norm": 2.984375,
      "learning_rate": 4.693280596796934e-05,
      "loss": 0.8687,
      "step": 11414
    },
    {
      "epoch": 0.1618883188984334,
      "grad_norm": 2.875,
      "learning_rate": 4.6931733614367213e-05,
      "loss": 0.8827,
      "step": 11416
    },
    {
      "epoch": 0.1619166805520596,
      "grad_norm": 2.953125,
      "learning_rate": 4.6930661085594856e-05,
      "loss": 0.8577,
      "step": 11418
    },
    {
      "epoch": 0.1619450422056858,
      "grad_norm": 3.375,
      "learning_rate": 4.6929588381660855e-05,
      "loss": 0.8423,
      "step": 11420
    },
    {
      "epoch": 0.16197340385931203,
      "grad_norm": 3.015625,
      "learning_rate": 4.692851550257377e-05,
      "loss": 0.838,
      "step": 11422
    },
    {
      "epoch": 0.16200176551293824,
      "grad_norm": 3.4375,
      "learning_rate": 4.6927442448342165e-05,
      "loss": 0.8198,
      "step": 11424
    },
    {
      "epoch": 0.16203012716656445,
      "grad_norm": 2.96875,
      "learning_rate": 4.6926369218974606e-05,
      "loss": 0.8818,
      "step": 11426
    },
    {
      "epoch": 0.16205848882019067,
      "grad_norm": 3.015625,
      "learning_rate": 4.6925295814479685e-05,
      "loss": 0.824,
      "step": 11428
    },
    {
      "epoch": 0.16208685047381688,
      "grad_norm": 2.890625,
      "learning_rate": 4.692422223486597e-05,
      "loss": 0.8132,
      "step": 11430
    },
    {
      "epoch": 0.1621152121274431,
      "grad_norm": 3.03125,
      "learning_rate": 4.692314848014202e-05,
      "loss": 0.8332,
      "step": 11432
    },
    {
      "epoch": 0.1621435737810693,
      "grad_norm": 3.3125,
      "learning_rate": 4.6922074550316416e-05,
      "loss": 0.9003,
      "step": 11434
    },
    {
      "epoch": 0.16217193543469552,
      "grad_norm": 3.265625,
      "learning_rate": 4.692100044539775e-05,
      "loss": 0.8576,
      "step": 11436
    },
    {
      "epoch": 0.16220029708832173,
      "grad_norm": 3.578125,
      "learning_rate": 4.691992616539459e-05,
      "loss": 0.8811,
      "step": 11438
    },
    {
      "epoch": 0.16222865874194795,
      "grad_norm": 3.3125,
      "learning_rate": 4.6918851710315516e-05,
      "loss": 0.8616,
      "step": 11440
    },
    {
      "epoch": 0.16225702039557416,
      "grad_norm": 3.765625,
      "learning_rate": 4.691777708016911e-05,
      "loss": 0.8868,
      "step": 11442
    },
    {
      "epoch": 0.16228538204920037,
      "grad_norm": 3.03125,
      "learning_rate": 4.6916702274963954e-05,
      "loss": 0.8452,
      "step": 11444
    },
    {
      "epoch": 0.16231374370282659,
      "grad_norm": 3.1875,
      "learning_rate": 4.691562729470864e-05,
      "loss": 0.8681,
      "step": 11446
    },
    {
      "epoch": 0.1623421053564528,
      "grad_norm": 3.453125,
      "learning_rate": 4.691455213941175e-05,
      "loss": 0.8583,
      "step": 11448
    },
    {
      "epoch": 0.162370467010079,
      "grad_norm": 3.203125,
      "learning_rate": 4.691347680908187e-05,
      "loss": 0.8464,
      "step": 11450
    },
    {
      "epoch": 0.16239882866370522,
      "grad_norm": 2.671875,
      "learning_rate": 4.691240130372758e-05,
      "loss": 0.8307,
      "step": 11452
    },
    {
      "epoch": 0.16242719031733147,
      "grad_norm": 3.109375,
      "learning_rate": 4.6911325623357486e-05,
      "loss": 0.8338,
      "step": 11454
    },
    {
      "epoch": 0.16245555197095768,
      "grad_norm": 3.078125,
      "learning_rate": 4.691024976798017e-05,
      "loss": 0.8442,
      "step": 11456
    },
    {
      "epoch": 0.1624839136245839,
      "grad_norm": 3.375,
      "learning_rate": 4.690917373760424e-05,
      "loss": 0.832,
      "step": 11458
    },
    {
      "epoch": 0.1625122752782101,
      "grad_norm": 3.265625,
      "learning_rate": 4.6908097532238266e-05,
      "loss": 0.8076,
      "step": 11460
    },
    {
      "epoch": 0.16254063693183632,
      "grad_norm": 3.40625,
      "learning_rate": 4.690702115189086e-05,
      "loss": 0.8543,
      "step": 11462
    },
    {
      "epoch": 0.16256899858546253,
      "grad_norm": 3.09375,
      "learning_rate": 4.690594459657061e-05,
      "loss": 0.8649,
      "step": 11464
    },
    {
      "epoch": 0.16259736023908875,
      "grad_norm": 3.34375,
      "learning_rate": 4.690486786628612e-05,
      "loss": 0.8624,
      "step": 11466
    },
    {
      "epoch": 0.16262572189271496,
      "grad_norm": 3.359375,
      "learning_rate": 4.690379096104599e-05,
      "loss": 0.903,
      "step": 11468
    },
    {
      "epoch": 0.16265408354634117,
      "grad_norm": 3.171875,
      "learning_rate": 4.690271388085883e-05,
      "loss": 0.8637,
      "step": 11470
    },
    {
      "epoch": 0.16268244519996738,
      "grad_norm": 2.921875,
      "learning_rate": 4.690163662573323e-05,
      "loss": 0.8222,
      "step": 11472
    },
    {
      "epoch": 0.1627108068535936,
      "grad_norm": 2.953125,
      "learning_rate": 4.690055919567778e-05,
      "loss": 0.8182,
      "step": 11474
    },
    {
      "epoch": 0.1627391685072198,
      "grad_norm": 2.90625,
      "learning_rate": 4.689948159070112e-05,
      "loss": 0.8298,
      "step": 11476
    },
    {
      "epoch": 0.16276753016084602,
      "grad_norm": 3.1875,
      "learning_rate": 4.689840381081184e-05,
      "loss": 0.8597,
      "step": 11478
    },
    {
      "epoch": 0.16279589181447224,
      "grad_norm": 2.765625,
      "learning_rate": 4.689732585601854e-05,
      "loss": 0.8127,
      "step": 11480
    },
    {
      "epoch": 0.16282425346809845,
      "grad_norm": 3.25,
      "learning_rate": 4.6896247726329846e-05,
      "loss": 0.8257,
      "step": 11482
    },
    {
      "epoch": 0.16285261512172466,
      "grad_norm": 2.78125,
      "learning_rate": 4.6895169421754353e-05,
      "loss": 0.8197,
      "step": 11484
    },
    {
      "epoch": 0.16288097677535088,
      "grad_norm": 3.21875,
      "learning_rate": 4.689409094230068e-05,
      "loss": 0.8919,
      "step": 11486
    },
    {
      "epoch": 0.1629093384289771,
      "grad_norm": 3.15625,
      "learning_rate": 4.689301228797746e-05,
      "loss": 0.8532,
      "step": 11488
    },
    {
      "epoch": 0.1629377000826033,
      "grad_norm": 2.859375,
      "learning_rate": 4.689193345879327e-05,
      "loss": 0.8087,
      "step": 11490
    },
    {
      "epoch": 0.16296606173622952,
      "grad_norm": 2.921875,
      "learning_rate": 4.689085445475676e-05,
      "loss": 0.8089,
      "step": 11492
    },
    {
      "epoch": 0.16299442338985576,
      "grad_norm": 3.125,
      "learning_rate": 4.688977527587652e-05,
      "loss": 0.8766,
      "step": 11494
    },
    {
      "epoch": 0.16302278504348197,
      "grad_norm": 2.671875,
      "learning_rate": 4.68886959221612e-05,
      "loss": 0.8046,
      "step": 11496
    },
    {
      "epoch": 0.16305114669710818,
      "grad_norm": 3.078125,
      "learning_rate": 4.688761639361941e-05,
      "loss": 0.8748,
      "step": 11498
    },
    {
      "epoch": 0.1630795083507344,
      "grad_norm": 3.21875,
      "learning_rate": 4.688653669025975e-05,
      "loss": 0.8427,
      "step": 11500
    },
    {
      "epoch": 0.1631078700043606,
      "grad_norm": 3.0625,
      "learning_rate": 4.688545681209087e-05,
      "loss": 0.8648,
      "step": 11502
    },
    {
      "epoch": 0.16313623165798682,
      "grad_norm": 2.765625,
      "learning_rate": 4.688437675912139e-05,
      "loss": 0.8057,
      "step": 11504
    },
    {
      "epoch": 0.16316459331161304,
      "grad_norm": 3.078125,
      "learning_rate": 4.688329653135993e-05,
      "loss": 0.8624,
      "step": 11506
    },
    {
      "epoch": 0.16319295496523925,
      "grad_norm": 2.984375,
      "learning_rate": 4.6882216128815124e-05,
      "loss": 0.9001,
      "step": 11508
    },
    {
      "epoch": 0.16322131661886546,
      "grad_norm": 3.203125,
      "learning_rate": 4.688113555149559e-05,
      "loss": 0.8948,
      "step": 11510
    },
    {
      "epoch": 0.16324967827249168,
      "grad_norm": 2.96875,
      "learning_rate": 4.6880054799409976e-05,
      "loss": 0.8707,
      "step": 11512
    },
    {
      "epoch": 0.1632780399261179,
      "grad_norm": 2.890625,
      "learning_rate": 4.68789738725669e-05,
      "loss": 0.878,
      "step": 11514
    },
    {
      "epoch": 0.1633064015797441,
      "grad_norm": 3.03125,
      "learning_rate": 4.6877892770975005e-05,
      "loss": 0.8464,
      "step": 11516
    },
    {
      "epoch": 0.16333476323337032,
      "grad_norm": 2.875,
      "learning_rate": 4.6876811494642916e-05,
      "loss": 0.858,
      "step": 11518
    },
    {
      "epoch": 0.16336312488699653,
      "grad_norm": 2.828125,
      "learning_rate": 4.687573004357928e-05,
      "loss": 0.8419,
      "step": 11520
    },
    {
      "epoch": 0.16339148654062274,
      "grad_norm": 2.8125,
      "learning_rate": 4.6874648417792724e-05,
      "loss": 0.8261,
      "step": 11522
    },
    {
      "epoch": 0.16341984819424896,
      "grad_norm": 2.875,
      "learning_rate": 4.6873566617291894e-05,
      "loss": 0.8312,
      "step": 11524
    },
    {
      "epoch": 0.16344820984787517,
      "grad_norm": 3.34375,
      "learning_rate": 4.687248464208542e-05,
      "loss": 0.8739,
      "step": 11526
    },
    {
      "epoch": 0.16347657150150138,
      "grad_norm": 3.109375,
      "learning_rate": 4.6871402492181964e-05,
      "loss": 0.8103,
      "step": 11528
    },
    {
      "epoch": 0.1635049331551276,
      "grad_norm": 3.359375,
      "learning_rate": 4.687032016759016e-05,
      "loss": 0.8238,
      "step": 11530
    },
    {
      "epoch": 0.16353329480875384,
      "grad_norm": 3.328125,
      "learning_rate": 4.686923766831864e-05,
      "loss": 0.8358,
      "step": 11532
    },
    {
      "epoch": 0.16356165646238005,
      "grad_norm": 2.828125,
      "learning_rate": 4.6868154994376064e-05,
      "loss": 0.7996,
      "step": 11534
    },
    {
      "epoch": 0.16359001811600626,
      "grad_norm": 3.421875,
      "learning_rate": 4.686707214577107e-05,
      "loss": 0.8327,
      "step": 11536
    },
    {
      "epoch": 0.16361837976963248,
      "grad_norm": 3.109375,
      "learning_rate": 4.686598912251232e-05,
      "loss": 0.8625,
      "step": 11538
    },
    {
      "epoch": 0.1636467414232587,
      "grad_norm": 3.09375,
      "learning_rate": 4.686490592460845e-05,
      "loss": 0.8517,
      "step": 11540
    },
    {
      "epoch": 0.1636751030768849,
      "grad_norm": 2.90625,
      "learning_rate": 4.686382255206813e-05,
      "loss": 0.8575,
      "step": 11542
    },
    {
      "epoch": 0.16370346473051112,
      "grad_norm": 3.125,
      "learning_rate": 4.686273900489999e-05,
      "loss": 0.8284,
      "step": 11544
    },
    {
      "epoch": 0.16373182638413733,
      "grad_norm": 3.40625,
      "learning_rate": 4.6861655283112695e-05,
      "loss": 0.8591,
      "step": 11546
    },
    {
      "epoch": 0.16376018803776354,
      "grad_norm": 3.171875,
      "learning_rate": 4.68605713867149e-05,
      "loss": 0.8571,
      "step": 11548
    },
    {
      "epoch": 0.16378854969138976,
      "grad_norm": 3.75,
      "learning_rate": 4.6859487315715273e-05,
      "loss": 0.8581,
      "step": 11550
    },
    {
      "epoch": 0.16381691134501597,
      "grad_norm": 3.09375,
      "learning_rate": 4.6858403070122456e-05,
      "loss": 0.8553,
      "step": 11552
    },
    {
      "epoch": 0.16384527299864218,
      "grad_norm": 3.25,
      "learning_rate": 4.6857318649945116e-05,
      "loss": 0.8583,
      "step": 11554
    },
    {
      "epoch": 0.1638736346522684,
      "grad_norm": 3.109375,
      "learning_rate": 4.6856234055191915e-05,
      "loss": 0.8615,
      "step": 11556
    },
    {
      "epoch": 0.1639019963058946,
      "grad_norm": 3.25,
      "learning_rate": 4.685514928587151e-05,
      "loss": 0.8148,
      "step": 11558
    },
    {
      "epoch": 0.16393035795952082,
      "grad_norm": 3.171875,
      "learning_rate": 4.685406434199258e-05,
      "loss": 0.8645,
      "step": 11560
    },
    {
      "epoch": 0.16395871961314704,
      "grad_norm": 3.171875,
      "learning_rate": 4.685297922356378e-05,
      "loss": 0.8507,
      "step": 11562
    },
    {
      "epoch": 0.16398708126677325,
      "grad_norm": 2.671875,
      "learning_rate": 4.685189393059377e-05,
      "loss": 0.8361,
      "step": 11564
    },
    {
      "epoch": 0.16401544292039946,
      "grad_norm": 2.765625,
      "learning_rate": 4.6850808463091226e-05,
      "loss": 0.8398,
      "step": 11566
    },
    {
      "epoch": 0.16404380457402568,
      "grad_norm": 3.3125,
      "learning_rate": 4.6849722821064814e-05,
      "loss": 0.8931,
      "step": 11568
    },
    {
      "epoch": 0.16407216622765192,
      "grad_norm": 3.734375,
      "learning_rate": 4.684863700452322e-05,
      "loss": 0.8418,
      "step": 11570
    },
    {
      "epoch": 0.16410052788127813,
      "grad_norm": 3.109375,
      "learning_rate": 4.68475510134751e-05,
      "loss": 0.8259,
      "step": 11572
    },
    {
      "epoch": 0.16412888953490434,
      "grad_norm": 2.875,
      "learning_rate": 4.684646484792913e-05,
      "loss": 0.8524,
      "step": 11574
    },
    {
      "epoch": 0.16415725118853056,
      "grad_norm": 2.984375,
      "learning_rate": 4.684537850789399e-05,
      "loss": 0.8189,
      "step": 11576
    },
    {
      "epoch": 0.16418561284215677,
      "grad_norm": 3.140625,
      "learning_rate": 4.684429199337835e-05,
      "loss": 0.895,
      "step": 11578
    },
    {
      "epoch": 0.16421397449578298,
      "grad_norm": 3.0625,
      "learning_rate": 4.68432053043909e-05,
      "loss": 0.8626,
      "step": 11580
    },
    {
      "epoch": 0.1642423361494092,
      "grad_norm": 2.984375,
      "learning_rate": 4.6842118440940306e-05,
      "loss": 0.8424,
      "step": 11582
    },
    {
      "epoch": 0.1642706978030354,
      "grad_norm": 2.75,
      "learning_rate": 4.6841031403035263e-05,
      "loss": 0.8172,
      "step": 11584
    },
    {
      "epoch": 0.16429905945666162,
      "grad_norm": 3.421875,
      "learning_rate": 4.6839944190684435e-05,
      "loss": 0.8304,
      "step": 11586
    },
    {
      "epoch": 0.16432742111028784,
      "grad_norm": 3.140625,
      "learning_rate": 4.683885680389652e-05,
      "loss": 0.8203,
      "step": 11588
    },
    {
      "epoch": 0.16435578276391405,
      "grad_norm": 3.328125,
      "learning_rate": 4.683776924268021e-05,
      "loss": 0.8517,
      "step": 11590
    },
    {
      "epoch": 0.16438414441754026,
      "grad_norm": 3.15625,
      "learning_rate": 4.683668150704417e-05,
      "loss": 0.8671,
      "step": 11592
    },
    {
      "epoch": 0.16441250607116648,
      "grad_norm": 3.15625,
      "learning_rate": 4.68355935969971e-05,
      "loss": 0.8732,
      "step": 11594
    },
    {
      "epoch": 0.1644408677247927,
      "grad_norm": 2.9375,
      "learning_rate": 4.683450551254768e-05,
      "loss": 0.7991,
      "step": 11596
    },
    {
      "epoch": 0.1644692293784189,
      "grad_norm": 2.8125,
      "learning_rate": 4.683341725370462e-05,
      "loss": 0.7983,
      "step": 11598
    },
    {
      "epoch": 0.16449759103204512,
      "grad_norm": 3.1875,
      "learning_rate": 4.683232882047659e-05,
      "loss": 0.8389,
      "step": 11600
    },
    {
      "epoch": 0.16452595268567133,
      "grad_norm": 3.203125,
      "learning_rate": 4.6831240212872305e-05,
      "loss": 0.8642,
      "step": 11602
    },
    {
      "epoch": 0.16455431433929754,
      "grad_norm": 3.25,
      "learning_rate": 4.6830151430900436e-05,
      "loss": 0.8493,
      "step": 11604
    },
    {
      "epoch": 0.16458267599292375,
      "grad_norm": 2.859375,
      "learning_rate": 4.68290624745697e-05,
      "loss": 0.8279,
      "step": 11606
    },
    {
      "epoch": 0.16461103764655,
      "grad_norm": 3.125,
      "learning_rate": 4.682797334388879e-05,
      "loss": 0.8293,
      "step": 11608
    },
    {
      "epoch": 0.1646393993001762,
      "grad_norm": 3.203125,
      "learning_rate": 4.682688403886639e-05,
      "loss": 0.8713,
      "step": 11610
    },
    {
      "epoch": 0.16466776095380242,
      "grad_norm": 2.953125,
      "learning_rate": 4.682579455951122e-05,
      "loss": 0.831,
      "step": 11612
    },
    {
      "epoch": 0.16469612260742864,
      "grad_norm": 3.140625,
      "learning_rate": 4.6824704905831965e-05,
      "loss": 0.847,
      "step": 11614
    },
    {
      "epoch": 0.16472448426105485,
      "grad_norm": 2.875,
      "learning_rate": 4.682361507783734e-05,
      "loss": 0.8625,
      "step": 11616
    },
    {
      "epoch": 0.16475284591468106,
      "grad_norm": 3.21875,
      "learning_rate": 4.682252507553604e-05,
      "loss": 0.8485,
      "step": 11618
    },
    {
      "epoch": 0.16478120756830728,
      "grad_norm": 2.78125,
      "learning_rate": 4.682143489893679e-05,
      "loss": 0.8866,
      "step": 11620
    },
    {
      "epoch": 0.1648095692219335,
      "grad_norm": 2.9375,
      "learning_rate": 4.682034454804827e-05,
      "loss": 0.8678,
      "step": 11622
    },
    {
      "epoch": 0.1648379308755597,
      "grad_norm": 2.6875,
      "learning_rate": 4.681925402287921e-05,
      "loss": 0.8255,
      "step": 11624
    },
    {
      "epoch": 0.16486629252918591,
      "grad_norm": 3.3125,
      "learning_rate": 4.6818163323438315e-05,
      "loss": 0.8098,
      "step": 11626
    },
    {
      "epoch": 0.16489465418281213,
      "grad_norm": 3.234375,
      "learning_rate": 4.681707244973429e-05,
      "loss": 0.8681,
      "step": 11628
    },
    {
      "epoch": 0.16492301583643834,
      "grad_norm": 3.265625,
      "learning_rate": 4.681598140177586e-05,
      "loss": 0.8719,
      "step": 11630
    },
    {
      "epoch": 0.16495137749006455,
      "grad_norm": 3.078125,
      "learning_rate": 4.6814890179571714e-05,
      "loss": 0.8267,
      "step": 11632
    },
    {
      "epoch": 0.16497973914369077,
      "grad_norm": 2.953125,
      "learning_rate": 4.68137987831306e-05,
      "loss": 0.8513,
      "step": 11634
    },
    {
      "epoch": 0.16500810079731698,
      "grad_norm": 2.890625,
      "learning_rate": 4.681270721246121e-05,
      "loss": 0.8709,
      "step": 11636
    },
    {
      "epoch": 0.1650364624509432,
      "grad_norm": 3.1875,
      "learning_rate": 4.681161546757228e-05,
      "loss": 0.8436,
      "step": 11638
    },
    {
      "epoch": 0.1650648241045694,
      "grad_norm": 3.109375,
      "learning_rate": 4.681052354847252e-05,
      "loss": 0.8277,
      "step": 11640
    },
    {
      "epoch": 0.16509318575819562,
      "grad_norm": 3.328125,
      "learning_rate": 4.680943145517066e-05,
      "loss": 0.8317,
      "step": 11642
    },
    {
      "epoch": 0.16512154741182183,
      "grad_norm": 3.015625,
      "learning_rate": 4.680833918767541e-05,
      "loss": 0.8115,
      "step": 11644
    },
    {
      "epoch": 0.16514990906544805,
      "grad_norm": 3.296875,
      "learning_rate": 4.6807246745995504e-05,
      "loss": 0.8505,
      "step": 11646
    },
    {
      "epoch": 0.1651782707190743,
      "grad_norm": 3.453125,
      "learning_rate": 4.680615413013966e-05,
      "loss": 0.8586,
      "step": 11648
    },
    {
      "epoch": 0.1652066323727005,
      "grad_norm": 2.734375,
      "learning_rate": 4.680506134011661e-05,
      "loss": 0.8146,
      "step": 11650
    },
    {
      "epoch": 0.16523499402632671,
      "grad_norm": 3.0625,
      "learning_rate": 4.6803968375935076e-05,
      "loss": 0.8066,
      "step": 11652
    },
    {
      "epoch": 0.16526335567995293,
      "grad_norm": 2.9375,
      "learning_rate": 4.680287523760379e-05,
      "loss": 0.8462,
      "step": 11654
    },
    {
      "epoch": 0.16529171733357914,
      "grad_norm": 3.65625,
      "learning_rate": 4.68017819251315e-05,
      "loss": 0.9049,
      "step": 11656
    },
    {
      "epoch": 0.16532007898720535,
      "grad_norm": 3.609375,
      "learning_rate": 4.680068843852691e-05,
      "loss": 0.8495,
      "step": 11658
    },
    {
      "epoch": 0.16534844064083157,
      "grad_norm": 3.359375,
      "learning_rate": 4.679959477779877e-05,
      "loss": 0.8622,
      "step": 11660
    },
    {
      "epoch": 0.16537680229445778,
      "grad_norm": 3.28125,
      "learning_rate": 4.679850094295581e-05,
      "loss": 0.8387,
      "step": 11662
    },
    {
      "epoch": 0.165405163948084,
      "grad_norm": 3.25,
      "learning_rate": 4.679740693400678e-05,
      "loss": 0.8058,
      "step": 11664
    },
    {
      "epoch": 0.1654335256017102,
      "grad_norm": 3.140625,
      "learning_rate": 4.67963127509604e-05,
      "loss": 0.8754,
      "step": 11666
    },
    {
      "epoch": 0.16546188725533642,
      "grad_norm": 2.984375,
      "learning_rate": 4.679521839382542e-05,
      "loss": 0.8166,
      "step": 11668
    },
    {
      "epoch": 0.16549024890896263,
      "grad_norm": 3.171875,
      "learning_rate": 4.679412386261057e-05,
      "loss": 0.8841,
      "step": 11670
    },
    {
      "epoch": 0.16551861056258885,
      "grad_norm": 3.171875,
      "learning_rate": 4.67930291573246e-05,
      "loss": 0.8433,
      "step": 11672
    },
    {
      "epoch": 0.16554697221621506,
      "grad_norm": 3.28125,
      "learning_rate": 4.679193427797626e-05,
      "loss": 0.8334,
      "step": 11674
    },
    {
      "epoch": 0.16557533386984127,
      "grad_norm": 3.265625,
      "learning_rate": 4.679083922457428e-05,
      "loss": 0.8449,
      "step": 11676
    },
    {
      "epoch": 0.1656036955234675,
      "grad_norm": 2.765625,
      "learning_rate": 4.678974399712742e-05,
      "loss": 0.8429,
      "step": 11678
    },
    {
      "epoch": 0.1656320571770937,
      "grad_norm": 2.921875,
      "learning_rate": 4.6788648595644417e-05,
      "loss": 0.7927,
      "step": 11680
    },
    {
      "epoch": 0.1656604188307199,
      "grad_norm": 2.984375,
      "learning_rate": 4.6787553020134025e-05,
      "loss": 0.8528,
      "step": 11682
    },
    {
      "epoch": 0.16568878048434613,
      "grad_norm": 3.015625,
      "learning_rate": 4.678645727060499e-05,
      "loss": 0.8949,
      "step": 11684
    },
    {
      "epoch": 0.16571714213797237,
      "grad_norm": 3.0,
      "learning_rate": 4.678536134706607e-05,
      "loss": 0.851,
      "step": 11686
    },
    {
      "epoch": 0.16574550379159858,
      "grad_norm": 3.359375,
      "learning_rate": 4.6784265249526016e-05,
      "loss": 0.8244,
      "step": 11688
    },
    {
      "epoch": 0.1657738654452248,
      "grad_norm": 3.234375,
      "learning_rate": 4.678316897799359e-05,
      "loss": 0.8746,
      "step": 11690
    },
    {
      "epoch": 0.165802227098851,
      "grad_norm": 3.328125,
      "learning_rate": 4.678207253247753e-05,
      "loss": 0.9013,
      "step": 11692
    },
    {
      "epoch": 0.16583058875247722,
      "grad_norm": 3.0625,
      "learning_rate": 4.678097591298661e-05,
      "loss": 0.8274,
      "step": 11694
    },
    {
      "epoch": 0.16585895040610343,
      "grad_norm": 3.171875,
      "learning_rate": 4.677987911952958e-05,
      "loss": 0.8901,
      "step": 11696
    },
    {
      "epoch": 0.16588731205972965,
      "grad_norm": 3.203125,
      "learning_rate": 4.6778782152115195e-05,
      "loss": 0.8614,
      "step": 11698
    },
    {
      "epoch": 0.16591567371335586,
      "grad_norm": 3.046875,
      "learning_rate": 4.677768501075224e-05,
      "loss": 0.9307,
      "step": 11700
    },
    {
      "epoch": 0.16594403536698207,
      "grad_norm": 3.421875,
      "learning_rate": 4.6776587695449455e-05,
      "loss": 0.8527,
      "step": 11702
    },
    {
      "epoch": 0.1659723970206083,
      "grad_norm": 2.921875,
      "learning_rate": 4.677549020621561e-05,
      "loss": 0.8012,
      "step": 11704
    },
    {
      "epoch": 0.1660007586742345,
      "grad_norm": 3.328125,
      "learning_rate": 4.6774392543059476e-05,
      "loss": 0.813,
      "step": 11706
    },
    {
      "epoch": 0.1660291203278607,
      "grad_norm": 2.9375,
      "learning_rate": 4.677329470598981e-05,
      "loss": 0.873,
      "step": 11708
    },
    {
      "epoch": 0.16605748198148693,
      "grad_norm": 2.828125,
      "learning_rate": 4.677219669501539e-05,
      "loss": 0.8658,
      "step": 11710
    },
    {
      "epoch": 0.16608584363511314,
      "grad_norm": 3.046875,
      "learning_rate": 4.6771098510144984e-05,
      "loss": 0.8266,
      "step": 11712
    },
    {
      "epoch": 0.16611420528873935,
      "grad_norm": 3.25,
      "learning_rate": 4.677000015138736e-05,
      "loss": 0.8378,
      "step": 11714
    },
    {
      "epoch": 0.16614256694236557,
      "grad_norm": 3.25,
      "learning_rate": 4.6768901618751295e-05,
      "loss": 0.8558,
      "step": 11716
    },
    {
      "epoch": 0.16617092859599178,
      "grad_norm": 3.140625,
      "learning_rate": 4.676780291224556e-05,
      "loss": 0.8853,
      "step": 11718
    },
    {
      "epoch": 0.166199290249618,
      "grad_norm": 3.203125,
      "learning_rate": 4.676670403187893e-05,
      "loss": 0.8853,
      "step": 11720
    },
    {
      "epoch": 0.1662276519032442,
      "grad_norm": 3.265625,
      "learning_rate": 4.676560497766019e-05,
      "loss": 0.8616,
      "step": 11722
    },
    {
      "epoch": 0.16625601355687045,
      "grad_norm": 3.046875,
      "learning_rate": 4.67645057495981e-05,
      "loss": 0.876,
      "step": 11724
    },
    {
      "epoch": 0.16628437521049666,
      "grad_norm": 3.046875,
      "learning_rate": 4.676340634770146e-05,
      "loss": 0.837,
      "step": 11726
    },
    {
      "epoch": 0.16631273686412287,
      "grad_norm": 3.328125,
      "learning_rate": 4.676230677197904e-05,
      "loss": 0.8328,
      "step": 11728
    },
    {
      "epoch": 0.1663410985177491,
      "grad_norm": 2.96875,
      "learning_rate": 4.6761207022439625e-05,
      "loss": 0.8458,
      "step": 11730
    },
    {
      "epoch": 0.1663694601713753,
      "grad_norm": 2.828125,
      "learning_rate": 4.6760107099091985e-05,
      "loss": 0.8434,
      "step": 11732
    },
    {
      "epoch": 0.1663978218250015,
      "grad_norm": 2.875,
      "learning_rate": 4.675900700194493e-05,
      "loss": 0.8469,
      "step": 11734
    },
    {
      "epoch": 0.16642618347862773,
      "grad_norm": 3.015625,
      "learning_rate": 4.675790673100724e-05,
      "loss": 0.7805,
      "step": 11736
    },
    {
      "epoch": 0.16645454513225394,
      "grad_norm": 3.078125,
      "learning_rate": 4.675680628628769e-05,
      "loss": 0.86,
      "step": 11738
    },
    {
      "epoch": 0.16648290678588015,
      "grad_norm": 2.703125,
      "learning_rate": 4.675570566779508e-05,
      "loss": 0.8214,
      "step": 11740
    },
    {
      "epoch": 0.16651126843950637,
      "grad_norm": 3.0,
      "learning_rate": 4.6754604875538187e-05,
      "loss": 0.7887,
      "step": 11742
    },
    {
      "epoch": 0.16653963009313258,
      "grad_norm": 3.046875,
      "learning_rate": 4.675350390952583e-05,
      "loss": 0.8604,
      "step": 11744
    },
    {
      "epoch": 0.1665679917467588,
      "grad_norm": 3.46875,
      "learning_rate": 4.675240276976678e-05,
      "loss": 0.8149,
      "step": 11746
    },
    {
      "epoch": 0.166596353400385,
      "grad_norm": 3.578125,
      "learning_rate": 4.675130145626984e-05,
      "loss": 0.9002,
      "step": 11748
    },
    {
      "epoch": 0.16662471505401122,
      "grad_norm": 3.296875,
      "learning_rate": 4.67501999690438e-05,
      "loss": 0.8644,
      "step": 11750
    },
    {
      "epoch": 0.16665307670763743,
      "grad_norm": 2.875,
      "learning_rate": 4.6749098308097464e-05,
      "loss": 0.8251,
      "step": 11752
    },
    {
      "epoch": 0.16668143836126365,
      "grad_norm": 3.203125,
      "learning_rate": 4.674799647343962e-05,
      "loss": 0.827,
      "step": 11754
    },
    {
      "epoch": 0.16670980001488986,
      "grad_norm": 3.015625,
      "learning_rate": 4.6746894465079094e-05,
      "loss": 0.8784,
      "step": 11756
    },
    {
      "epoch": 0.16673816166851607,
      "grad_norm": 2.765625,
      "learning_rate": 4.674579228302466e-05,
      "loss": 0.8642,
      "step": 11758
    },
    {
      "epoch": 0.16676652332214228,
      "grad_norm": 3.28125,
      "learning_rate": 4.6744689927285134e-05,
      "loss": 0.919,
      "step": 11760
    },
    {
      "epoch": 0.16679488497576853,
      "grad_norm": 3.203125,
      "learning_rate": 4.674358739786933e-05,
      "loss": 0.8714,
      "step": 11762
    },
    {
      "epoch": 0.16682324662939474,
      "grad_norm": 2.953125,
      "learning_rate": 4.6742484694786024e-05,
      "loss": 0.8562,
      "step": 11764
    },
    {
      "epoch": 0.16685160828302095,
      "grad_norm": 3.015625,
      "learning_rate": 4.674138181804406e-05,
      "loss": 0.8466,
      "step": 11766
    },
    {
      "epoch": 0.16687996993664717,
      "grad_norm": 3.09375,
      "learning_rate": 4.674027876765222e-05,
      "loss": 0.8277,
      "step": 11768
    },
    {
      "epoch": 0.16690833159027338,
      "grad_norm": 2.875,
      "learning_rate": 4.673917554361933e-05,
      "loss": 0.8278,
      "step": 11770
    },
    {
      "epoch": 0.1669366932438996,
      "grad_norm": 3.34375,
      "learning_rate": 4.673807214595419e-05,
      "loss": 0.854,
      "step": 11772
    },
    {
      "epoch": 0.1669650548975258,
      "grad_norm": 3.125,
      "learning_rate": 4.6736968574665616e-05,
      "loss": 0.8949,
      "step": 11774
    },
    {
      "epoch": 0.16699341655115202,
      "grad_norm": 3.234375,
      "learning_rate": 4.673586482976243e-05,
      "loss": 0.8712,
      "step": 11776
    },
    {
      "epoch": 0.16702177820477823,
      "grad_norm": 3.015625,
      "learning_rate": 4.6734760911253436e-05,
      "loss": 0.8431,
      "step": 11778
    },
    {
      "epoch": 0.16705013985840445,
      "grad_norm": 3.171875,
      "learning_rate": 4.673365681914746e-05,
      "loss": 0.8148,
      "step": 11780
    },
    {
      "epoch": 0.16707850151203066,
      "grad_norm": 3.0625,
      "learning_rate": 4.6732552553453316e-05,
      "loss": 0.8347,
      "step": 11782
    },
    {
      "epoch": 0.16710686316565687,
      "grad_norm": 3.140625,
      "learning_rate": 4.673144811417982e-05,
      "loss": 0.877,
      "step": 11784
    },
    {
      "epoch": 0.16713522481928308,
      "grad_norm": 2.96875,
      "learning_rate": 4.67303435013358e-05,
      "loss": 0.8252,
      "step": 11786
    },
    {
      "epoch": 0.1671635864729093,
      "grad_norm": 3.078125,
      "learning_rate": 4.672923871493009e-05,
      "loss": 0.8834,
      "step": 11788
    },
    {
      "epoch": 0.1671919481265355,
      "grad_norm": 3.484375,
      "learning_rate": 4.672813375497149e-05,
      "loss": 0.8938,
      "step": 11790
    },
    {
      "epoch": 0.16722030978016172,
      "grad_norm": 3.46875,
      "learning_rate": 4.672702862146884e-05,
      "loss": 0.8909,
      "step": 11792
    },
    {
      "epoch": 0.16724867143378794,
      "grad_norm": 3.484375,
      "learning_rate": 4.6725923314430965e-05,
      "loss": 0.8957,
      "step": 11794
    },
    {
      "epoch": 0.16727703308741415,
      "grad_norm": 3.390625,
      "learning_rate": 4.6724817833866684e-05,
      "loss": 0.872,
      "step": 11796
    },
    {
      "epoch": 0.16730539474104036,
      "grad_norm": 2.921875,
      "learning_rate": 4.672371217978484e-05,
      "loss": 0.8108,
      "step": 11798
    },
    {
      "epoch": 0.16733375639466658,
      "grad_norm": 2.6875,
      "learning_rate": 4.6722606352194255e-05,
      "loss": 0.771,
      "step": 11800
    },
    {
      "epoch": 0.16736211804829282,
      "grad_norm": 2.84375,
      "learning_rate": 4.6721500351103766e-05,
      "loss": 0.8661,
      "step": 11802
    },
    {
      "epoch": 0.16739047970191903,
      "grad_norm": 3.453125,
      "learning_rate": 4.6720394176522206e-05,
      "loss": 0.8279,
      "step": 11804
    },
    {
      "epoch": 0.16741884135554524,
      "grad_norm": 2.96875,
      "learning_rate": 4.6719287828458405e-05,
      "loss": 0.8817,
      "step": 11806
    },
    {
      "epoch": 0.16744720300917146,
      "grad_norm": 3.09375,
      "learning_rate": 4.671818130692121e-05,
      "loss": 0.8429,
      "step": 11808
    },
    {
      "epoch": 0.16747556466279767,
      "grad_norm": 3.3125,
      "learning_rate": 4.671707461191944e-05,
      "loss": 0.8783,
      "step": 11810
    },
    {
      "epoch": 0.16750392631642388,
      "grad_norm": 3.296875,
      "learning_rate": 4.671596774346196e-05,
      "loss": 0.8274,
      "step": 11812
    },
    {
      "epoch": 0.1675322879700501,
      "grad_norm": 4.53125,
      "learning_rate": 4.6714860701557586e-05,
      "loss": 0.8945,
      "step": 11814
    },
    {
      "epoch": 0.1675606496236763,
      "grad_norm": 3.0625,
      "learning_rate": 4.6713753486215185e-05,
      "loss": 0.8565,
      "step": 11816
    },
    {
      "epoch": 0.16758901127730252,
      "grad_norm": 3.359375,
      "learning_rate": 4.6712646097443575e-05,
      "loss": 0.849,
      "step": 11818
    },
    {
      "epoch": 0.16761737293092874,
      "grad_norm": 3.109375,
      "learning_rate": 4.671153853525162e-05,
      "loss": 0.8912,
      "step": 11820
    },
    {
      "epoch": 0.16764573458455495,
      "grad_norm": 2.796875,
      "learning_rate": 4.671043079964815e-05,
      "loss": 0.8166,
      "step": 11822
    },
    {
      "epoch": 0.16767409623818116,
      "grad_norm": 2.78125,
      "learning_rate": 4.670932289064203e-05,
      "loss": 0.8272,
      "step": 11824
    },
    {
      "epoch": 0.16770245789180738,
      "grad_norm": 2.84375,
      "learning_rate": 4.670821480824209e-05,
      "loss": 0.8401,
      "step": 11826
    },
    {
      "epoch": 0.1677308195454336,
      "grad_norm": 3.0625,
      "learning_rate": 4.670710655245719e-05,
      "loss": 0.8439,
      "step": 11828
    },
    {
      "epoch": 0.1677591811990598,
      "grad_norm": 3.25,
      "learning_rate": 4.6705998123296194e-05,
      "loss": 0.851,
      "step": 11830
    },
    {
      "epoch": 0.16778754285268602,
      "grad_norm": 3.21875,
      "learning_rate": 4.6704889520767935e-05,
      "loss": 0.826,
      "step": 11832
    },
    {
      "epoch": 0.16781590450631223,
      "grad_norm": 2.984375,
      "learning_rate": 4.670378074488128e-05,
      "loss": 0.8485,
      "step": 11834
    },
    {
      "epoch": 0.16784426615993844,
      "grad_norm": 3.09375,
      "learning_rate": 4.6702671795645074e-05,
      "loss": 0.8362,
      "step": 11836
    },
    {
      "epoch": 0.16787262781356466,
      "grad_norm": 3.25,
      "learning_rate": 4.670156267306819e-05,
      "loss": 0.8829,
      "step": 11838
    },
    {
      "epoch": 0.1679009894671909,
      "grad_norm": 3.0625,
      "learning_rate": 4.670045337715947e-05,
      "loss": 0.8554,
      "step": 11840
    },
    {
      "epoch": 0.1679293511208171,
      "grad_norm": 3.375,
      "learning_rate": 4.6699343907927785e-05,
      "loss": 0.9396,
      "step": 11842
    },
    {
      "epoch": 0.16795771277444332,
      "grad_norm": 3.03125,
      "learning_rate": 4.669823426538199e-05,
      "loss": 0.8417,
      "step": 11844
    },
    {
      "epoch": 0.16798607442806954,
      "grad_norm": 2.796875,
      "learning_rate": 4.669712444953095e-05,
      "loss": 0.8867,
      "step": 11846
    },
    {
      "epoch": 0.16801443608169575,
      "grad_norm": 3.484375,
      "learning_rate": 4.669601446038353e-05,
      "loss": 0.8677,
      "step": 11848
    },
    {
      "epoch": 0.16804279773532196,
      "grad_norm": 3.25,
      "learning_rate": 4.6694904297948606e-05,
      "loss": 0.8429,
      "step": 11850
    },
    {
      "epoch": 0.16807115938894818,
      "grad_norm": 3.28125,
      "learning_rate": 4.669379396223502e-05,
      "loss": 0.8423,
      "step": 11852
    },
    {
      "epoch": 0.1680995210425744,
      "grad_norm": 2.96875,
      "learning_rate": 4.669268345325166e-05,
      "loss": 0.8341,
      "step": 11854
    },
    {
      "epoch": 0.1681278826962006,
      "grad_norm": 3.078125,
      "learning_rate": 4.669157277100739e-05,
      "loss": 0.8422,
      "step": 11856
    },
    {
      "epoch": 0.16815624434982682,
      "grad_norm": 2.9375,
      "learning_rate": 4.6690461915511085e-05,
      "loss": 0.8446,
      "step": 11858
    },
    {
      "epoch": 0.16818460600345303,
      "grad_norm": 2.625,
      "learning_rate": 4.668935088677161e-05,
      "loss": 0.8158,
      "step": 11860
    },
    {
      "epoch": 0.16821296765707924,
      "grad_norm": 3.03125,
      "learning_rate": 4.668823968479784e-05,
      "loss": 0.8574,
      "step": 11862
    },
    {
      "epoch": 0.16824132931070546,
      "grad_norm": 3.03125,
      "learning_rate": 4.668712830959866e-05,
      "loss": 0.8233,
      "step": 11864
    },
    {
      "epoch": 0.16826969096433167,
      "grad_norm": 3.09375,
      "learning_rate": 4.6686016761182936e-05,
      "loss": 0.8841,
      "step": 11866
    },
    {
      "epoch": 0.16829805261795788,
      "grad_norm": 3.09375,
      "learning_rate": 4.6684905039559554e-05,
      "loss": 0.8682,
      "step": 11868
    },
    {
      "epoch": 0.1683264142715841,
      "grad_norm": 3.078125,
      "learning_rate": 4.668379314473739e-05,
      "loss": 0.819,
      "step": 11870
    },
    {
      "epoch": 0.1683547759252103,
      "grad_norm": 3.015625,
      "learning_rate": 4.668268107672531e-05,
      "loss": 0.8432,
      "step": 11872
    },
    {
      "epoch": 0.16838313757883652,
      "grad_norm": 3.078125,
      "learning_rate": 4.668156883553222e-05,
      "loss": 0.835,
      "step": 11874
    },
    {
      "epoch": 0.16841149923246274,
      "grad_norm": 3.015625,
      "learning_rate": 4.6680456421166995e-05,
      "loss": 0.8565,
      "step": 11876
    },
    {
      "epoch": 0.16843986088608898,
      "grad_norm": 3.3125,
      "learning_rate": 4.6679343833638513e-05,
      "loss": 0.8787,
      "step": 11878
    },
    {
      "epoch": 0.1684682225397152,
      "grad_norm": 2.8125,
      "learning_rate": 4.667823107295567e-05,
      "loss": 0.8169,
      "step": 11880
    },
    {
      "epoch": 0.1684965841933414,
      "grad_norm": 3.140625,
      "learning_rate": 4.6677118139127354e-05,
      "loss": 0.8376,
      "step": 11882
    },
    {
      "epoch": 0.16852494584696762,
      "grad_norm": 3.359375,
      "learning_rate": 4.667600503216244e-05,
      "loss": 0.8218,
      "step": 11884
    },
    {
      "epoch": 0.16855330750059383,
      "grad_norm": 2.921875,
      "learning_rate": 4.6674891752069825e-05,
      "loss": 0.8295,
      "step": 11886
    },
    {
      "epoch": 0.16858166915422004,
      "grad_norm": 3.546875,
      "learning_rate": 4.667377829885842e-05,
      "loss": 0.8859,
      "step": 11888
    },
    {
      "epoch": 0.16861003080784626,
      "grad_norm": 3.265625,
      "learning_rate": 4.6672664672537085e-05,
      "loss": 0.8312,
      "step": 11890
    },
    {
      "epoch": 0.16863839246147247,
      "grad_norm": 2.875,
      "learning_rate": 4.6671550873114744e-05,
      "loss": 0.7894,
      "step": 11892
    },
    {
      "epoch": 0.16866675411509868,
      "grad_norm": 3.3125,
      "learning_rate": 4.667043690060027e-05,
      "loss": 0.8461,
      "step": 11894
    },
    {
      "epoch": 0.1686951157687249,
      "grad_norm": 3.53125,
      "learning_rate": 4.6669322755002584e-05,
      "loss": 0.8659,
      "step": 11896
    },
    {
      "epoch": 0.1687234774223511,
      "grad_norm": 2.671875,
      "learning_rate": 4.666820843633056e-05,
      "loss": 0.804,
      "step": 11898
    },
    {
      "epoch": 0.16875183907597732,
      "grad_norm": 3.390625,
      "learning_rate": 4.666709394459312e-05,
      "loss": 0.8645,
      "step": 11900
    },
    {
      "epoch": 0.16878020072960354,
      "grad_norm": 3.21875,
      "learning_rate": 4.6665979279799146e-05,
      "loss": 0.8576,
      "step": 11902
    },
    {
      "epoch": 0.16880856238322975,
      "grad_norm": 2.921875,
      "learning_rate": 4.6664864441957556e-05,
      "loss": 0.786,
      "step": 11904
    },
    {
      "epoch": 0.16883692403685596,
      "grad_norm": 2.765625,
      "learning_rate": 4.6663749431077245e-05,
      "loss": 0.8135,
      "step": 11906
    },
    {
      "epoch": 0.16886528569048218,
      "grad_norm": 3.59375,
      "learning_rate": 4.6662634247167123e-05,
      "loss": 0.8875,
      "step": 11908
    },
    {
      "epoch": 0.1688936473441084,
      "grad_norm": 3.296875,
      "learning_rate": 4.66615188902361e-05,
      "loss": 0.9018,
      "step": 11910
    },
    {
      "epoch": 0.1689220089977346,
      "grad_norm": 3.3125,
      "learning_rate": 4.666040336029308e-05,
      "loss": 0.8483,
      "step": 11912
    },
    {
      "epoch": 0.16895037065136081,
      "grad_norm": 3.0,
      "learning_rate": 4.665928765734697e-05,
      "loss": 0.813,
      "step": 11914
    },
    {
      "epoch": 0.16897873230498706,
      "grad_norm": 3.0625,
      "learning_rate": 4.665817178140669e-05,
      "loss": 0.8794,
      "step": 11916
    },
    {
      "epoch": 0.16900709395861327,
      "grad_norm": 2.90625,
      "learning_rate": 4.665705573248115e-05,
      "loss": 0.8425,
      "step": 11918
    },
    {
      "epoch": 0.16903545561223948,
      "grad_norm": 3.09375,
      "learning_rate": 4.665593951057925e-05,
      "loss": 0.8495,
      "step": 11920
    },
    {
      "epoch": 0.1690638172658657,
      "grad_norm": 3.09375,
      "learning_rate": 4.665482311570992e-05,
      "loss": 0.8202,
      "step": 11922
    },
    {
      "epoch": 0.1690921789194919,
      "grad_norm": 2.890625,
      "learning_rate": 4.665370654788208e-05,
      "loss": 0.8607,
      "step": 11924
    },
    {
      "epoch": 0.16912054057311812,
      "grad_norm": 3.3125,
      "learning_rate": 4.665258980710464e-05,
      "loss": 0.8583,
      "step": 11926
    },
    {
      "epoch": 0.16914890222674434,
      "grad_norm": 2.828125,
      "learning_rate": 4.6651472893386516e-05,
      "loss": 0.8575,
      "step": 11928
    },
    {
      "epoch": 0.16917726388037055,
      "grad_norm": 3.015625,
      "learning_rate": 4.6650355806736636e-05,
      "loss": 0.8092,
      "step": 11930
    },
    {
      "epoch": 0.16920562553399676,
      "grad_norm": 2.84375,
      "learning_rate": 4.664923854716392e-05,
      "loss": 0.8544,
      "step": 11932
    },
    {
      "epoch": 0.16923398718762298,
      "grad_norm": 3.03125,
      "learning_rate": 4.664812111467729e-05,
      "loss": 0.8056,
      "step": 11934
    },
    {
      "epoch": 0.1692623488412492,
      "grad_norm": 3.078125,
      "learning_rate": 4.664700350928567e-05,
      "loss": 0.8264,
      "step": 11936
    },
    {
      "epoch": 0.1692907104948754,
      "grad_norm": 3.296875,
      "learning_rate": 4.6645885730998e-05,
      "loss": 0.8973,
      "step": 11938
    },
    {
      "epoch": 0.16931907214850161,
      "grad_norm": 3.046875,
      "learning_rate": 4.664476777982319e-05,
      "loss": 0.8087,
      "step": 11940
    },
    {
      "epoch": 0.16934743380212783,
      "grad_norm": 2.640625,
      "learning_rate": 4.664364965577018e-05,
      "loss": 0.8382,
      "step": 11942
    },
    {
      "epoch": 0.16937579545575404,
      "grad_norm": 3.21875,
      "learning_rate": 4.664253135884789e-05,
      "loss": 0.8228,
      "step": 11944
    },
    {
      "epoch": 0.16940415710938025,
      "grad_norm": 3.09375,
      "learning_rate": 4.664141288906526e-05,
      "loss": 0.8257,
      "step": 11946
    },
    {
      "epoch": 0.16943251876300647,
      "grad_norm": 2.953125,
      "learning_rate": 4.6640294246431224e-05,
      "loss": 0.8633,
      "step": 11948
    },
    {
      "epoch": 0.16946088041663268,
      "grad_norm": 2.703125,
      "learning_rate": 4.663917543095472e-05,
      "loss": 0.7782,
      "step": 11950
    },
    {
      "epoch": 0.1694892420702589,
      "grad_norm": 3.109375,
      "learning_rate": 4.663805644264467e-05,
      "loss": 0.8466,
      "step": 11952
    },
    {
      "epoch": 0.1695176037238851,
      "grad_norm": 3.03125,
      "learning_rate": 4.663693728151002e-05,
      "loss": 0.8613,
      "step": 11954
    },
    {
      "epoch": 0.16954596537751135,
      "grad_norm": 2.984375,
      "learning_rate": 4.663581794755971e-05,
      "loss": 0.8326,
      "step": 11956
    },
    {
      "epoch": 0.16957432703113756,
      "grad_norm": 2.84375,
      "learning_rate": 4.6634698440802685e-05,
      "loss": 0.846,
      "step": 11958
    },
    {
      "epoch": 0.16960268868476377,
      "grad_norm": 2.96875,
      "learning_rate": 4.663357876124787e-05,
      "loss": 0.882,
      "step": 11960
    },
    {
      "epoch": 0.16963105033839,
      "grad_norm": 3.109375,
      "learning_rate": 4.663245890890423e-05,
      "loss": 0.8149,
      "step": 11962
    },
    {
      "epoch": 0.1696594119920162,
      "grad_norm": 2.84375,
      "learning_rate": 4.6631338883780685e-05,
      "loss": 0.8539,
      "step": 11964
    },
    {
      "epoch": 0.16968777364564241,
      "grad_norm": 3.390625,
      "learning_rate": 4.66302186858862e-05,
      "loss": 0.8202,
      "step": 11966
    },
    {
      "epoch": 0.16971613529926863,
      "grad_norm": 2.75,
      "learning_rate": 4.662909831522972e-05,
      "loss": 0.8443,
      "step": 11968
    },
    {
      "epoch": 0.16974449695289484,
      "grad_norm": 3.25,
      "learning_rate": 4.662797777182018e-05,
      "loss": 0.8539,
      "step": 11970
    },
    {
      "epoch": 0.16977285860652105,
      "grad_norm": 2.875,
      "learning_rate": 4.6626857055666546e-05,
      "loss": 0.827,
      "step": 11972
    },
    {
      "epoch": 0.16980122026014727,
      "grad_norm": 3.125,
      "learning_rate": 4.662573616677775e-05,
      "loss": 0.7641,
      "step": 11974
    },
    {
      "epoch": 0.16982958191377348,
      "grad_norm": 3.171875,
      "learning_rate": 4.662461510516278e-05,
      "loss": 0.8039,
      "step": 11976
    },
    {
      "epoch": 0.1698579435673997,
      "grad_norm": 3.234375,
      "learning_rate": 4.662349387083055e-05,
      "loss": 0.8898,
      "step": 11978
    },
    {
      "epoch": 0.1698863052210259,
      "grad_norm": 3.5625,
      "learning_rate": 4.662237246379004e-05,
      "loss": 0.8806,
      "step": 11980
    },
    {
      "epoch": 0.16991466687465212,
      "grad_norm": 2.953125,
      "learning_rate": 4.6621250884050195e-05,
      "loss": 0.7659,
      "step": 11982
    },
    {
      "epoch": 0.16994302852827833,
      "grad_norm": 3.234375,
      "learning_rate": 4.662012913161997e-05,
      "loss": 0.8905,
      "step": 11984
    },
    {
      "epoch": 0.16997139018190455,
      "grad_norm": 2.90625,
      "learning_rate": 4.6619007206508335e-05,
      "loss": 0.8382,
      "step": 11986
    },
    {
      "epoch": 0.16999975183553076,
      "grad_norm": 2.984375,
      "learning_rate": 4.661788510872426e-05,
      "loss": 0.8666,
      "step": 11988
    },
    {
      "epoch": 0.17002811348915697,
      "grad_norm": 3.125,
      "learning_rate": 4.6616762838276687e-05,
      "loss": 0.816,
      "step": 11990
    },
    {
      "epoch": 0.1700564751427832,
      "grad_norm": 3.140625,
      "learning_rate": 4.661564039517458e-05,
      "loss": 0.801,
      "step": 11992
    },
    {
      "epoch": 0.17008483679640943,
      "grad_norm": 3.0,
      "learning_rate": 4.6614517779426917e-05,
      "loss": 0.8468,
      "step": 11994
    },
    {
      "epoch": 0.17011319845003564,
      "grad_norm": 2.890625,
      "learning_rate": 4.661339499104266e-05,
      "loss": 0.91,
      "step": 11996
    },
    {
      "epoch": 0.17014156010366185,
      "grad_norm": 3.265625,
      "learning_rate": 4.661227203003077e-05,
      "loss": 0.8211,
      "step": 11998
    },
    {
      "epoch": 0.17016992175728807,
      "grad_norm": 2.84375,
      "learning_rate": 4.661114889640023e-05,
      "loss": 0.8023,
      "step": 12000
    },
    {
      "epoch": 0.17019828341091428,
      "grad_norm": 2.765625,
      "learning_rate": 4.661002559016e-05,
      "loss": 0.8685,
      "step": 12002
    },
    {
      "epoch": 0.1702266450645405,
      "grad_norm": 3.109375,
      "learning_rate": 4.6608902111319054e-05,
      "loss": 0.861,
      "step": 12004
    },
    {
      "epoch": 0.1702550067181667,
      "grad_norm": 3.21875,
      "learning_rate": 4.6607778459886365e-05,
      "loss": 0.8641,
      "step": 12006
    },
    {
      "epoch": 0.17028336837179292,
      "grad_norm": 2.8125,
      "learning_rate": 4.66066546358709e-05,
      "loss": 0.8237,
      "step": 12008
    },
    {
      "epoch": 0.17031173002541913,
      "grad_norm": 2.609375,
      "learning_rate": 4.660553063928166e-05,
      "loss": 0.8554,
      "step": 12010
    },
    {
      "epoch": 0.17034009167904535,
      "grad_norm": 2.96875,
      "learning_rate": 4.660440647012759e-05,
      "loss": 0.8075,
      "step": 12012
    },
    {
      "epoch": 0.17036845333267156,
      "grad_norm": 3.078125,
      "learning_rate": 4.660328212841769e-05,
      "loss": 0.8381,
      "step": 12014
    },
    {
      "epoch": 0.17039681498629777,
      "grad_norm": 3.453125,
      "learning_rate": 4.660215761416093e-05,
      "loss": 0.8715,
      "step": 12016
    },
    {
      "epoch": 0.170425176639924,
      "grad_norm": 2.984375,
      "learning_rate": 4.660103292736631e-05,
      "loss": 0.8527,
      "step": 12018
    },
    {
      "epoch": 0.1704535382935502,
      "grad_norm": 3.46875,
      "learning_rate": 4.659990806804279e-05,
      "loss": 0.8506,
      "step": 12020
    },
    {
      "epoch": 0.1704818999471764,
      "grad_norm": 3.125,
      "learning_rate": 4.659878303619937e-05,
      "loss": 0.8983,
      "step": 12022
    },
    {
      "epoch": 0.17051026160080263,
      "grad_norm": 3.46875,
      "learning_rate": 4.659765783184502e-05,
      "loss": 0.8089,
      "step": 12024
    },
    {
      "epoch": 0.17053862325442884,
      "grad_norm": 3.046875,
      "learning_rate": 4.659653245498874e-05,
      "loss": 0.8553,
      "step": 12026
    },
    {
      "epoch": 0.17056698490805505,
      "grad_norm": 3.0625,
      "learning_rate": 4.659540690563952e-05,
      "loss": 0.8351,
      "step": 12028
    },
    {
      "epoch": 0.17059534656168127,
      "grad_norm": 3.015625,
      "learning_rate": 4.659428118380633e-05,
      "loss": 0.8277,
      "step": 12030
    },
    {
      "epoch": 0.1706237082153075,
      "grad_norm": 2.890625,
      "learning_rate": 4.659315528949819e-05,
      "loss": 0.837,
      "step": 12032
    },
    {
      "epoch": 0.17065206986893372,
      "grad_norm": 3.09375,
      "learning_rate": 4.659202922272408e-05,
      "loss": 0.8581,
      "step": 12034
    },
    {
      "epoch": 0.17068043152255993,
      "grad_norm": 3.09375,
      "learning_rate": 4.659090298349298e-05,
      "loss": 0.834,
      "step": 12036
    },
    {
      "epoch": 0.17070879317618615,
      "grad_norm": 2.90625,
      "learning_rate": 4.6589776571813904e-05,
      "loss": 0.8573,
      "step": 12038
    },
    {
      "epoch": 0.17073715482981236,
      "grad_norm": 3.34375,
      "learning_rate": 4.6588649987695843e-05,
      "loss": 0.9004,
      "step": 12040
    },
    {
      "epoch": 0.17076551648343857,
      "grad_norm": 2.84375,
      "learning_rate": 4.658752323114779e-05,
      "loss": 0.8496,
      "step": 12042
    },
    {
      "epoch": 0.1707938781370648,
      "grad_norm": 3.0625,
      "learning_rate": 4.658639630217876e-05,
      "loss": 0.8618,
      "step": 12044
    },
    {
      "epoch": 0.170822239790691,
      "grad_norm": 3.03125,
      "learning_rate": 4.658526920079773e-05,
      "loss": 0.8557,
      "step": 12046
    },
    {
      "epoch": 0.1708506014443172,
      "grad_norm": 3.0625,
      "learning_rate": 4.658414192701373e-05,
      "loss": 0.8539,
      "step": 12048
    },
    {
      "epoch": 0.17087896309794343,
      "grad_norm": 3.4375,
      "learning_rate": 4.658301448083573e-05,
      "loss": 0.8704,
      "step": 12050
    },
    {
      "epoch": 0.17090732475156964,
      "grad_norm": 3.375,
      "learning_rate": 4.658188686227277e-05,
      "loss": 0.8164,
      "step": 12052
    },
    {
      "epoch": 0.17093568640519585,
      "grad_norm": 3.265625,
      "learning_rate": 4.658075907133383e-05,
      "loss": 0.8432,
      "step": 12054
    },
    {
      "epoch": 0.17096404805882207,
      "grad_norm": 3.34375,
      "learning_rate": 4.6579631108027934e-05,
      "loss": 0.8348,
      "step": 12056
    },
    {
      "epoch": 0.17099240971244828,
      "grad_norm": 3.046875,
      "learning_rate": 4.657850297236408e-05,
      "loss": 0.8454,
      "step": 12058
    },
    {
      "epoch": 0.1710207713660745,
      "grad_norm": 3.203125,
      "learning_rate": 4.6577374664351286e-05,
      "loss": 0.8803,
      "step": 12060
    },
    {
      "epoch": 0.1710491330197007,
      "grad_norm": 2.796875,
      "learning_rate": 4.6576246183998564e-05,
      "loss": 0.8578,
      "step": 12062
    },
    {
      "epoch": 0.17107749467332692,
      "grad_norm": 3.0625,
      "learning_rate": 4.6575117531314915e-05,
      "loss": 0.7894,
      "step": 12064
    },
    {
      "epoch": 0.17110585632695313,
      "grad_norm": 2.890625,
      "learning_rate": 4.657398870630937e-05,
      "loss": 0.8139,
      "step": 12066
    },
    {
      "epoch": 0.17113421798057935,
      "grad_norm": 3.015625,
      "learning_rate": 4.657285970899094e-05,
      "loss": 0.8054,
      "step": 12068
    },
    {
      "epoch": 0.17116257963420559,
      "grad_norm": 2.984375,
      "learning_rate": 4.657173053936864e-05,
      "loss": 0.8545,
      "step": 12070
    },
    {
      "epoch": 0.1711909412878318,
      "grad_norm": 2.8125,
      "learning_rate": 4.657060119745149e-05,
      "loss": 0.7978,
      "step": 12072
    },
    {
      "epoch": 0.171219302941458,
      "grad_norm": 2.8125,
      "learning_rate": 4.6569471683248504e-05,
      "loss": 0.8334,
      "step": 12074
    },
    {
      "epoch": 0.17124766459508423,
      "grad_norm": 2.921875,
      "learning_rate": 4.656834199676872e-05,
      "loss": 0.8115,
      "step": 12076
    },
    {
      "epoch": 0.17127602624871044,
      "grad_norm": 3.296875,
      "learning_rate": 4.6567212138021135e-05,
      "loss": 0.854,
      "step": 12078
    },
    {
      "epoch": 0.17130438790233665,
      "grad_norm": 3.390625,
      "learning_rate": 4.65660821070148e-05,
      "loss": 0.8539,
      "step": 12080
    },
    {
      "epoch": 0.17133274955596287,
      "grad_norm": 2.828125,
      "learning_rate": 4.656495190375872e-05,
      "loss": 0.8037,
      "step": 12082
    },
    {
      "epoch": 0.17136111120958908,
      "grad_norm": 3.328125,
      "learning_rate": 4.656382152826194e-05,
      "loss": 0.8373,
      "step": 12084
    },
    {
      "epoch": 0.1713894728632153,
      "grad_norm": 2.875,
      "learning_rate": 4.656269098053347e-05,
      "loss": 0.8812,
      "step": 12086
    },
    {
      "epoch": 0.1714178345168415,
      "grad_norm": 3.203125,
      "learning_rate": 4.6561560260582356e-05,
      "loss": 0.8568,
      "step": 12088
    },
    {
      "epoch": 0.17144619617046772,
      "grad_norm": 2.90625,
      "learning_rate": 4.656042936841761e-05,
      "loss": 0.8384,
      "step": 12090
    },
    {
      "epoch": 0.17147455782409393,
      "grad_norm": 3.0625,
      "learning_rate": 4.655929830404829e-05,
      "loss": 0.7781,
      "step": 12092
    },
    {
      "epoch": 0.17150291947772014,
      "grad_norm": 3.078125,
      "learning_rate": 4.6558167067483414e-05,
      "loss": 0.8153,
      "step": 12094
    },
    {
      "epoch": 0.17153128113134636,
      "grad_norm": 3.125,
      "learning_rate": 4.655703565873202e-05,
      "loss": 0.8655,
      "step": 12096
    },
    {
      "epoch": 0.17155964278497257,
      "grad_norm": 3.296875,
      "learning_rate": 4.655590407780314e-05,
      "loss": 0.91,
      "step": 12098
    },
    {
      "epoch": 0.17158800443859878,
      "grad_norm": 3.15625,
      "learning_rate": 4.655477232470582e-05,
      "loss": 0.8305,
      "step": 12100
    },
    {
      "epoch": 0.171616366092225,
      "grad_norm": 3.40625,
      "learning_rate": 4.655364039944909e-05,
      "loss": 0.8632,
      "step": 12102
    },
    {
      "epoch": 0.1716447277458512,
      "grad_norm": 2.9375,
      "learning_rate": 4.6552508302042e-05,
      "loss": 0.833,
      "step": 12104
    },
    {
      "epoch": 0.17167308939947742,
      "grad_norm": 3.046875,
      "learning_rate": 4.655137603249359e-05,
      "loss": 0.8381,
      "step": 12106
    },
    {
      "epoch": 0.17170145105310364,
      "grad_norm": 3.09375,
      "learning_rate": 4.6550243590812905e-05,
      "loss": 0.9034,
      "step": 12108
    },
    {
      "epoch": 0.17172981270672988,
      "grad_norm": 3.0,
      "learning_rate": 4.654911097700898e-05,
      "loss": 0.8615,
      "step": 12110
    },
    {
      "epoch": 0.1717581743603561,
      "grad_norm": 3.3125,
      "learning_rate": 4.654797819109087e-05,
      "loss": 0.8611,
      "step": 12112
    },
    {
      "epoch": 0.1717865360139823,
      "grad_norm": 3.0625,
      "learning_rate": 4.6546845233067623e-05,
      "loss": 0.8215,
      "step": 12114
    },
    {
      "epoch": 0.17181489766760852,
      "grad_norm": 3.234375,
      "learning_rate": 4.654571210294829e-05,
      "loss": 0.8661,
      "step": 12116
    },
    {
      "epoch": 0.17184325932123473,
      "grad_norm": 2.921875,
      "learning_rate": 4.6544578800741914e-05,
      "loss": 0.8596,
      "step": 12118
    },
    {
      "epoch": 0.17187162097486094,
      "grad_norm": 2.921875,
      "learning_rate": 4.6543445326457545e-05,
      "loss": 0.8393,
      "step": 12120
    },
    {
      "epoch": 0.17189998262848716,
      "grad_norm": 3.234375,
      "learning_rate": 4.654231168010425e-05,
      "loss": 0.8367,
      "step": 12122
    },
    {
      "epoch": 0.17192834428211337,
      "grad_norm": 3.0625,
      "learning_rate": 4.654117786169107e-05,
      "loss": 0.8503,
      "step": 12124
    },
    {
      "epoch": 0.17195670593573958,
      "grad_norm": 2.84375,
      "learning_rate": 4.6540043871227064e-05,
      "loss": 0.8454,
      "step": 12126
    },
    {
      "epoch": 0.1719850675893658,
      "grad_norm": 3.1875,
      "learning_rate": 4.653890970872129e-05,
      "loss": 0.8781,
      "step": 12128
    },
    {
      "epoch": 0.172013429242992,
      "grad_norm": 2.9375,
      "learning_rate": 4.6537775374182813e-05,
      "loss": 0.8093,
      "step": 12130
    },
    {
      "epoch": 0.17204179089661822,
      "grad_norm": 3.15625,
      "learning_rate": 4.6536640867620686e-05,
      "loss": 0.8489,
      "step": 12132
    },
    {
      "epoch": 0.17207015255024444,
      "grad_norm": 2.953125,
      "learning_rate": 4.653550618904396e-05,
      "loss": 0.8283,
      "step": 12134
    },
    {
      "epoch": 0.17209851420387065,
      "grad_norm": 3.28125,
      "learning_rate": 4.653437133846172e-05,
      "loss": 0.9028,
      "step": 12136
    },
    {
      "epoch": 0.17212687585749686,
      "grad_norm": 3.1875,
      "learning_rate": 4.653323631588302e-05,
      "loss": 0.8341,
      "step": 12138
    },
    {
      "epoch": 0.17215523751112308,
      "grad_norm": 2.8125,
      "learning_rate": 4.653210112131693e-05,
      "loss": 0.8365,
      "step": 12140
    },
    {
      "epoch": 0.1721835991647493,
      "grad_norm": 3.34375,
      "learning_rate": 4.65309657547725e-05,
      "loss": 0.8941,
      "step": 12142
    },
    {
      "epoch": 0.1722119608183755,
      "grad_norm": 3.734375,
      "learning_rate": 4.652983021625883e-05,
      "loss": 0.8741,
      "step": 12144
    },
    {
      "epoch": 0.17224032247200172,
      "grad_norm": 2.90625,
      "learning_rate": 4.652869450578495e-05,
      "loss": 0.821,
      "step": 12146
    },
    {
      "epoch": 0.17226868412562796,
      "grad_norm": 3.1875,
      "learning_rate": 4.652755862335996e-05,
      "loss": 0.9094,
      "step": 12148
    },
    {
      "epoch": 0.17229704577925417,
      "grad_norm": 3.09375,
      "learning_rate": 4.652642256899292e-05,
      "loss": 0.8,
      "step": 12150
    },
    {
      "epoch": 0.17232540743288038,
      "grad_norm": 3.34375,
      "learning_rate": 4.652528634269291e-05,
      "loss": 0.8423,
      "step": 12152
    },
    {
      "epoch": 0.1723537690865066,
      "grad_norm": 2.796875,
      "learning_rate": 4.6524149944469e-05,
      "loss": 0.849,
      "step": 12154
    },
    {
      "epoch": 0.1723821307401328,
      "grad_norm": 3.28125,
      "learning_rate": 4.652301337433027e-05,
      "loss": 0.9289,
      "step": 12156
    },
    {
      "epoch": 0.17241049239375902,
      "grad_norm": 3.1875,
      "learning_rate": 4.6521876632285796e-05,
      "loss": 0.8313,
      "step": 12158
    },
    {
      "epoch": 0.17243885404738524,
      "grad_norm": 3.21875,
      "learning_rate": 4.6520739718344665e-05,
      "loss": 0.8727,
      "step": 12160
    },
    {
      "epoch": 0.17246721570101145,
      "grad_norm": 3.1875,
      "learning_rate": 4.651960263251594e-05,
      "loss": 0.8212,
      "step": 12162
    },
    {
      "epoch": 0.17249557735463766,
      "grad_norm": 3.03125,
      "learning_rate": 4.6518465374808714e-05,
      "loss": 0.8639,
      "step": 12164
    },
    {
      "epoch": 0.17252393900826388,
      "grad_norm": 3.25,
      "learning_rate": 4.6517327945232075e-05,
      "loss": 0.8575,
      "step": 12166
    },
    {
      "epoch": 0.1725523006618901,
      "grad_norm": 3.359375,
      "learning_rate": 4.65161903437951e-05,
      "loss": 0.8691,
      "step": 12168
    },
    {
      "epoch": 0.1725806623155163,
      "grad_norm": 3.234375,
      "learning_rate": 4.651505257050688e-05,
      "loss": 0.8713,
      "step": 12170
    },
    {
      "epoch": 0.17260902396914252,
      "grad_norm": 3.359375,
      "learning_rate": 4.65139146253765e-05,
      "loss": 0.8117,
      "step": 12172
    },
    {
      "epoch": 0.17263738562276873,
      "grad_norm": 3.234375,
      "learning_rate": 4.651277650841305e-05,
      "loss": 0.873,
      "step": 12174
    },
    {
      "epoch": 0.17266574727639494,
      "grad_norm": 2.75,
      "learning_rate": 4.651163821962561e-05,
      "loss": 0.867,
      "step": 12176
    },
    {
      "epoch": 0.17269410893002116,
      "grad_norm": 3.09375,
      "learning_rate": 4.651049975902328e-05,
      "loss": 0.8184,
      "step": 12178
    },
    {
      "epoch": 0.17272247058364737,
      "grad_norm": 2.875,
      "learning_rate": 4.650936112661517e-05,
      "loss": 0.8054,
      "step": 12180
    },
    {
      "epoch": 0.17275083223727358,
      "grad_norm": 2.78125,
      "learning_rate": 4.650822232241034e-05,
      "loss": 0.8438,
      "step": 12182
    },
    {
      "epoch": 0.1727791938908998,
      "grad_norm": 3.0625,
      "learning_rate": 4.6507083346417916e-05,
      "loss": 0.7824,
      "step": 12184
    },
    {
      "epoch": 0.17280755554452604,
      "grad_norm": 2.921875,
      "learning_rate": 4.650594419864698e-05,
      "loss": 0.7867,
      "step": 12186
    },
    {
      "epoch": 0.17283591719815225,
      "grad_norm": 3.046875,
      "learning_rate": 4.6504804879106636e-05,
      "loss": 0.8093,
      "step": 12188
    },
    {
      "epoch": 0.17286427885177846,
      "grad_norm": 3.0,
      "learning_rate": 4.6503665387805975e-05,
      "loss": 0.869,
      "step": 12190
    },
    {
      "epoch": 0.17289264050540468,
      "grad_norm": 2.84375,
      "learning_rate": 4.650252572475411e-05,
      "loss": 0.8349,
      "step": 12192
    },
    {
      "epoch": 0.1729210021590309,
      "grad_norm": 3.09375,
      "learning_rate": 4.650138588996013e-05,
      "loss": 0.8225,
      "step": 12194
    },
    {
      "epoch": 0.1729493638126571,
      "grad_norm": 2.984375,
      "learning_rate": 4.650024588343315e-05,
      "loss": 0.8305,
      "step": 12196
    },
    {
      "epoch": 0.17297772546628332,
      "grad_norm": 2.78125,
      "learning_rate": 4.6499105705182275e-05,
      "loss": 0.8393,
      "step": 12198
    },
    {
      "epoch": 0.17300608711990953,
      "grad_norm": 2.84375,
      "learning_rate": 4.649796535521661e-05,
      "loss": 0.8002,
      "step": 12200
    },
    {
      "epoch": 0.17303444877353574,
      "grad_norm": 3.3125,
      "learning_rate": 4.649682483354525e-05,
      "loss": 0.8751,
      "step": 12202
    },
    {
      "epoch": 0.17306281042716196,
      "grad_norm": 2.921875,
      "learning_rate": 4.649568414017733e-05,
      "loss": 0.805,
      "step": 12204
    },
    {
      "epoch": 0.17309117208078817,
      "grad_norm": 3.25,
      "learning_rate": 4.649454327512194e-05,
      "loss": 0.822,
      "step": 12206
    },
    {
      "epoch": 0.17311953373441438,
      "grad_norm": 3.140625,
      "learning_rate": 4.6493402238388204e-05,
      "loss": 0.8436,
      "step": 12208
    },
    {
      "epoch": 0.1731478953880406,
      "grad_norm": 3.3125,
      "learning_rate": 4.6492261029985226e-05,
      "loss": 0.9227,
      "step": 12210
    },
    {
      "epoch": 0.1731762570416668,
      "grad_norm": 2.875,
      "learning_rate": 4.6491119649922124e-05,
      "loss": 0.8329,
      "step": 12212
    },
    {
      "epoch": 0.17320461869529302,
      "grad_norm": 3.125,
      "learning_rate": 4.648997809820802e-05,
      "loss": 0.8423,
      "step": 12214
    },
    {
      "epoch": 0.17323298034891924,
      "grad_norm": 3.140625,
      "learning_rate": 4.648883637485203e-05,
      "loss": 0.8236,
      "step": 12216
    },
    {
      "epoch": 0.17326134200254545,
      "grad_norm": 3.1875,
      "learning_rate": 4.648769447986326e-05,
      "loss": 0.8733,
      "step": 12218
    },
    {
      "epoch": 0.17328970365617166,
      "grad_norm": 2.9375,
      "learning_rate": 4.648655241325085e-05,
      "loss": 0.8388,
      "step": 12220
    },
    {
      "epoch": 0.17331806530979788,
      "grad_norm": 3.0625,
      "learning_rate": 4.648541017502392e-05,
      "loss": 0.8709,
      "step": 12222
    },
    {
      "epoch": 0.17334642696342412,
      "grad_norm": 3.015625,
      "learning_rate": 4.648426776519157e-05,
      "loss": 0.8962,
      "step": 12224
    },
    {
      "epoch": 0.17337478861705033,
      "grad_norm": 3.234375,
      "learning_rate": 4.648312518376295e-05,
      "loss": 0.8839,
      "step": 12226
    },
    {
      "epoch": 0.17340315027067654,
      "grad_norm": 2.703125,
      "learning_rate": 4.648198243074717e-05,
      "loss": 0.8691,
      "step": 12228
    },
    {
      "epoch": 0.17343151192430276,
      "grad_norm": 3.140625,
      "learning_rate": 4.6480839506153364e-05,
      "loss": 0.8455,
      "step": 12230
    },
    {
      "epoch": 0.17345987357792897,
      "grad_norm": 3.5625,
      "learning_rate": 4.647969640999066e-05,
      "loss": 0.855,
      "step": 12232
    },
    {
      "epoch": 0.17348823523155518,
      "grad_norm": 2.84375,
      "learning_rate": 4.64785531422682e-05,
      "loss": 0.9109,
      "step": 12234
    },
    {
      "epoch": 0.1735165968851814,
      "grad_norm": 2.90625,
      "learning_rate": 4.647740970299509e-05,
      "loss": 0.8638,
      "step": 12236
    },
    {
      "epoch": 0.1735449585388076,
      "grad_norm": 3.0625,
      "learning_rate": 4.647626609218048e-05,
      "loss": 0.8612,
      "step": 12238
    },
    {
      "epoch": 0.17357332019243382,
      "grad_norm": 2.859375,
      "learning_rate": 4.6475122309833504e-05,
      "loss": 0.8732,
      "step": 12240
    },
    {
      "epoch": 0.17360168184606004,
      "grad_norm": 3.5,
      "learning_rate": 4.647397835596329e-05,
      "loss": 0.8411,
      "step": 12242
    },
    {
      "epoch": 0.17363004349968625,
      "grad_norm": 2.890625,
      "learning_rate": 4.6472834230578977e-05,
      "loss": 0.8358,
      "step": 12244
    },
    {
      "epoch": 0.17365840515331246,
      "grad_norm": 3.078125,
      "learning_rate": 4.6471689933689704e-05,
      "loss": 0.8974,
      "step": 12246
    },
    {
      "epoch": 0.17368676680693867,
      "grad_norm": 3.25,
      "learning_rate": 4.647054546530462e-05,
      "loss": 0.8865,
      "step": 12248
    },
    {
      "epoch": 0.1737151284605649,
      "grad_norm": 2.875,
      "learning_rate": 4.646940082543285e-05,
      "loss": 0.7892,
      "step": 12250
    },
    {
      "epoch": 0.1737434901141911,
      "grad_norm": 3.28125,
      "learning_rate": 4.6468256014083546e-05,
      "loss": 0.8558,
      "step": 12252
    },
    {
      "epoch": 0.17377185176781731,
      "grad_norm": 3.015625,
      "learning_rate": 4.646711103126584e-05,
      "loss": 0.8111,
      "step": 12254
    },
    {
      "epoch": 0.17380021342144353,
      "grad_norm": 3.03125,
      "learning_rate": 4.64659658769889e-05,
      "loss": 0.9268,
      "step": 12256
    },
    {
      "epoch": 0.17382857507506974,
      "grad_norm": 2.65625,
      "learning_rate": 4.646482055126185e-05,
      "loss": 0.8588,
      "step": 12258
    },
    {
      "epoch": 0.17385693672869595,
      "grad_norm": 3.125,
      "learning_rate": 4.646367505409385e-05,
      "loss": 0.8747,
      "step": 12260
    },
    {
      "epoch": 0.17388529838232217,
      "grad_norm": 3.5,
      "learning_rate": 4.646252938549405e-05,
      "loss": 0.8417,
      "step": 12262
    },
    {
      "epoch": 0.1739136600359484,
      "grad_norm": 2.890625,
      "learning_rate": 4.646138354547159e-05,
      "loss": 0.8216,
      "step": 12264
    },
    {
      "epoch": 0.17394202168957462,
      "grad_norm": 3.21875,
      "learning_rate": 4.6460237534035626e-05,
      "loss": 0.8357,
      "step": 12266
    },
    {
      "epoch": 0.17397038334320084,
      "grad_norm": 3.0,
      "learning_rate": 4.6459091351195316e-05,
      "loss": 0.7958,
      "step": 12268
    },
    {
      "epoch": 0.17399874499682705,
      "grad_norm": 3.40625,
      "learning_rate": 4.645794499695981e-05,
      "loss": 0.8411,
      "step": 12270
    },
    {
      "epoch": 0.17402710665045326,
      "grad_norm": 3.203125,
      "learning_rate": 4.645679847133827e-05,
      "loss": 0.8601,
      "step": 12272
    },
    {
      "epoch": 0.17405546830407947,
      "grad_norm": 2.921875,
      "learning_rate": 4.645565177433985e-05,
      "loss": 0.8439,
      "step": 12274
    },
    {
      "epoch": 0.1740838299577057,
      "grad_norm": 3.265625,
      "learning_rate": 4.645450490597371e-05,
      "loss": 0.8395,
      "step": 12276
    },
    {
      "epoch": 0.1741121916113319,
      "grad_norm": 3.28125,
      "learning_rate": 4.6453357866249e-05,
      "loss": 0.8503,
      "step": 12278
    },
    {
      "epoch": 0.17414055326495811,
      "grad_norm": 3.71875,
      "learning_rate": 4.64522106551749e-05,
      "loss": 0.8124,
      "step": 12280
    },
    {
      "epoch": 0.17416891491858433,
      "grad_norm": 3.046875,
      "learning_rate": 4.645106327276056e-05,
      "loss": 0.8141,
      "step": 12282
    },
    {
      "epoch": 0.17419727657221054,
      "grad_norm": 3.1875,
      "learning_rate": 4.6449915719015144e-05,
      "loss": 0.7862,
      "step": 12284
    },
    {
      "epoch": 0.17422563822583675,
      "grad_norm": 3.359375,
      "learning_rate": 4.6448767993947815e-05,
      "loss": 0.8373,
      "step": 12286
    },
    {
      "epoch": 0.17425399987946297,
      "grad_norm": 2.921875,
      "learning_rate": 4.6447620097567754e-05,
      "loss": 0.8462,
      "step": 12288
    },
    {
      "epoch": 0.17428236153308918,
      "grad_norm": 3.125,
      "learning_rate": 4.644647202988412e-05,
      "loss": 0.902,
      "step": 12290
    },
    {
      "epoch": 0.1743107231867154,
      "grad_norm": 2.921875,
      "learning_rate": 4.644532379090608e-05,
      "loss": 0.7617,
      "step": 12292
    },
    {
      "epoch": 0.1743390848403416,
      "grad_norm": 3.265625,
      "learning_rate": 4.6444175380642807e-05,
      "loss": 0.8468,
      "step": 12294
    },
    {
      "epoch": 0.17436744649396782,
      "grad_norm": 3.296875,
      "learning_rate": 4.6443026799103486e-05,
      "loss": 0.818,
      "step": 12296
    },
    {
      "epoch": 0.17439580814759403,
      "grad_norm": 3.09375,
      "learning_rate": 4.644187804629727e-05,
      "loss": 0.9021,
      "step": 12298
    },
    {
      "epoch": 0.17442416980122025,
      "grad_norm": 3.015625,
      "learning_rate": 4.644072912223334e-05,
      "loss": 0.8118,
      "step": 12300
    },
    {
      "epoch": 0.1744525314548465,
      "grad_norm": 3.078125,
      "learning_rate": 4.643958002692088e-05,
      "loss": 0.8486,
      "step": 12302
    },
    {
      "epoch": 0.1744808931084727,
      "grad_norm": 2.96875,
      "learning_rate": 4.6438430760369064e-05,
      "loss": 0.8269,
      "step": 12304
    },
    {
      "epoch": 0.17450925476209891,
      "grad_norm": 2.90625,
      "learning_rate": 4.643728132258708e-05,
      "loss": 0.8216,
      "step": 12306
    },
    {
      "epoch": 0.17453761641572513,
      "grad_norm": 3.0,
      "learning_rate": 4.643613171358409e-05,
      "loss": 0.8519,
      "step": 12308
    },
    {
      "epoch": 0.17456597806935134,
      "grad_norm": 3.21875,
      "learning_rate": 4.6434981933369294e-05,
      "loss": 0.8493,
      "step": 12310
    },
    {
      "epoch": 0.17459433972297755,
      "grad_norm": 2.953125,
      "learning_rate": 4.643383198195186e-05,
      "loss": 0.8004,
      "step": 12312
    },
    {
      "epoch": 0.17462270137660377,
      "grad_norm": 3.03125,
      "learning_rate": 4.643268185934099e-05,
      "loss": 0.7851,
      "step": 12314
    },
    {
      "epoch": 0.17465106303022998,
      "grad_norm": 3.1875,
      "learning_rate": 4.643153156554585e-05,
      "loss": 0.8573,
      "step": 12316
    },
    {
      "epoch": 0.1746794246838562,
      "grad_norm": 3.3125,
      "learning_rate": 4.643038110057564e-05,
      "loss": 0.8299,
      "step": 12318
    },
    {
      "epoch": 0.1747077863374824,
      "grad_norm": 3.078125,
      "learning_rate": 4.6429230464439544e-05,
      "loss": 0.8255,
      "step": 12320
    },
    {
      "epoch": 0.17473614799110862,
      "grad_norm": 3.234375,
      "learning_rate": 4.642807965714676e-05,
      "loss": 0.8639,
      "step": 12322
    },
    {
      "epoch": 0.17476450964473483,
      "grad_norm": 3.140625,
      "learning_rate": 4.642692867870648e-05,
      "loss": 0.8382,
      "step": 12324
    },
    {
      "epoch": 0.17479287129836105,
      "grad_norm": 3.296875,
      "learning_rate": 4.6425777529127885e-05,
      "loss": 0.8247,
      "step": 12326
    },
    {
      "epoch": 0.17482123295198726,
      "grad_norm": 2.921875,
      "learning_rate": 4.6424626208420175e-05,
      "loss": 0.8779,
      "step": 12328
    },
    {
      "epoch": 0.17484959460561347,
      "grad_norm": 3.0625,
      "learning_rate": 4.6423474716592546e-05,
      "loss": 0.8107,
      "step": 12330
    },
    {
      "epoch": 0.1748779562592397,
      "grad_norm": 3.125,
      "learning_rate": 4.6422323053654205e-05,
      "loss": 0.8488,
      "step": 12332
    },
    {
      "epoch": 0.1749063179128659,
      "grad_norm": 2.921875,
      "learning_rate": 4.642117121961432e-05,
      "loss": 0.8343,
      "step": 12334
    },
    {
      "epoch": 0.1749346795664921,
      "grad_norm": 2.96875,
      "learning_rate": 4.642001921448214e-05,
      "loss": 0.8628,
      "step": 12336
    },
    {
      "epoch": 0.17496304122011833,
      "grad_norm": 3.59375,
      "learning_rate": 4.6418867038266814e-05,
      "loss": 0.8656,
      "step": 12338
    },
    {
      "epoch": 0.17499140287374457,
      "grad_norm": 3.15625,
      "learning_rate": 4.6417714690977575e-05,
      "loss": 0.8346,
      "step": 12340
    },
    {
      "epoch": 0.17501976452737078,
      "grad_norm": 3.015625,
      "learning_rate": 4.641656217262362e-05,
      "loss": 0.8389,
      "step": 12342
    },
    {
      "epoch": 0.175048126180997,
      "grad_norm": 2.75,
      "learning_rate": 4.6415409483214164e-05,
      "loss": 0.836,
      "step": 12344
    },
    {
      "epoch": 0.1750764878346232,
      "grad_norm": 2.90625,
      "learning_rate": 4.6414256622758396e-05,
      "loss": 0.9025,
      "step": 12346
    },
    {
      "epoch": 0.17510484948824942,
      "grad_norm": 2.765625,
      "learning_rate": 4.641310359126553e-05,
      "loss": 0.8001,
      "step": 12348
    },
    {
      "epoch": 0.17513321114187563,
      "grad_norm": 2.984375,
      "learning_rate": 4.641195038874479e-05,
      "loss": 0.8434,
      "step": 12350
    },
    {
      "epoch": 0.17516157279550185,
      "grad_norm": 2.734375,
      "learning_rate": 4.641079701520535e-05,
      "loss": 0.8375,
      "step": 12352
    },
    {
      "epoch": 0.17518993444912806,
      "grad_norm": 3.078125,
      "learning_rate": 4.640964347065646e-05,
      "loss": 0.8381,
      "step": 12354
    },
    {
      "epoch": 0.17521829610275427,
      "grad_norm": 3.6875,
      "learning_rate": 4.640848975510732e-05,
      "loss": 0.8247,
      "step": 12356
    },
    {
      "epoch": 0.1752466577563805,
      "grad_norm": 3.265625,
      "learning_rate": 4.640733586856715e-05,
      "loss": 0.8165,
      "step": 12358
    },
    {
      "epoch": 0.1752750194100067,
      "grad_norm": 2.9375,
      "learning_rate": 4.640618181104516e-05,
      "loss": 0.8395,
      "step": 12360
    },
    {
      "epoch": 0.1753033810636329,
      "grad_norm": 3.484375,
      "learning_rate": 4.6405027582550556e-05,
      "loss": 0.8736,
      "step": 12362
    },
    {
      "epoch": 0.17533174271725913,
      "grad_norm": 3.15625,
      "learning_rate": 4.640387318309258e-05,
      "loss": 0.8534,
      "step": 12364
    },
    {
      "epoch": 0.17536010437088534,
      "grad_norm": 3.1875,
      "learning_rate": 4.640271861268043e-05,
      "loss": 0.8289,
      "step": 12366
    },
    {
      "epoch": 0.17538846602451155,
      "grad_norm": 2.703125,
      "learning_rate": 4.640156387132335e-05,
      "loss": 0.7691,
      "step": 12368
    },
    {
      "epoch": 0.17541682767813777,
      "grad_norm": 3.03125,
      "learning_rate": 4.640040895903054e-05,
      "loss": 0.8178,
      "step": 12370
    },
    {
      "epoch": 0.17544518933176398,
      "grad_norm": 3.359375,
      "learning_rate": 4.639925387581125e-05,
      "loss": 0.8038,
      "step": 12372
    },
    {
      "epoch": 0.1754735509853902,
      "grad_norm": 2.953125,
      "learning_rate": 4.639809862167468e-05,
      "loss": 0.8212,
      "step": 12374
    },
    {
      "epoch": 0.1755019126390164,
      "grad_norm": 2.984375,
      "learning_rate": 4.639694319663008e-05,
      "loss": 0.8392,
      "step": 12376
    },
    {
      "epoch": 0.17553027429264265,
      "grad_norm": 2.984375,
      "learning_rate": 4.639578760068666e-05,
      "loss": 0.7977,
      "step": 12378
    },
    {
      "epoch": 0.17555863594626886,
      "grad_norm": 2.890625,
      "learning_rate": 4.6394631833853666e-05,
      "loss": 0.869,
      "step": 12380
    },
    {
      "epoch": 0.17558699759989507,
      "grad_norm": 3.453125,
      "learning_rate": 4.6393475896140306e-05,
      "loss": 0.8748,
      "step": 12382
    },
    {
      "epoch": 0.17561535925352129,
      "grad_norm": 3.203125,
      "learning_rate": 4.639231978755584e-05,
      "loss": 0.8879,
      "step": 12384
    },
    {
      "epoch": 0.1756437209071475,
      "grad_norm": 2.8125,
      "learning_rate": 4.639116350810948e-05,
      "loss": 0.8267,
      "step": 12386
    },
    {
      "epoch": 0.1756720825607737,
      "grad_norm": 3.5,
      "learning_rate": 4.639000705781048e-05,
      "loss": 0.8497,
      "step": 12388
    },
    {
      "epoch": 0.17570044421439993,
      "grad_norm": 3.5,
      "learning_rate": 4.6388850436668065e-05,
      "loss": 0.8637,
      "step": 12390
    },
    {
      "epoch": 0.17572880586802614,
      "grad_norm": 2.828125,
      "learning_rate": 4.6387693644691464e-05,
      "loss": 0.8515,
      "step": 12392
    },
    {
      "epoch": 0.17575716752165235,
      "grad_norm": 3.21875,
      "learning_rate": 4.6386536681889934e-05,
      "loss": 0.7947,
      "step": 12394
    },
    {
      "epoch": 0.17578552917527857,
      "grad_norm": 3.125,
      "learning_rate": 4.638537954827271e-05,
      "loss": 0.8897,
      "step": 12396
    },
    {
      "epoch": 0.17581389082890478,
      "grad_norm": 3.171875,
      "learning_rate": 4.638422224384904e-05,
      "loss": 0.8315,
      "step": 12398
    },
    {
      "epoch": 0.175842252482531,
      "grad_norm": 2.78125,
      "learning_rate": 4.638306476862815e-05,
      "loss": 0.847,
      "step": 12400
    },
    {
      "epoch": 0.1758706141361572,
      "grad_norm": 2.96875,
      "learning_rate": 4.63819071226193e-05,
      "loss": 0.8284,
      "step": 12402
    },
    {
      "epoch": 0.17589897578978342,
      "grad_norm": 3.078125,
      "learning_rate": 4.638074930583173e-05,
      "loss": 0.8503,
      "step": 12404
    },
    {
      "epoch": 0.17592733744340963,
      "grad_norm": 3.078125,
      "learning_rate": 4.6379591318274684e-05,
      "loss": 0.8233,
      "step": 12406
    },
    {
      "epoch": 0.17595569909703584,
      "grad_norm": 3.8125,
      "learning_rate": 4.637843315995742e-05,
      "loss": 0.8389,
      "step": 12408
    },
    {
      "epoch": 0.17598406075066206,
      "grad_norm": 2.90625,
      "learning_rate": 4.637727483088919e-05,
      "loss": 0.8146,
      "step": 12410
    },
    {
      "epoch": 0.17601242240428827,
      "grad_norm": 3.015625,
      "learning_rate": 4.6376116331079235e-05,
      "loss": 0.8279,
      "step": 12412
    },
    {
      "epoch": 0.17604078405791448,
      "grad_norm": 3.0625,
      "learning_rate": 4.6374957660536816e-05,
      "loss": 0.8128,
      "step": 12414
    },
    {
      "epoch": 0.1760691457115407,
      "grad_norm": 3.65625,
      "learning_rate": 4.6373798819271186e-05,
      "loss": 0.8043,
      "step": 12416
    },
    {
      "epoch": 0.17609750736516694,
      "grad_norm": 3.1875,
      "learning_rate": 4.637263980729159e-05,
      "loss": 0.8207,
      "step": 12418
    },
    {
      "epoch": 0.17612586901879315,
      "grad_norm": 3.015625,
      "learning_rate": 4.63714806246073e-05,
      "loss": 0.8652,
      "step": 12420
    },
    {
      "epoch": 0.17615423067241937,
      "grad_norm": 3.0625,
      "learning_rate": 4.637032127122757e-05,
      "loss": 0.8387,
      "step": 12422
    },
    {
      "epoch": 0.17618259232604558,
      "grad_norm": 2.78125,
      "learning_rate": 4.636916174716166e-05,
      "loss": 0.8219,
      "step": 12424
    },
    {
      "epoch": 0.1762109539796718,
      "grad_norm": 3.203125,
      "learning_rate": 4.636800205241882e-05,
      "loss": 0.8478,
      "step": 12426
    },
    {
      "epoch": 0.176239315633298,
      "grad_norm": 2.84375,
      "learning_rate": 4.6366842187008333e-05,
      "loss": 0.8115,
      "step": 12428
    },
    {
      "epoch": 0.17626767728692422,
      "grad_norm": 3.3125,
      "learning_rate": 4.636568215093945e-05,
      "loss": 0.8798,
      "step": 12430
    },
    {
      "epoch": 0.17629603894055043,
      "grad_norm": 3.34375,
      "learning_rate": 4.636452194422144e-05,
      "loss": 0.8187,
      "step": 12432
    },
    {
      "epoch": 0.17632440059417664,
      "grad_norm": 3.25,
      "learning_rate": 4.636336156686356e-05,
      "loss": 0.8794,
      "step": 12434
    },
    {
      "epoch": 0.17635276224780286,
      "grad_norm": 3.1875,
      "learning_rate": 4.63622010188751e-05,
      "loss": 0.8367,
      "step": 12436
    },
    {
      "epoch": 0.17638112390142907,
      "grad_norm": 3.015625,
      "learning_rate": 4.63610403002653e-05,
      "loss": 0.8112,
      "step": 12438
    },
    {
      "epoch": 0.17640948555505528,
      "grad_norm": 2.921875,
      "learning_rate": 4.635987941104346e-05,
      "loss": 0.8639,
      "step": 12440
    },
    {
      "epoch": 0.1764378472086815,
      "grad_norm": 3.4375,
      "learning_rate": 4.635871835121883e-05,
      "loss": 0.8322,
      "step": 12442
    },
    {
      "epoch": 0.1764662088623077,
      "grad_norm": 2.84375,
      "learning_rate": 4.63575571208007e-05,
      "loss": 0.8443,
      "step": 12444
    },
    {
      "epoch": 0.17649457051593392,
      "grad_norm": 3.203125,
      "learning_rate": 4.635639571979832e-05,
      "loss": 0.8351,
      "step": 12446
    },
    {
      "epoch": 0.17652293216956014,
      "grad_norm": 3.046875,
      "learning_rate": 4.6355234148221e-05,
      "loss": 0.8675,
      "step": 12448
    },
    {
      "epoch": 0.17655129382318635,
      "grad_norm": 2.734375,
      "learning_rate": 4.635407240607799e-05,
      "loss": 0.7643,
      "step": 12450
    },
    {
      "epoch": 0.17657965547681256,
      "grad_norm": 2.859375,
      "learning_rate": 4.635291049337859e-05,
      "loss": 0.8354,
      "step": 12452
    },
    {
      "epoch": 0.17660801713043878,
      "grad_norm": 2.84375,
      "learning_rate": 4.6351748410132055e-05,
      "loss": 0.8055,
      "step": 12454
    },
    {
      "epoch": 0.17663637878406502,
      "grad_norm": 2.734375,
      "learning_rate": 4.63505861563477e-05,
      "loss": 0.8122,
      "step": 12456
    },
    {
      "epoch": 0.17666474043769123,
      "grad_norm": 3.171875,
      "learning_rate": 4.634942373203477e-05,
      "loss": 0.8549,
      "step": 12458
    },
    {
      "epoch": 0.17669310209131744,
      "grad_norm": 3.125,
      "learning_rate": 4.634826113720258e-05,
      "loss": 0.8179,
      "step": 12460
    },
    {
      "epoch": 0.17672146374494366,
      "grad_norm": 3.03125,
      "learning_rate": 4.6347098371860396e-05,
      "loss": 0.8378,
      "step": 12462
    },
    {
      "epoch": 0.17674982539856987,
      "grad_norm": 2.84375,
      "learning_rate": 4.634593543601752e-05,
      "loss": 0.8412,
      "step": 12464
    },
    {
      "epoch": 0.17677818705219608,
      "grad_norm": 3.484375,
      "learning_rate": 4.634477232968324e-05,
      "loss": 0.8585,
      "step": 12466
    },
    {
      "epoch": 0.1768065487058223,
      "grad_norm": 3.375,
      "learning_rate": 4.634360905286683e-05,
      "loss": 0.8448,
      "step": 12468
    },
    {
      "epoch": 0.1768349103594485,
      "grad_norm": 3.125,
      "learning_rate": 4.6342445605577595e-05,
      "loss": 0.87,
      "step": 12470
    },
    {
      "epoch": 0.17686327201307472,
      "grad_norm": 3.171875,
      "learning_rate": 4.6341281987824817e-05,
      "loss": 0.8332,
      "step": 12472
    },
    {
      "epoch": 0.17689163366670094,
      "grad_norm": 3.703125,
      "learning_rate": 4.6340118199617805e-05,
      "loss": 0.8791,
      "step": 12474
    },
    {
      "epoch": 0.17691999532032715,
      "grad_norm": 2.71875,
      "learning_rate": 4.6338954240965836e-05,
      "loss": 0.8119,
      "step": 12476
    },
    {
      "epoch": 0.17694835697395336,
      "grad_norm": 3.125,
      "learning_rate": 4.633779011187823e-05,
      "loss": 0.8894,
      "step": 12478
    },
    {
      "epoch": 0.17697671862757958,
      "grad_norm": 2.890625,
      "learning_rate": 4.633662581236426e-05,
      "loss": 0.8229,
      "step": 12480
    },
    {
      "epoch": 0.1770050802812058,
      "grad_norm": 3.328125,
      "learning_rate": 4.633546134243324e-05,
      "loss": 0.8509,
      "step": 12482
    },
    {
      "epoch": 0.177033441934832,
      "grad_norm": 2.890625,
      "learning_rate": 4.633429670209447e-05,
      "loss": 0.7919,
      "step": 12484
    },
    {
      "epoch": 0.17706180358845822,
      "grad_norm": 3.109375,
      "learning_rate": 4.633313189135724e-05,
      "loss": 0.9022,
      "step": 12486
    },
    {
      "epoch": 0.17709016524208443,
      "grad_norm": 3.140625,
      "learning_rate": 4.633196691023087e-05,
      "loss": 0.9055,
      "step": 12488
    },
    {
      "epoch": 0.17711852689571064,
      "grad_norm": 3.0,
      "learning_rate": 4.6330801758724656e-05,
      "loss": 0.8362,
      "step": 12490
    },
    {
      "epoch": 0.17714688854933686,
      "grad_norm": 3.015625,
      "learning_rate": 4.632963643684791e-05,
      "loss": 0.8349,
      "step": 12492
    },
    {
      "epoch": 0.1771752502029631,
      "grad_norm": 3.375,
      "learning_rate": 4.632847094460992e-05,
      "loss": 0.8317,
      "step": 12494
    },
    {
      "epoch": 0.1772036118565893,
      "grad_norm": 3.21875,
      "learning_rate": 4.6327305282020026e-05,
      "loss": 0.8622,
      "step": 12496
    },
    {
      "epoch": 0.17723197351021552,
      "grad_norm": 3.390625,
      "learning_rate": 4.6326139449087514e-05,
      "loss": 0.8526,
      "step": 12498
    },
    {
      "epoch": 0.17726033516384174,
      "grad_norm": 2.96875,
      "learning_rate": 4.6324973445821705e-05,
      "loss": 0.8261,
      "step": 12500
    },
    {
      "epoch": 0.17728869681746795,
      "grad_norm": 2.75,
      "learning_rate": 4.6323807272231915e-05,
      "loss": 0.848,
      "step": 12502
    },
    {
      "epoch": 0.17731705847109416,
      "grad_norm": 3.015625,
      "learning_rate": 4.6322640928327455e-05,
      "loss": 0.8271,
      "step": 12504
    },
    {
      "epoch": 0.17734542012472038,
      "grad_norm": 3.15625,
      "learning_rate": 4.6321474414117636e-05,
      "loss": 0.84,
      "step": 12506
    },
    {
      "epoch": 0.1773737817783466,
      "grad_norm": 3.125,
      "learning_rate": 4.6320307729611775e-05,
      "loss": 0.8298,
      "step": 12508
    },
    {
      "epoch": 0.1774021434319728,
      "grad_norm": 3.0625,
      "learning_rate": 4.63191408748192e-05,
      "loss": 0.8465,
      "step": 12510
    },
    {
      "epoch": 0.17743050508559902,
      "grad_norm": 3.203125,
      "learning_rate": 4.631797384974922e-05,
      "loss": 0.7893,
      "step": 12512
    },
    {
      "epoch": 0.17745886673922523,
      "grad_norm": 3.0625,
      "learning_rate": 4.631680665441116e-05,
      "loss": 0.8321,
      "step": 12514
    },
    {
      "epoch": 0.17748722839285144,
      "grad_norm": 3.265625,
      "learning_rate": 4.631563928881435e-05,
      "loss": 0.8563,
      "step": 12516
    },
    {
      "epoch": 0.17751559004647766,
      "grad_norm": 3.1875,
      "learning_rate": 4.63144717529681e-05,
      "loss": 0.8354,
      "step": 12518
    },
    {
      "epoch": 0.17754395170010387,
      "grad_norm": 3.203125,
      "learning_rate": 4.6313304046881756e-05,
      "loss": 0.8669,
      "step": 12520
    },
    {
      "epoch": 0.17757231335373008,
      "grad_norm": 3.171875,
      "learning_rate": 4.631213617056462e-05,
      "loss": 0.8339,
      "step": 12522
    },
    {
      "epoch": 0.1776006750073563,
      "grad_norm": 3.390625,
      "learning_rate": 4.631096812402603e-05,
      "loss": 0.8774,
      "step": 12524
    },
    {
      "epoch": 0.1776290366609825,
      "grad_norm": 2.828125,
      "learning_rate": 4.630979990727532e-05,
      "loss": 0.8019,
      "step": 12526
    },
    {
      "epoch": 0.17765739831460872,
      "grad_norm": 3.265625,
      "learning_rate": 4.630863152032181e-05,
      "loss": 0.8873,
      "step": 12528
    },
    {
      "epoch": 0.17768575996823494,
      "grad_norm": 3.015625,
      "learning_rate": 4.6307462963174845e-05,
      "loss": 0.8402,
      "step": 12530
    },
    {
      "epoch": 0.17771412162186118,
      "grad_norm": 3.203125,
      "learning_rate": 4.630629423584376e-05,
      "loss": 0.8487,
      "step": 12532
    },
    {
      "epoch": 0.1777424832754874,
      "grad_norm": 2.765625,
      "learning_rate": 4.630512533833787e-05,
      "loss": 0.8458,
      "step": 12534
    },
    {
      "epoch": 0.1777708449291136,
      "grad_norm": 2.90625,
      "learning_rate": 4.630395627066653e-05,
      "loss": 0.822,
      "step": 12536
    },
    {
      "epoch": 0.17779920658273982,
      "grad_norm": 2.90625,
      "learning_rate": 4.630278703283906e-05,
      "loss": 0.849,
      "step": 12538
    },
    {
      "epoch": 0.17782756823636603,
      "grad_norm": 2.859375,
      "learning_rate": 4.6301617624864815e-05,
      "loss": 0.8411,
      "step": 12540
    },
    {
      "epoch": 0.17785592988999224,
      "grad_norm": 3.078125,
      "learning_rate": 4.630044804675313e-05,
      "loss": 0.8556,
      "step": 12542
    },
    {
      "epoch": 0.17788429154361846,
      "grad_norm": 2.828125,
      "learning_rate": 4.629927829851334e-05,
      "loss": 0.8855,
      "step": 12544
    },
    {
      "epoch": 0.17791265319724467,
      "grad_norm": 2.953125,
      "learning_rate": 4.62981083801548e-05,
      "loss": 0.8526,
      "step": 12546
    },
    {
      "epoch": 0.17794101485087088,
      "grad_norm": 3.140625,
      "learning_rate": 4.6296938291686845e-05,
      "loss": 0.8358,
      "step": 12548
    },
    {
      "epoch": 0.1779693765044971,
      "grad_norm": 2.765625,
      "learning_rate": 4.629576803311882e-05,
      "loss": 0.8437,
      "step": 12550
    },
    {
      "epoch": 0.1779977381581233,
      "grad_norm": 3.078125,
      "learning_rate": 4.6294597604460086e-05,
      "loss": 0.8169,
      "step": 12552
    },
    {
      "epoch": 0.17802609981174952,
      "grad_norm": 3.046875,
      "learning_rate": 4.629342700571998e-05,
      "loss": 0.8293,
      "step": 12554
    },
    {
      "epoch": 0.17805446146537574,
      "grad_norm": 3.0625,
      "learning_rate": 4.629225623690784e-05,
      "loss": 0.8989,
      "step": 12556
    },
    {
      "epoch": 0.17808282311900195,
      "grad_norm": 2.671875,
      "learning_rate": 4.6291085298033035e-05,
      "loss": 0.8461,
      "step": 12558
    },
    {
      "epoch": 0.17811118477262816,
      "grad_norm": 3.21875,
      "learning_rate": 4.6289914189104914e-05,
      "loss": 0.8541,
      "step": 12560
    },
    {
      "epoch": 0.17813954642625437,
      "grad_norm": 2.890625,
      "learning_rate": 4.6288742910132834e-05,
      "loss": 0.8325,
      "step": 12562
    },
    {
      "epoch": 0.1781679080798806,
      "grad_norm": 3.234375,
      "learning_rate": 4.6287571461126134e-05,
      "loss": 0.8721,
      "step": 12564
    },
    {
      "epoch": 0.1781962697335068,
      "grad_norm": 3.15625,
      "learning_rate": 4.6286399842094185e-05,
      "loss": 0.8826,
      "step": 12566
    },
    {
      "epoch": 0.17822463138713301,
      "grad_norm": 3.109375,
      "learning_rate": 4.628522805304634e-05,
      "loss": 0.832,
      "step": 12568
    },
    {
      "epoch": 0.17825299304075923,
      "grad_norm": 3.078125,
      "learning_rate": 4.628405609399196e-05,
      "loss": 0.8381,
      "step": 12570
    },
    {
      "epoch": 0.17828135469438547,
      "grad_norm": 2.96875,
      "learning_rate": 4.62828839649404e-05,
      "loss": 0.8607,
      "step": 12572
    },
    {
      "epoch": 0.17830971634801168,
      "grad_norm": 3.078125,
      "learning_rate": 4.628171166590103e-05,
      "loss": 0.8604,
      "step": 12574
    },
    {
      "epoch": 0.1783380780016379,
      "grad_norm": 3.34375,
      "learning_rate": 4.628053919688321e-05,
      "loss": 0.8196,
      "step": 12576
    },
    {
      "epoch": 0.1783664396552641,
      "grad_norm": 2.96875,
      "learning_rate": 4.6279366557896296e-05,
      "loss": 0.818,
      "step": 12578
    },
    {
      "epoch": 0.17839480130889032,
      "grad_norm": 3.0,
      "learning_rate": 4.627819374894967e-05,
      "loss": 0.8581,
      "step": 12580
    },
    {
      "epoch": 0.17842316296251653,
      "grad_norm": 3.40625,
      "learning_rate": 4.6277020770052695e-05,
      "loss": 0.8901,
      "step": 12582
    },
    {
      "epoch": 0.17845152461614275,
      "grad_norm": 2.984375,
      "learning_rate": 4.627584762121473e-05,
      "loss": 0.832,
      "step": 12584
    },
    {
      "epoch": 0.17847988626976896,
      "grad_norm": 2.984375,
      "learning_rate": 4.627467430244515e-05,
      "loss": 0.8414,
      "step": 12586
    },
    {
      "epoch": 0.17850824792339517,
      "grad_norm": 3.1875,
      "learning_rate": 4.6273500813753324e-05,
      "loss": 0.8246,
      "step": 12588
    },
    {
      "epoch": 0.1785366095770214,
      "grad_norm": 2.875,
      "learning_rate": 4.6272327155148634e-05,
      "loss": 0.8164,
      "step": 12590
    },
    {
      "epoch": 0.1785649712306476,
      "grad_norm": 2.828125,
      "learning_rate": 4.627115332664045e-05,
      "loss": 0.8129,
      "step": 12592
    },
    {
      "epoch": 0.17859333288427381,
      "grad_norm": 2.859375,
      "learning_rate": 4.626997932823815e-05,
      "loss": 0.8421,
      "step": 12594
    },
    {
      "epoch": 0.17862169453790003,
      "grad_norm": 3.28125,
      "learning_rate": 4.6268805159951086e-05,
      "loss": 0.8196,
      "step": 12596
    },
    {
      "epoch": 0.17865005619152624,
      "grad_norm": 2.90625,
      "learning_rate": 4.6267630821788675e-05,
      "loss": 0.8847,
      "step": 12598
    },
    {
      "epoch": 0.17867841784515245,
      "grad_norm": 3.328125,
      "learning_rate": 4.626645631376027e-05,
      "loss": 0.9046,
      "step": 12600
    },
    {
      "epoch": 0.17870677949877867,
      "grad_norm": 3.71875,
      "learning_rate": 4.626528163587527e-05,
      "loss": 0.8962,
      "step": 12602
    },
    {
      "epoch": 0.17873514115240488,
      "grad_norm": 3.453125,
      "learning_rate": 4.6264106788143035e-05,
      "loss": 0.8703,
      "step": 12604
    },
    {
      "epoch": 0.1787635028060311,
      "grad_norm": 3.140625,
      "learning_rate": 4.626293177057296e-05,
      "loss": 0.8129,
      "step": 12606
    },
    {
      "epoch": 0.1787918644596573,
      "grad_norm": 3.1875,
      "learning_rate": 4.6261756583174443e-05,
      "loss": 0.8194,
      "step": 12608
    },
    {
      "epoch": 0.17882022611328355,
      "grad_norm": 3.0625,
      "learning_rate": 4.6260581225956844e-05,
      "loss": 0.8133,
      "step": 12610
    },
    {
      "epoch": 0.17884858776690976,
      "grad_norm": 2.796875,
      "learning_rate": 4.625940569892958e-05,
      "loss": 0.8832,
      "step": 12612
    },
    {
      "epoch": 0.17887694942053597,
      "grad_norm": 3.015625,
      "learning_rate": 4.625823000210201e-05,
      "loss": 0.8218,
      "step": 12614
    },
    {
      "epoch": 0.1789053110741622,
      "grad_norm": 2.84375,
      "learning_rate": 4.6257054135483546e-05,
      "loss": 0.844,
      "step": 12616
    },
    {
      "epoch": 0.1789336727277884,
      "grad_norm": 2.78125,
      "learning_rate": 4.6255878099083575e-05,
      "loss": 0.8746,
      "step": 12618
    },
    {
      "epoch": 0.17896203438141461,
      "grad_norm": 3.140625,
      "learning_rate": 4.625470189291148e-05,
      "loss": 0.839,
      "step": 12620
    },
    {
      "epoch": 0.17899039603504083,
      "grad_norm": 3.078125,
      "learning_rate": 4.625352551697667e-05,
      "loss": 0.8041,
      "step": 12622
    },
    {
      "epoch": 0.17901875768866704,
      "grad_norm": 3.21875,
      "learning_rate": 4.625234897128854e-05,
      "loss": 0.8834,
      "step": 12624
    },
    {
      "epoch": 0.17904711934229325,
      "grad_norm": 2.953125,
      "learning_rate": 4.6251172255856465e-05,
      "loss": 0.7874,
      "step": 12626
    },
    {
      "epoch": 0.17907548099591947,
      "grad_norm": 3.140625,
      "learning_rate": 4.624999537068987e-05,
      "loss": 0.8547,
      "step": 12628
    },
    {
      "epoch": 0.17910384264954568,
      "grad_norm": 2.96875,
      "learning_rate": 4.624881831579815e-05,
      "loss": 0.8339,
      "step": 12630
    },
    {
      "epoch": 0.1791322043031719,
      "grad_norm": 3.34375,
      "learning_rate": 4.624764109119069e-05,
      "loss": 0.8424,
      "step": 12632
    },
    {
      "epoch": 0.1791605659567981,
      "grad_norm": 3.28125,
      "learning_rate": 4.624646369687691e-05,
      "loss": 0.8683,
      "step": 12634
    },
    {
      "epoch": 0.17918892761042432,
      "grad_norm": 3.0625,
      "learning_rate": 4.62452861328662e-05,
      "loss": 0.8303,
      "step": 12636
    },
    {
      "epoch": 0.17921728926405053,
      "grad_norm": 2.6875,
      "learning_rate": 4.624410839916798e-05,
      "loss": 0.8176,
      "step": 12638
    },
    {
      "epoch": 0.17924565091767675,
      "grad_norm": 3.078125,
      "learning_rate": 4.6242930495791646e-05,
      "loss": 0.8482,
      "step": 12640
    },
    {
      "epoch": 0.17927401257130296,
      "grad_norm": 2.921875,
      "learning_rate": 4.624175242274661e-05,
      "loss": 0.8685,
      "step": 12642
    },
    {
      "epoch": 0.17930237422492917,
      "grad_norm": 3.390625,
      "learning_rate": 4.624057418004228e-05,
      "loss": 0.8555,
      "step": 12644
    },
    {
      "epoch": 0.1793307358785554,
      "grad_norm": 3.625,
      "learning_rate": 4.623939576768807e-05,
      "loss": 0.8978,
      "step": 12646
    },
    {
      "epoch": 0.17935909753218163,
      "grad_norm": 3.1875,
      "learning_rate": 4.623821718569338e-05,
      "loss": 0.8696,
      "step": 12648
    },
    {
      "epoch": 0.17938745918580784,
      "grad_norm": 3.109375,
      "learning_rate": 4.623703843406764e-05,
      "loss": 0.8355,
      "step": 12650
    },
    {
      "epoch": 0.17941582083943405,
      "grad_norm": 3.015625,
      "learning_rate": 4.623585951282026e-05,
      "loss": 0.8115,
      "step": 12652
    },
    {
      "epoch": 0.17944418249306027,
      "grad_norm": 3.03125,
      "learning_rate": 4.6234680421960655e-05,
      "loss": 0.8517,
      "step": 12654
    },
    {
      "epoch": 0.17947254414668648,
      "grad_norm": 3.15625,
      "learning_rate": 4.623350116149823e-05,
      "loss": 0.8207,
      "step": 12656
    },
    {
      "epoch": 0.1795009058003127,
      "grad_norm": 2.828125,
      "learning_rate": 4.623232173144242e-05,
      "loss": 0.8364,
      "step": 12658
    },
    {
      "epoch": 0.1795292674539389,
      "grad_norm": 2.9375,
      "learning_rate": 4.623114213180264e-05,
      "loss": 0.8285,
      "step": 12660
    },
    {
      "epoch": 0.17955762910756512,
      "grad_norm": 3.359375,
      "learning_rate": 4.622996236258832e-05,
      "loss": 0.8542,
      "step": 12662
    },
    {
      "epoch": 0.17958599076119133,
      "grad_norm": 3.078125,
      "learning_rate": 4.622878242380887e-05,
      "loss": 0.8826,
      "step": 12664
    },
    {
      "epoch": 0.17961435241481755,
      "grad_norm": 3.390625,
      "learning_rate": 4.622760231547371e-05,
      "loss": 0.8536,
      "step": 12666
    },
    {
      "epoch": 0.17964271406844376,
      "grad_norm": 2.90625,
      "learning_rate": 4.6226422037592286e-05,
      "loss": 0.8177,
      "step": 12668
    },
    {
      "epoch": 0.17967107572206997,
      "grad_norm": 2.71875,
      "learning_rate": 4.6225241590174005e-05,
      "loss": 0.7825,
      "step": 12670
    },
    {
      "epoch": 0.17969943737569619,
      "grad_norm": 2.71875,
      "learning_rate": 4.6224060973228314e-05,
      "loss": 0.7484,
      "step": 12672
    },
    {
      "epoch": 0.1797277990293224,
      "grad_norm": 2.96875,
      "learning_rate": 4.622288018676462e-05,
      "loss": 0.8645,
      "step": 12674
    },
    {
      "epoch": 0.1797561606829486,
      "grad_norm": 3.296875,
      "learning_rate": 4.622169923079237e-05,
      "loss": 0.8663,
      "step": 12676
    },
    {
      "epoch": 0.17978452233657483,
      "grad_norm": 3.09375,
      "learning_rate": 4.6220518105320996e-05,
      "loss": 0.86,
      "step": 12678
    },
    {
      "epoch": 0.17981288399020104,
      "grad_norm": 3.046875,
      "learning_rate": 4.621933681035992e-05,
      "loss": 0.8226,
      "step": 12680
    },
    {
      "epoch": 0.17984124564382725,
      "grad_norm": 2.640625,
      "learning_rate": 4.62181553459186e-05,
      "loss": 0.8311,
      "step": 12682
    },
    {
      "epoch": 0.17986960729745347,
      "grad_norm": 3.140625,
      "learning_rate": 4.621697371200645e-05,
      "loss": 0.8647,
      "step": 12684
    },
    {
      "epoch": 0.1798979689510797,
      "grad_norm": 3.546875,
      "learning_rate": 4.621579190863292e-05,
      "loss": 0.8611,
      "step": 12686
    },
    {
      "epoch": 0.17992633060470592,
      "grad_norm": 2.875,
      "learning_rate": 4.621460993580744e-05,
      "loss": 0.832,
      "step": 12688
    },
    {
      "epoch": 0.17995469225833213,
      "grad_norm": 2.859375,
      "learning_rate": 4.6213427793539457e-05,
      "loss": 0.7857,
      "step": 12690
    },
    {
      "epoch": 0.17998305391195835,
      "grad_norm": 3.15625,
      "learning_rate": 4.621224548183841e-05,
      "loss": 0.8165,
      "step": 12692
    },
    {
      "epoch": 0.18001141556558456,
      "grad_norm": 3.5625,
      "learning_rate": 4.621106300071374e-05,
      "loss": 0.8272,
      "step": 12694
    },
    {
      "epoch": 0.18003977721921077,
      "grad_norm": 2.859375,
      "learning_rate": 4.6209880350174904e-05,
      "loss": 0.7935,
      "step": 12696
    },
    {
      "epoch": 0.18006813887283699,
      "grad_norm": 3.25,
      "learning_rate": 4.620869753023133e-05,
      "loss": 0.8143,
      "step": 12698
    },
    {
      "epoch": 0.1800965005264632,
      "grad_norm": 3.234375,
      "learning_rate": 4.6207514540892485e-05,
      "loss": 0.8784,
      "step": 12700
    },
    {
      "epoch": 0.1801248621800894,
      "grad_norm": 2.78125,
      "learning_rate": 4.62063313821678e-05,
      "loss": 0.8044,
      "step": 12702
    },
    {
      "epoch": 0.18015322383371563,
      "grad_norm": 3.25,
      "learning_rate": 4.620514805406673e-05,
      "loss": 0.8564,
      "step": 12704
    },
    {
      "epoch": 0.18018158548734184,
      "grad_norm": 2.640625,
      "learning_rate": 4.6203964556598734e-05,
      "loss": 0.813,
      "step": 12706
    },
    {
      "epoch": 0.18020994714096805,
      "grad_norm": 3.09375,
      "learning_rate": 4.620278088977325e-05,
      "loss": 0.8727,
      "step": 12708
    },
    {
      "epoch": 0.18023830879459427,
      "grad_norm": 3.140625,
      "learning_rate": 4.620159705359975e-05,
      "loss": 0.8759,
      "step": 12710
    },
    {
      "epoch": 0.18026667044822048,
      "grad_norm": 2.875,
      "learning_rate": 4.620041304808767e-05,
      "loss": 0.8334,
      "step": 12712
    },
    {
      "epoch": 0.1802950321018467,
      "grad_norm": 2.859375,
      "learning_rate": 4.619922887324649e-05,
      "loss": 0.7911,
      "step": 12714
    },
    {
      "epoch": 0.1803233937554729,
      "grad_norm": 3.203125,
      "learning_rate": 4.619804452908565e-05,
      "loss": 0.839,
      "step": 12716
    },
    {
      "epoch": 0.18035175540909912,
      "grad_norm": 3.234375,
      "learning_rate": 4.619686001561461e-05,
      "loss": 0.7925,
      "step": 12718
    },
    {
      "epoch": 0.18038011706272533,
      "grad_norm": 3.015625,
      "learning_rate": 4.619567533284284e-05,
      "loss": 0.8259,
      "step": 12720
    },
    {
      "epoch": 0.18040847871635154,
      "grad_norm": 3.03125,
      "learning_rate": 4.619449048077979e-05,
      "loss": 0.855,
      "step": 12722
    },
    {
      "epoch": 0.18043684036997776,
      "grad_norm": 2.75,
      "learning_rate": 4.619330545943493e-05,
      "loss": 0.8324,
      "step": 12724
    },
    {
      "epoch": 0.180465202023604,
      "grad_norm": 3.078125,
      "learning_rate": 4.619212026881773e-05,
      "loss": 0.8099,
      "step": 12726
    },
    {
      "epoch": 0.1804935636772302,
      "grad_norm": 3.109375,
      "learning_rate": 4.619093490893766e-05,
      "loss": 0.8235,
      "step": 12728
    },
    {
      "epoch": 0.18052192533085643,
      "grad_norm": 3.03125,
      "learning_rate": 4.618974937980417e-05,
      "loss": 0.7999,
      "step": 12730
    },
    {
      "epoch": 0.18055028698448264,
      "grad_norm": 3.0625,
      "learning_rate": 4.618856368142674e-05,
      "loss": 0.8385,
      "step": 12732
    },
    {
      "epoch": 0.18057864863810885,
      "grad_norm": 3.0,
      "learning_rate": 4.6187377813814844e-05,
      "loss": 0.8238,
      "step": 12734
    },
    {
      "epoch": 0.18060701029173507,
      "grad_norm": 2.796875,
      "learning_rate": 4.6186191776977946e-05,
      "loss": 0.791,
      "step": 12736
    },
    {
      "epoch": 0.18063537194536128,
      "grad_norm": 2.859375,
      "learning_rate": 4.6185005570925515e-05,
      "loss": 0.8129,
      "step": 12738
    },
    {
      "epoch": 0.1806637335989875,
      "grad_norm": 3.0625,
      "learning_rate": 4.618381919566703e-05,
      "loss": 0.8641,
      "step": 12740
    },
    {
      "epoch": 0.1806920952526137,
      "grad_norm": 3.109375,
      "learning_rate": 4.6182632651211976e-05,
      "loss": 0.8208,
      "step": 12742
    },
    {
      "epoch": 0.18072045690623992,
      "grad_norm": 3.25,
      "learning_rate": 4.618144593756982e-05,
      "loss": 0.8637,
      "step": 12744
    },
    {
      "epoch": 0.18074881855986613,
      "grad_norm": 2.96875,
      "learning_rate": 4.618025905475004e-05,
      "loss": 0.8398,
      "step": 12746
    },
    {
      "epoch": 0.18077718021349234,
      "grad_norm": 3.125,
      "learning_rate": 4.6179072002762123e-05,
      "loss": 0.8371,
      "step": 12748
    },
    {
      "epoch": 0.18080554186711856,
      "grad_norm": 3.078125,
      "learning_rate": 4.617788478161554e-05,
      "loss": 0.814,
      "step": 12750
    },
    {
      "epoch": 0.18083390352074477,
      "grad_norm": 3.0625,
      "learning_rate": 4.617669739131979e-05,
      "loss": 0.8517,
      "step": 12752
    },
    {
      "epoch": 0.18086226517437098,
      "grad_norm": 3.375,
      "learning_rate": 4.6175509831884336e-05,
      "loss": 0.827,
      "step": 12754
    },
    {
      "epoch": 0.1808906268279972,
      "grad_norm": 3.578125,
      "learning_rate": 4.617432210331867e-05,
      "loss": 0.7785,
      "step": 12756
    },
    {
      "epoch": 0.1809189884816234,
      "grad_norm": 2.828125,
      "learning_rate": 4.6173134205632284e-05,
      "loss": 0.805,
      "step": 12758
    },
    {
      "epoch": 0.18094735013524962,
      "grad_norm": 2.96875,
      "learning_rate": 4.6171946138834667e-05,
      "loss": 0.7827,
      "step": 12760
    },
    {
      "epoch": 0.18097571178887584,
      "grad_norm": 2.96875,
      "learning_rate": 4.6170757902935296e-05,
      "loss": 0.8323,
      "step": 12762
    },
    {
      "epoch": 0.18100407344250208,
      "grad_norm": 3.046875,
      "learning_rate": 4.6169569497943674e-05,
      "loss": 0.8319,
      "step": 12764
    },
    {
      "epoch": 0.1810324350961283,
      "grad_norm": 3.15625,
      "learning_rate": 4.616838092386929e-05,
      "loss": 0.8897,
      "step": 12766
    },
    {
      "epoch": 0.1810607967497545,
      "grad_norm": 3.25,
      "learning_rate": 4.616719218072163e-05,
      "loss": 0.8245,
      "step": 12768
    },
    {
      "epoch": 0.18108915840338072,
      "grad_norm": 2.96875,
      "learning_rate": 4.61660032685102e-05,
      "loss": 0.8604,
      "step": 12770
    },
    {
      "epoch": 0.18111752005700693,
      "grad_norm": 3.46875,
      "learning_rate": 4.616481418724449e-05,
      "loss": 0.9098,
      "step": 12772
    },
    {
      "epoch": 0.18114588171063314,
      "grad_norm": 3.15625,
      "learning_rate": 4.6163624936933994e-05,
      "loss": 0.8468,
      "step": 12774
    },
    {
      "epoch": 0.18117424336425936,
      "grad_norm": 3.203125,
      "learning_rate": 4.6162435517588216e-05,
      "loss": 0.884,
      "step": 12776
    },
    {
      "epoch": 0.18120260501788557,
      "grad_norm": 3.40625,
      "learning_rate": 4.6161245929216655e-05,
      "loss": 0.875,
      "step": 12778
    },
    {
      "epoch": 0.18123096667151178,
      "grad_norm": 3.0625,
      "learning_rate": 4.616005617182881e-05,
      "loss": 0.7976,
      "step": 12780
    },
    {
      "epoch": 0.181259328325138,
      "grad_norm": 3.25,
      "learning_rate": 4.615886624543418e-05,
      "loss": 0.8438,
      "step": 12782
    },
    {
      "epoch": 0.1812876899787642,
      "grad_norm": 2.90625,
      "learning_rate": 4.6157676150042286e-05,
      "loss": 0.8343,
      "step": 12784
    },
    {
      "epoch": 0.18131605163239042,
      "grad_norm": 2.921875,
      "learning_rate": 4.6156485885662606e-05,
      "loss": 0.8782,
      "step": 12786
    },
    {
      "epoch": 0.18134441328601664,
      "grad_norm": 3.375,
      "learning_rate": 4.615529545230467e-05,
      "loss": 0.8824,
      "step": 12788
    },
    {
      "epoch": 0.18137277493964285,
      "grad_norm": 3.15625,
      "learning_rate": 4.615410484997798e-05,
      "loss": 0.8476,
      "step": 12790
    },
    {
      "epoch": 0.18140113659326906,
      "grad_norm": 2.8125,
      "learning_rate": 4.6152914078692046e-05,
      "loss": 0.8412,
      "step": 12792
    },
    {
      "epoch": 0.18142949824689528,
      "grad_norm": 3.265625,
      "learning_rate": 4.615172313845637e-05,
      "loss": 0.8553,
      "step": 12794
    },
    {
      "epoch": 0.1814578599005215,
      "grad_norm": 3.5,
      "learning_rate": 4.615053202928047e-05,
      "loss": 0.8119,
      "step": 12796
    },
    {
      "epoch": 0.1814862215541477,
      "grad_norm": 3.0625,
      "learning_rate": 4.614934075117386e-05,
      "loss": 0.8376,
      "step": 12798
    },
    {
      "epoch": 0.18151458320777392,
      "grad_norm": 2.984375,
      "learning_rate": 4.614814930414606e-05,
      "loss": 0.8349,
      "step": 12800
    },
    {
      "epoch": 0.18154294486140016,
      "grad_norm": 3.1875,
      "learning_rate": 4.6146957688206585e-05,
      "loss": 0.8453,
      "step": 12802
    },
    {
      "epoch": 0.18157130651502637,
      "grad_norm": 3.015625,
      "learning_rate": 4.614576590336493e-05,
      "loss": 0.861,
      "step": 12804
    },
    {
      "epoch": 0.18159966816865258,
      "grad_norm": 3.296875,
      "learning_rate": 4.614457394963065e-05,
      "loss": 0.8354,
      "step": 12806
    },
    {
      "epoch": 0.1816280298222788,
      "grad_norm": 3.046875,
      "learning_rate": 4.6143381827013246e-05,
      "loss": 0.8669,
      "step": 12808
    },
    {
      "epoch": 0.181656391475905,
      "grad_norm": 3.125,
      "learning_rate": 4.614218953552223e-05,
      "loss": 0.8659,
      "step": 12810
    },
    {
      "epoch": 0.18168475312953122,
      "grad_norm": 3.140625,
      "learning_rate": 4.614099707516715e-05,
      "loss": 0.8282,
      "step": 12812
    },
    {
      "epoch": 0.18171311478315744,
      "grad_norm": 3.609375,
      "learning_rate": 4.61398044459575e-05,
      "loss": 0.8183,
      "step": 12814
    },
    {
      "epoch": 0.18174147643678365,
      "grad_norm": 3.453125,
      "learning_rate": 4.613861164790283e-05,
      "loss": 0.8283,
      "step": 12816
    },
    {
      "epoch": 0.18176983809040986,
      "grad_norm": 2.828125,
      "learning_rate": 4.613741868101266e-05,
      "loss": 0.8666,
      "step": 12818
    },
    {
      "epoch": 0.18179819974403608,
      "grad_norm": 3.0625,
      "learning_rate": 4.613622554529652e-05,
      "loss": 0.8464,
      "step": 12820
    },
    {
      "epoch": 0.1818265613976623,
      "grad_norm": 3.328125,
      "learning_rate": 4.613503224076393e-05,
      "loss": 0.8237,
      "step": 12822
    },
    {
      "epoch": 0.1818549230512885,
      "grad_norm": 3.296875,
      "learning_rate": 4.613383876742443e-05,
      "loss": 0.8372,
      "step": 12824
    },
    {
      "epoch": 0.18188328470491472,
      "grad_norm": 3.046875,
      "learning_rate": 4.613264512528755e-05,
      "loss": 0.8469,
      "step": 12826
    },
    {
      "epoch": 0.18191164635854093,
      "grad_norm": 2.84375,
      "learning_rate": 4.613145131436283e-05,
      "loss": 0.8517,
      "step": 12828
    },
    {
      "epoch": 0.18194000801216714,
      "grad_norm": 3.25,
      "learning_rate": 4.6130257334659794e-05,
      "loss": 0.8702,
      "step": 12830
    },
    {
      "epoch": 0.18196836966579336,
      "grad_norm": 3.59375,
      "learning_rate": 4.612906318618798e-05,
      "loss": 0.8935,
      "step": 12832
    },
    {
      "epoch": 0.18199673131941957,
      "grad_norm": 3.03125,
      "learning_rate": 4.612786886895693e-05,
      "loss": 0.8124,
      "step": 12834
    },
    {
      "epoch": 0.18202509297304578,
      "grad_norm": 3.3125,
      "learning_rate": 4.6126674382976185e-05,
      "loss": 0.8502,
      "step": 12836
    },
    {
      "epoch": 0.182053454626672,
      "grad_norm": 3.015625,
      "learning_rate": 4.6125479728255286e-05,
      "loss": 0.8418,
      "step": 12838
    },
    {
      "epoch": 0.18208181628029824,
      "grad_norm": 3.421875,
      "learning_rate": 4.612428490480376e-05,
      "loss": 0.8009,
      "step": 12840
    },
    {
      "epoch": 0.18211017793392445,
      "grad_norm": 3.3125,
      "learning_rate": 4.612308991263118e-05,
      "loss": 0.8816,
      "step": 12842
    },
    {
      "epoch": 0.18213853958755066,
      "grad_norm": 3.078125,
      "learning_rate": 4.612189475174705e-05,
      "loss": 0.8444,
      "step": 12844
    },
    {
      "epoch": 0.18216690124117688,
      "grad_norm": 2.953125,
      "learning_rate": 4.6120699422160954e-05,
      "loss": 0.8561,
      "step": 12846
    },
    {
      "epoch": 0.1821952628948031,
      "grad_norm": 2.75,
      "learning_rate": 4.6119503923882416e-05,
      "loss": 0.8057,
      "step": 12848
    },
    {
      "epoch": 0.1822236245484293,
      "grad_norm": 3.109375,
      "learning_rate": 4.6118308256920994e-05,
      "loss": 0.8083,
      "step": 12850
    },
    {
      "epoch": 0.18225198620205552,
      "grad_norm": 2.953125,
      "learning_rate": 4.6117112421286235e-05,
      "loss": 0.8472,
      "step": 12852
    },
    {
      "epoch": 0.18228034785568173,
      "grad_norm": 2.96875,
      "learning_rate": 4.6115916416987686e-05,
      "loss": 0.8956,
      "step": 12854
    },
    {
      "epoch": 0.18230870950930794,
      "grad_norm": 2.890625,
      "learning_rate": 4.611472024403491e-05,
      "loss": 0.8338,
      "step": 12856
    },
    {
      "epoch": 0.18233707116293416,
      "grad_norm": 3.734375,
      "learning_rate": 4.611352390243745e-05,
      "loss": 0.8513,
      "step": 12858
    },
    {
      "epoch": 0.18236543281656037,
      "grad_norm": 3.359375,
      "learning_rate": 4.611232739220487e-05,
      "loss": 0.782,
      "step": 12860
    },
    {
      "epoch": 0.18239379447018658,
      "grad_norm": 2.984375,
      "learning_rate": 4.611113071334673e-05,
      "loss": 0.8203,
      "step": 12862
    },
    {
      "epoch": 0.1824221561238128,
      "grad_norm": 3.453125,
      "learning_rate": 4.610993386587257e-05,
      "loss": 0.7996,
      "step": 12864
    },
    {
      "epoch": 0.182450517777439,
      "grad_norm": 3.015625,
      "learning_rate": 4.610873684979196e-05,
      "loss": 0.8971,
      "step": 12866
    },
    {
      "epoch": 0.18247887943106522,
      "grad_norm": 3.21875,
      "learning_rate": 4.610753966511446e-05,
      "loss": 0.8564,
      "step": 12868
    },
    {
      "epoch": 0.18250724108469144,
      "grad_norm": 3.296875,
      "learning_rate": 4.610634231184964e-05,
      "loss": 0.8376,
      "step": 12870
    },
    {
      "epoch": 0.18253560273831765,
      "grad_norm": 3.28125,
      "learning_rate": 4.610514479000706e-05,
      "loss": 0.8547,
      "step": 12872
    },
    {
      "epoch": 0.18256396439194386,
      "grad_norm": 3.34375,
      "learning_rate": 4.610394709959627e-05,
      "loss": 0.8569,
      "step": 12874
    },
    {
      "epoch": 0.18259232604557007,
      "grad_norm": 3.359375,
      "learning_rate": 4.610274924062685e-05,
      "loss": 0.8305,
      "step": 12876
    },
    {
      "epoch": 0.1826206876991963,
      "grad_norm": 2.953125,
      "learning_rate": 4.6101551213108365e-05,
      "loss": 0.8287,
      "step": 12878
    },
    {
      "epoch": 0.18264904935282253,
      "grad_norm": 2.859375,
      "learning_rate": 4.610035301705038e-05,
      "loss": 0.8344,
      "step": 12880
    },
    {
      "epoch": 0.18267741100644874,
      "grad_norm": 3.125,
      "learning_rate": 4.6099154652462474e-05,
      "loss": 0.859,
      "step": 12882
    },
    {
      "epoch": 0.18270577266007496,
      "grad_norm": 3.296875,
      "learning_rate": 4.609795611935421e-05,
      "loss": 0.8725,
      "step": 12884
    },
    {
      "epoch": 0.18273413431370117,
      "grad_norm": 3.1875,
      "learning_rate": 4.609675741773516e-05,
      "loss": 0.8796,
      "step": 12886
    },
    {
      "epoch": 0.18276249596732738,
      "grad_norm": 3.265625,
      "learning_rate": 4.60955585476149e-05,
      "loss": 0.8429,
      "step": 12888
    },
    {
      "epoch": 0.1827908576209536,
      "grad_norm": 2.828125,
      "learning_rate": 4.6094359509003014e-05,
      "loss": 0.827,
      "step": 12890
    },
    {
      "epoch": 0.1828192192745798,
      "grad_norm": 3.0625,
      "learning_rate": 4.609316030190906e-05,
      "loss": 0.8486,
      "step": 12892
    },
    {
      "epoch": 0.18284758092820602,
      "grad_norm": 2.90625,
      "learning_rate": 4.6091960926342644e-05,
      "loss": 0.8078,
      "step": 12894
    },
    {
      "epoch": 0.18287594258183223,
      "grad_norm": 3.1875,
      "learning_rate": 4.609076138231332e-05,
      "loss": 0.8252,
      "step": 12896
    },
    {
      "epoch": 0.18290430423545845,
      "grad_norm": 3.0625,
      "learning_rate": 4.608956166983068e-05,
      "loss": 0.8628,
      "step": 12898
    },
    {
      "epoch": 0.18293266588908466,
      "grad_norm": 3.0625,
      "learning_rate": 4.60883617889043e-05,
      "loss": 0.8249,
      "step": 12900
    },
    {
      "epoch": 0.18296102754271087,
      "grad_norm": 3.0,
      "learning_rate": 4.608716173954377e-05,
      "loss": 0.8377,
      "step": 12902
    },
    {
      "epoch": 0.1829893891963371,
      "grad_norm": 3.0625,
      "learning_rate": 4.608596152175868e-05,
      "loss": 0.8572,
      "step": 12904
    },
    {
      "epoch": 0.1830177508499633,
      "grad_norm": 2.875,
      "learning_rate": 4.60847611355586e-05,
      "loss": 0.8088,
      "step": 12906
    },
    {
      "epoch": 0.18304611250358951,
      "grad_norm": 3.015625,
      "learning_rate": 4.608356058095313e-05,
      "loss": 0.8655,
      "step": 12908
    },
    {
      "epoch": 0.18307447415721573,
      "grad_norm": 3.0625,
      "learning_rate": 4.608235985795185e-05,
      "loss": 0.8257,
      "step": 12910
    },
    {
      "epoch": 0.18310283581084194,
      "grad_norm": 3.21875,
      "learning_rate": 4.608115896656437e-05,
      "loss": 0.8127,
      "step": 12912
    },
    {
      "epoch": 0.18313119746446815,
      "grad_norm": 3.140625,
      "learning_rate": 4.607995790680025e-05,
      "loss": 0.8644,
      "step": 12914
    },
    {
      "epoch": 0.18315955911809437,
      "grad_norm": 3.40625,
      "learning_rate": 4.607875667866911e-05,
      "loss": 0.851,
      "step": 12916
    },
    {
      "epoch": 0.1831879207717206,
      "grad_norm": 3.25,
      "learning_rate": 4.6077555282180527e-05,
      "loss": 0.8592,
      "step": 12918
    },
    {
      "epoch": 0.18321628242534682,
      "grad_norm": 2.671875,
      "learning_rate": 4.607635371734411e-05,
      "loss": 0.8004,
      "step": 12920
    },
    {
      "epoch": 0.18324464407897303,
      "grad_norm": 2.75,
      "learning_rate": 4.607515198416945e-05,
      "loss": 0.8097,
      "step": 12922
    },
    {
      "epoch": 0.18327300573259925,
      "grad_norm": 2.890625,
      "learning_rate": 4.607395008266615e-05,
      "loss": 0.8322,
      "step": 12924
    },
    {
      "epoch": 0.18330136738622546,
      "grad_norm": 3.625,
      "learning_rate": 4.60727480128438e-05,
      "loss": 0.8453,
      "step": 12926
    },
    {
      "epoch": 0.18332972903985167,
      "grad_norm": 3.125,
      "learning_rate": 4.6071545774712e-05,
      "loss": 0.8569,
      "step": 12928
    },
    {
      "epoch": 0.1833580906934779,
      "grad_norm": 3.0625,
      "learning_rate": 4.607034336828037e-05,
      "loss": 0.8407,
      "step": 12930
    },
    {
      "epoch": 0.1833864523471041,
      "grad_norm": 3.25,
      "learning_rate": 4.6069140793558495e-05,
      "loss": 0.8226,
      "step": 12932
    },
    {
      "epoch": 0.18341481400073031,
      "grad_norm": 2.9375,
      "learning_rate": 4.6067938050555985e-05,
      "loss": 0.869,
      "step": 12934
    },
    {
      "epoch": 0.18344317565435653,
      "grad_norm": 3.265625,
      "learning_rate": 4.606673513928245e-05,
      "loss": 0.8623,
      "step": 12936
    },
    {
      "epoch": 0.18347153730798274,
      "grad_norm": 3.1875,
      "learning_rate": 4.60655320597475e-05,
      "loss": 0.8535,
      "step": 12938
    },
    {
      "epoch": 0.18349989896160895,
      "grad_norm": 3.484375,
      "learning_rate": 4.606432881196074e-05,
      "loss": 0.8589,
      "step": 12940
    },
    {
      "epoch": 0.18352826061523517,
      "grad_norm": 2.921875,
      "learning_rate": 4.606312539593178e-05,
      "loss": 0.8367,
      "step": 12942
    },
    {
      "epoch": 0.18355662226886138,
      "grad_norm": 3.171875,
      "learning_rate": 4.606192181167023e-05,
      "loss": 0.8597,
      "step": 12944
    },
    {
      "epoch": 0.1835849839224876,
      "grad_norm": 3.625,
      "learning_rate": 4.606071805918571e-05,
      "loss": 0.8607,
      "step": 12946
    },
    {
      "epoch": 0.1836133455761138,
      "grad_norm": 3.21875,
      "learning_rate": 4.605951413848783e-05,
      "loss": 0.8627,
      "step": 12948
    },
    {
      "epoch": 0.18364170722974002,
      "grad_norm": 3.234375,
      "learning_rate": 4.6058310049586196e-05,
      "loss": 0.8562,
      "step": 12950
    },
    {
      "epoch": 0.18367006888336623,
      "grad_norm": 3.328125,
      "learning_rate": 4.6057105792490446e-05,
      "loss": 0.8422,
      "step": 12952
    },
    {
      "epoch": 0.18369843053699245,
      "grad_norm": 2.875,
      "learning_rate": 4.6055901367210185e-05,
      "loss": 0.7802,
      "step": 12954
    },
    {
      "epoch": 0.1837267921906187,
      "grad_norm": 3.0,
      "learning_rate": 4.6054696773755036e-05,
      "loss": 0.8314,
      "step": 12956
    },
    {
      "epoch": 0.1837551538442449,
      "grad_norm": 3.46875,
      "learning_rate": 4.6053492012134615e-05,
      "loss": 0.7939,
      "step": 12958
    },
    {
      "epoch": 0.1837835154978711,
      "grad_norm": 3.15625,
      "learning_rate": 4.6052287082358556e-05,
      "loss": 0.8382,
      "step": 12960
    },
    {
      "epoch": 0.18381187715149733,
      "grad_norm": 3.140625,
      "learning_rate": 4.605108198443647e-05,
      "loss": 0.8241,
      "step": 12962
    },
    {
      "epoch": 0.18384023880512354,
      "grad_norm": 3.40625,
      "learning_rate": 4.604987671837799e-05,
      "loss": 0.8548,
      "step": 12964
    },
    {
      "epoch": 0.18386860045874975,
      "grad_norm": 3.140625,
      "learning_rate": 4.604867128419273e-05,
      "loss": 0.8871,
      "step": 12966
    },
    {
      "epoch": 0.18389696211237597,
      "grad_norm": 3.125,
      "learning_rate": 4.604746568189034e-05,
      "loss": 0.8224,
      "step": 12968
    },
    {
      "epoch": 0.18392532376600218,
      "grad_norm": 3.6875,
      "learning_rate": 4.6046259911480446e-05,
      "loss": 0.862,
      "step": 12970
    },
    {
      "epoch": 0.1839536854196284,
      "grad_norm": 3.171875,
      "learning_rate": 4.6045053972972654e-05,
      "loss": 0.8243,
      "step": 12972
    },
    {
      "epoch": 0.1839820470732546,
      "grad_norm": 3.15625,
      "learning_rate": 4.604384786637662e-05,
      "loss": 0.813,
      "step": 12974
    },
    {
      "epoch": 0.18401040872688082,
      "grad_norm": 2.96875,
      "learning_rate": 4.604264159170197e-05,
      "loss": 0.8009,
      "step": 12976
    },
    {
      "epoch": 0.18403877038050703,
      "grad_norm": 3.140625,
      "learning_rate": 4.604143514895833e-05,
      "loss": 0.865,
      "step": 12978
    },
    {
      "epoch": 0.18406713203413325,
      "grad_norm": 3.1875,
      "learning_rate": 4.604022853815536e-05,
      "loss": 0.8174,
      "step": 12980
    },
    {
      "epoch": 0.18409549368775946,
      "grad_norm": 2.953125,
      "learning_rate": 4.603902175930267e-05,
      "loss": 0.8475,
      "step": 12982
    },
    {
      "epoch": 0.18412385534138567,
      "grad_norm": 3.515625,
      "learning_rate": 4.6037814812409906e-05,
      "loss": 0.8608,
      "step": 12984
    },
    {
      "epoch": 0.18415221699501189,
      "grad_norm": 2.984375,
      "learning_rate": 4.6036607697486724e-05,
      "loss": 0.8211,
      "step": 12986
    },
    {
      "epoch": 0.1841805786486381,
      "grad_norm": 3.15625,
      "learning_rate": 4.6035400414542745e-05,
      "loss": 0.8025,
      "step": 12988
    },
    {
      "epoch": 0.1842089403022643,
      "grad_norm": 3.109375,
      "learning_rate": 4.603419296358763e-05,
      "loss": 0.8281,
      "step": 12990
    },
    {
      "epoch": 0.18423730195589053,
      "grad_norm": 2.984375,
      "learning_rate": 4.6032985344631e-05,
      "loss": 0.829,
      "step": 12992
    },
    {
      "epoch": 0.18426566360951677,
      "grad_norm": 3.03125,
      "learning_rate": 4.6031777557682514e-05,
      "loss": 0.8284,
      "step": 12994
    },
    {
      "epoch": 0.18429402526314298,
      "grad_norm": 3.078125,
      "learning_rate": 4.6030569602751825e-05,
      "loss": 0.8627,
      "step": 12996
    },
    {
      "epoch": 0.1843223869167692,
      "grad_norm": 3.484375,
      "learning_rate": 4.6029361479848566e-05,
      "loss": 0.835,
      "step": 12998
    },
    {
      "epoch": 0.1843507485703954,
      "grad_norm": 3.203125,
      "learning_rate": 4.60281531889824e-05,
      "loss": 0.8228,
      "step": 13000
    },
    {
      "epoch": 0.18437911022402162,
      "grad_norm": 3.0625,
      "learning_rate": 4.602694473016297e-05,
      "loss": 0.8422,
      "step": 13002
    },
    {
      "epoch": 0.18440747187764783,
      "grad_norm": 2.96875,
      "learning_rate": 4.602573610339993e-05,
      "loss": 0.8327,
      "step": 13004
    },
    {
      "epoch": 0.18443583353127405,
      "grad_norm": 2.859375,
      "learning_rate": 4.602452730870293e-05,
      "loss": 0.8415,
      "step": 13006
    },
    {
      "epoch": 0.18446419518490026,
      "grad_norm": 3.0,
      "learning_rate": 4.6023318346081636e-05,
      "loss": 0.7983,
      "step": 13008
    },
    {
      "epoch": 0.18449255683852647,
      "grad_norm": 3.34375,
      "learning_rate": 4.602210921554569e-05,
      "loss": 0.8707,
      "step": 13010
    },
    {
      "epoch": 0.18452091849215269,
      "grad_norm": 3.265625,
      "learning_rate": 4.602089991710475e-05,
      "loss": 0.8581,
      "step": 13012
    },
    {
      "epoch": 0.1845492801457789,
      "grad_norm": 3.203125,
      "learning_rate": 4.6019690450768493e-05,
      "loss": 0.8545,
      "step": 13014
    },
    {
      "epoch": 0.1845776417994051,
      "grad_norm": 3.40625,
      "learning_rate": 4.601848081654656e-05,
      "loss": 0.8699,
      "step": 13016
    },
    {
      "epoch": 0.18460600345303133,
      "grad_norm": 3.421875,
      "learning_rate": 4.6017271014448625e-05,
      "loss": 0.8446,
      "step": 13018
    },
    {
      "epoch": 0.18463436510665754,
      "grad_norm": 3.203125,
      "learning_rate": 4.6016061044484335e-05,
      "loss": 0.8681,
      "step": 13020
    },
    {
      "epoch": 0.18466272676028375,
      "grad_norm": 3.3125,
      "learning_rate": 4.601485090666337e-05,
      "loss": 0.8604,
      "step": 13022
    },
    {
      "epoch": 0.18469108841390997,
      "grad_norm": 2.796875,
      "learning_rate": 4.601364060099539e-05,
      "loss": 0.854,
      "step": 13024
    },
    {
      "epoch": 0.18471945006753618,
      "grad_norm": 2.953125,
      "learning_rate": 4.6012430127490055e-05,
      "loss": 0.8177,
      "step": 13026
    },
    {
      "epoch": 0.1847478117211624,
      "grad_norm": 3.234375,
      "learning_rate": 4.601121948615704e-05,
      "loss": 0.8287,
      "step": 13028
    },
    {
      "epoch": 0.1847761733747886,
      "grad_norm": 3.15625,
      "learning_rate": 4.6010008677006015e-05,
      "loss": 0.8152,
      "step": 13030
    },
    {
      "epoch": 0.18480453502841482,
      "grad_norm": 3.03125,
      "learning_rate": 4.6008797700046647e-05,
      "loss": 0.8466,
      "step": 13032
    },
    {
      "epoch": 0.18483289668204106,
      "grad_norm": 2.84375,
      "learning_rate": 4.6007586555288615e-05,
      "loss": 0.8034,
      "step": 13034
    },
    {
      "epoch": 0.18486125833566727,
      "grad_norm": 2.734375,
      "learning_rate": 4.600637524274158e-05,
      "loss": 0.8526,
      "step": 13036
    },
    {
      "epoch": 0.18488961998929349,
      "grad_norm": 2.859375,
      "learning_rate": 4.600516376241524e-05,
      "loss": 0.8214,
      "step": 13038
    },
    {
      "epoch": 0.1849179816429197,
      "grad_norm": 3.28125,
      "learning_rate": 4.600395211431925e-05,
      "loss": 0.8014,
      "step": 13040
    },
    {
      "epoch": 0.1849463432965459,
      "grad_norm": 3.109375,
      "learning_rate": 4.600274029846329e-05,
      "loss": 0.8617,
      "step": 13042
    },
    {
      "epoch": 0.18497470495017213,
      "grad_norm": 2.6875,
      "learning_rate": 4.6001528314857034e-05,
      "loss": 0.7872,
      "step": 13044
    },
    {
      "epoch": 0.18500306660379834,
      "grad_norm": 3.109375,
      "learning_rate": 4.600031616351018e-05,
      "loss": 0.8183,
      "step": 13046
    },
    {
      "epoch": 0.18503142825742455,
      "grad_norm": 3.078125,
      "learning_rate": 4.59991038444324e-05,
      "loss": 0.802,
      "step": 13048
    },
    {
      "epoch": 0.18505978991105076,
      "grad_norm": 3.046875,
      "learning_rate": 4.5997891357633374e-05,
      "loss": 0.7783,
      "step": 13050
    },
    {
      "epoch": 0.18508815156467698,
      "grad_norm": 2.96875,
      "learning_rate": 4.5996678703122794e-05,
      "loss": 0.8105,
      "step": 13052
    },
    {
      "epoch": 0.1851165132183032,
      "grad_norm": 3.140625,
      "learning_rate": 4.5995465880910336e-05,
      "loss": 0.8528,
      "step": 13054
    },
    {
      "epoch": 0.1851448748719294,
      "grad_norm": 2.859375,
      "learning_rate": 4.59942528910057e-05,
      "loss": 0.8125,
      "step": 13056
    },
    {
      "epoch": 0.18517323652555562,
      "grad_norm": 2.734375,
      "learning_rate": 4.599303973341856e-05,
      "loss": 0.8686,
      "step": 13058
    },
    {
      "epoch": 0.18520159817918183,
      "grad_norm": 3.1875,
      "learning_rate": 4.5991826408158614e-05,
      "loss": 0.7863,
      "step": 13060
    },
    {
      "epoch": 0.18522995983280804,
      "grad_norm": 3.265625,
      "learning_rate": 4.5990612915235545e-05,
      "loss": 0.8668,
      "step": 13062
    },
    {
      "epoch": 0.18525832148643426,
      "grad_norm": 2.765625,
      "learning_rate": 4.598939925465905e-05,
      "loss": 0.8114,
      "step": 13064
    },
    {
      "epoch": 0.18528668314006047,
      "grad_norm": 3.265625,
      "learning_rate": 4.5988185426438826e-05,
      "loss": 0.8909,
      "step": 13066
    },
    {
      "epoch": 0.18531504479368668,
      "grad_norm": 3.09375,
      "learning_rate": 4.5986971430584556e-05,
      "loss": 0.8368,
      "step": 13068
    },
    {
      "epoch": 0.1853434064473129,
      "grad_norm": 3.375,
      "learning_rate": 4.598575726710596e-05,
      "loss": 0.8269,
      "step": 13070
    },
    {
      "epoch": 0.18537176810093914,
      "grad_norm": 3.421875,
      "learning_rate": 4.5984542936012716e-05,
      "loss": 0.7916,
      "step": 13072
    },
    {
      "epoch": 0.18540012975456535,
      "grad_norm": 2.921875,
      "learning_rate": 4.5983328437314526e-05,
      "loss": 0.8511,
      "step": 13074
    },
    {
      "epoch": 0.18542849140819156,
      "grad_norm": 2.875,
      "learning_rate": 4.5982113771021095e-05,
      "loss": 0.8668,
      "step": 13076
    },
    {
      "epoch": 0.18545685306181778,
      "grad_norm": 3.328125,
      "learning_rate": 4.598089893714212e-05,
      "loss": 0.8231,
      "step": 13078
    },
    {
      "epoch": 0.185485214715444,
      "grad_norm": 3.171875,
      "learning_rate": 4.597968393568731e-05,
      "loss": 0.8385,
      "step": 13080
    },
    {
      "epoch": 0.1855135763690702,
      "grad_norm": 3.25,
      "learning_rate": 4.597846876666635e-05,
      "loss": 0.8211,
      "step": 13082
    },
    {
      "epoch": 0.18554193802269642,
      "grad_norm": 3.140625,
      "learning_rate": 4.597725343008898e-05,
      "loss": 0.8548,
      "step": 13084
    },
    {
      "epoch": 0.18557029967632263,
      "grad_norm": 3.203125,
      "learning_rate": 4.5976037925964874e-05,
      "loss": 0.8151,
      "step": 13086
    },
    {
      "epoch": 0.18559866132994884,
      "grad_norm": 3.140625,
      "learning_rate": 4.597482225430376e-05,
      "loss": 0.8955,
      "step": 13088
    },
    {
      "epoch": 0.18562702298357506,
      "grad_norm": 3.25,
      "learning_rate": 4.597360641511534e-05,
      "loss": 0.8102,
      "step": 13090
    },
    {
      "epoch": 0.18565538463720127,
      "grad_norm": 3.0625,
      "learning_rate": 4.597239040840933e-05,
      "loss": 0.8371,
      "step": 13092
    },
    {
      "epoch": 0.18568374629082748,
      "grad_norm": 3.171875,
      "learning_rate": 4.5971174234195435e-05,
      "loss": 0.8388,
      "step": 13094
    },
    {
      "epoch": 0.1857121079444537,
      "grad_norm": 3.265625,
      "learning_rate": 4.596995789248337e-05,
      "loss": 0.8481,
      "step": 13096
    },
    {
      "epoch": 0.1857404695980799,
      "grad_norm": 2.90625,
      "learning_rate": 4.596874138328287e-05,
      "loss": 0.7828,
      "step": 13098
    },
    {
      "epoch": 0.18576883125170612,
      "grad_norm": 2.890625,
      "learning_rate": 4.5967524706603615e-05,
      "loss": 0.8554,
      "step": 13100
    },
    {
      "epoch": 0.18579719290533234,
      "grad_norm": 3.03125,
      "learning_rate": 4.5966307862455344e-05,
      "loss": 0.8659,
      "step": 13102
    },
    {
      "epoch": 0.18582555455895855,
      "grad_norm": 3.53125,
      "learning_rate": 4.596509085084778e-05,
      "loss": 0.848,
      "step": 13104
    },
    {
      "epoch": 0.18585391621258476,
      "grad_norm": 3.203125,
      "learning_rate": 4.596387367179063e-05,
      "loss": 0.8382,
      "step": 13106
    },
    {
      "epoch": 0.18588227786621098,
      "grad_norm": 2.953125,
      "learning_rate": 4.596265632529363e-05,
      "loss": 0.8609,
      "step": 13108
    },
    {
      "epoch": 0.18591063951983722,
      "grad_norm": 3.3125,
      "learning_rate": 4.5961438811366484e-05,
      "loss": 0.8494,
      "step": 13110
    },
    {
      "epoch": 0.18593900117346343,
      "grad_norm": 3.109375,
      "learning_rate": 4.5960221130018946e-05,
      "loss": 0.8323,
      "step": 13112
    },
    {
      "epoch": 0.18596736282708964,
      "grad_norm": 3.484375,
      "learning_rate": 4.595900328126071e-05,
      "loss": 0.8719,
      "step": 13114
    },
    {
      "epoch": 0.18599572448071586,
      "grad_norm": 3.03125,
      "learning_rate": 4.5957785265101514e-05,
      "loss": 0.8155,
      "step": 13116
    },
    {
      "epoch": 0.18602408613434207,
      "grad_norm": 2.984375,
      "learning_rate": 4.595656708155111e-05,
      "loss": 0.7966,
      "step": 13118
    },
    {
      "epoch": 0.18605244778796828,
      "grad_norm": 2.953125,
      "learning_rate": 4.595534873061919e-05,
      "loss": 0.8396,
      "step": 13120
    },
    {
      "epoch": 0.1860808094415945,
      "grad_norm": 2.640625,
      "learning_rate": 4.59541302123155e-05,
      "loss": 0.7848,
      "step": 13122
    },
    {
      "epoch": 0.1861091710952207,
      "grad_norm": 2.609375,
      "learning_rate": 4.595291152664978e-05,
      "loss": 0.8058,
      "step": 13124
    },
    {
      "epoch": 0.18613753274884692,
      "grad_norm": 3.109375,
      "learning_rate": 4.595169267363175e-05,
      "loss": 0.8243,
      "step": 13126
    },
    {
      "epoch": 0.18616589440247314,
      "grad_norm": 3.09375,
      "learning_rate": 4.5950473653271165e-05,
      "loss": 0.8273,
      "step": 13128
    },
    {
      "epoch": 0.18619425605609935,
      "grad_norm": 2.984375,
      "learning_rate": 4.594925446557774e-05,
      "loss": 0.8088,
      "step": 13130
    },
    {
      "epoch": 0.18622261770972556,
      "grad_norm": 2.96875,
      "learning_rate": 4.5948035110561236e-05,
      "loss": 0.805,
      "step": 13132
    },
    {
      "epoch": 0.18625097936335178,
      "grad_norm": 3.296875,
      "learning_rate": 4.594681558823136e-05,
      "loss": 0.8682,
      "step": 13134
    },
    {
      "epoch": 0.186279341016978,
      "grad_norm": 3.015625,
      "learning_rate": 4.594559589859789e-05,
      "loss": 0.8321,
      "step": 13136
    },
    {
      "epoch": 0.1863077026706042,
      "grad_norm": 2.84375,
      "learning_rate": 4.594437604167053e-05,
      "loss": 0.8459,
      "step": 13138
    },
    {
      "epoch": 0.18633606432423042,
      "grad_norm": 3.484375,
      "learning_rate": 4.5943156017459054e-05,
      "loss": 0.8633,
      "step": 13140
    },
    {
      "epoch": 0.18636442597785663,
      "grad_norm": 3.046875,
      "learning_rate": 4.594193582597319e-05,
      "loss": 0.8347,
      "step": 13142
    },
    {
      "epoch": 0.18639278763148284,
      "grad_norm": 2.953125,
      "learning_rate": 4.594071546722269e-05,
      "loss": 0.8528,
      "step": 13144
    },
    {
      "epoch": 0.18642114928510906,
      "grad_norm": 2.921875,
      "learning_rate": 4.59394949412173e-05,
      "loss": 0.7814,
      "step": 13146
    },
    {
      "epoch": 0.1864495109387353,
      "grad_norm": 3.0625,
      "learning_rate": 4.593827424796676e-05,
      "loss": 0.8238,
      "step": 13148
    },
    {
      "epoch": 0.1864778725923615,
      "grad_norm": 3.203125,
      "learning_rate": 4.5937053387480835e-05,
      "loss": 0.8767,
      "step": 13150
    },
    {
      "epoch": 0.18650623424598772,
      "grad_norm": 2.90625,
      "learning_rate": 4.593583235976926e-05,
      "loss": 0.8338,
      "step": 13152
    },
    {
      "epoch": 0.18653459589961394,
      "grad_norm": 2.984375,
      "learning_rate": 4.59346111648418e-05,
      "loss": 0.8379,
      "step": 13154
    },
    {
      "epoch": 0.18656295755324015,
      "grad_norm": 2.59375,
      "learning_rate": 4.593338980270821e-05,
      "loss": 0.8138,
      "step": 13156
    },
    {
      "epoch": 0.18659131920686636,
      "grad_norm": 2.875,
      "learning_rate": 4.593216827337823e-05,
      "loss": 0.8153,
      "step": 13158
    },
    {
      "epoch": 0.18661968086049258,
      "grad_norm": 3.109375,
      "learning_rate": 4.593094657686162e-05,
      "loss": 0.852,
      "step": 13160
    },
    {
      "epoch": 0.1866480425141188,
      "grad_norm": 3.453125,
      "learning_rate": 4.592972471316815e-05,
      "loss": 0.8778,
      "step": 13162
    },
    {
      "epoch": 0.186676404167745,
      "grad_norm": 2.84375,
      "learning_rate": 4.592850268230758e-05,
      "loss": 0.8269,
      "step": 13164
    },
    {
      "epoch": 0.18670476582137122,
      "grad_norm": 2.953125,
      "learning_rate": 4.592728048428966e-05,
      "loss": 0.8166,
      "step": 13166
    },
    {
      "epoch": 0.18673312747499743,
      "grad_norm": 3.15625,
      "learning_rate": 4.592605811912415e-05,
      "loss": 0.8613,
      "step": 13168
    },
    {
      "epoch": 0.18676148912862364,
      "grad_norm": 3.09375,
      "learning_rate": 4.592483558682083e-05,
      "loss": 0.829,
      "step": 13170
    },
    {
      "epoch": 0.18678985078224986,
      "grad_norm": 2.953125,
      "learning_rate": 4.592361288738945e-05,
      "loss": 0.8649,
      "step": 13172
    },
    {
      "epoch": 0.18681821243587607,
      "grad_norm": 2.921875,
      "learning_rate": 4.592239002083976e-05,
      "loss": 0.8067,
      "step": 13174
    },
    {
      "epoch": 0.18684657408950228,
      "grad_norm": 3.015625,
      "learning_rate": 4.5921166987181566e-05,
      "loss": 0.8051,
      "step": 13176
    },
    {
      "epoch": 0.1868749357431285,
      "grad_norm": 3.296875,
      "learning_rate": 4.59199437864246e-05,
      "loss": 0.8701,
      "step": 13178
    },
    {
      "epoch": 0.1869032973967547,
      "grad_norm": 2.859375,
      "learning_rate": 4.591872041857865e-05,
      "loss": 0.8331,
      "step": 13180
    },
    {
      "epoch": 0.18693165905038092,
      "grad_norm": 2.9375,
      "learning_rate": 4.591749688365349e-05,
      "loss": 0.855,
      "step": 13182
    },
    {
      "epoch": 0.18696002070400713,
      "grad_norm": 3.296875,
      "learning_rate": 4.591627318165889e-05,
      "loss": 0.8416,
      "step": 13184
    },
    {
      "epoch": 0.18698838235763335,
      "grad_norm": 2.734375,
      "learning_rate": 4.591504931260462e-05,
      "loss": 0.8072,
      "step": 13186
    },
    {
      "epoch": 0.1870167440112596,
      "grad_norm": 2.828125,
      "learning_rate": 4.591382527650046e-05,
      "loss": 0.8383,
      "step": 13188
    },
    {
      "epoch": 0.1870451056648858,
      "grad_norm": 2.59375,
      "learning_rate": 4.5912601073356164e-05,
      "loss": 0.8402,
      "step": 13190
    },
    {
      "epoch": 0.18707346731851202,
      "grad_norm": 2.828125,
      "learning_rate": 4.591137670318155e-05,
      "loss": 0.857,
      "step": 13192
    },
    {
      "epoch": 0.18710182897213823,
      "grad_norm": 2.78125,
      "learning_rate": 4.591015216598637e-05,
      "loss": 0.8345,
      "step": 13194
    },
    {
      "epoch": 0.18713019062576444,
      "grad_norm": 3.0,
      "learning_rate": 4.590892746178039e-05,
      "loss": 0.8868,
      "step": 13196
    },
    {
      "epoch": 0.18715855227939066,
      "grad_norm": 3.15625,
      "learning_rate": 4.5907702590573434e-05,
      "loss": 0.8614,
      "step": 13198
    },
    {
      "epoch": 0.18718691393301687,
      "grad_norm": 3.015625,
      "learning_rate": 4.590647755237526e-05,
      "loss": 0.7963,
      "step": 13200
    },
    {
      "epoch": 0.18721527558664308,
      "grad_norm": 2.921875,
      "learning_rate": 4.590525234719565e-05,
      "loss": 0.8295,
      "step": 13202
    },
    {
      "epoch": 0.1872436372402693,
      "grad_norm": 3.140625,
      "learning_rate": 4.5904026975044404e-05,
      "loss": 0.8039,
      "step": 13204
    },
    {
      "epoch": 0.1872719988938955,
      "grad_norm": 3.203125,
      "learning_rate": 4.5902801435931286e-05,
      "loss": 0.8509,
      "step": 13206
    },
    {
      "epoch": 0.18730036054752172,
      "grad_norm": 3.140625,
      "learning_rate": 4.590157572986611e-05,
      "loss": 0.8388,
      "step": 13208
    },
    {
      "epoch": 0.18732872220114793,
      "grad_norm": 3.40625,
      "learning_rate": 4.590034985685865e-05,
      "loss": 0.8909,
      "step": 13210
    },
    {
      "epoch": 0.18735708385477415,
      "grad_norm": 2.9375,
      "learning_rate": 4.58991238169187e-05,
      "loss": 0.8036,
      "step": 13212
    },
    {
      "epoch": 0.18738544550840036,
      "grad_norm": 3.0625,
      "learning_rate": 4.589789761005606e-05,
      "loss": 0.8134,
      "step": 13214
    },
    {
      "epoch": 0.18741380716202657,
      "grad_norm": 3.046875,
      "learning_rate": 4.589667123628051e-05,
      "loss": 0.8743,
      "step": 13216
    },
    {
      "epoch": 0.1874421688156528,
      "grad_norm": 3.46875,
      "learning_rate": 4.5895444695601855e-05,
      "loss": 0.8436,
      "step": 13218
    },
    {
      "epoch": 0.187470530469279,
      "grad_norm": 3.40625,
      "learning_rate": 4.5894217988029894e-05,
      "loss": 0.8654,
      "step": 13220
    },
    {
      "epoch": 0.18749889212290521,
      "grad_norm": 2.984375,
      "learning_rate": 4.589299111357441e-05,
      "loss": 0.8007,
      "step": 13222
    },
    {
      "epoch": 0.18752725377653143,
      "grad_norm": 2.9375,
      "learning_rate": 4.5891764072245224e-05,
      "loss": 0.8077,
      "step": 13224
    },
    {
      "epoch": 0.18755561543015767,
      "grad_norm": 3.265625,
      "learning_rate": 4.5890536864052125e-05,
      "loss": 0.7967,
      "step": 13226
    },
    {
      "epoch": 0.18758397708378388,
      "grad_norm": 2.984375,
      "learning_rate": 4.588930948900491e-05,
      "loss": 0.8254,
      "step": 13228
    },
    {
      "epoch": 0.1876123387374101,
      "grad_norm": 2.71875,
      "learning_rate": 4.5888081947113387e-05,
      "loss": 0.8194,
      "step": 13230
    },
    {
      "epoch": 0.1876407003910363,
      "grad_norm": 3.265625,
      "learning_rate": 4.5886854238387364e-05,
      "loss": 0.819,
      "step": 13232
    },
    {
      "epoch": 0.18766906204466252,
      "grad_norm": 3.125,
      "learning_rate": 4.588562636283664e-05,
      "loss": 0.8626,
      "step": 13234
    },
    {
      "epoch": 0.18769742369828873,
      "grad_norm": 2.90625,
      "learning_rate": 4.5884398320471025e-05,
      "loss": 0.8365,
      "step": 13236
    },
    {
      "epoch": 0.18772578535191495,
      "grad_norm": 3.375,
      "learning_rate": 4.5883170111300325e-05,
      "loss": 0.8916,
      "step": 13238
    },
    {
      "epoch": 0.18775414700554116,
      "grad_norm": 3.0625,
      "learning_rate": 4.588194173533436e-05,
      "loss": 0.8426,
      "step": 13240
    },
    {
      "epoch": 0.18778250865916737,
      "grad_norm": 3.359375,
      "learning_rate": 4.588071319258293e-05,
      "loss": 0.8109,
      "step": 13242
    },
    {
      "epoch": 0.1878108703127936,
      "grad_norm": 3.359375,
      "learning_rate": 4.5879484483055846e-05,
      "loss": 0.8049,
      "step": 13244
    },
    {
      "epoch": 0.1878392319664198,
      "grad_norm": 3.0625,
      "learning_rate": 4.5878255606762925e-05,
      "loss": 0.8489,
      "step": 13246
    },
    {
      "epoch": 0.18786759362004601,
      "grad_norm": 3.015625,
      "learning_rate": 4.587702656371399e-05,
      "loss": 0.864,
      "step": 13248
    },
    {
      "epoch": 0.18789595527367223,
      "grad_norm": 3.203125,
      "learning_rate": 4.587579735391885e-05,
      "loss": 0.8556,
      "step": 13250
    },
    {
      "epoch": 0.18792431692729844,
      "grad_norm": 2.96875,
      "learning_rate": 4.5874567977387326e-05,
      "loss": 0.7831,
      "step": 13252
    },
    {
      "epoch": 0.18795267858092465,
      "grad_norm": 3.21875,
      "learning_rate": 4.587333843412923e-05,
      "loss": 0.8611,
      "step": 13254
    },
    {
      "epoch": 0.18798104023455087,
      "grad_norm": 3.28125,
      "learning_rate": 4.587210872415439e-05,
      "loss": 0.8093,
      "step": 13256
    },
    {
      "epoch": 0.18800940188817708,
      "grad_norm": 3.03125,
      "learning_rate": 4.587087884747263e-05,
      "loss": 0.8137,
      "step": 13258
    },
    {
      "epoch": 0.1880377635418033,
      "grad_norm": 2.9375,
      "learning_rate": 4.586964880409377e-05,
      "loss": 0.8307,
      "step": 13260
    },
    {
      "epoch": 0.1880661251954295,
      "grad_norm": 3.453125,
      "learning_rate": 4.586841859402763e-05,
      "loss": 0.8623,
      "step": 13262
    },
    {
      "epoch": 0.18809448684905575,
      "grad_norm": 3.3125,
      "learning_rate": 4.586718821728403e-05,
      "loss": 0.8325,
      "step": 13264
    },
    {
      "epoch": 0.18812284850268196,
      "grad_norm": 2.9375,
      "learning_rate": 4.5865957673872816e-05,
      "loss": 0.8541,
      "step": 13266
    },
    {
      "epoch": 0.18815121015630817,
      "grad_norm": 3.546875,
      "learning_rate": 4.5864726963803804e-05,
      "loss": 0.8937,
      "step": 13268
    },
    {
      "epoch": 0.1881795718099344,
      "grad_norm": 2.84375,
      "learning_rate": 4.586349608708682e-05,
      "loss": 0.7934,
      "step": 13270
    },
    {
      "epoch": 0.1882079334635606,
      "grad_norm": 3.21875,
      "learning_rate": 4.586226504373171e-05,
      "loss": 0.8646,
      "step": 13272
    },
    {
      "epoch": 0.1882362951171868,
      "grad_norm": 2.6875,
      "learning_rate": 4.5861033833748296e-05,
      "loss": 0.8557,
      "step": 13274
    },
    {
      "epoch": 0.18826465677081303,
      "grad_norm": 2.8125,
      "learning_rate": 4.585980245714641e-05,
      "loss": 0.8433,
      "step": 13276
    },
    {
      "epoch": 0.18829301842443924,
      "grad_norm": 2.84375,
      "learning_rate": 4.585857091393589e-05,
      "loss": 0.886,
      "step": 13278
    },
    {
      "epoch": 0.18832138007806545,
      "grad_norm": 3.140625,
      "learning_rate": 4.585733920412658e-05,
      "loss": 0.7842,
      "step": 13280
    },
    {
      "epoch": 0.18834974173169167,
      "grad_norm": 2.890625,
      "learning_rate": 4.5856107327728305e-05,
      "loss": 0.8127,
      "step": 13282
    },
    {
      "epoch": 0.18837810338531788,
      "grad_norm": 3.28125,
      "learning_rate": 4.585487528475091e-05,
      "loss": 0.8308,
      "step": 13284
    },
    {
      "epoch": 0.1884064650389441,
      "grad_norm": 3.09375,
      "learning_rate": 4.5853643075204235e-05,
      "loss": 0.8094,
      "step": 13286
    },
    {
      "epoch": 0.1884348266925703,
      "grad_norm": 3.296875,
      "learning_rate": 4.585241069909812e-05,
      "loss": 0.8379,
      "step": 13288
    },
    {
      "epoch": 0.18846318834619652,
      "grad_norm": 3.0625,
      "learning_rate": 4.585117815644241e-05,
      "loss": 0.8627,
      "step": 13290
    },
    {
      "epoch": 0.18849154999982273,
      "grad_norm": 2.984375,
      "learning_rate": 4.584994544724695e-05,
      "loss": 0.8446,
      "step": 13292
    },
    {
      "epoch": 0.18851991165344895,
      "grad_norm": 2.9375,
      "learning_rate": 4.584871257152158e-05,
      "loss": 0.8755,
      "step": 13294
    },
    {
      "epoch": 0.18854827330707516,
      "grad_norm": 2.90625,
      "learning_rate": 4.584747952927617e-05,
      "loss": 0.8244,
      "step": 13296
    },
    {
      "epoch": 0.18857663496070137,
      "grad_norm": 3.109375,
      "learning_rate": 4.5846246320520534e-05,
      "loss": 0.8497,
      "step": 13298
    },
    {
      "epoch": 0.18860499661432759,
      "grad_norm": 2.9375,
      "learning_rate": 4.584501294526454e-05,
      "loss": 0.8321,
      "step": 13300
    },
    {
      "epoch": 0.18863335826795383,
      "grad_norm": 2.765625,
      "learning_rate": 4.584377940351804e-05,
      "loss": 0.8601,
      "step": 13302
    },
    {
      "epoch": 0.18866171992158004,
      "grad_norm": 3.09375,
      "learning_rate": 4.584254569529088e-05,
      "loss": 0.8426,
      "step": 13304
    },
    {
      "epoch": 0.18869008157520625,
      "grad_norm": 3.0,
      "learning_rate": 4.584131182059292e-05,
      "loss": 0.8461,
      "step": 13306
    },
    {
      "epoch": 0.18871844322883247,
      "grad_norm": 3.0,
      "learning_rate": 4.584007777943402e-05,
      "loss": 0.832,
      "step": 13308
    },
    {
      "epoch": 0.18874680488245868,
      "grad_norm": 2.984375,
      "learning_rate": 4.583884357182402e-05,
      "loss": 0.8127,
      "step": 13310
    },
    {
      "epoch": 0.1887751665360849,
      "grad_norm": 3.203125,
      "learning_rate": 4.583760919777279e-05,
      "loss": 0.85,
      "step": 13312
    },
    {
      "epoch": 0.1888035281897111,
      "grad_norm": 3.15625,
      "learning_rate": 4.5836374657290184e-05,
      "loss": 0.7969,
      "step": 13314
    },
    {
      "epoch": 0.18883188984333732,
      "grad_norm": 2.921875,
      "learning_rate": 4.583513995038606e-05,
      "loss": 0.8127,
      "step": 13316
    },
    {
      "epoch": 0.18886025149696353,
      "grad_norm": 2.90625,
      "learning_rate": 4.58339050770703e-05,
      "loss": 0.7802,
      "step": 13318
    },
    {
      "epoch": 0.18888861315058975,
      "grad_norm": 2.921875,
      "learning_rate": 4.583267003735273e-05,
      "loss": 0.85,
      "step": 13320
    },
    {
      "epoch": 0.18891697480421596,
      "grad_norm": 3.453125,
      "learning_rate": 4.583143483124324e-05,
      "loss": 0.9022,
      "step": 13322
    },
    {
      "epoch": 0.18894533645784217,
      "grad_norm": 3.171875,
      "learning_rate": 4.583019945875169e-05,
      "loss": 0.8715,
      "step": 13324
    },
    {
      "epoch": 0.18897369811146839,
      "grad_norm": 3.109375,
      "learning_rate": 4.5828963919887955e-05,
      "loss": 0.7876,
      "step": 13326
    },
    {
      "epoch": 0.1890020597650946,
      "grad_norm": 3.3125,
      "learning_rate": 4.582772821466189e-05,
      "loss": 0.7808,
      "step": 13328
    },
    {
      "epoch": 0.1890304214187208,
      "grad_norm": 3.15625,
      "learning_rate": 4.5826492343083374e-05,
      "loss": 0.9097,
      "step": 13330
    },
    {
      "epoch": 0.18905878307234703,
      "grad_norm": 3.0,
      "learning_rate": 4.582525630516227e-05,
      "loss": 0.7945,
      "step": 13332
    },
    {
      "epoch": 0.18908714472597324,
      "grad_norm": 2.9375,
      "learning_rate": 4.582402010090845e-05,
      "loss": 0.8638,
      "step": 13334
    },
    {
      "epoch": 0.18911550637959945,
      "grad_norm": 2.984375,
      "learning_rate": 4.582278373033181e-05,
      "loss": 0.8251,
      "step": 13336
    },
    {
      "epoch": 0.18914386803322566,
      "grad_norm": 3.3125,
      "learning_rate": 4.5821547193442196e-05,
      "loss": 0.9171,
      "step": 13338
    },
    {
      "epoch": 0.18917222968685188,
      "grad_norm": 3.0,
      "learning_rate": 4.5820310490249497e-05,
      "loss": 0.7947,
      "step": 13340
    },
    {
      "epoch": 0.18920059134047812,
      "grad_norm": 3.15625,
      "learning_rate": 4.581907362076359e-05,
      "loss": 0.8604,
      "step": 13342
    },
    {
      "epoch": 0.18922895299410433,
      "grad_norm": 3.28125,
      "learning_rate": 4.581783658499435e-05,
      "loss": 0.8572,
      "step": 13344
    },
    {
      "epoch": 0.18925731464773055,
      "grad_norm": 2.90625,
      "learning_rate": 4.581659938295166e-05,
      "loss": 0.8633,
      "step": 13346
    },
    {
      "epoch": 0.18928567630135676,
      "grad_norm": 3.21875,
      "learning_rate": 4.581536201464541e-05,
      "loss": 0.8596,
      "step": 13348
    },
    {
      "epoch": 0.18931403795498297,
      "grad_norm": 2.890625,
      "learning_rate": 4.5814124480085465e-05,
      "loss": 0.8283,
      "step": 13350
    },
    {
      "epoch": 0.18934239960860919,
      "grad_norm": 3.5625,
      "learning_rate": 4.581288677928173e-05,
      "loss": 0.8768,
      "step": 13352
    },
    {
      "epoch": 0.1893707612622354,
      "grad_norm": 3.40625,
      "learning_rate": 4.5811648912244074e-05,
      "loss": 0.859,
      "step": 13354
    },
    {
      "epoch": 0.1893991229158616,
      "grad_norm": 3.40625,
      "learning_rate": 4.581041087898239e-05,
      "loss": 0.8265,
      "step": 13356
    },
    {
      "epoch": 0.18942748456948783,
      "grad_norm": 3.0625,
      "learning_rate": 4.5809172679506566e-05,
      "loss": 0.8739,
      "step": 13358
    },
    {
      "epoch": 0.18945584622311404,
      "grad_norm": 3.046875,
      "learning_rate": 4.580793431382649e-05,
      "loss": 0.8043,
      "step": 13360
    },
    {
      "epoch": 0.18948420787674025,
      "grad_norm": 3.375,
      "learning_rate": 4.580669578195206e-05,
      "loss": 0.8729,
      "step": 13362
    },
    {
      "epoch": 0.18951256953036646,
      "grad_norm": 2.828125,
      "learning_rate": 4.580545708389317e-05,
      "loss": 0.8522,
      "step": 13364
    },
    {
      "epoch": 0.18954093118399268,
      "grad_norm": 3.0625,
      "learning_rate": 4.580421821965969e-05,
      "loss": 0.8505,
      "step": 13366
    },
    {
      "epoch": 0.1895692928376189,
      "grad_norm": 3.109375,
      "learning_rate": 4.580297918926154e-05,
      "loss": 0.841,
      "step": 13368
    },
    {
      "epoch": 0.1895976544912451,
      "grad_norm": 3.21875,
      "learning_rate": 4.5801739992708607e-05,
      "loss": 0.8498,
      "step": 13370
    },
    {
      "epoch": 0.18962601614487132,
      "grad_norm": 3.046875,
      "learning_rate": 4.580050063001079e-05,
      "loss": 0.834,
      "step": 13372
    },
    {
      "epoch": 0.18965437779849753,
      "grad_norm": 3.09375,
      "learning_rate": 4.5799261101177984e-05,
      "loss": 0.8731,
      "step": 13374
    },
    {
      "epoch": 0.18968273945212374,
      "grad_norm": 2.765625,
      "learning_rate": 4.579802140622009e-05,
      "loss": 0.8383,
      "step": 13376
    },
    {
      "epoch": 0.18971110110574996,
      "grad_norm": 3.078125,
      "learning_rate": 4.579678154514702e-05,
      "loss": 0.8743,
      "step": 13378
    },
    {
      "epoch": 0.1897394627593762,
      "grad_norm": 3.125,
      "learning_rate": 4.579554151796867e-05,
      "loss": 0.837,
      "step": 13380
    },
    {
      "epoch": 0.1897678244130024,
      "grad_norm": 3.296875,
      "learning_rate": 4.5794301324694934e-05,
      "loss": 0.8268,
      "step": 13382
    },
    {
      "epoch": 0.18979618606662862,
      "grad_norm": 3.09375,
      "learning_rate": 4.579306096533573e-05,
      "loss": 0.823,
      "step": 13384
    },
    {
      "epoch": 0.18982454772025484,
      "grad_norm": 3.078125,
      "learning_rate": 4.579182043990096e-05,
      "loss": 0.8011,
      "step": 13386
    },
    {
      "epoch": 0.18985290937388105,
      "grad_norm": 3.75,
      "learning_rate": 4.579057974840054e-05,
      "loss": 0.8512,
      "step": 13388
    },
    {
      "epoch": 0.18988127102750726,
      "grad_norm": 3.109375,
      "learning_rate": 4.578933889084436e-05,
      "loss": 0.8128,
      "step": 13390
    },
    {
      "epoch": 0.18990963268113348,
      "grad_norm": 2.875,
      "learning_rate": 4.5788097867242355e-05,
      "loss": 0.828,
      "step": 13392
    },
    {
      "epoch": 0.1899379943347597,
      "grad_norm": 3.125,
      "learning_rate": 4.5786856677604416e-05,
      "loss": 0.8378,
      "step": 13394
    },
    {
      "epoch": 0.1899663559883859,
      "grad_norm": 2.8125,
      "learning_rate": 4.578561532194048e-05,
      "loss": 0.8487,
      "step": 13396
    },
    {
      "epoch": 0.18999471764201212,
      "grad_norm": 3.09375,
      "learning_rate": 4.578437380026043e-05,
      "loss": 0.7965,
      "step": 13398
    },
    {
      "epoch": 0.19002307929563833,
      "grad_norm": 3.0625,
      "learning_rate": 4.578313211257421e-05,
      "loss": 0.8712,
      "step": 13400
    },
    {
      "epoch": 0.19005144094926454,
      "grad_norm": 3.125,
      "learning_rate": 4.578189025889173e-05,
      "loss": 0.8758,
      "step": 13402
    },
    {
      "epoch": 0.19007980260289076,
      "grad_norm": 2.671875,
      "learning_rate": 4.57806482392229e-05,
      "loss": 0.835,
      "step": 13404
    },
    {
      "epoch": 0.19010816425651697,
      "grad_norm": 3.015625,
      "learning_rate": 4.5779406053577653e-05,
      "loss": 0.8259,
      "step": 13406
    },
    {
      "epoch": 0.19013652591014318,
      "grad_norm": 3.59375,
      "learning_rate": 4.5778163701965904e-05,
      "loss": 0.8892,
      "step": 13408
    },
    {
      "epoch": 0.1901648875637694,
      "grad_norm": 3.03125,
      "learning_rate": 4.577692118439757e-05,
      "loss": 0.8482,
      "step": 13410
    },
    {
      "epoch": 0.1901932492173956,
      "grad_norm": 3.078125,
      "learning_rate": 4.577567850088258e-05,
      "loss": 0.8033,
      "step": 13412
    },
    {
      "epoch": 0.19022161087102182,
      "grad_norm": 3.765625,
      "learning_rate": 4.577443565143087e-05,
      "loss": 0.8294,
      "step": 13414
    },
    {
      "epoch": 0.19024997252464804,
      "grad_norm": 3.078125,
      "learning_rate": 4.577319263605235e-05,
      "loss": 0.8551,
      "step": 13416
    },
    {
      "epoch": 0.19027833417827428,
      "grad_norm": 2.875,
      "learning_rate": 4.5771949454756955e-05,
      "loss": 0.855,
      "step": 13418
    },
    {
      "epoch": 0.1903066958319005,
      "grad_norm": 3.125,
      "learning_rate": 4.577070610755461e-05,
      "loss": 0.8671,
      "step": 13420
    },
    {
      "epoch": 0.1903350574855267,
      "grad_norm": 2.890625,
      "learning_rate": 4.5769462594455256e-05,
      "loss": 0.9328,
      "step": 13422
    },
    {
      "epoch": 0.19036341913915292,
      "grad_norm": 3.203125,
      "learning_rate": 4.576821891546882e-05,
      "loss": 0.8645,
      "step": 13424
    },
    {
      "epoch": 0.19039178079277913,
      "grad_norm": 3.0,
      "learning_rate": 4.5766975070605227e-05,
      "loss": 0.838,
      "step": 13426
    },
    {
      "epoch": 0.19042014244640534,
      "grad_norm": 3.21875,
      "learning_rate": 4.576573105987442e-05,
      "loss": 0.8534,
      "step": 13428
    },
    {
      "epoch": 0.19044850410003156,
      "grad_norm": 3.015625,
      "learning_rate": 4.576448688328634e-05,
      "loss": 0.8261,
      "step": 13430
    },
    {
      "epoch": 0.19047686575365777,
      "grad_norm": 3.234375,
      "learning_rate": 4.576324254085092e-05,
      "loss": 0.8201,
      "step": 13432
    },
    {
      "epoch": 0.19050522740728398,
      "grad_norm": 3.421875,
      "learning_rate": 4.576199803257809e-05,
      "loss": 0.8686,
      "step": 13434
    },
    {
      "epoch": 0.1905335890609102,
      "grad_norm": 3.25,
      "learning_rate": 4.5760753358477796e-05,
      "loss": 0.8248,
      "step": 13436
    },
    {
      "epoch": 0.1905619507145364,
      "grad_norm": 2.859375,
      "learning_rate": 4.5759508518559984e-05,
      "loss": 0.8624,
      "step": 13438
    },
    {
      "epoch": 0.19059031236816262,
      "grad_norm": 2.984375,
      "learning_rate": 4.575826351283459e-05,
      "loss": 0.794,
      "step": 13440
    },
    {
      "epoch": 0.19061867402178884,
      "grad_norm": 3.21875,
      "learning_rate": 4.5757018341311565e-05,
      "loss": 0.8272,
      "step": 13442
    },
    {
      "epoch": 0.19064703567541505,
      "grad_norm": 3.265625,
      "learning_rate": 4.575577300400085e-05,
      "loss": 0.8793,
      "step": 13444
    },
    {
      "epoch": 0.19067539732904126,
      "grad_norm": 2.828125,
      "learning_rate": 4.5754527500912386e-05,
      "loss": 0.8291,
      "step": 13446
    },
    {
      "epoch": 0.19070375898266748,
      "grad_norm": 2.640625,
      "learning_rate": 4.5753281832056127e-05,
      "loss": 0.8333,
      "step": 13448
    },
    {
      "epoch": 0.1907321206362937,
      "grad_norm": 3.109375,
      "learning_rate": 4.5752035997442025e-05,
      "loss": 0.8739,
      "step": 13450
    },
    {
      "epoch": 0.1907604822899199,
      "grad_norm": 3.296875,
      "learning_rate": 4.5750789997080035e-05,
      "loss": 0.7886,
      "step": 13452
    },
    {
      "epoch": 0.19078884394354612,
      "grad_norm": 3.15625,
      "learning_rate": 4.574954383098009e-05,
      "loss": 0.8612,
      "step": 13454
    },
    {
      "epoch": 0.19081720559717236,
      "grad_norm": 2.984375,
      "learning_rate": 4.574829749915216e-05,
      "loss": 0.8282,
      "step": 13456
    },
    {
      "epoch": 0.19084556725079857,
      "grad_norm": 3.171875,
      "learning_rate": 4.574705100160619e-05,
      "loss": 0.885,
      "step": 13458
    },
    {
      "epoch": 0.19087392890442478,
      "grad_norm": 3.03125,
      "learning_rate": 4.574580433835214e-05,
      "loss": 0.7724,
      "step": 13460
    },
    {
      "epoch": 0.190902290558051,
      "grad_norm": 2.84375,
      "learning_rate": 4.574455750939997e-05,
      "loss": 0.8258,
      "step": 13462
    },
    {
      "epoch": 0.1909306522116772,
      "grad_norm": 2.65625,
      "learning_rate": 4.574331051475963e-05,
      "loss": 0.7976,
      "step": 13464
    },
    {
      "epoch": 0.19095901386530342,
      "grad_norm": 2.9375,
      "learning_rate": 4.574206335444109e-05,
      "loss": 0.8289,
      "step": 13466
    },
    {
      "epoch": 0.19098737551892964,
      "grad_norm": 3.109375,
      "learning_rate": 4.57408160284543e-05,
      "loss": 0.8247,
      "step": 13468
    },
    {
      "epoch": 0.19101573717255585,
      "grad_norm": 3.171875,
      "learning_rate": 4.573956853680924e-05,
      "loss": 0.8351,
      "step": 13470
    },
    {
      "epoch": 0.19104409882618206,
      "grad_norm": 3.078125,
      "learning_rate": 4.573832087951586e-05,
      "loss": 0.8374,
      "step": 13472
    },
    {
      "epoch": 0.19107246047980828,
      "grad_norm": 3.15625,
      "learning_rate": 4.5737073056584124e-05,
      "loss": 0.8595,
      "step": 13474
    },
    {
      "epoch": 0.1911008221334345,
      "grad_norm": 3.09375,
      "learning_rate": 4.5735825068024e-05,
      "loss": 0.7851,
      "step": 13476
    },
    {
      "epoch": 0.1911291837870607,
      "grad_norm": 3.140625,
      "learning_rate": 4.5734576913845465e-05,
      "loss": 0.8746,
      "step": 13478
    },
    {
      "epoch": 0.19115754544068692,
      "grad_norm": 3.234375,
      "learning_rate": 4.5733328594058475e-05,
      "loss": 0.7864,
      "step": 13480
    },
    {
      "epoch": 0.19118590709431313,
      "grad_norm": 3.59375,
      "learning_rate": 4.5732080108673007e-05,
      "loss": 0.8418,
      "step": 13482
    },
    {
      "epoch": 0.19121426874793934,
      "grad_norm": 3.34375,
      "learning_rate": 4.573083145769904e-05,
      "loss": 0.8358,
      "step": 13484
    },
    {
      "epoch": 0.19124263040156556,
      "grad_norm": 3.265625,
      "learning_rate": 4.572958264114653e-05,
      "loss": 0.8298,
      "step": 13486
    },
    {
      "epoch": 0.19127099205519177,
      "grad_norm": 3.453125,
      "learning_rate": 4.572833365902547e-05,
      "loss": 0.8836,
      "step": 13488
    },
    {
      "epoch": 0.19129935370881798,
      "grad_norm": 3.0625,
      "learning_rate": 4.572708451134582e-05,
      "loss": 0.8458,
      "step": 13490
    },
    {
      "epoch": 0.1913277153624442,
      "grad_norm": 3.28125,
      "learning_rate": 4.572583519811756e-05,
      "loss": 0.9055,
      "step": 13492
    },
    {
      "epoch": 0.1913560770160704,
      "grad_norm": 2.90625,
      "learning_rate": 4.572458571935068e-05,
      "loss": 0.771,
      "step": 13494
    },
    {
      "epoch": 0.19138443866969665,
      "grad_norm": 2.6875,
      "learning_rate": 4.5723336075055144e-05,
      "loss": 0.8415,
      "step": 13496
    },
    {
      "epoch": 0.19141280032332286,
      "grad_norm": 3.21875,
      "learning_rate": 4.572208626524095e-05,
      "loss": 0.8388,
      "step": 13498
    },
    {
      "epoch": 0.19144116197694908,
      "grad_norm": 3.578125,
      "learning_rate": 4.572083628991807e-05,
      "loss": 0.8264,
      "step": 13500
    },
    {
      "epoch": 0.1914695236305753,
      "grad_norm": 3.265625,
      "learning_rate": 4.571958614909648e-05,
      "loss": 0.8362,
      "step": 13502
    },
    {
      "epoch": 0.1914978852842015,
      "grad_norm": 3.140625,
      "learning_rate": 4.5718335842786176e-05,
      "loss": 0.7967,
      "step": 13504
    },
    {
      "epoch": 0.19152624693782772,
      "grad_norm": 3.109375,
      "learning_rate": 4.571708537099714e-05,
      "loss": 0.8293,
      "step": 13506
    },
    {
      "epoch": 0.19155460859145393,
      "grad_norm": 4.28125,
      "learning_rate": 4.571583473373937e-05,
      "loss": 0.8175,
      "step": 13508
    },
    {
      "epoch": 0.19158297024508014,
      "grad_norm": 3.203125,
      "learning_rate": 4.5714583931022835e-05,
      "loss": 0.8071,
      "step": 13510
    },
    {
      "epoch": 0.19161133189870636,
      "grad_norm": 3.25,
      "learning_rate": 4.571333296285755e-05,
      "loss": 0.8241,
      "step": 13512
    },
    {
      "epoch": 0.19163969355233257,
      "grad_norm": 2.984375,
      "learning_rate": 4.571208182925348e-05,
      "loss": 0.802,
      "step": 13514
    },
    {
      "epoch": 0.19166805520595878,
      "grad_norm": 3.125,
      "learning_rate": 4.571083053022064e-05,
      "loss": 0.8576,
      "step": 13516
    },
    {
      "epoch": 0.191696416859585,
      "grad_norm": 3.109375,
      "learning_rate": 4.5709579065769e-05,
      "loss": 0.8417,
      "step": 13518
    },
    {
      "epoch": 0.1917247785132112,
      "grad_norm": 3.578125,
      "learning_rate": 4.570832743590859e-05,
      "loss": 0.9205,
      "step": 13520
    },
    {
      "epoch": 0.19175314016683742,
      "grad_norm": 3.1875,
      "learning_rate": 4.570707564064938e-05,
      "loss": 0.8363,
      "step": 13522
    },
    {
      "epoch": 0.19178150182046363,
      "grad_norm": 3.28125,
      "learning_rate": 4.570582368000137e-05,
      "loss": 0.798,
      "step": 13524
    },
    {
      "epoch": 0.19180986347408985,
      "grad_norm": 3.1875,
      "learning_rate": 4.5704571553974574e-05,
      "loss": 0.8799,
      "step": 13526
    },
    {
      "epoch": 0.19183822512771606,
      "grad_norm": 2.9375,
      "learning_rate": 4.570331926257898e-05,
      "loss": 0.7609,
      "step": 13528
    },
    {
      "epoch": 0.19186658678134227,
      "grad_norm": 3.171875,
      "learning_rate": 4.5702066805824596e-05,
      "loss": 0.8474,
      "step": 13530
    },
    {
      "epoch": 0.1918949484349685,
      "grad_norm": 2.921875,
      "learning_rate": 4.570081418372142e-05,
      "loss": 0.8562,
      "step": 13532
    },
    {
      "epoch": 0.19192331008859473,
      "grad_norm": 2.875,
      "learning_rate": 4.5699561396279464e-05,
      "loss": 0.8165,
      "step": 13534
    },
    {
      "epoch": 0.19195167174222094,
      "grad_norm": 2.9375,
      "learning_rate": 4.569830844350873e-05,
      "loss": 0.8489,
      "step": 13536
    },
    {
      "epoch": 0.19198003339584715,
      "grad_norm": 3.515625,
      "learning_rate": 4.569705532541923e-05,
      "loss": 0.8313,
      "step": 13538
    },
    {
      "epoch": 0.19200839504947337,
      "grad_norm": 2.828125,
      "learning_rate": 4.5695802042020955e-05,
      "loss": 0.812,
      "step": 13540
    },
    {
      "epoch": 0.19203675670309958,
      "grad_norm": 3.25,
      "learning_rate": 4.569454859332394e-05,
      "loss": 0.839,
      "step": 13542
    },
    {
      "epoch": 0.1920651183567258,
      "grad_norm": 2.953125,
      "learning_rate": 4.5693294979338174e-05,
      "loss": 0.8769,
      "step": 13544
    },
    {
      "epoch": 0.192093480010352,
      "grad_norm": 3.09375,
      "learning_rate": 4.5692041200073696e-05,
      "loss": 0.8314,
      "step": 13546
    },
    {
      "epoch": 0.19212184166397822,
      "grad_norm": 3.203125,
      "learning_rate": 4.569078725554049e-05,
      "loss": 0.8544,
      "step": 13548
    },
    {
      "epoch": 0.19215020331760443,
      "grad_norm": 3.078125,
      "learning_rate": 4.56895331457486e-05,
      "loss": 0.8675,
      "step": 13550
    },
    {
      "epoch": 0.19217856497123065,
      "grad_norm": 3.09375,
      "learning_rate": 4.568827887070802e-05,
      "loss": 0.8087,
      "step": 13552
    },
    {
      "epoch": 0.19220692662485686,
      "grad_norm": 3.234375,
      "learning_rate": 4.568702443042878e-05,
      "loss": 0.82,
      "step": 13554
    },
    {
      "epoch": 0.19223528827848307,
      "grad_norm": 2.90625,
      "learning_rate": 4.56857698249209e-05,
      "loss": 0.8295,
      "step": 13556
    },
    {
      "epoch": 0.1922636499321093,
      "grad_norm": 3.1875,
      "learning_rate": 4.568451505419439e-05,
      "loss": 0.8292,
      "step": 13558
    },
    {
      "epoch": 0.1922920115857355,
      "grad_norm": 3.296875,
      "learning_rate": 4.568326011825927e-05,
      "loss": 0.8522,
      "step": 13560
    },
    {
      "epoch": 0.1923203732393617,
      "grad_norm": 3.28125,
      "learning_rate": 4.5682005017125584e-05,
      "loss": 0.8151,
      "step": 13562
    },
    {
      "epoch": 0.19234873489298793,
      "grad_norm": 3.15625,
      "learning_rate": 4.568074975080334e-05,
      "loss": 0.8472,
      "step": 13564
    },
    {
      "epoch": 0.19237709654661414,
      "grad_norm": 2.921875,
      "learning_rate": 4.5679494319302573e-05,
      "loss": 0.8505,
      "step": 13566
    },
    {
      "epoch": 0.19240545820024035,
      "grad_norm": 3.21875,
      "learning_rate": 4.567823872263331e-05,
      "loss": 0.7962,
      "step": 13568
    },
    {
      "epoch": 0.19243381985386657,
      "grad_norm": 2.640625,
      "learning_rate": 4.567698296080557e-05,
      "loss": 0.8168,
      "step": 13570
    },
    {
      "epoch": 0.1924621815074928,
      "grad_norm": 3.3125,
      "learning_rate": 4.5675727033829386e-05,
      "loss": 0.8491,
      "step": 13572
    },
    {
      "epoch": 0.19249054316111902,
      "grad_norm": 3.046875,
      "learning_rate": 4.567447094171479e-05,
      "loss": 0.8174,
      "step": 13574
    },
    {
      "epoch": 0.19251890481474523,
      "grad_norm": 3.28125,
      "learning_rate": 4.5673214684471814e-05,
      "loss": 0.869,
      "step": 13576
    },
    {
      "epoch": 0.19254726646837145,
      "grad_norm": 2.96875,
      "learning_rate": 4.56719582621105e-05,
      "loss": 0.868,
      "step": 13578
    },
    {
      "epoch": 0.19257562812199766,
      "grad_norm": 2.78125,
      "learning_rate": 4.567070167464087e-05,
      "loss": 0.8353,
      "step": 13580
    },
    {
      "epoch": 0.19260398977562387,
      "grad_norm": 3.5625,
      "learning_rate": 4.5669444922072965e-05,
      "loss": 0.848,
      "step": 13582
    },
    {
      "epoch": 0.1926323514292501,
      "grad_norm": 3.484375,
      "learning_rate": 4.5668188004416834e-05,
      "loss": 0.8841,
      "step": 13584
    },
    {
      "epoch": 0.1926607130828763,
      "grad_norm": 3.4375,
      "learning_rate": 4.5666930921682507e-05,
      "loss": 0.8356,
      "step": 13586
    },
    {
      "epoch": 0.1926890747365025,
      "grad_norm": 3.03125,
      "learning_rate": 4.566567367388002e-05,
      "loss": 0.8326,
      "step": 13588
    },
    {
      "epoch": 0.19271743639012873,
      "grad_norm": 3.171875,
      "learning_rate": 4.5664416261019416e-05,
      "loss": 0.8111,
      "step": 13590
    },
    {
      "epoch": 0.19274579804375494,
      "grad_norm": 2.984375,
      "learning_rate": 4.566315868311074e-05,
      "loss": 0.867,
      "step": 13592
    },
    {
      "epoch": 0.19277415969738115,
      "grad_norm": 2.96875,
      "learning_rate": 4.566190094016404e-05,
      "loss": 0.8006,
      "step": 13594
    },
    {
      "epoch": 0.19280252135100737,
      "grad_norm": 2.859375,
      "learning_rate": 4.5660643032189365e-05,
      "loss": 0.8367,
      "step": 13596
    },
    {
      "epoch": 0.19283088300463358,
      "grad_norm": 3.1875,
      "learning_rate": 4.565938495919675e-05,
      "loss": 0.7735,
      "step": 13598
    },
    {
      "epoch": 0.1928592446582598,
      "grad_norm": 3.375,
      "learning_rate": 4.565812672119625e-05,
      "loss": 0.9016,
      "step": 13600
    },
    {
      "epoch": 0.192887606311886,
      "grad_norm": 2.90625,
      "learning_rate": 4.5656868318197914e-05,
      "loss": 0.8102,
      "step": 13602
    },
    {
      "epoch": 0.19291596796551222,
      "grad_norm": 3.296875,
      "learning_rate": 4.565560975021179e-05,
      "loss": 0.8075,
      "step": 13604
    },
    {
      "epoch": 0.19294432961913843,
      "grad_norm": 2.96875,
      "learning_rate": 4.565435101724794e-05,
      "loss": 0.8305,
      "step": 13606
    },
    {
      "epoch": 0.19297269127276465,
      "grad_norm": 3.109375,
      "learning_rate": 4.565309211931641e-05,
      "loss": 0.8243,
      "step": 13608
    },
    {
      "epoch": 0.1930010529263909,
      "grad_norm": 3.15625,
      "learning_rate": 4.5651833056427254e-05,
      "loss": 0.8478,
      "step": 13610
    },
    {
      "epoch": 0.1930294145800171,
      "grad_norm": 2.765625,
      "learning_rate": 4.5650573828590525e-05,
      "loss": 0.8312,
      "step": 13612
    },
    {
      "epoch": 0.1930577762336433,
      "grad_norm": 3.015625,
      "learning_rate": 4.5649314435816295e-05,
      "loss": 0.9103,
      "step": 13614
    },
    {
      "epoch": 0.19308613788726953,
      "grad_norm": 3.1875,
      "learning_rate": 4.5648054878114606e-05,
      "loss": 0.8661,
      "step": 13616
    },
    {
      "epoch": 0.19311449954089574,
      "grad_norm": 3.203125,
      "learning_rate": 4.564679515549553e-05,
      "loss": 0.8761,
      "step": 13618
    },
    {
      "epoch": 0.19314286119452195,
      "grad_norm": 3.109375,
      "learning_rate": 4.5645535267969116e-05,
      "loss": 0.7999,
      "step": 13620
    },
    {
      "epoch": 0.19317122284814817,
      "grad_norm": 2.796875,
      "learning_rate": 4.564427521554544e-05,
      "loss": 0.8543,
      "step": 13622
    },
    {
      "epoch": 0.19319958450177438,
      "grad_norm": 2.734375,
      "learning_rate": 4.564301499823456e-05,
      "loss": 0.8345,
      "step": 13624
    },
    {
      "epoch": 0.1932279461554006,
      "grad_norm": 2.828125,
      "learning_rate": 4.564175461604654e-05,
      "loss": 0.8269,
      "step": 13626
    },
    {
      "epoch": 0.1932563078090268,
      "grad_norm": 3.4375,
      "learning_rate": 4.564049406899146e-05,
      "loss": 0.9168,
      "step": 13628
    },
    {
      "epoch": 0.19328466946265302,
      "grad_norm": 3.21875,
      "learning_rate": 4.563923335707936e-05,
      "loss": 0.8391,
      "step": 13630
    },
    {
      "epoch": 0.19331303111627923,
      "grad_norm": 3.203125,
      "learning_rate": 4.563797248032034e-05,
      "loss": 0.8275,
      "step": 13632
    },
    {
      "epoch": 0.19334139276990545,
      "grad_norm": 3.125,
      "learning_rate": 4.563671143872445e-05,
      "loss": 0.8301,
      "step": 13634
    },
    {
      "epoch": 0.19336975442353166,
      "grad_norm": 3.09375,
      "learning_rate": 4.563545023230177e-05,
      "loss": 0.8992,
      "step": 13636
    },
    {
      "epoch": 0.19339811607715787,
      "grad_norm": 3.09375,
      "learning_rate": 4.563418886106238e-05,
      "loss": 0.8563,
      "step": 13638
    },
    {
      "epoch": 0.19342647773078409,
      "grad_norm": 3.234375,
      "learning_rate": 4.563292732501634e-05,
      "loss": 0.8374,
      "step": 13640
    },
    {
      "epoch": 0.1934548393844103,
      "grad_norm": 3.0,
      "learning_rate": 4.563166562417374e-05,
      "loss": 0.83,
      "step": 13642
    },
    {
      "epoch": 0.1934832010380365,
      "grad_norm": 2.796875,
      "learning_rate": 4.5630403758544646e-05,
      "loss": 0.8717,
      "step": 13644
    },
    {
      "epoch": 0.19351156269166273,
      "grad_norm": 3.09375,
      "learning_rate": 4.562914172813913e-05,
      "loss": 0.8788,
      "step": 13646
    },
    {
      "epoch": 0.19353992434528894,
      "grad_norm": 3.328125,
      "learning_rate": 4.56278795329673e-05,
      "loss": 0.8499,
      "step": 13648
    },
    {
      "epoch": 0.19356828599891518,
      "grad_norm": 3.234375,
      "learning_rate": 4.562661717303921e-05,
      "loss": 0.7898,
      "step": 13650
    },
    {
      "epoch": 0.1935966476525414,
      "grad_norm": 3.40625,
      "learning_rate": 4.562535464836496e-05,
      "loss": 0.873,
      "step": 13652
    },
    {
      "epoch": 0.1936250093061676,
      "grad_norm": 3.375,
      "learning_rate": 4.562409195895463e-05,
      "loss": 0.8257,
      "step": 13654
    },
    {
      "epoch": 0.19365337095979382,
      "grad_norm": 3.265625,
      "learning_rate": 4.5622829104818287e-05,
      "loss": 0.8552,
      "step": 13656
    },
    {
      "epoch": 0.19368173261342003,
      "grad_norm": 3.3125,
      "learning_rate": 4.562156608596604e-05,
      "loss": 0.8391,
      "step": 13658
    },
    {
      "epoch": 0.19371009426704625,
      "grad_norm": 2.8125,
      "learning_rate": 4.562030290240798e-05,
      "loss": 0.7875,
      "step": 13660
    },
    {
      "epoch": 0.19373845592067246,
      "grad_norm": 2.671875,
      "learning_rate": 4.561903955415417e-05,
      "loss": 0.796,
      "step": 13662
    },
    {
      "epoch": 0.19376681757429867,
      "grad_norm": 3.078125,
      "learning_rate": 4.561777604121472e-05,
      "loss": 0.8196,
      "step": 13664
    },
    {
      "epoch": 0.19379517922792489,
      "grad_norm": 3.078125,
      "learning_rate": 4.561651236359972e-05,
      "loss": 0.8133,
      "step": 13666
    },
    {
      "epoch": 0.1938235408815511,
      "grad_norm": 2.90625,
      "learning_rate": 4.5615248521319256e-05,
      "loss": 0.8728,
      "step": 13668
    },
    {
      "epoch": 0.1938519025351773,
      "grad_norm": 3.078125,
      "learning_rate": 4.5613984514383434e-05,
      "loss": 0.8194,
      "step": 13670
    },
    {
      "epoch": 0.19388026418880352,
      "grad_norm": 2.890625,
      "learning_rate": 4.561272034280234e-05,
      "loss": 0.812,
      "step": 13672
    },
    {
      "epoch": 0.19390862584242974,
      "grad_norm": 3.234375,
      "learning_rate": 4.5611456006586076e-05,
      "loss": 0.866,
      "step": 13674
    },
    {
      "epoch": 0.19393698749605595,
      "grad_norm": 3.03125,
      "learning_rate": 4.561019150574473e-05,
      "loss": 0.8266,
      "step": 13676
    },
    {
      "epoch": 0.19396534914968216,
      "grad_norm": 3.296875,
      "learning_rate": 4.560892684028841e-05,
      "loss": 0.8297,
      "step": 13678
    },
    {
      "epoch": 0.19399371080330838,
      "grad_norm": 2.90625,
      "learning_rate": 4.560766201022723e-05,
      "loss": 0.8646,
      "step": 13680
    },
    {
      "epoch": 0.1940220724569346,
      "grad_norm": 3.078125,
      "learning_rate": 4.560639701557127e-05,
      "loss": 0.8831,
      "step": 13682
    },
    {
      "epoch": 0.1940504341105608,
      "grad_norm": 2.9375,
      "learning_rate": 4.560513185633064e-05,
      "loss": 0.8065,
      "step": 13684
    },
    {
      "epoch": 0.19407879576418702,
      "grad_norm": 2.875,
      "learning_rate": 4.560386653251545e-05,
      "loss": 0.8051,
      "step": 13686
    },
    {
      "epoch": 0.19410715741781326,
      "grad_norm": 3.0,
      "learning_rate": 4.560260104413581e-05,
      "loss": 0.8603,
      "step": 13688
    },
    {
      "epoch": 0.19413551907143947,
      "grad_norm": 2.984375,
      "learning_rate": 4.560133539120181e-05,
      "loss": 0.8588,
      "step": 13690
    },
    {
      "epoch": 0.19416388072506569,
      "grad_norm": 3.078125,
      "learning_rate": 4.5600069573723577e-05,
      "loss": 0.8204,
      "step": 13692
    },
    {
      "epoch": 0.1941922423786919,
      "grad_norm": 2.796875,
      "learning_rate": 4.559880359171121e-05,
      "loss": 0.8171,
      "step": 13694
    },
    {
      "epoch": 0.1942206040323181,
      "grad_norm": 3.15625,
      "learning_rate": 4.5597537445174835e-05,
      "loss": 0.7763,
      "step": 13696
    },
    {
      "epoch": 0.19424896568594432,
      "grad_norm": 3.109375,
      "learning_rate": 4.559627113412454e-05,
      "loss": 0.8849,
      "step": 13698
    },
    {
      "epoch": 0.19427732733957054,
      "grad_norm": 3.078125,
      "learning_rate": 4.559500465857046e-05,
      "loss": 0.8396,
      "step": 13700
    },
    {
      "epoch": 0.19430568899319675,
      "grad_norm": 2.90625,
      "learning_rate": 4.559373801852271e-05,
      "loss": 0.8361,
      "step": 13702
    },
    {
      "epoch": 0.19433405064682296,
      "grad_norm": 2.921875,
      "learning_rate": 4.55924712139914e-05,
      "loss": 0.7922,
      "step": 13704
    },
    {
      "epoch": 0.19436241230044918,
      "grad_norm": 2.703125,
      "learning_rate": 4.5591204244986644e-05,
      "loss": 0.8029,
      "step": 13706
    },
    {
      "epoch": 0.1943907739540754,
      "grad_norm": 3.140625,
      "learning_rate": 4.558993711151856e-05,
      "loss": 0.8688,
      "step": 13708
    },
    {
      "epoch": 0.1944191356077016,
      "grad_norm": 2.9375,
      "learning_rate": 4.5588669813597285e-05,
      "loss": 0.8106,
      "step": 13710
    },
    {
      "epoch": 0.19444749726132782,
      "grad_norm": 2.953125,
      "learning_rate": 4.558740235123292e-05,
      "loss": 0.7667,
      "step": 13712
    },
    {
      "epoch": 0.19447585891495403,
      "grad_norm": 3.359375,
      "learning_rate": 4.5586134724435614e-05,
      "loss": 0.8804,
      "step": 13714
    },
    {
      "epoch": 0.19450422056858024,
      "grad_norm": 2.953125,
      "learning_rate": 4.5584866933215465e-05,
      "loss": 0.8249,
      "step": 13716
    },
    {
      "epoch": 0.19453258222220646,
      "grad_norm": 3.234375,
      "learning_rate": 4.558359897758262e-05,
      "loss": 0.861,
      "step": 13718
    },
    {
      "epoch": 0.19456094387583267,
      "grad_norm": 3.09375,
      "learning_rate": 4.558233085754719e-05,
      "loss": 0.8596,
      "step": 13720
    },
    {
      "epoch": 0.19458930552945888,
      "grad_norm": 3.609375,
      "learning_rate": 4.558106257311932e-05,
      "loss": 0.8829,
      "step": 13722
    },
    {
      "epoch": 0.1946176671830851,
      "grad_norm": 3.203125,
      "learning_rate": 4.5579794124309116e-05,
      "loss": 0.8759,
      "step": 13724
    },
    {
      "epoch": 0.19464602883671134,
      "grad_norm": 3.09375,
      "learning_rate": 4.557852551112673e-05,
      "loss": 0.829,
      "step": 13726
    },
    {
      "epoch": 0.19467439049033755,
      "grad_norm": 3.015625,
      "learning_rate": 4.557725673358229e-05,
      "loss": 0.8493,
      "step": 13728
    },
    {
      "epoch": 0.19470275214396376,
      "grad_norm": 3.21875,
      "learning_rate": 4.557598779168593e-05,
      "loss": 0.8296,
      "step": 13730
    },
    {
      "epoch": 0.19473111379758998,
      "grad_norm": 2.953125,
      "learning_rate": 4.5574718685447784e-05,
      "loss": 0.7857,
      "step": 13732
    },
    {
      "epoch": 0.1947594754512162,
      "grad_norm": 3.34375,
      "learning_rate": 4.557344941487798e-05,
      "loss": 0.8104,
      "step": 13734
    },
    {
      "epoch": 0.1947878371048424,
      "grad_norm": 3.109375,
      "learning_rate": 4.557217997998666e-05,
      "loss": 0.8318,
      "step": 13736
    },
    {
      "epoch": 0.19481619875846862,
      "grad_norm": 3.21875,
      "learning_rate": 4.557091038078398e-05,
      "loss": 0.8497,
      "step": 13738
    },
    {
      "epoch": 0.19484456041209483,
      "grad_norm": 3.203125,
      "learning_rate": 4.556964061728006e-05,
      "loss": 0.8762,
      "step": 13740
    },
    {
      "epoch": 0.19487292206572104,
      "grad_norm": 3.15625,
      "learning_rate": 4.556837068948505e-05,
      "loss": 0.8858,
      "step": 13742
    },
    {
      "epoch": 0.19490128371934726,
      "grad_norm": 3.09375,
      "learning_rate": 4.5567100597409085e-05,
      "loss": 0.7808,
      "step": 13744
    },
    {
      "epoch": 0.19492964537297347,
      "grad_norm": 2.875,
      "learning_rate": 4.556583034106232e-05,
      "loss": 0.8354,
      "step": 13746
    },
    {
      "epoch": 0.19495800702659968,
      "grad_norm": 3.265625,
      "learning_rate": 4.55645599204549e-05,
      "loss": 0.8572,
      "step": 13748
    },
    {
      "epoch": 0.1949863686802259,
      "grad_norm": 3.0,
      "learning_rate": 4.556328933559696e-05,
      "loss": 0.8329,
      "step": 13750
    },
    {
      "epoch": 0.1950147303338521,
      "grad_norm": 3.203125,
      "learning_rate": 4.556201858649867e-05,
      "loss": 0.8736,
      "step": 13752
    },
    {
      "epoch": 0.19504309198747832,
      "grad_norm": 3.28125,
      "learning_rate": 4.556074767317016e-05,
      "loss": 0.8634,
      "step": 13754
    },
    {
      "epoch": 0.19507145364110454,
      "grad_norm": 2.84375,
      "learning_rate": 4.555947659562158e-05,
      "loss": 0.8357,
      "step": 13756
    },
    {
      "epoch": 0.19509981529473075,
      "grad_norm": 2.90625,
      "learning_rate": 4.555820535386309e-05,
      "loss": 0.8555,
      "step": 13758
    },
    {
      "epoch": 0.19512817694835696,
      "grad_norm": 3.171875,
      "learning_rate": 4.555693394790485e-05,
      "loss": 0.8446,
      "step": 13760
    },
    {
      "epoch": 0.19515653860198318,
      "grad_norm": 3.21875,
      "learning_rate": 4.5555662377757e-05,
      "loss": 0.8843,
      "step": 13762
    },
    {
      "epoch": 0.19518490025560942,
      "grad_norm": 3.203125,
      "learning_rate": 4.555439064342971e-05,
      "loss": 0.8879,
      "step": 13764
    },
    {
      "epoch": 0.19521326190923563,
      "grad_norm": 3.09375,
      "learning_rate": 4.555311874493313e-05,
      "loss": 0.8431,
      "step": 13766
    },
    {
      "epoch": 0.19524162356286184,
      "grad_norm": 3.078125,
      "learning_rate": 4.5551846682277415e-05,
      "loss": 0.8254,
      "step": 13768
    },
    {
      "epoch": 0.19526998521648806,
      "grad_norm": 3.125,
      "learning_rate": 4.5550574455472736e-05,
      "loss": 0.7993,
      "step": 13770
    },
    {
      "epoch": 0.19529834687011427,
      "grad_norm": 3.09375,
      "learning_rate": 4.554930206452924e-05,
      "loss": 0.8603,
      "step": 13772
    },
    {
      "epoch": 0.19532670852374048,
      "grad_norm": 2.84375,
      "learning_rate": 4.5548029509457105e-05,
      "loss": 0.7865,
      "step": 13774
    },
    {
      "epoch": 0.1953550701773667,
      "grad_norm": 3.484375,
      "learning_rate": 4.554675679026649e-05,
      "loss": 0.8493,
      "step": 13776
    },
    {
      "epoch": 0.1953834318309929,
      "grad_norm": 2.859375,
      "learning_rate": 4.554548390696755e-05,
      "loss": 0.8078,
      "step": 13778
    },
    {
      "epoch": 0.19541179348461912,
      "grad_norm": 3.0625,
      "learning_rate": 4.554421085957046e-05,
      "loss": 0.859,
      "step": 13780
    },
    {
      "epoch": 0.19544015513824534,
      "grad_norm": 2.921875,
      "learning_rate": 4.5542937648085394e-05,
      "loss": 0.8576,
      "step": 13782
    },
    {
      "epoch": 0.19546851679187155,
      "grad_norm": 3.21875,
      "learning_rate": 4.554166427252251e-05,
      "loss": 0.871,
      "step": 13784
    },
    {
      "epoch": 0.19549687844549776,
      "grad_norm": 3.359375,
      "learning_rate": 4.5540390732891986e-05,
      "loss": 0.8773,
      "step": 13786
    },
    {
      "epoch": 0.19552524009912398,
      "grad_norm": 2.71875,
      "learning_rate": 4.553911702920399e-05,
      "loss": 0.8572,
      "step": 13788
    },
    {
      "epoch": 0.1955536017527502,
      "grad_norm": 3.71875,
      "learning_rate": 4.55378431614687e-05,
      "loss": 0.8114,
      "step": 13790
    },
    {
      "epoch": 0.1955819634063764,
      "grad_norm": 3.359375,
      "learning_rate": 4.553656912969628e-05,
      "loss": 0.8716,
      "step": 13792
    },
    {
      "epoch": 0.19561032506000262,
      "grad_norm": 3.1875,
      "learning_rate": 4.553529493389691e-05,
      "loss": 0.8173,
      "step": 13794
    },
    {
      "epoch": 0.19563868671362883,
      "grad_norm": 2.78125,
      "learning_rate": 4.553402057408078e-05,
      "loss": 0.8162,
      "step": 13796
    },
    {
      "epoch": 0.19566704836725504,
      "grad_norm": 2.921875,
      "learning_rate": 4.553274605025805e-05,
      "loss": 0.886,
      "step": 13798
    },
    {
      "epoch": 0.19569541002088126,
      "grad_norm": 3.28125,
      "learning_rate": 4.553147136243892e-05,
      "loss": 0.8876,
      "step": 13800
    },
    {
      "epoch": 0.19572377167450747,
      "grad_norm": 2.765625,
      "learning_rate": 4.553019651063354e-05,
      "loss": 0.8142,
      "step": 13802
    },
    {
      "epoch": 0.1957521333281337,
      "grad_norm": 2.890625,
      "learning_rate": 4.552892149485212e-05,
      "loss": 0.7927,
      "step": 13804
    },
    {
      "epoch": 0.19578049498175992,
      "grad_norm": 3.21875,
      "learning_rate": 4.5527646315104835e-05,
      "loss": 0.8089,
      "step": 13806
    },
    {
      "epoch": 0.19580885663538614,
      "grad_norm": 3.203125,
      "learning_rate": 4.552637097140187e-05,
      "loss": 0.8527,
      "step": 13808
    },
    {
      "epoch": 0.19583721828901235,
      "grad_norm": 3.046875,
      "learning_rate": 4.552509546375341e-05,
      "loss": 0.8326,
      "step": 13810
    },
    {
      "epoch": 0.19586557994263856,
      "grad_norm": 3.015625,
      "learning_rate": 4.5523819792169646e-05,
      "loss": 0.8095,
      "step": 13812
    },
    {
      "epoch": 0.19589394159626478,
      "grad_norm": 2.984375,
      "learning_rate": 4.552254395666077e-05,
      "loss": 0.8843,
      "step": 13814
    },
    {
      "epoch": 0.195922303249891,
      "grad_norm": 3.03125,
      "learning_rate": 4.552126795723695e-05,
      "loss": 0.8622,
      "step": 13816
    },
    {
      "epoch": 0.1959506649035172,
      "grad_norm": 3.0625,
      "learning_rate": 4.551999179390841e-05,
      "loss": 0.8287,
      "step": 13818
    },
    {
      "epoch": 0.19597902655714342,
      "grad_norm": 2.75,
      "learning_rate": 4.5518715466685314e-05,
      "loss": 0.826,
      "step": 13820
    },
    {
      "epoch": 0.19600738821076963,
      "grad_norm": 3.171875,
      "learning_rate": 4.551743897557788e-05,
      "loss": 0.8949,
      "step": 13822
    },
    {
      "epoch": 0.19603574986439584,
      "grad_norm": 3.046875,
      "learning_rate": 4.5516162320596284e-05,
      "loss": 0.8615,
      "step": 13824
    },
    {
      "epoch": 0.19606411151802206,
      "grad_norm": 3.125,
      "learning_rate": 4.551488550175074e-05,
      "loss": 0.8313,
      "step": 13826
    },
    {
      "epoch": 0.19609247317164827,
      "grad_norm": 3.140625,
      "learning_rate": 4.551360851905143e-05,
      "loss": 0.8419,
      "step": 13828
    },
    {
      "epoch": 0.19612083482527448,
      "grad_norm": 3.59375,
      "learning_rate": 4.551233137250856e-05,
      "loss": 0.8335,
      "step": 13830
    },
    {
      "epoch": 0.1961491964789007,
      "grad_norm": 3.046875,
      "learning_rate": 4.551105406213233e-05,
      "loss": 0.8553,
      "step": 13832
    },
    {
      "epoch": 0.1961775581325269,
      "grad_norm": 3.203125,
      "learning_rate": 4.550977658793295e-05,
      "loss": 0.8536,
      "step": 13834
    },
    {
      "epoch": 0.19620591978615312,
      "grad_norm": 3.234375,
      "learning_rate": 4.550849894992061e-05,
      "loss": 0.858,
      "step": 13836
    },
    {
      "epoch": 0.19623428143977933,
      "grad_norm": 2.78125,
      "learning_rate": 4.550722114810553e-05,
      "loss": 0.8121,
      "step": 13838
    },
    {
      "epoch": 0.19626264309340555,
      "grad_norm": 3.34375,
      "learning_rate": 4.55059431824979e-05,
      "loss": 0.861,
      "step": 13840
    },
    {
      "epoch": 0.1962910047470318,
      "grad_norm": 3.234375,
      "learning_rate": 4.550466505310793e-05,
      "loss": 0.8931,
      "step": 13842
    },
    {
      "epoch": 0.196319366400658,
      "grad_norm": 2.828125,
      "learning_rate": 4.550338675994584e-05,
      "loss": 0.8087,
      "step": 13844
    },
    {
      "epoch": 0.19634772805428422,
      "grad_norm": 3.03125,
      "learning_rate": 4.550210830302183e-05,
      "loss": 0.8787,
      "step": 13846
    },
    {
      "epoch": 0.19637608970791043,
      "grad_norm": 3.328125,
      "learning_rate": 4.5500829682346116e-05,
      "loss": 0.8898,
      "step": 13848
    },
    {
      "epoch": 0.19640445136153664,
      "grad_norm": 3.015625,
      "learning_rate": 4.54995508979289e-05,
      "loss": 0.8772,
      "step": 13850
    },
    {
      "epoch": 0.19643281301516285,
      "grad_norm": 2.984375,
      "learning_rate": 4.5498271949780414e-05,
      "loss": 0.7862,
      "step": 13852
    },
    {
      "epoch": 0.19646117466878907,
      "grad_norm": 3.015625,
      "learning_rate": 4.5496992837910856e-05,
      "loss": 0.8032,
      "step": 13854
    },
    {
      "epoch": 0.19648953632241528,
      "grad_norm": 2.703125,
      "learning_rate": 4.549571356233045e-05,
      "loss": 0.8171,
      "step": 13856
    },
    {
      "epoch": 0.1965178979760415,
      "grad_norm": 3.171875,
      "learning_rate": 4.5494434123049414e-05,
      "loss": 0.8677,
      "step": 13858
    },
    {
      "epoch": 0.1965462596296677,
      "grad_norm": 3.0,
      "learning_rate": 4.5493154520077966e-05,
      "loss": 0.8341,
      "step": 13860
    },
    {
      "epoch": 0.19657462128329392,
      "grad_norm": 3.171875,
      "learning_rate": 4.549187475342632e-05,
      "loss": 0.8545,
      "step": 13862
    },
    {
      "epoch": 0.19660298293692013,
      "grad_norm": 2.796875,
      "learning_rate": 4.549059482310471e-05,
      "loss": 0.8293,
      "step": 13864
    },
    {
      "epoch": 0.19663134459054635,
      "grad_norm": 3.453125,
      "learning_rate": 4.548931472912335e-05,
      "loss": 0.8482,
      "step": 13866
    },
    {
      "epoch": 0.19665970624417256,
      "grad_norm": 2.8125,
      "learning_rate": 4.548803447149247e-05,
      "loss": 0.8486,
      "step": 13868
    },
    {
      "epoch": 0.19668806789779877,
      "grad_norm": 3.0625,
      "learning_rate": 4.548675405022229e-05,
      "loss": 0.8538,
      "step": 13870
    },
    {
      "epoch": 0.196716429551425,
      "grad_norm": 3.375,
      "learning_rate": 4.5485473465323035e-05,
      "loss": 0.8593,
      "step": 13872
    },
    {
      "epoch": 0.1967447912050512,
      "grad_norm": 2.734375,
      "learning_rate": 4.5484192716804944e-05,
      "loss": 0.8004,
      "step": 13874
    },
    {
      "epoch": 0.1967731528586774,
      "grad_norm": 3.28125,
      "learning_rate": 4.548291180467824e-05,
      "loss": 0.8399,
      "step": 13876
    },
    {
      "epoch": 0.19680151451230363,
      "grad_norm": 3.28125,
      "learning_rate": 4.5481630728953136e-05,
      "loss": 0.8608,
      "step": 13878
    },
    {
      "epoch": 0.19682987616592987,
      "grad_norm": 2.875,
      "learning_rate": 4.5480349489639904e-05,
      "loss": 0.7961,
      "step": 13880
    },
    {
      "epoch": 0.19685823781955608,
      "grad_norm": 3.109375,
      "learning_rate": 4.5479068086748746e-05,
      "loss": 0.8559,
      "step": 13882
    },
    {
      "epoch": 0.1968865994731823,
      "grad_norm": 3.078125,
      "learning_rate": 4.54777865202899e-05,
      "loss": 0.8349,
      "step": 13884
    },
    {
      "epoch": 0.1969149611268085,
      "grad_norm": 2.953125,
      "learning_rate": 4.547650479027361e-05,
      "loss": 0.8214,
      "step": 13886
    },
    {
      "epoch": 0.19694332278043472,
      "grad_norm": 3.109375,
      "learning_rate": 4.547522289671011e-05,
      "loss": 0.8419,
      "step": 13888
    },
    {
      "epoch": 0.19697168443406093,
      "grad_norm": 3.671875,
      "learning_rate": 4.547394083960964e-05,
      "loss": 0.8859,
      "step": 13890
    },
    {
      "epoch": 0.19700004608768715,
      "grad_norm": 3.234375,
      "learning_rate": 4.5472658618982446e-05,
      "loss": 0.8351,
      "step": 13892
    },
    {
      "epoch": 0.19702840774131336,
      "grad_norm": 2.921875,
      "learning_rate": 4.5471376234838754e-05,
      "loss": 0.7962,
      "step": 13894
    },
    {
      "epoch": 0.19705676939493957,
      "grad_norm": 3.046875,
      "learning_rate": 4.5470093687188815e-05,
      "loss": 0.8598,
      "step": 13896
    },
    {
      "epoch": 0.1970851310485658,
      "grad_norm": 2.859375,
      "learning_rate": 4.546881097604288e-05,
      "loss": 0.8238,
      "step": 13898
    },
    {
      "epoch": 0.197113492702192,
      "grad_norm": 3.125,
      "learning_rate": 4.546752810141117e-05,
      "loss": 0.8107,
      "step": 13900
    },
    {
      "epoch": 0.1971418543558182,
      "grad_norm": 3.234375,
      "learning_rate": 4.546624506330396e-05,
      "loss": 0.8252,
      "step": 13902
    },
    {
      "epoch": 0.19717021600944443,
      "grad_norm": 3.625,
      "learning_rate": 4.546496186173148e-05,
      "loss": 0.7961,
      "step": 13904
    },
    {
      "epoch": 0.19719857766307064,
      "grad_norm": 2.90625,
      "learning_rate": 4.546367849670399e-05,
      "loss": 0.8548,
      "step": 13906
    },
    {
      "epoch": 0.19722693931669685,
      "grad_norm": 2.875,
      "learning_rate": 4.546239496823174e-05,
      "loss": 0.8434,
      "step": 13908
    },
    {
      "epoch": 0.19725530097032307,
      "grad_norm": 2.90625,
      "learning_rate": 4.5461111276324964e-05,
      "loss": 0.8626,
      "step": 13910
    },
    {
      "epoch": 0.19728366262394928,
      "grad_norm": 2.953125,
      "learning_rate": 4.545982742099394e-05,
      "loss": 0.8576,
      "step": 13912
    },
    {
      "epoch": 0.1973120242775755,
      "grad_norm": 3.15625,
      "learning_rate": 4.5458543402248895e-05,
      "loss": 0.8932,
      "step": 13914
    },
    {
      "epoch": 0.1973403859312017,
      "grad_norm": 3.015625,
      "learning_rate": 4.545725922010011e-05,
      "loss": 0.8027,
      "step": 13916
    },
    {
      "epoch": 0.19736874758482795,
      "grad_norm": 3.078125,
      "learning_rate": 4.545597487455783e-05,
      "loss": 0.8015,
      "step": 13918
    },
    {
      "epoch": 0.19739710923845416,
      "grad_norm": 2.96875,
      "learning_rate": 4.545469036563231e-05,
      "loss": 0.8285,
      "step": 13920
    },
    {
      "epoch": 0.19742547089208037,
      "grad_norm": 3.0625,
      "learning_rate": 4.545340569333382e-05,
      "loss": 0.8165,
      "step": 13922
    },
    {
      "epoch": 0.1974538325457066,
      "grad_norm": 3.40625,
      "learning_rate": 4.545212085767261e-05,
      "loss": 0.8747,
      "step": 13924
    },
    {
      "epoch": 0.1974821941993328,
      "grad_norm": 3.203125,
      "learning_rate": 4.5450835858658946e-05,
      "loss": 0.7792,
      "step": 13926
    },
    {
      "epoch": 0.197510555852959,
      "grad_norm": 3.046875,
      "learning_rate": 4.544955069630309e-05,
      "loss": 0.8908,
      "step": 13928
    },
    {
      "epoch": 0.19753891750658523,
      "grad_norm": 2.984375,
      "learning_rate": 4.54482653706153e-05,
      "loss": 0.887,
      "step": 13930
    },
    {
      "epoch": 0.19756727916021144,
      "grad_norm": 3.328125,
      "learning_rate": 4.5446979881605874e-05,
      "loss": 0.8407,
      "step": 13932
    },
    {
      "epoch": 0.19759564081383765,
      "grad_norm": 3.03125,
      "learning_rate": 4.5445694229285044e-05,
      "loss": 0.7915,
      "step": 13934
    },
    {
      "epoch": 0.19762400246746387,
      "grad_norm": 3.078125,
      "learning_rate": 4.5444408413663085e-05,
      "loss": 0.8103,
      "step": 13936
    },
    {
      "epoch": 0.19765236412109008,
      "grad_norm": 2.859375,
      "learning_rate": 4.5443122434750284e-05,
      "loss": 0.8414,
      "step": 13938
    },
    {
      "epoch": 0.1976807257747163,
      "grad_norm": 3.125,
      "learning_rate": 4.5441836292556894e-05,
      "loss": 0.8512,
      "step": 13940
    },
    {
      "epoch": 0.1977090874283425,
      "grad_norm": 3.1875,
      "learning_rate": 4.544054998709319e-05,
      "loss": 0.8175,
      "step": 13942
    },
    {
      "epoch": 0.19773744908196872,
      "grad_norm": 3.171875,
      "learning_rate": 4.5439263518369455e-05,
      "loss": 0.8902,
      "step": 13944
    },
    {
      "epoch": 0.19776581073559493,
      "grad_norm": 2.984375,
      "learning_rate": 4.543797688639596e-05,
      "loss": 0.8388,
      "step": 13946
    },
    {
      "epoch": 0.19779417238922115,
      "grad_norm": 3.3125,
      "learning_rate": 4.543669009118298e-05,
      "loss": 0.8277,
      "step": 13948
    },
    {
      "epoch": 0.19782253404284736,
      "grad_norm": 3.234375,
      "learning_rate": 4.5435403132740787e-05,
      "loss": 0.877,
      "step": 13950
    },
    {
      "epoch": 0.19785089569647357,
      "grad_norm": 3.796875,
      "learning_rate": 4.5434116011079675e-05,
      "loss": 0.8509,
      "step": 13952
    },
    {
      "epoch": 0.19787925735009979,
      "grad_norm": 3.078125,
      "learning_rate": 4.543282872620991e-05,
      "loss": 0.8072,
      "step": 13954
    },
    {
      "epoch": 0.197907619003726,
      "grad_norm": 3.265625,
      "learning_rate": 4.543154127814178e-05,
      "loss": 0.831,
      "step": 13956
    },
    {
      "epoch": 0.19793598065735224,
      "grad_norm": 3.234375,
      "learning_rate": 4.543025366688557e-05,
      "loss": 0.7641,
      "step": 13958
    },
    {
      "epoch": 0.19796434231097845,
      "grad_norm": 3.15625,
      "learning_rate": 4.542896589245156e-05,
      "loss": 0.8415,
      "step": 13960
    },
    {
      "epoch": 0.19799270396460467,
      "grad_norm": 3.484375,
      "learning_rate": 4.542767795485003e-05,
      "loss": 0.8234,
      "step": 13962
    },
    {
      "epoch": 0.19802106561823088,
      "grad_norm": 3.078125,
      "learning_rate": 4.542638985409128e-05,
      "loss": 0.8215,
      "step": 13964
    },
    {
      "epoch": 0.1980494272718571,
      "grad_norm": 3.046875,
      "learning_rate": 4.542510159018559e-05,
      "loss": 0.8773,
      "step": 13966
    },
    {
      "epoch": 0.1980777889254833,
      "grad_norm": 3.21875,
      "learning_rate": 4.5423813163143254e-05,
      "loss": 0.8258,
      "step": 13968
    },
    {
      "epoch": 0.19810615057910952,
      "grad_norm": 3.171875,
      "learning_rate": 4.5422524572974556e-05,
      "loss": 0.8509,
      "step": 13970
    },
    {
      "epoch": 0.19813451223273573,
      "grad_norm": 2.75,
      "learning_rate": 4.5421235819689796e-05,
      "loss": 0.8229,
      "step": 13972
    },
    {
      "epoch": 0.19816287388636195,
      "grad_norm": 2.875,
      "learning_rate": 4.541994690329926e-05,
      "loss": 0.8346,
      "step": 13974
    },
    {
      "epoch": 0.19819123553998816,
      "grad_norm": 3.078125,
      "learning_rate": 4.541865782381325e-05,
      "loss": 0.825,
      "step": 13976
    },
    {
      "epoch": 0.19821959719361437,
      "grad_norm": 2.71875,
      "learning_rate": 4.5417368581242054e-05,
      "loss": 0.7776,
      "step": 13978
    },
    {
      "epoch": 0.19824795884724059,
      "grad_norm": 2.921875,
      "learning_rate": 4.541607917559597e-05,
      "loss": 0.837,
      "step": 13980
    },
    {
      "epoch": 0.1982763205008668,
      "grad_norm": 3.09375,
      "learning_rate": 4.541478960688531e-05,
      "loss": 0.8399,
      "step": 13982
    },
    {
      "epoch": 0.198304682154493,
      "grad_norm": 3.0,
      "learning_rate": 4.5413499875120355e-05,
      "loss": 0.8735,
      "step": 13984
    },
    {
      "epoch": 0.19833304380811922,
      "grad_norm": 3.046875,
      "learning_rate": 4.541220998031142e-05,
      "loss": 0.8168,
      "step": 13986
    },
    {
      "epoch": 0.19836140546174544,
      "grad_norm": 3.046875,
      "learning_rate": 4.5410919922468795e-05,
      "loss": 0.8119,
      "step": 13988
    },
    {
      "epoch": 0.19838976711537165,
      "grad_norm": 2.984375,
      "learning_rate": 4.540962970160281e-05,
      "loss": 0.8397,
      "step": 13990
    },
    {
      "epoch": 0.19841812876899786,
      "grad_norm": 3.203125,
      "learning_rate": 4.540833931772373e-05,
      "loss": 0.8275,
      "step": 13992
    },
    {
      "epoch": 0.19844649042262408,
      "grad_norm": 2.734375,
      "learning_rate": 4.540704877084189e-05,
      "loss": 0.8753,
      "step": 13994
    },
    {
      "epoch": 0.19847485207625032,
      "grad_norm": 3.3125,
      "learning_rate": 4.54057580609676e-05,
      "loss": 0.833,
      "step": 13996
    },
    {
      "epoch": 0.19850321372987653,
      "grad_norm": 2.8125,
      "learning_rate": 4.540446718811114e-05,
      "loss": 0.8535,
      "step": 13998
    },
    {
      "epoch": 0.19853157538350275,
      "grad_norm": 3.375,
      "learning_rate": 4.5403176152282856e-05,
      "loss": 0.8728,
      "step": 14000
    }
  ],
  "logging_steps": 2,
  "max_steps": 70517,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.860275745291921e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}