{
  "best_global_step": 12168,
  "best_metric": 0.0602399967610836,
  "best_model_checkpoint": "./models/whisper-small-ml-baseline-high-lr/checkpoint-12168",
  "epoch": 5.0,
  "eval_steps": 4056,
  "global_step": 20280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_loss": 2.034205198287964,
      "eval_runtime": 65.0644,
      "eval_samples_per_second": 244.819,
      "eval_steps_per_second": 1.921,
      "step": 0
    },
    {
      "epoch": 0.0004930966469428008,
      "grad_norm": 23.125,
      "learning_rate": 9.861932938856017e-08,
      "loss": 1.9732,
      "step": 2
    },
    {
      "epoch": 0.0009861932938856016,
      "grad_norm": 23.125,
      "learning_rate": 2.958579881656805e-07,
      "loss": 1.9901,
      "step": 4
    },
    {
      "epoch": 0.0014792899408284023,
      "grad_norm": 23.25,
      "learning_rate": 4.930966469428008e-07,
      "loss": 1.9936,
      "step": 6
    },
    {
      "epoch": 0.0019723865877712033,
      "grad_norm": 23.75,
      "learning_rate": 6.903353057199211e-07,
      "loss": 1.9993,
      "step": 8
    },
    {
      "epoch": 0.002465483234714004,
      "grad_norm": 22.25,
      "learning_rate": 8.875739644970415e-07,
      "loss": 1.953,
      "step": 10
    },
    {
      "epoch": 0.0029585798816568047,
      "grad_norm": 22.125,
      "learning_rate": 1.0848126232741618e-06,
      "loss": 1.972,
      "step": 12
    },
    {
      "epoch": 0.0034516765285996054,
      "grad_norm": 22.875,
      "learning_rate": 1.282051282051282e-06,
      "loss": 1.9888,
      "step": 14
    },
    {
      "epoch": 0.0039447731755424065,
      "grad_norm": 22.75,
      "learning_rate": 1.4792899408284024e-06,
      "loss": 1.9878,
      "step": 16
    },
    {
      "epoch": 0.004437869822485207,
      "grad_norm": 21.5,
      "learning_rate": 1.6765285996055227e-06,
      "loss": 1.9712,
      "step": 18
    },
    {
      "epoch": 0.004930966469428008,
      "grad_norm": 21.375,
      "learning_rate": 1.8737672583826429e-06,
      "loss": 1.9812,
      "step": 20
    },
    {
      "epoch": 0.005424063116370809,
      "grad_norm": 20.5,
      "learning_rate": 2.0710059171597635e-06,
      "loss": 1.9411,
      "step": 22
    },
    {
      "epoch": 0.005917159763313609,
      "grad_norm": 19.75,
      "learning_rate": 2.268244575936884e-06,
      "loss": 1.9677,
      "step": 24
    },
    {
      "epoch": 0.00641025641025641,
      "grad_norm": 18.875,
      "learning_rate": 2.465483234714004e-06,
      "loss": 1.9734,
      "step": 26
    },
    {
      "epoch": 0.006903353057199211,
      "grad_norm": 18.25,
      "learning_rate": 2.6627218934911246e-06,
      "loss": 1.9409,
      "step": 28
    },
    {
      "epoch": 0.0073964497041420114,
      "grad_norm": 17.625,
      "learning_rate": 2.859960552268245e-06,
      "loss": 1.9655,
      "step": 30
    },
    {
      "epoch": 0.007889546351084813,
      "grad_norm": 15.6875,
      "learning_rate": 3.057199211045365e-06,
      "loss": 1.8875,
      "step": 32
    },
    {
      "epoch": 0.008382642998027613,
      "grad_norm": 15.3125,
      "learning_rate": 3.2544378698224853e-06,
      "loss": 1.8834,
      "step": 34
    },
    {
      "epoch": 0.008875739644970414,
      "grad_norm": 15.5625,
      "learning_rate": 3.451676528599606e-06,
      "loss": 1.8718,
      "step": 36
    },
    {
      "epoch": 0.009368836291913214,
      "grad_norm": 15.8125,
      "learning_rate": 3.648915187376726e-06,
      "loss": 1.8831,
      "step": 38
    },
    {
      "epoch": 0.009861932938856016,
      "grad_norm": 15.0,
      "learning_rate": 3.846153846153847e-06,
      "loss": 1.8631,
      "step": 40
    },
    {
      "epoch": 0.010355029585798817,
      "grad_norm": 14.6875,
      "learning_rate": 4.043392504930966e-06,
      "loss": 1.8634,
      "step": 42
    },
    {
      "epoch": 0.010848126232741617,
      "grad_norm": 13.3125,
      "learning_rate": 4.2406311637080875e-06,
      "loss": 1.8314,
      "step": 44
    },
    {
      "epoch": 0.011341222879684419,
      "grad_norm": 12.0625,
      "learning_rate": 4.437869822485207e-06,
      "loss": 1.8107,
      "step": 46
    },
    {
      "epoch": 0.011834319526627219,
      "grad_norm": 11.4375,
      "learning_rate": 4.6351084812623274e-06,
      "loss": 1.7942,
      "step": 48
    },
    {
      "epoch": 0.01232741617357002,
      "grad_norm": 10.4375,
      "learning_rate": 4.832347140039448e-06,
      "loss": 1.7604,
      "step": 50
    },
    {
      "epoch": 0.01282051282051282,
      "grad_norm": 9.5625,
      "learning_rate": 5.029585798816568e-06,
      "loss": 1.7598,
      "step": 52
    },
    {
      "epoch": 0.013313609467455622,
      "grad_norm": 9.0625,
      "learning_rate": 5.2268244575936885e-06,
      "loss": 1.7374,
      "step": 54
    },
    {
      "epoch": 0.013806706114398421,
      "grad_norm": 8.4375,
      "learning_rate": 5.424063116370809e-06,
      "loss": 1.7313,
      "step": 56
    },
    {
      "epoch": 0.014299802761341223,
      "grad_norm": 8.125,
      "learning_rate": 5.621301775147929e-06,
      "loss": 1.7077,
      "step": 58
    },
    {
      "epoch": 0.014792899408284023,
      "grad_norm": 7.5625,
      "learning_rate": 5.81854043392505e-06,
      "loss": 1.677,
      "step": 60
    },
    {
      "epoch": 0.015285996055226824,
      "grad_norm": 7.0,
      "learning_rate": 6.01577909270217e-06,
      "loss": 1.6691,
      "step": 62
    },
    {
      "epoch": 0.015779092702169626,
      "grad_norm": 7.03125,
      "learning_rate": 6.21301775147929e-06,
      "loss": 1.6344,
      "step": 64
    },
    {
      "epoch": 0.016272189349112426,
      "grad_norm": 6.0625,
      "learning_rate": 6.41025641025641e-06,
      "loss": 1.5967,
      "step": 66
    },
    {
      "epoch": 0.016765285996055226,
      "grad_norm": 6.1875,
      "learning_rate": 6.607495069033531e-06,
      "loss": 1.6006,
      "step": 68
    },
    {
      "epoch": 0.01725838264299803,
      "grad_norm": 6.28125,
      "learning_rate": 6.8047337278106515e-06,
      "loss": 1.5758,
      "step": 70
    },
    {
      "epoch": 0.01775147928994083,
      "grad_norm": 5.8125,
      "learning_rate": 7.001972386587771e-06,
      "loss": 1.5521,
      "step": 72
    },
    {
      "epoch": 0.01824457593688363,
      "grad_norm": 5.0625,
      "learning_rate": 7.199211045364892e-06,
      "loss": 1.5081,
      "step": 74
    },
    {
      "epoch": 0.01873767258382643,
      "grad_norm": 5.125,
      "learning_rate": 7.396449704142013e-06,
      "loss": 1.481,
      "step": 76
    },
    {
      "epoch": 0.019230769230769232,
      "grad_norm": 5.1875,
      "learning_rate": 7.593688362919132e-06,
      "loss": 1.4716,
      "step": 78
    },
    {
      "epoch": 0.01972386587771203,
      "grad_norm": 4.1875,
      "learning_rate": 7.790927021696252e-06,
      "loss": 1.4429,
      "step": 80
    },
    {
      "epoch": 0.02021696252465483,
      "grad_norm": 4.125,
      "learning_rate": 7.988165680473373e-06,
      "loss": 1.425,
      "step": 82
    },
    {
      "epoch": 0.020710059171597635,
      "grad_norm": 3.921875,
      "learning_rate": 8.185404339250494e-06,
      "loss": 1.405,
      "step": 84
    },
    {
      "epoch": 0.021203155818540435,
      "grad_norm": 4.34375,
      "learning_rate": 8.382642998027614e-06,
      "loss": 1.3888,
      "step": 86
    },
    {
      "epoch": 0.021696252465483234,
      "grad_norm": 5.3125,
      "learning_rate": 8.579881656804733e-06,
      "loss": 1.3612,
      "step": 88
    },
    {
      "epoch": 0.022189349112426034,
      "grad_norm": 4.1875,
      "learning_rate": 8.777120315581854e-06,
      "loss": 1.3386,
      "step": 90
    },
    {
      "epoch": 0.022682445759368838,
      "grad_norm": 3.515625,
      "learning_rate": 8.974358974358976e-06,
      "loss": 1.3304,
      "step": 92
    },
    {
      "epoch": 0.023175542406311637,
      "grad_norm": 3.859375,
      "learning_rate": 9.171597633136095e-06,
      "loss": 1.3084,
      "step": 94
    },
    {
      "epoch": 0.023668639053254437,
      "grad_norm": 4.1875,
      "learning_rate": 9.368836291913216e-06,
      "loss": 1.2826,
      "step": 96
    },
    {
      "epoch": 0.024161735700197237,
      "grad_norm": 3.875,
      "learning_rate": 9.566074950690336e-06,
      "loss": 1.2691,
      "step": 98
    },
    {
      "epoch": 0.02465483234714004,
      "grad_norm": 4.40625,
      "learning_rate": 9.763313609467455e-06,
      "loss": 1.2355,
      "step": 100
    },
    {
      "epoch": 0.02514792899408284,
      "grad_norm": 4.5,
      "learning_rate": 9.960552268244577e-06,
      "loss": 1.2235,
      "step": 102
    },
    {
      "epoch": 0.02564102564102564,
      "grad_norm": 3.984375,
      "learning_rate": 1.0157790927021698e-05,
      "loss": 1.1906,
      "step": 104
    },
    {
      "epoch": 0.026134122287968443,
      "grad_norm": 4.96875,
      "learning_rate": 1.0355029585798817e-05,
      "loss": 1.1732,
      "step": 106
    },
    {
      "epoch": 0.026627218934911243,
      "grad_norm": 4.34375,
      "learning_rate": 1.0552268244575937e-05,
      "loss": 1.1496,
      "step": 108
    },
    {
      "epoch": 0.027120315581854043,
      "grad_norm": 4.21875,
      "learning_rate": 1.0749506903353056e-05,
      "loss": 1.1229,
      "step": 110
    },
    {
      "epoch": 0.027613412228796843,
      "grad_norm": 4.25,
      "learning_rate": 1.094674556213018e-05,
      "loss": 1.0984,
      "step": 112
    },
    {
      "epoch": 0.028106508875739646,
      "grad_norm": 4.53125,
      "learning_rate": 1.1143984220907299e-05,
      "loss": 1.0504,
      "step": 114
    },
    {
      "epoch": 0.028599605522682446,
      "grad_norm": 5.71875,
      "learning_rate": 1.1341222879684418e-05,
      "loss": 1.0159,
      "step": 116
    },
    {
      "epoch": 0.029092702169625246,
      "grad_norm": 4.6875,
      "learning_rate": 1.153846153846154e-05,
      "loss": 0.9707,
      "step": 118
    },
    {
      "epoch": 0.029585798816568046,
      "grad_norm": 4.6875,
      "learning_rate": 1.1735700197238659e-05,
      "loss": 0.9146,
      "step": 120
    },
    {
      "epoch": 0.03007889546351085,
      "grad_norm": 4.21875,
      "learning_rate": 1.193293885601578e-05,
      "loss": 0.8875,
      "step": 122
    },
    {
      "epoch": 0.03057199211045365,
      "grad_norm": 4.3125,
      "learning_rate": 1.21301775147929e-05,
      "loss": 0.8699,
      "step": 124
    },
    {
      "epoch": 0.03106508875739645,
      "grad_norm": 4.46875,
      "learning_rate": 1.2327416173570021e-05,
      "loss": 0.8266,
      "step": 126
    },
    {
      "epoch": 0.03155818540433925,
      "grad_norm": 4.625,
      "learning_rate": 1.252465483234714e-05,
      "loss": 0.7786,
      "step": 128
    },
    {
      "epoch": 0.03205128205128205,
      "grad_norm": 4.1875,
      "learning_rate": 1.2721893491124262e-05,
      "loss": 0.7451,
      "step": 130
    },
    {
      "epoch": 0.03254437869822485,
      "grad_norm": 4.1875,
      "learning_rate": 1.2919132149901381e-05,
      "loss": 0.7227,
      "step": 132
    },
    {
      "epoch": 0.033037475345167655,
      "grad_norm": 4.5625,
      "learning_rate": 1.3116370808678502e-05,
      "loss": 0.6885,
      "step": 134
    },
    {
      "epoch": 0.03353057199211045,
      "grad_norm": 3.359375,
      "learning_rate": 1.3313609467455624e-05,
      "loss": 0.6587,
      "step": 136
    },
    {
      "epoch": 0.034023668639053255,
      "grad_norm": 4.6875,
      "learning_rate": 1.3510848126232742e-05,
      "loss": 0.6341,
      "step": 138
    },
    {
      "epoch": 0.03451676528599606,
      "grad_norm": 5.03125,
      "learning_rate": 1.3708086785009863e-05,
      "loss": 0.6167,
      "step": 140
    },
    {
      "epoch": 0.035009861932938854,
      "grad_norm": 4.96875,
      "learning_rate": 1.3905325443786982e-05,
      "loss": 0.5948,
      "step": 142
    },
    {
      "epoch": 0.03550295857988166,
      "grad_norm": 3.1875,
      "learning_rate": 1.4102564102564104e-05,
      "loss": 0.5768,
      "step": 144
    },
    {
      "epoch": 0.03599605522682446,
      "grad_norm": 3.15625,
      "learning_rate": 1.4299802761341225e-05,
      "loss": 0.5624,
      "step": 146
    },
    {
      "epoch": 0.03648915187376726,
      "grad_norm": 3.921875,
      "learning_rate": 1.4497041420118343e-05,
      "loss": 0.5311,
      "step": 148
    },
    {
      "epoch": 0.03698224852071006,
      "grad_norm": 3.25,
      "learning_rate": 1.4694280078895464e-05,
      "loss": 0.5299,
      "step": 150
    },
    {
      "epoch": 0.03747534516765286,
      "grad_norm": 3.453125,
      "learning_rate": 1.4891518737672585e-05,
      "loss": 0.4947,
      "step": 152
    },
    {
      "epoch": 0.03796844181459566,
      "grad_norm": 2.96875,
      "learning_rate": 1.5088757396449705e-05,
      "loss": 0.4862,
      "step": 154
    },
    {
      "epoch": 0.038461538461538464,
      "grad_norm": 3.453125,
      "learning_rate": 1.5285996055226824e-05,
      "loss": 0.4677,
      "step": 156
    },
    {
      "epoch": 0.03895463510848126,
      "grad_norm": 3.265625,
      "learning_rate": 1.5483234714003947e-05,
      "loss": 0.4577,
      "step": 158
    },
    {
      "epoch": 0.03944773175542406,
      "grad_norm": 3.21875,
      "learning_rate": 1.5680473372781066e-05,
      "loss": 0.4456,
      "step": 160
    },
    {
      "epoch": 0.03994082840236687,
      "grad_norm": 3.03125,
      "learning_rate": 1.5877712031558186e-05,
      "loss": 0.4459,
      "step": 162
    },
    {
      "epoch": 0.04043392504930966,
      "grad_norm": 3.078125,
      "learning_rate": 1.6074950690335306e-05,
      "loss": 0.4249,
      "step": 164
    },
    {
      "epoch": 0.040927021696252466,
      "grad_norm": 3.859375,
      "learning_rate": 1.6272189349112425e-05,
      "loss": 0.4159,
      "step": 166
    },
    {
      "epoch": 0.04142011834319527,
      "grad_norm": 3.125,
      "learning_rate": 1.6469428007889548e-05,
      "loss": 0.4136,
      "step": 168
    },
    {
      "epoch": 0.041913214990138066,
      "grad_norm": 3.0625,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.3989,
      "step": 170
    },
    {
      "epoch": 0.04240631163708087,
      "grad_norm": 3.71875,
      "learning_rate": 1.6863905325443787e-05,
      "loss": 0.391,
      "step": 172
    },
    {
      "epoch": 0.042899408284023666,
      "grad_norm": 3.78125,
      "learning_rate": 1.706114398422091e-05,
      "loss": 0.3874,
      "step": 174
    },
    {
      "epoch": 0.04339250493096647,
      "grad_norm": 3.546875,
      "learning_rate": 1.725838264299803e-05,
      "loss": 0.384,
      "step": 176
    },
    {
      "epoch": 0.04388560157790927,
      "grad_norm": 3.0625,
      "learning_rate": 1.745562130177515e-05,
      "loss": 0.3746,
      "step": 178
    },
    {
      "epoch": 0.04437869822485207,
      "grad_norm": 3.9375,
      "learning_rate": 1.7652859960552272e-05,
      "loss": 0.3623,
      "step": 180
    },
    {
      "epoch": 0.04487179487179487,
      "grad_norm": 2.453125,
      "learning_rate": 1.7850098619329388e-05,
      "loss": 0.3748,
      "step": 182
    },
    {
      "epoch": 0.045364891518737675,
      "grad_norm": 2.453125,
      "learning_rate": 1.804733727810651e-05,
      "loss": 0.3672,
      "step": 184
    },
    {
      "epoch": 0.04585798816568047,
      "grad_norm": 2.8125,
      "learning_rate": 1.824457593688363e-05,
      "loss": 0.3505,
      "step": 186
    },
    {
      "epoch": 0.046351084812623275,
      "grad_norm": 2.921875,
      "learning_rate": 1.844181459566075e-05,
      "loss": 0.3425,
      "step": 188
    },
    {
      "epoch": 0.04684418145956608,
      "grad_norm": 2.546875,
      "learning_rate": 1.8639053254437873e-05,
      "loss": 0.3356,
      "step": 190
    },
    {
      "epoch": 0.047337278106508875,
      "grad_norm": 2.71875,
      "learning_rate": 1.883629191321499e-05,
      "loss": 0.3191,
      "step": 192
    },
    {
      "epoch": 0.04783037475345168,
      "grad_norm": 3.046875,
      "learning_rate": 1.9033530571992112e-05,
      "loss": 0.3223,
      "step": 194
    },
    {
      "epoch": 0.048323471400394474,
      "grad_norm": 3.09375,
      "learning_rate": 1.923076923076923e-05,
      "loss": 0.3332,
      "step": 196
    },
    {
      "epoch": 0.04881656804733728,
      "grad_norm": 2.75,
      "learning_rate": 1.942800788954635e-05,
      "loss": 0.3058,
      "step": 198
    },
    {
      "epoch": 0.04930966469428008,
      "grad_norm": 2.703125,
      "learning_rate": 1.9625246548323474e-05,
      "loss": 0.3213,
      "step": 200
    },
    {
      "epoch": 0.04980276134122288,
      "grad_norm": 2.5,
      "learning_rate": 1.9822485207100593e-05,
      "loss": 0.3116,
      "step": 202
    },
    {
      "epoch": 0.05029585798816568,
      "grad_norm": 3.0,
      "learning_rate": 2.0019723865877713e-05,
      "loss": 0.308,
      "step": 204
    },
    {
      "epoch": 0.050788954635108484,
      "grad_norm": 2.625,
      "learning_rate": 2.0216962524654832e-05,
      "loss": 0.3035,
      "step": 206
    },
    {
      "epoch": 0.05128205128205128,
      "grad_norm": 2.546875,
      "learning_rate": 2.0414201183431952e-05,
      "loss": 0.3024,
      "step": 208
    },
    {
      "epoch": 0.051775147928994084,
      "grad_norm": 2.96875,
      "learning_rate": 2.0611439842209075e-05,
      "loss": 0.2876,
      "step": 210
    },
    {
      "epoch": 0.05226824457593689,
      "grad_norm": 2.984375,
      "learning_rate": 2.0808678500986194e-05,
      "loss": 0.2927,
      "step": 212
    },
    {
      "epoch": 0.05276134122287968,
      "grad_norm": 2.390625,
      "learning_rate": 2.1005917159763314e-05,
      "loss": 0.2899,
      "step": 214
    },
    {
      "epoch": 0.05325443786982249,
      "grad_norm": 3.203125,
      "learning_rate": 2.1203155818540433e-05,
      "loss": 0.2813,
      "step": 216
    },
    {
      "epoch": 0.05374753451676528,
      "grad_norm": 3.140625,
      "learning_rate": 2.1400394477317556e-05,
      "loss": 0.2734,
      "step": 218
    },
    {
      "epoch": 0.054240631163708086,
      "grad_norm": 2.953125,
      "learning_rate": 2.1597633136094676e-05,
      "loss": 0.2813,
      "step": 220
    },
    {
      "epoch": 0.05473372781065089,
      "grad_norm": 2.640625,
      "learning_rate": 2.1794871794871795e-05,
      "loss": 0.2807,
      "step": 222
    },
    {
      "epoch": 0.055226824457593686,
      "grad_norm": 3.109375,
      "learning_rate": 2.199211045364892e-05,
      "loss": 0.2588,
      "step": 224
    },
    {
      "epoch": 0.05571992110453649,
      "grad_norm": 3.875,
      "learning_rate": 2.2189349112426034e-05,
      "loss": 0.2633,
      "step": 226
    },
    {
      "epoch": 0.05621301775147929,
      "grad_norm": 3.875,
      "learning_rate": 2.2386587771203157e-05,
      "loss": 0.2616,
      "step": 228
    },
    {
      "epoch": 0.05670611439842209,
      "grad_norm": 2.390625,
      "learning_rate": 2.2583826429980277e-05,
      "loss": 0.2529,
      "step": 230
    },
    {
      "epoch": 0.05719921104536489,
      "grad_norm": 2.921875,
      "learning_rate": 2.2781065088757396e-05,
      "loss": 0.2525,
      "step": 232
    },
    {
      "epoch": 0.057692307692307696,
      "grad_norm": 2.390625,
      "learning_rate": 2.297830374753452e-05,
      "loss": 0.2611,
      "step": 234
    },
    {
      "epoch": 0.05818540433925049,
      "grad_norm": 2.6875,
      "learning_rate": 2.317554240631164e-05,
      "loss": 0.2349,
      "step": 236
    },
    {
      "epoch": 0.058678500986193295,
      "grad_norm": 2.46875,
      "learning_rate": 2.337278106508876e-05,
      "loss": 0.2351,
      "step": 238
    },
    {
      "epoch": 0.05917159763313609,
      "grad_norm": 2.421875,
      "learning_rate": 2.357001972386588e-05,
      "loss": 0.2436,
      "step": 240
    },
    {
      "epoch": 0.059664694280078895,
      "grad_norm": 2.609375,
      "learning_rate": 2.3767258382642997e-05,
      "loss": 0.2303,
      "step": 242
    },
    {
      "epoch": 0.0601577909270217,
      "grad_norm": 2.5625,
      "learning_rate": 2.396449704142012e-05,
      "loss": 0.2366,
      "step": 244
    },
    {
      "epoch": 0.060650887573964495,
      "grad_norm": 2.25,
      "learning_rate": 2.416173570019724e-05,
      "loss": 0.2173,
      "step": 246
    },
    {
      "epoch": 0.0611439842209073,
      "grad_norm": 2.953125,
      "learning_rate": 2.435897435897436e-05,
      "loss": 0.2331,
      "step": 248
    },
    {
      "epoch": 0.0616370808678501,
      "grad_norm": 3.546875,
      "learning_rate": 2.4556213017751482e-05,
      "loss": 0.2223,
      "step": 250
    },
    {
      "epoch": 0.0621301775147929,
      "grad_norm": 2.890625,
      "learning_rate": 2.47534516765286e-05,
      "loss": 0.2215,
      "step": 252
    },
    {
      "epoch": 0.0626232741617357,
      "grad_norm": 2.140625,
      "learning_rate": 2.495069033530572e-05,
      "loss": 0.223,
      "step": 254
    },
    {
      "epoch": 0.0631163708086785,
      "grad_norm": 2.21875,
      "learning_rate": 2.514792899408284e-05,
      "loss": 0.2176,
      "step": 256
    },
    {
      "epoch": 0.06360946745562131,
      "grad_norm": 2.59375,
      "learning_rate": 2.5345167652859964e-05,
      "loss": 0.2097,
      "step": 258
    },
    {
      "epoch": 0.0641025641025641,
      "grad_norm": 3.015625,
      "learning_rate": 2.5542406311637083e-05,
      "loss": 0.2082,
      "step": 260
    },
    {
      "epoch": 0.0645956607495069,
      "grad_norm": 3.65625,
      "learning_rate": 2.57396449704142e-05,
      "loss": 0.1992,
      "step": 262
    },
    {
      "epoch": 0.0650887573964497,
      "grad_norm": 2.546875,
      "learning_rate": 2.5936883629191322e-05,
      "loss": 0.2171,
      "step": 264
    },
    {
      "epoch": 0.0655818540433925,
      "grad_norm": 2.375,
      "learning_rate": 2.6134122287968442e-05,
      "loss": 0.2123,
      "step": 266
    },
    {
      "epoch": 0.06607495069033531,
      "grad_norm": 2.96875,
      "learning_rate": 2.6331360946745565e-05,
      "loss": 0.1963,
      "step": 268
    },
    {
      "epoch": 0.06656804733727811,
      "grad_norm": 2.71875,
      "learning_rate": 2.6528599605522688e-05,
      "loss": 0.1946,
      "step": 270
    },
    {
      "epoch": 0.0670611439842209,
      "grad_norm": 2.546875,
      "learning_rate": 2.67258382642998e-05,
      "loss": 0.2032,
      "step": 272
    },
    {
      "epoch": 0.0675542406311637,
      "grad_norm": 1.890625,
      "learning_rate": 2.6923076923076923e-05,
      "loss": 0.1975,
      "step": 274
    },
    {
      "epoch": 0.06804733727810651,
      "grad_norm": 3.328125,
      "learning_rate": 2.7120315581854043e-05,
      "loss": 0.1951,
      "step": 276
    },
    {
      "epoch": 0.06854043392504931,
      "grad_norm": 3.53125,
      "learning_rate": 2.7317554240631166e-05,
      "loss": 0.2055,
      "step": 278
    },
    {
      "epoch": 0.06903353057199212,
      "grad_norm": 3.796875,
      "learning_rate": 2.751479289940829e-05,
      "loss": 0.1879,
      "step": 280
    },
    {
      "epoch": 0.0695266272189349,
      "grad_norm": 2.359375,
      "learning_rate": 2.7712031558185408e-05,
      "loss": 0.1937,
      "step": 282
    },
    {
      "epoch": 0.07001972386587771,
      "grad_norm": 3.984375,
      "learning_rate": 2.7909270216962524e-05,
      "loss": 0.1867,
      "step": 284
    },
    {
      "epoch": 0.07051282051282051,
      "grad_norm": 2.75,
      "learning_rate": 2.8106508875739644e-05,
      "loss": 0.1838,
      "step": 286
    },
    {
      "epoch": 0.07100591715976332,
      "grad_norm": 3.078125,
      "learning_rate": 2.8303747534516767e-05,
      "loss": 0.1846,
      "step": 288
    },
    {
      "epoch": 0.07149901380670612,
      "grad_norm": 1.703125,
      "learning_rate": 2.850098619329389e-05,
      "loss": 0.1946,
      "step": 290
    },
    {
      "epoch": 0.07199211045364892,
      "grad_norm": 2.375,
      "learning_rate": 2.869822485207101e-05,
      "loss": 0.1826,
      "step": 292
    },
    {
      "epoch": 0.07248520710059171,
      "grad_norm": 2.265625,
      "learning_rate": 2.8895463510848125e-05,
      "loss": 0.1804,
      "step": 294
    },
    {
      "epoch": 0.07297830374753451,
      "grad_norm": 3.484375,
      "learning_rate": 2.9092702169625248e-05,
      "loss": 0.1924,
      "step": 296
    },
    {
      "epoch": 0.07347140039447732,
      "grad_norm": 2.390625,
      "learning_rate": 2.9289940828402368e-05,
      "loss": 0.1799,
      "step": 298
    },
    {
      "epoch": 0.07396449704142012,
      "grad_norm": 1.984375,
      "learning_rate": 2.948717948717949e-05,
      "loss": 0.1816,
      "step": 300
    },
    {
      "epoch": 0.07445759368836292,
      "grad_norm": 1.6171875,
      "learning_rate": 2.968441814595661e-05,
      "loss": 0.1845,
      "step": 302
    },
    {
      "epoch": 0.07495069033530571,
      "grad_norm": 1.4453125,
      "learning_rate": 2.9881656804733733e-05,
      "loss": 0.1782,
      "step": 304
    },
    {
      "epoch": 0.07544378698224852,
      "grad_norm": 2.0625,
      "learning_rate": 3.007889546351085e-05,
      "loss": 0.1648,
      "step": 306
    },
    {
      "epoch": 0.07593688362919132,
      "grad_norm": 2.6875,
      "learning_rate": 3.027613412228797e-05,
      "loss": 0.1746,
      "step": 308
    },
    {
      "epoch": 0.07642998027613412,
      "grad_norm": 2.640625,
      "learning_rate": 3.047337278106509e-05,
      "loss": 0.1809,
      "step": 310
    },
    {
      "epoch": 0.07692307692307693,
      "grad_norm": 2.140625,
      "learning_rate": 3.0670611439842215e-05,
      "loss": 0.1694,
      "step": 312
    },
    {
      "epoch": 0.07741617357001973,
      "grad_norm": 2.5,
      "learning_rate": 3.0867850098619334e-05,
      "loss": 0.1771,
      "step": 314
    },
    {
      "epoch": 0.07790927021696252,
      "grad_norm": 3.390625,
      "learning_rate": 3.106508875739645e-05,
      "loss": 0.1746,
      "step": 316
    },
    {
      "epoch": 0.07840236686390532,
      "grad_norm": 2.953125,
      "learning_rate": 3.126232741617357e-05,
      "loss": 0.174,
      "step": 318
    },
    {
      "epoch": 0.07889546351084813,
      "grad_norm": 3.9375,
      "learning_rate": 3.145956607495069e-05,
      "loss": 0.1634,
      "step": 320
    },
    {
      "epoch": 0.07938856015779093,
      "grad_norm": 2.5,
      "learning_rate": 3.165680473372781e-05,
      "loss": 0.1693,
      "step": 322
    },
    {
      "epoch": 0.07988165680473373,
      "grad_norm": 2.296875,
      "learning_rate": 3.185404339250493e-05,
      "loss": 0.1669,
      "step": 324
    },
    {
      "epoch": 0.08037475345167652,
      "grad_norm": 5.34375,
      "learning_rate": 3.205128205128206e-05,
      "loss": 0.1747,
      "step": 326
    },
    {
      "epoch": 0.08086785009861933,
      "grad_norm": 4.5,
      "learning_rate": 3.224852071005917e-05,
      "loss": 0.1636,
      "step": 328
    },
    {
      "epoch": 0.08136094674556213,
      "grad_norm": 2.21875,
      "learning_rate": 3.244575936883629e-05,
      "loss": 0.1737,
      "step": 330
    },
    {
      "epoch": 0.08185404339250493,
      "grad_norm": 7.28125,
      "learning_rate": 3.2642998027613417e-05,
      "loss": 0.1793,
      "step": 332
    },
    {
      "epoch": 0.08234714003944774,
      "grad_norm": 4.65625,
      "learning_rate": 3.2840236686390536e-05,
      "loss": 0.1743,
      "step": 334
    },
    {
      "epoch": 0.08284023668639054,
      "grad_norm": 4.21875,
      "learning_rate": 3.3037475345167656e-05,
      "loss": 0.1681,
      "step": 336
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 3.171875,
      "learning_rate": 3.3234714003944775e-05,
      "loss": 0.156,
      "step": 338
    },
    {
      "epoch": 0.08382642998027613,
      "grad_norm": 2.265625,
      "learning_rate": 3.3431952662721895e-05,
      "loss": 0.1664,
      "step": 340
    },
    {
      "epoch": 0.08431952662721894,
      "grad_norm": 2.09375,
      "learning_rate": 3.3629191321499014e-05,
      "loss": 0.1654,
      "step": 342
    },
    {
      "epoch": 0.08481262327416174,
      "grad_norm": 2.3125,
      "learning_rate": 3.3826429980276134e-05,
      "loss": 0.1582,
      "step": 344
    },
    {
      "epoch": 0.08530571992110454,
      "grad_norm": 3.859375,
      "learning_rate": 3.402366863905326e-05,
      "loss": 0.1593,
      "step": 346
    },
    {
      "epoch": 0.08579881656804733,
      "grad_norm": 3.0625,
      "learning_rate": 3.422090729783038e-05,
      "loss": 0.1671,
      "step": 348
    },
    {
      "epoch": 0.08629191321499013,
      "grad_norm": 1.9921875,
      "learning_rate": 3.441814595660749e-05,
      "loss": 0.1556,
      "step": 350
    },
    {
      "epoch": 0.08678500986193294,
      "grad_norm": 3.03125,
      "learning_rate": 3.461538461538462e-05,
      "loss": 0.1636,
      "step": 352
    },
    {
      "epoch": 0.08727810650887574,
      "grad_norm": 3.4375,
      "learning_rate": 3.481262327416174e-05,
      "loss": 0.1489,
      "step": 354
    },
    {
      "epoch": 0.08777120315581854,
      "grad_norm": 1.90625,
      "learning_rate": 3.500986193293886e-05,
      "loss": 0.1595,
      "step": 356
    },
    {
      "epoch": 0.08826429980276135,
      "grad_norm": 2.5625,
      "learning_rate": 3.520710059171598e-05,
      "loss": 0.1552,
      "step": 358
    },
    {
      "epoch": 0.08875739644970414,
      "grad_norm": 2.359375,
      "learning_rate": 3.54043392504931e-05,
      "loss": 0.1651,
      "step": 360
    },
    {
      "epoch": 0.08925049309664694,
      "grad_norm": 2.53125,
      "learning_rate": 3.5601577909270216e-05,
      "loss": 0.1632,
      "step": 362
    },
    {
      "epoch": 0.08974358974358974,
      "grad_norm": 2.296875,
      "learning_rate": 3.5798816568047336e-05,
      "loss": 0.1528,
      "step": 364
    },
    {
      "epoch": 0.09023668639053255,
      "grad_norm": 2.078125,
      "learning_rate": 3.599605522682446e-05,
      "loss": 0.15,
      "step": 366
    },
    {
      "epoch": 0.09072978303747535,
      "grad_norm": 1.8828125,
      "learning_rate": 3.619329388560158e-05,
      "loss": 0.1598,
      "step": 368
    },
    {
      "epoch": 0.09122287968441814,
      "grad_norm": 3.109375,
      "learning_rate": 3.63905325443787e-05,
      "loss": 0.1555,
      "step": 370
    },
    {
      "epoch": 0.09171597633136094,
      "grad_norm": 1.9609375,
      "learning_rate": 3.658777120315582e-05,
      "loss": 0.1466,
      "step": 372
    },
    {
      "epoch": 0.09220907297830375,
      "grad_norm": 3.25,
      "learning_rate": 3.678500986193294e-05,
      "loss": 0.1545,
      "step": 374
    },
    {
      "epoch": 0.09270216962524655,
      "grad_norm": 3.015625,
      "learning_rate": 3.698224852071006e-05,
      "loss": 0.1581,
      "step": 376
    },
    {
      "epoch": 0.09319526627218935,
      "grad_norm": 2.609375,
      "learning_rate": 3.717948717948718e-05,
      "loss": 0.1592,
      "step": 378
    },
    {
      "epoch": 0.09368836291913216,
      "grad_norm": 1.78125,
      "learning_rate": 3.7376725838264305e-05,
      "loss": 0.152,
      "step": 380
    },
    {
      "epoch": 0.09418145956607495,
      "grad_norm": 1.875,
      "learning_rate": 3.757396449704142e-05,
      "loss": 0.1502,
      "step": 382
    },
    {
      "epoch": 0.09467455621301775,
      "grad_norm": 2.140625,
      "learning_rate": 3.777120315581854e-05,
      "loss": 0.1627,
      "step": 384
    },
    {
      "epoch": 0.09516765285996055,
      "grad_norm": 1.65625,
      "learning_rate": 3.7968441814595664e-05,
      "loss": 0.1508,
      "step": 386
    },
    {
      "epoch": 0.09566074950690336,
      "grad_norm": 1.96875,
      "learning_rate": 3.8165680473372784e-05,
      "loss": 0.1508,
      "step": 388
    },
    {
      "epoch": 0.09615384615384616,
      "grad_norm": 2.390625,
      "learning_rate": 3.83629191321499e-05,
      "loss": 0.171,
      "step": 390
    },
    {
      "epoch": 0.09664694280078895,
      "grad_norm": 1.890625,
      "learning_rate": 3.856015779092703e-05,
      "loss": 0.1912,
      "step": 392
    },
    {
      "epoch": 0.09714003944773175,
      "grad_norm": 1.546875,
      "learning_rate": 3.875739644970414e-05,
      "loss": 0.1458,
      "step": 394
    },
    {
      "epoch": 0.09763313609467456,
      "grad_norm": 2.015625,
      "learning_rate": 3.895463510848126e-05,
      "loss": 0.1531,
      "step": 396
    },
    {
      "epoch": 0.09812623274161736,
      "grad_norm": 2.171875,
      "learning_rate": 3.915187376725839e-05,
      "loss": 0.2027,
      "step": 398
    },
    {
      "epoch": 0.09861932938856016,
      "grad_norm": 2.078125,
      "learning_rate": 3.934911242603551e-05,
      "loss": 0.1979,
      "step": 400
    },
    {
      "epoch": 0.09911242603550297,
      "grad_norm": 1.8203125,
      "learning_rate": 3.954635108481263e-05,
      "loss": 0.1393,
      "step": 402
    },
    {
      "epoch": 0.09960552268244575,
      "grad_norm": 2.65625,
      "learning_rate": 3.974358974358974e-05,
      "loss": 0.1335,
      "step": 404
    },
    {
      "epoch": 0.10009861932938856,
      "grad_norm": 2.578125,
      "learning_rate": 3.9940828402366866e-05,
      "loss": 0.1869,
      "step": 406
    },
    {
      "epoch": 0.10059171597633136,
      "grad_norm": 2.90625,
      "learning_rate": 4.0138067061143986e-05,
      "loss": 0.2058,
      "step": 408
    },
    {
      "epoch": 0.10108481262327416,
      "grad_norm": 2.65625,
      "learning_rate": 4.0335305719921105e-05,
      "loss": 0.133,
      "step": 410
    },
    {
      "epoch": 0.10157790927021697,
      "grad_norm": 2.09375,
      "learning_rate": 4.053254437869823e-05,
      "loss": 0.1377,
      "step": 412
    },
    {
      "epoch": 0.10207100591715976,
      "grad_norm": 2.15625,
      "learning_rate": 4.072978303747535e-05,
      "loss": 0.1909,
      "step": 414
    },
    {
      "epoch": 0.10256410256410256,
      "grad_norm": 2.25,
      "learning_rate": 4.0927021696252464e-05,
      "loss": 0.2018,
      "step": 416
    },
    {
      "epoch": 0.10305719921104536,
      "grad_norm": 3.28125,
      "learning_rate": 4.112426035502959e-05,
      "loss": 0.1427,
      "step": 418
    },
    {
      "epoch": 0.10355029585798817,
      "grad_norm": 2.578125,
      "learning_rate": 4.132149901380671e-05,
      "loss": 0.1332,
      "step": 420
    },
    {
      "epoch": 0.10404339250493097,
      "grad_norm": 2.734375,
      "learning_rate": 4.151873767258383e-05,
      "loss": 0.2182,
      "step": 422
    },
    {
      "epoch": 0.10453648915187377,
      "grad_norm": 1.640625,
      "learning_rate": 4.171597633136095e-05,
      "loss": 0.226,
      "step": 424
    },
    {
      "epoch": 0.10502958579881656,
      "grad_norm": 1.6328125,
      "learning_rate": 4.191321499013807e-05,
      "loss": 0.1437,
      "step": 426
    },
    {
      "epoch": 0.10552268244575937,
      "grad_norm": 1.390625,
      "learning_rate": 4.211045364891519e-05,
      "loss": 0.1255,
      "step": 428
    },
    {
      "epoch": 0.10601577909270217,
      "grad_norm": 1.6484375,
      "learning_rate": 4.230769230769231e-05,
      "loss": 0.2089,
      "step": 430
    },
    {
      "epoch": 0.10650887573964497,
      "grad_norm": 2.1875,
      "learning_rate": 4.2504930966469433e-05,
      "loss": 0.2151,
      "step": 432
    },
    {
      "epoch": 0.10700197238658778,
      "grad_norm": 1.4140625,
      "learning_rate": 4.270216962524655e-05,
      "loss": 0.136,
      "step": 434
    },
    {
      "epoch": 0.10749506903353057,
      "grad_norm": 2.015625,
      "learning_rate": 4.289940828402367e-05,
      "loss": 0.1381,
      "step": 436
    },
    {
      "epoch": 0.10798816568047337,
      "grad_norm": 2.8125,
      "learning_rate": 4.309664694280079e-05,
      "loss": 0.2191,
      "step": 438
    },
    {
      "epoch": 0.10848126232741617,
      "grad_norm": 2.875,
      "learning_rate": 4.329388560157791e-05,
      "loss": 0.217,
      "step": 440
    },
    {
      "epoch": 0.10897435897435898,
      "grad_norm": 2.109375,
      "learning_rate": 4.349112426035503e-05,
      "loss": 0.1527,
      "step": 442
    },
    {
      "epoch": 0.10946745562130178,
      "grad_norm": 1.4375,
      "learning_rate": 4.368836291913215e-05,
      "loss": 0.1538,
      "step": 444
    },
    {
      "epoch": 0.10996055226824458,
      "grad_norm": 2.046875,
      "learning_rate": 4.388560157790928e-05,
      "loss": 0.2244,
      "step": 446
    },
    {
      "epoch": 0.11045364891518737,
      "grad_norm": 2.453125,
      "learning_rate": 4.408284023668639e-05,
      "loss": 0.2452,
      "step": 448
    },
    {
      "epoch": 0.11094674556213018,
      "grad_norm": 1.921875,
      "learning_rate": 4.428007889546351e-05,
      "loss": 0.1577,
      "step": 450
    },
    {
      "epoch": 0.11143984220907298,
      "grad_norm": 2.21875,
      "learning_rate": 4.4477317554240635e-05,
      "loss": 0.1576,
      "step": 452
    },
    {
      "epoch": 0.11193293885601578,
      "grad_norm": 2.734375,
      "learning_rate": 4.4674556213017755e-05,
      "loss": 0.2118,
      "step": 454
    },
    {
      "epoch": 0.11242603550295859,
      "grad_norm": 2.296875,
      "learning_rate": 4.4871794871794874e-05,
      "loss": 0.2099,
      "step": 456
    },
    {
      "epoch": 0.11291913214990137,
      "grad_norm": 1.453125,
      "learning_rate": 4.5069033530571994e-05,
      "loss": 0.1398,
      "step": 458
    },
    {
      "epoch": 0.11341222879684418,
      "grad_norm": 1.6953125,
      "learning_rate": 4.5266272189349114e-05,
      "loss": 0.1405,
      "step": 460
    },
    {
      "epoch": 0.11390532544378698,
      "grad_norm": 2.28125,
      "learning_rate": 4.546351084812623e-05,
      "loss": 0.2018,
      "step": 462
    },
    {
      "epoch": 0.11439842209072978,
      "grad_norm": 2.125,
      "learning_rate": 4.566074950690335e-05,
      "loss": 0.1951,
      "step": 464
    },
    {
      "epoch": 0.11489151873767259,
      "grad_norm": 1.390625,
      "learning_rate": 4.585798816568048e-05,
      "loss": 0.1505,
      "step": 466
    },
    {
      "epoch": 0.11538461538461539,
      "grad_norm": 1.5546875,
      "learning_rate": 4.60552268244576e-05,
      "loss": 0.1527,
      "step": 468
    },
    {
      "epoch": 0.11587771203155818,
      "grad_norm": 2.203125,
      "learning_rate": 4.625246548323471e-05,
      "loss": 0.2001,
      "step": 470
    },
    {
      "epoch": 0.11637080867850098,
      "grad_norm": 3.140625,
      "learning_rate": 4.644970414201184e-05,
      "loss": 0.2114,
      "step": 472
    },
    {
      "epoch": 0.11686390532544379,
      "grad_norm": 2.9375,
      "learning_rate": 4.664694280078896e-05,
      "loss": 0.1362,
      "step": 474
    },
    {
      "epoch": 0.11735700197238659,
      "grad_norm": 3.0,
      "learning_rate": 4.6844181459566076e-05,
      "loss": 0.1709,
      "step": 476
    },
    {
      "epoch": 0.1178500986193294,
      "grad_norm": 1.640625,
      "learning_rate": 4.7041420118343196e-05,
      "loss": 0.1914,
      "step": 478
    },
    {
      "epoch": 0.11834319526627218,
      "grad_norm": 3.0,
      "learning_rate": 4.723865877712032e-05,
      "loss": 0.1906,
      "step": 480
    },
    {
      "epoch": 0.11883629191321499,
      "grad_norm": 2.46875,
      "learning_rate": 4.7435897435897435e-05,
      "loss": 0.1796,
      "step": 482
    },
    {
      "epoch": 0.11932938856015779,
      "grad_norm": 3.1875,
      "learning_rate": 4.7633136094674555e-05,
      "loss": 0.1555,
      "step": 484
    },
    {
      "epoch": 0.11982248520710059,
      "grad_norm": 3.234375,
      "learning_rate": 4.783037475345168e-05,
      "loss": 0.2049,
      "step": 486
    },
    {
      "epoch": 0.1203155818540434,
      "grad_norm": 2.046875,
      "learning_rate": 4.80276134122288e-05,
      "loss": 0.1921,
      "step": 488
    },
    {
      "epoch": 0.1208086785009862,
      "grad_norm": 1.78125,
      "learning_rate": 4.822485207100592e-05,
      "loss": 0.1553,
      "step": 490
    },
    {
      "epoch": 0.12130177514792899,
      "grad_norm": 2.140625,
      "learning_rate": 4.842209072978304e-05,
      "loss": 0.166,
      "step": 492
    },
    {
      "epoch": 0.12179487179487179,
      "grad_norm": 2.0625,
      "learning_rate": 4.861932938856016e-05,
      "loss": 0.1908,
      "step": 494
    },
    {
      "epoch": 0.1222879684418146,
      "grad_norm": 1.7734375,
      "learning_rate": 4.881656804733728e-05,
      "loss": 0.1874,
      "step": 496
    },
    {
      "epoch": 0.1227810650887574,
      "grad_norm": 1.7109375,
      "learning_rate": 4.90138067061144e-05,
      "loss": 0.1648,
      "step": 498
    },
    {
      "epoch": 0.1232741617357002,
      "grad_norm": 1.8515625,
      "learning_rate": 4.9211045364891524e-05,
      "loss": 0.16,
      "step": 500
    },
    {
      "epoch": 0.12376725838264299,
      "grad_norm": 1.828125,
      "learning_rate": 4.9408284023668644e-05,
      "loss": 0.196,
      "step": 502
    },
    {
      "epoch": 0.1242603550295858,
      "grad_norm": 2.625,
      "learning_rate": 4.9605522682445757e-05,
      "loss": 0.1931,
      "step": 504
    },
    {
      "epoch": 0.1247534516765286,
      "grad_norm": 1.734375,
      "learning_rate": 4.980276134122288e-05,
      "loss": 0.1683,
      "step": 506
    },
    {
      "epoch": 0.1252465483234714,
      "grad_norm": 1.9453125,
      "learning_rate": 5e-05,
      "loss": 0.1598,
      "step": 508
    },
    {
      "epoch": 0.1257396449704142,
      "grad_norm": 2.234375,
      "learning_rate": 5.019723865877712e-05,
      "loss": 0.1901,
      "step": 510
    },
    {
      "epoch": 0.126232741617357,
      "grad_norm": 2.703125,
      "learning_rate": 5.039447731755425e-05,
      "loss": 0.1916,
      "step": 512
    },
    {
      "epoch": 0.1267258382642998,
      "grad_norm": 2.375,
      "learning_rate": 5.059171597633137e-05,
      "loss": 0.1795,
      "step": 514
    },
    {
      "epoch": 0.12721893491124261,
      "grad_norm": 2.25,
      "learning_rate": 5.078895463510849e-05,
      "loss": 0.1779,
      "step": 516
    },
    {
      "epoch": 0.12771203155818542,
      "grad_norm": 2.703125,
      "learning_rate": 5.098619329388561e-05,
      "loss": 0.1942,
      "step": 518
    },
    {
      "epoch": 0.1282051282051282,
      "grad_norm": 3.015625,
      "learning_rate": 5.118343195266272e-05,
      "loss": 0.2064,
      "step": 520
    },
    {
      "epoch": 0.128698224852071,
      "grad_norm": 3.015625,
      "learning_rate": 5.138067061143984e-05,
      "loss": 0.1712,
      "step": 522
    },
    {
      "epoch": 0.1291913214990138,
      "grad_norm": 2.265625,
      "learning_rate": 5.157790927021696e-05,
      "loss": 0.1746,
      "step": 524
    },
    {
      "epoch": 0.1296844181459566,
      "grad_norm": 1.578125,
      "learning_rate": 5.1775147928994085e-05,
      "loss": 0.1744,
      "step": 526
    },
    {
      "epoch": 0.1301775147928994,
      "grad_norm": 2.515625,
      "learning_rate": 5.1972386587771204e-05,
      "loss": 0.1891,
      "step": 528
    },
    {
      "epoch": 0.1306706114398422,
      "grad_norm": 1.828125,
      "learning_rate": 5.2169625246548324e-05,
      "loss": 0.1584,
      "step": 530
    },
    {
      "epoch": 0.131163708086785,
      "grad_norm": 2.109375,
      "learning_rate": 5.236686390532545e-05,
      "loss": 0.1558,
      "step": 532
    },
    {
      "epoch": 0.13165680473372782,
      "grad_norm": 1.6484375,
      "learning_rate": 5.256410256410257e-05,
      "loss": 0.1771,
      "step": 534
    },
    {
      "epoch": 0.13214990138067062,
      "grad_norm": 1.8203125,
      "learning_rate": 5.276134122287969e-05,
      "loss": 0.1821,
      "step": 536
    },
    {
      "epoch": 0.13264299802761342,
      "grad_norm": 1.71875,
      "learning_rate": 5.295857988165681e-05,
      "loss": 0.167,
      "step": 538
    },
    {
      "epoch": 0.13313609467455623,
      "grad_norm": 1.8125,
      "learning_rate": 5.3155818540433935e-05,
      "loss": 0.1688,
      "step": 540
    },
    {
      "epoch": 0.133629191321499,
      "grad_norm": 1.734375,
      "learning_rate": 5.335305719921104e-05,
      "loss": 0.1876,
      "step": 542
    },
    {
      "epoch": 0.1341222879684418,
      "grad_norm": 1.671875,
      "learning_rate": 5.355029585798817e-05,
      "loss": 0.187,
      "step": 544
    },
    {
      "epoch": 0.1346153846153846,
      "grad_norm": 2.875,
      "learning_rate": 5.374753451676529e-05,
      "loss": 0.2008,
      "step": 546
    },
    {
      "epoch": 0.1351084812623274,
      "grad_norm": 3.34375,
      "learning_rate": 5.3944773175542406e-05,
      "loss": 0.1844,
      "step": 548
    },
    {
      "epoch": 0.13560157790927022,
      "grad_norm": 2.921875,
      "learning_rate": 5.4142011834319526e-05,
      "loss": 0.1954,
      "step": 550
    },
    {
      "epoch": 0.13609467455621302,
      "grad_norm": 2.6875,
      "learning_rate": 5.433925049309665e-05,
      "loss": 0.1857,
      "step": 552
    },
    {
      "epoch": 0.13658777120315582,
      "grad_norm": 2.21875,
      "learning_rate": 5.453648915187377e-05,
      "loss": 0.1744,
      "step": 554
    },
    {
      "epoch": 0.13708086785009863,
      "grad_norm": 2.40625,
      "learning_rate": 5.473372781065089e-05,
      "loss": 0.1661,
      "step": 556
    },
    {
      "epoch": 0.13757396449704143,
      "grad_norm": 3.46875,
      "learning_rate": 5.493096646942801e-05,
      "loss": 0.1865,
      "step": 558
    },
    {
      "epoch": 0.13806706114398423,
      "grad_norm": 1.7265625,
      "learning_rate": 5.512820512820514e-05,
      "loss": 0.1814,
      "step": 560
    },
    {
      "epoch": 0.13856015779092704,
      "grad_norm": 1.8046875,
      "learning_rate": 5.532544378698226e-05,
      "loss": 0.1552,
      "step": 562
    },
    {
      "epoch": 0.1390532544378698,
      "grad_norm": 1.8984375,
      "learning_rate": 5.552268244575937e-05,
      "loss": 0.1633,
      "step": 564
    },
    {
      "epoch": 0.13954635108481261,
      "grad_norm": 2.453125,
      "learning_rate": 5.571992110453649e-05,
      "loss": 0.1767,
      "step": 566
    },
    {
      "epoch": 0.14003944773175542,
      "grad_norm": 2.328125,
      "learning_rate": 5.591715976331361e-05,
      "loss": 0.184,
      "step": 568
    },
    {
      "epoch": 0.14053254437869822,
      "grad_norm": 2.328125,
      "learning_rate": 5.611439842209073e-05,
      "loss": 0.1704,
      "step": 570
    },
    {
      "epoch": 0.14102564102564102,
      "grad_norm": 1.984375,
      "learning_rate": 5.6311637080867854e-05,
      "loss": 0.1725,
      "step": 572
    },
    {
      "epoch": 0.14151873767258383,
      "grad_norm": 2.453125,
      "learning_rate": 5.6508875739644974e-05,
      "loss": 0.1971,
      "step": 574
    },
    {
      "epoch": 0.14201183431952663,
      "grad_norm": 1.375,
      "learning_rate": 5.670611439842209e-05,
      "loss": 0.1814,
      "step": 576
    },
    {
      "epoch": 0.14250493096646943,
      "grad_norm": 1.5390625,
      "learning_rate": 5.690335305719921e-05,
      "loss": 0.1756,
      "step": 578
    },
    {
      "epoch": 0.14299802761341224,
      "grad_norm": 1.5,
      "learning_rate": 5.710059171597634e-05,
      "loss": 0.18,
      "step": 580
    },
    {
      "epoch": 0.14349112426035504,
      "grad_norm": 2.203125,
      "learning_rate": 5.729783037475346e-05,
      "loss": 0.1866,
      "step": 582
    },
    {
      "epoch": 0.14398422090729784,
      "grad_norm": 1.8828125,
      "learning_rate": 5.749506903353058e-05,
      "loss": 0.1948,
      "step": 584
    },
    {
      "epoch": 0.14447731755424062,
      "grad_norm": 2.046875,
      "learning_rate": 5.769230769230769e-05,
      "loss": 0.1744,
      "step": 586
    },
    {
      "epoch": 0.14497041420118342,
      "grad_norm": 2.46875,
      "learning_rate": 5.788954635108481e-05,
      "loss": 0.1688,
      "step": 588
    },
    {
      "epoch": 0.14546351084812623,
      "grad_norm": 2.078125,
      "learning_rate": 5.808678500986193e-05,
      "loss": 0.1758,
      "step": 590
    },
    {
      "epoch": 0.14595660749506903,
      "grad_norm": 1.9765625,
      "learning_rate": 5.8284023668639056e-05,
      "loss": 0.1717,
      "step": 592
    },
    {
      "epoch": 0.14644970414201183,
      "grad_norm": 1.6953125,
      "learning_rate": 5.8481262327416176e-05,
      "loss": 0.1642,
      "step": 594
    },
    {
      "epoch": 0.14694280078895464,
      "grad_norm": 1.7734375,
      "learning_rate": 5.8678500986193295e-05,
      "loss": 0.1563,
      "step": 596
    },
    {
      "epoch": 0.14743589743589744,
      "grad_norm": 1.828125,
      "learning_rate": 5.8875739644970415e-05,
      "loss": 0.1838,
      "step": 598
    },
    {
      "epoch": 0.14792899408284024,
      "grad_norm": 1.8828125,
      "learning_rate": 5.907297830374754e-05,
      "loss": 0.1779,
      "step": 600
    },
    {
      "epoch": 0.14842209072978305,
      "grad_norm": 1.7890625,
      "learning_rate": 5.927021696252466e-05,
      "loss": 0.1821,
      "step": 602
    },
    {
      "epoch": 0.14891518737672585,
      "grad_norm": 1.8671875,
      "learning_rate": 5.946745562130178e-05,
      "loss": 0.1742,
      "step": 604
    },
    {
      "epoch": 0.14940828402366865,
      "grad_norm": 3.0,
      "learning_rate": 5.9664694280078906e-05,
      "loss": 0.1834,
      "step": 606
    },
    {
      "epoch": 0.14990138067061143,
      "grad_norm": 3.078125,
      "learning_rate": 5.986193293885601e-05,
      "loss": 0.1985,
      "step": 608
    },
    {
      "epoch": 0.15039447731755423,
      "grad_norm": 2.953125,
      "learning_rate": 6.005917159763313e-05,
      "loss": 0.168,
      "step": 610
    },
    {
      "epoch": 0.15088757396449703,
      "grad_norm": 2.234375,
      "learning_rate": 6.025641025641026e-05,
      "loss": 0.1617,
      "step": 612
    },
    {
      "epoch": 0.15138067061143984,
      "grad_norm": 2.21875,
      "learning_rate": 6.045364891518738e-05,
      "loss": 0.175,
      "step": 614
    },
    {
      "epoch": 0.15187376725838264,
      "grad_norm": 2.5,
      "learning_rate": 6.06508875739645e-05,
      "loss": 0.1662,
      "step": 616
    },
    {
      "epoch": 0.15236686390532544,
      "grad_norm": 2.859375,
      "learning_rate": 6.084812623274162e-05,
      "loss": 0.1673,
      "step": 618
    },
    {
      "epoch": 0.15285996055226825,
      "grad_norm": 1.8359375,
      "learning_rate": 6.104536489151874e-05,
      "loss": 0.1618,
      "step": 620
    },
    {
      "epoch": 0.15335305719921105,
      "grad_norm": 1.6484375,
      "learning_rate": 6.124260355029586e-05,
      "loss": 0.1645,
      "step": 622
    },
    {
      "epoch": 0.15384615384615385,
      "grad_norm": 2.171875,
      "learning_rate": 6.143984220907298e-05,
      "loss": 0.1726,
      "step": 624
    },
    {
      "epoch": 0.15433925049309666,
      "grad_norm": 2.875,
      "learning_rate": 6.16370808678501e-05,
      "loss": 0.1656,
      "step": 626
    },
    {
      "epoch": 0.15483234714003946,
      "grad_norm": 2.46875,
      "learning_rate": 6.183431952662722e-05,
      "loss": 0.1638,
      "step": 628
    },
    {
      "epoch": 0.15532544378698224,
      "grad_norm": 2.25,
      "learning_rate": 6.203155818540434e-05,
      "loss": 0.1641,
      "step": 630
    },
    {
      "epoch": 0.15581854043392504,
      "grad_norm": 2.546875,
      "learning_rate": 6.222879684418146e-05,
      "loss": 0.1713,
      "step": 632
    },
    {
      "epoch": 0.15631163708086784,
      "grad_norm": 2.984375,
      "learning_rate": 6.242603550295858e-05,
      "loss": 0.151,
      "step": 634
    },
    {
      "epoch": 0.15680473372781065,
      "grad_norm": 2.140625,
      "learning_rate": 6.26232741617357e-05,
      "loss": 0.1677,
      "step": 636
    },
    {
      "epoch": 0.15729783037475345,
      "grad_norm": 1.6875,
      "learning_rate": 6.282051282051282e-05,
      "loss": 0.1569,
      "step": 638
    },
    {
      "epoch": 0.15779092702169625,
      "grad_norm": 5.0625,
      "learning_rate": 6.301775147928994e-05,
      "loss": 0.1723,
      "step": 640
    },
    {
      "epoch": 0.15828402366863906,
      "grad_norm": 4.21875,
      "learning_rate": 6.321499013806707e-05,
      "loss": 0.1635,
      "step": 642
    },
    {
      "epoch": 0.15877712031558186,
      "grad_norm": 3.578125,
      "learning_rate": 6.341222879684419e-05,
      "loss": 0.1714,
      "step": 644
    },
    {
      "epoch": 0.15927021696252466,
      "grad_norm": 4.6875,
      "learning_rate": 6.360946745562131e-05,
      "loss": 0.1865,
      "step": 646
    },
    {
      "epoch": 0.15976331360946747,
      "grad_norm": 2.640625,
      "learning_rate": 6.380670611439843e-05,
      "loss": 0.1748,
      "step": 648
    },
    {
      "epoch": 0.16025641025641027,
      "grad_norm": 1.90625,
      "learning_rate": 6.400394477317555e-05,
      "loss": 0.16,
      "step": 650
    },
    {
      "epoch": 0.16074950690335305,
      "grad_norm": 3.484375,
      "learning_rate": 6.420118343195266e-05,
      "loss": 0.17,
      "step": 652
    },
    {
      "epoch": 0.16124260355029585,
      "grad_norm": 2.890625,
      "learning_rate": 6.439842209072979e-05,
      "loss": 0.1669,
      "step": 654
    },
    {
      "epoch": 0.16173570019723865,
      "grad_norm": 2.515625,
      "learning_rate": 6.459566074950691e-05,
      "loss": 0.1806,
      "step": 656
    },
    {
      "epoch": 0.16222879684418146,
      "grad_norm": 1.296875,
      "learning_rate": 6.479289940828403e-05,
      "loss": 0.1513,
      "step": 658
    },
    {
      "epoch": 0.16272189349112426,
      "grad_norm": 1.609375,
      "learning_rate": 6.499013806706115e-05,
      "loss": 0.1556,
      "step": 660
    },
    {
      "epoch": 0.16321499013806706,
      "grad_norm": 3.484375,
      "learning_rate": 6.518737672583827e-05,
      "loss": 0.1687,
      "step": 662
    },
    {
      "epoch": 0.16370808678500987,
      "grad_norm": 3.265625,
      "learning_rate": 6.538461538461539e-05,
      "loss": 0.1745,
      "step": 664
    },
    {
      "epoch": 0.16420118343195267,
      "grad_norm": 2.21875,
      "learning_rate": 6.55818540433925e-05,
      "loss": 0.1599,
      "step": 666
    },
    {
      "epoch": 0.16469428007889547,
      "grad_norm": 1.5390625,
      "learning_rate": 6.577909270216963e-05,
      "loss": 0.1574,
      "step": 668
    },
    {
      "epoch": 0.16518737672583828,
      "grad_norm": 1.7734375,
      "learning_rate": 6.597633136094676e-05,
      "loss": 0.1672,
      "step": 670
    },
    {
      "epoch": 0.16568047337278108,
      "grad_norm": 2.453125,
      "learning_rate": 6.617357001972388e-05,
      "loss": 0.1692,
      "step": 672
    },
    {
      "epoch": 0.16617357001972385,
      "grad_norm": 1.953125,
      "learning_rate": 6.637080867850098e-05,
      "loss": 0.1529,
      "step": 674
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 1.7734375,
      "learning_rate": 6.65680473372781e-05,
      "loss": 0.1656,
      "step": 676
    },
    {
      "epoch": 0.16715976331360946,
      "grad_norm": 1.4375,
      "learning_rate": 6.676528599605522e-05,
      "loss": 0.1726,
      "step": 678
    },
    {
      "epoch": 0.16765285996055226,
      "grad_norm": 1.3984375,
      "learning_rate": 6.696252465483234e-05,
      "loss": 0.1582,
      "step": 680
    },
    {
      "epoch": 0.16814595660749507,
      "grad_norm": 1.3984375,
      "learning_rate": 6.715976331360948e-05,
      "loss": 0.1487,
      "step": 682
    },
    {
      "epoch": 0.16863905325443787,
      "grad_norm": 1.9140625,
      "learning_rate": 6.73570019723866e-05,
      "loss": 0.1647,
      "step": 684
    },
    {
      "epoch": 0.16913214990138067,
      "grad_norm": 2.359375,
      "learning_rate": 6.755424063116371e-05,
      "loss": 0.168,
      "step": 686
    },
    {
      "epoch": 0.16962524654832348,
      "grad_norm": 3.171875,
      "learning_rate": 6.775147928994083e-05,
      "loss": 0.1776,
      "step": 688
    },
    {
      "epoch": 0.17011834319526628,
      "grad_norm": 2.625,
      "learning_rate": 6.794871794871795e-05,
      "loss": 0.167,
      "step": 690
    },
    {
      "epoch": 0.17061143984220908,
      "grad_norm": 3.4375,
      "learning_rate": 6.814595660749507e-05,
      "loss": 0.166,
      "step": 692
    },
    {
      "epoch": 0.1711045364891519,
      "grad_norm": 1.6953125,
      "learning_rate": 6.834319526627219e-05,
      "loss": 0.1699,
      "step": 694
    },
    {
      "epoch": 0.17159763313609466,
      "grad_norm": 5.9375,
      "learning_rate": 6.854043392504931e-05,
      "loss": 0.1815,
      "step": 696
    },
    {
      "epoch": 0.17209072978303747,
      "grad_norm": 2.65625,
      "learning_rate": 6.873767258382643e-05,
      "loss": 0.1712,
      "step": 698
    },
    {
      "epoch": 0.17258382642998027,
      "grad_norm": 2.28125,
      "learning_rate": 6.893491124260355e-05,
      "loss": 0.1548,
      "step": 700
    },
    {
      "epoch": 0.17307692307692307,
      "grad_norm": 3.140625,
      "learning_rate": 6.913214990138067e-05,
      "loss": 0.1726,
      "step": 702
    },
    {
      "epoch": 0.17357001972386588,
      "grad_norm": 4.15625,
      "learning_rate": 6.932938856015779e-05,
      "loss": 0.1736,
      "step": 704
    },
    {
      "epoch": 0.17406311637080868,
      "grad_norm": 2.015625,
      "learning_rate": 6.952662721893491e-05,
      "loss": 0.1688,
      "step": 706
    },
    {
      "epoch": 0.17455621301775148,
      "grad_norm": 3.8125,
      "learning_rate": 6.972386587771203e-05,
      "loss": 0.1568,
      "step": 708
    },
    {
      "epoch": 0.17504930966469429,
      "grad_norm": 2.90625,
      "learning_rate": 6.992110453648916e-05,
      "loss": 0.1761,
      "step": 710
    },
    {
      "epoch": 0.1755424063116371,
      "grad_norm": 2.1875,
      "learning_rate": 7.011834319526628e-05,
      "loss": 0.1637,
      "step": 712
    },
    {
      "epoch": 0.1760355029585799,
      "grad_norm": 2.84375,
      "learning_rate": 7.03155818540434e-05,
      "loss": 0.1639,
      "step": 714
    },
    {
      "epoch": 0.1765285996055227,
      "grad_norm": 3.453125,
      "learning_rate": 7.051282051282052e-05,
      "loss": 0.1599,
      "step": 716
    },
    {
      "epoch": 0.17702169625246547,
      "grad_norm": 2.6875,
      "learning_rate": 7.071005917159763e-05,
      "loss": 0.1634,
      "step": 718
    },
    {
      "epoch": 0.17751479289940827,
      "grad_norm": 4.0,
      "learning_rate": 7.090729783037475e-05,
      "loss": 0.1687,
      "step": 720
    },
    {
      "epoch": 0.17800788954635108,
      "grad_norm": 2.671875,
      "learning_rate": 7.110453648915188e-05,
      "loss": 0.1428,
      "step": 722
    },
    {
      "epoch": 0.17850098619329388,
      "grad_norm": 1.7734375,
      "learning_rate": 7.1301775147929e-05,
      "loss": 0.1493,
      "step": 724
    },
    {
      "epoch": 0.17899408284023668,
      "grad_norm": 2.703125,
      "learning_rate": 7.149901380670612e-05,
      "loss": 0.1508,
      "step": 726
    },
    {
      "epoch": 0.1794871794871795,
      "grad_norm": 2.859375,
      "learning_rate": 7.169625246548324e-05,
      "loss": 0.1567,
      "step": 728
    },
    {
      "epoch": 0.1799802761341223,
      "grad_norm": 1.703125,
      "learning_rate": 7.189349112426036e-05,
      "loss": 0.1558,
      "step": 730
    },
    {
      "epoch": 0.1804733727810651,
      "grad_norm": 3.515625,
      "learning_rate": 7.209072978303748e-05,
      "loss": 0.1595,
      "step": 732
    },
    {
      "epoch": 0.1809664694280079,
      "grad_norm": 2.75,
      "learning_rate": 7.22879684418146e-05,
      "loss": 0.1588,
      "step": 734
    },
    {
      "epoch": 0.1814595660749507,
      "grad_norm": 1.5078125,
      "learning_rate": 7.248520710059173e-05,
      "loss": 0.1501,
      "step": 736
    },
    {
      "epoch": 0.1819526627218935,
      "grad_norm": 3.734375,
      "learning_rate": 7.268244575936885e-05,
      "loss": 0.151,
      "step": 738
    },
    {
      "epoch": 0.18244575936883628,
      "grad_norm": 2.59375,
      "learning_rate": 7.287968441814596e-05,
      "loss": 0.1475,
      "step": 740
    },
    {
      "epoch": 0.18293885601577908,
      "grad_norm": 2.015625,
      "learning_rate": 7.307692307692307e-05,
      "loss": 0.163,
      "step": 742
    },
    {
      "epoch": 0.1834319526627219,
      "grad_norm": 3.015625,
      "learning_rate": 7.32741617357002e-05,
      "loss": 0.1619,
      "step": 744
    },
    {
      "epoch": 0.1839250493096647,
      "grad_norm": 2.15625,
      "learning_rate": 7.347140039447731e-05,
      "loss": 0.1452,
      "step": 746
    },
    {
      "epoch": 0.1844181459566075,
      "grad_norm": 1.453125,
      "learning_rate": 7.366863905325445e-05,
      "loss": 0.1522,
      "step": 748
    },
    {
      "epoch": 0.1849112426035503,
      "grad_norm": 2.484375,
      "learning_rate": 7.386587771203157e-05,
      "loss": 0.1681,
      "step": 750
    },
    {
      "epoch": 0.1854043392504931,
      "grad_norm": 2.71875,
      "learning_rate": 7.406311637080869e-05,
      "loss": 0.1552,
      "step": 752
    },
    {
      "epoch": 0.1858974358974359,
      "grad_norm": 1.5234375,
      "learning_rate": 7.42603550295858e-05,
      "loss": 0.1655,
      "step": 754
    },
    {
      "epoch": 0.1863905325443787,
      "grad_norm": 1.5546875,
      "learning_rate": 7.445759368836292e-05,
      "loss": 0.1508,
      "step": 756
    },
    {
      "epoch": 0.1868836291913215,
      "grad_norm": 1.7578125,
      "learning_rate": 7.465483234714004e-05,
      "loss": 0.1639,
      "step": 758
    },
    {
      "epoch": 0.1873767258382643,
      "grad_norm": 1.734375,
      "learning_rate": 7.485207100591716e-05,
      "loss": 0.1648,
      "step": 760
    },
    {
      "epoch": 0.1878698224852071,
      "grad_norm": 1.6953125,
      "learning_rate": 7.504930966469428e-05,
      "loss": 0.1603,
      "step": 762
    },
    {
      "epoch": 0.1883629191321499,
      "grad_norm": 1.4609375,
      "learning_rate": 7.52465483234714e-05,
      "loss": 0.1508,
      "step": 764
    },
    {
      "epoch": 0.1888560157790927,
      "grad_norm": 1.578125,
      "learning_rate": 7.544378698224852e-05,
      "loss": 0.1593,
      "step": 766
    },
    {
      "epoch": 0.1893491124260355,
      "grad_norm": 1.8515625,
      "learning_rate": 7.564102564102564e-05,
      "loss": 0.1725,
      "step": 768
    },
    {
      "epoch": 0.1898422090729783,
      "grad_norm": 1.75,
      "learning_rate": 7.583826429980276e-05,
      "loss": 0.1699,
      "step": 770
    },
    {
      "epoch": 0.1903353057199211,
      "grad_norm": 1.90625,
      "learning_rate": 7.603550295857988e-05,
      "loss": 0.1525,
      "step": 772
    },
    {
      "epoch": 0.1908284023668639,
      "grad_norm": 2.296875,
      "learning_rate": 7.6232741617357e-05,
      "loss": 0.1733,
      "step": 774
    },
    {
      "epoch": 0.1913214990138067,
      "grad_norm": 1.7109375,
      "learning_rate": 7.642998027613413e-05,
      "loss": 0.1543,
      "step": 776
    },
    {
      "epoch": 0.19181459566074952,
      "grad_norm": 1.8828125,
      "learning_rate": 7.662721893491125e-05,
      "loss": 0.1601,
      "step": 778
    },
    {
      "epoch": 0.19230769230769232,
      "grad_norm": 1.4375,
      "learning_rate": 7.682445759368837e-05,
      "loss": 0.155,
      "step": 780
    },
    {
      "epoch": 0.19280078895463512,
      "grad_norm": 1.3125,
      "learning_rate": 7.702169625246549e-05,
      "loss": 0.1585,
      "step": 782
    },
    {
      "epoch": 0.1932938856015779,
      "grad_norm": 1.515625,
      "learning_rate": 7.72189349112426e-05,
      "loss": 0.1551,
      "step": 784
    },
    {
      "epoch": 0.1937869822485207,
      "grad_norm": 1.5,
      "learning_rate": 7.741617357001972e-05,
      "loss": 0.1723,
      "step": 786
    },
    {
      "epoch": 0.1942800788954635,
      "grad_norm": 1.796875,
      "learning_rate": 7.761341222879685e-05,
      "loss": 0.1625,
      "step": 788
    },
    {
      "epoch": 0.1947731755424063,
      "grad_norm": 1.5859375,
      "learning_rate": 7.781065088757397e-05,
      "loss": 0.1655,
      "step": 790
    },
    {
      "epoch": 0.1952662721893491,
      "grad_norm": 1.4921875,
      "learning_rate": 7.800788954635109e-05,
      "loss": 0.1629,
      "step": 792
    },
    {
      "epoch": 0.1957593688362919,
      "grad_norm": 1.5234375,
      "learning_rate": 7.820512820512821e-05,
      "loss": 0.154,
      "step": 794
    },
    {
      "epoch": 0.19625246548323472,
      "grad_norm": 1.65625,
      "learning_rate": 7.840236686390533e-05,
      "loss": 0.1611,
      "step": 796
    },
    {
      "epoch": 0.19674556213017752,
      "grad_norm": 1.5859375,
      "learning_rate": 7.859960552268245e-05,
      "loss": 0.1661,
      "step": 798
    },
    {
      "epoch": 0.19723865877712032,
      "grad_norm": 1.5078125,
      "learning_rate": 7.879684418145957e-05,
      "loss": 0.1709,
      "step": 800
    },
    {
      "epoch": 0.19773175542406313,
      "grad_norm": 1.5078125,
      "learning_rate": 7.899408284023669e-05,
      "loss": 0.1501,
      "step": 802
    },
    {
      "epoch": 0.19822485207100593,
      "grad_norm": 1.2890625,
      "learning_rate": 7.919132149901382e-05,
      "loss": 0.1397,
      "step": 804
    },
    {
      "epoch": 0.1987179487179487,
      "grad_norm": 2.34375,
      "learning_rate": 7.938856015779093e-05,
      "loss": 0.1559,
      "step": 806
    },
    {
      "epoch": 0.1992110453648915,
      "grad_norm": 2.421875,
      "learning_rate": 7.958579881656805e-05,
      "loss": 0.1514,
      "step": 808
    },
    {
      "epoch": 0.1997041420118343,
      "grad_norm": 2.421875,
      "learning_rate": 7.978303747534517e-05,
      "loss": 0.1556,
      "step": 810
    },
    {
      "epoch": 0.20019723865877712,
      "grad_norm": 1.8671875,
      "learning_rate": 7.998027613412229e-05,
      "loss": 0.1565,
      "step": 812
    },
    {
      "epoch": 0.20069033530571992,
      "grad_norm": 1.65625,
      "learning_rate": 8.01775147928994e-05,
      "loss": 0.1601,
      "step": 814
    },
    {
      "epoch": 0.20118343195266272,
      "grad_norm": 1.8125,
      "learning_rate": 8.037475345167654e-05,
      "loss": 0.1457,
      "step": 816
    },
    {
      "epoch": 0.20167652859960553,
      "grad_norm": 1.5078125,
      "learning_rate": 8.057199211045366e-05,
      "loss": 0.1414,
      "step": 818
    },
    {
      "epoch": 0.20216962524654833,
      "grad_norm": 2.15625,
      "learning_rate": 8.076923076923078e-05,
      "loss": 0.1593,
      "step": 820
    },
    {
      "epoch": 0.20266272189349113,
      "grad_norm": 2.578125,
      "learning_rate": 8.09664694280079e-05,
      "loss": 0.1635,
      "step": 822
    },
    {
      "epoch": 0.20315581854043394,
      "grad_norm": 2.328125,
      "learning_rate": 8.116370808678502e-05,
      "loss": 0.152,
      "step": 824
    },
    {
      "epoch": 0.20364891518737674,
      "grad_norm": 1.9375,
      "learning_rate": 8.136094674556214e-05,
      "loss": 0.1458,
      "step": 826
    },
    {
      "epoch": 0.20414201183431951,
      "grad_norm": 1.78125,
      "learning_rate": 8.155818540433925e-05,
      "loss": 0.1348,
      "step": 828
    },
    {
      "epoch": 0.20463510848126232,
      "grad_norm": 1.5625,
      "learning_rate": 8.175542406311637e-05,
      "loss": 0.1542,
      "step": 830
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 1.7265625,
      "learning_rate": 8.19526627218935e-05,
      "loss": 0.1471,
      "step": 832
    },
    {
      "epoch": 0.20562130177514792,
      "grad_norm": 2.671875,
      "learning_rate": 8.214990138067061e-05,
      "loss": 0.1475,
      "step": 834
    },
    {
      "epoch": 0.20611439842209073,
      "grad_norm": 4.625,
      "learning_rate": 8.234714003944773e-05,
      "loss": 0.1558,
      "step": 836
    },
    {
      "epoch": 0.20660749506903353,
      "grad_norm": 3.125,
      "learning_rate": 8.254437869822485e-05,
      "loss": 0.1556,
      "step": 838
    },
    {
      "epoch": 0.20710059171597633,
      "grad_norm": 3.5625,
      "learning_rate": 8.274161735700197e-05,
      "loss": 0.157,
      "step": 840
    },
    {
      "epoch": 0.20759368836291914,
      "grad_norm": 2.3125,
      "learning_rate": 8.29388560157791e-05,
      "loss": 0.1477,
      "step": 842
    },
    {
      "epoch": 0.20808678500986194,
      "grad_norm": 1.28125,
      "learning_rate": 8.313609467455622e-05,
      "loss": 0.1421,
      "step": 844
    },
    {
      "epoch": 0.20857988165680474,
      "grad_norm": 3.515625,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.1524,
      "step": 846
    },
    {
      "epoch": 0.20907297830374755,
      "grad_norm": 4.4375,
      "learning_rate": 8.353057199211046e-05,
      "loss": 0.166,
      "step": 848
    },
    {
      "epoch": 0.20956607495069032,
      "grad_norm": 3.1875,
      "learning_rate": 8.372781065088757e-05,
      "loss": 0.1594,
      "step": 850
    },
    {
      "epoch": 0.21005917159763313,
      "grad_norm": 2.453125,
      "learning_rate": 8.392504930966469e-05,
      "loss": 0.1515,
      "step": 852
    },
    {
      "epoch": 0.21055226824457593,
      "grad_norm": 1.8203125,
      "learning_rate": 8.412228796844181e-05,
      "loss": 0.1649,
      "step": 854
    },
    {
      "epoch": 0.21104536489151873,
      "grad_norm": 1.0625,
      "learning_rate": 8.431952662721894e-05,
      "loss": 0.1576,
      "step": 856
    },
    {
      "epoch": 0.21153846153846154,
      "grad_norm": 2.890625,
      "learning_rate": 8.451676528599606e-05,
      "loss": 0.1414,
      "step": 858
    },
    {
      "epoch": 0.21203155818540434,
      "grad_norm": 3.71875,
      "learning_rate": 8.471400394477318e-05,
      "loss": 0.1465,
      "step": 860
    },
    {
      "epoch": 0.21252465483234714,
      "grad_norm": 2.296875,
      "learning_rate": 8.49112426035503e-05,
      "loss": 0.1575,
      "step": 862
    },
    {
      "epoch": 0.21301775147928995,
      "grad_norm": 3.109375,
      "learning_rate": 8.510848126232742e-05,
      "loss": 0.169,
      "step": 864
    },
    {
      "epoch": 0.21351084812623275,
      "grad_norm": 2.21875,
      "learning_rate": 8.530571992110454e-05,
      "loss": 0.1543,
      "step": 866
    },
    {
      "epoch": 0.21400394477317555,
      "grad_norm": 1.5390625,
      "learning_rate": 8.550295857988166e-05,
      "loss": 0.1435,
      "step": 868
    },
    {
      "epoch": 0.21449704142011836,
      "grad_norm": 2.078125,
      "learning_rate": 8.570019723865879e-05,
      "loss": 0.1407,
      "step": 870
    },
    {
      "epoch": 0.21499013806706113,
      "grad_norm": 2.21875,
      "learning_rate": 8.58974358974359e-05,
      "loss": 0.1517,
      "step": 872
    },
    {
      "epoch": 0.21548323471400394,
      "grad_norm": 1.0625,
      "learning_rate": 8.609467455621302e-05,
      "loss": 0.1461,
      "step": 874
    },
    {
      "epoch": 0.21597633136094674,
      "grad_norm": 1.5,
      "learning_rate": 8.629191321499014e-05,
      "loss": 0.1499,
      "step": 876
    },
    {
      "epoch": 0.21646942800788954,
      "grad_norm": 1.578125,
      "learning_rate": 8.648915187376726e-05,
      "loss": 0.1478,
      "step": 878
    },
    {
      "epoch": 0.21696252465483234,
      "grad_norm": 1.390625,
      "learning_rate": 8.668639053254438e-05,
      "loss": 0.147,
      "step": 880
    },
    {
      "epoch": 0.21745562130177515,
      "grad_norm": 1.671875,
      "learning_rate": 8.688362919132151e-05,
      "loss": 0.1499,
      "step": 882
    },
    {
      "epoch": 0.21794871794871795,
      "grad_norm": 1.265625,
      "learning_rate": 8.708086785009863e-05,
      "loss": 0.1557,
      "step": 884
    },
    {
      "epoch": 0.21844181459566075,
      "grad_norm": 2.28125,
      "learning_rate": 8.727810650887575e-05,
      "loss": 0.156,
      "step": 886
    },
    {
      "epoch": 0.21893491124260356,
      "grad_norm": 1.9921875,
      "learning_rate": 8.747534516765287e-05,
      "loss": 0.1606,
      "step": 888
    },
    {
      "epoch": 0.21942800788954636,
      "grad_norm": 1.8046875,
      "learning_rate": 8.767258382642999e-05,
      "loss": 0.1562,
      "step": 890
    },
    {
      "epoch": 0.21992110453648916,
      "grad_norm": 1.359375,
      "learning_rate": 8.78698224852071e-05,
      "loss": 0.1494,
      "step": 892
    },
    {
      "epoch": 0.22041420118343194,
      "grad_norm": 1.671875,
      "learning_rate": 8.806706114398423e-05,
      "loss": 0.1458,
      "step": 894
    },
    {
      "epoch": 0.22090729783037474,
      "grad_norm": 1.4609375,
      "learning_rate": 8.826429980276135e-05,
      "loss": 0.1437,
      "step": 896
    },
    {
      "epoch": 0.22140039447731755,
      "grad_norm": 1.2890625,
      "learning_rate": 8.846153846153847e-05,
      "loss": 0.1496,
      "step": 898
    },
    {
      "epoch": 0.22189349112426035,
      "grad_norm": 1.4140625,
      "learning_rate": 8.865877712031558e-05,
      "loss": 0.1478,
      "step": 900
    },
    {
      "epoch": 0.22238658777120315,
      "grad_norm": 1.3671875,
      "learning_rate": 8.88560157790927e-05,
      "loss": 0.1476,
      "step": 902
    },
    {
      "epoch": 0.22287968441814596,
      "grad_norm": 1.7734375,
      "learning_rate": 8.905325443786982e-05,
      "loss": 0.1506,
      "step": 904
    },
    {
      "epoch": 0.22337278106508876,
      "grad_norm": 1.2265625,
      "learning_rate": 8.925049309664694e-05,
      "loss": 0.1537,
      "step": 906
    },
    {
      "epoch": 0.22386587771203156,
      "grad_norm": 1.1484375,
      "learning_rate": 8.944773175542406e-05,
      "loss": 0.1467,
      "step": 908
    },
    {
      "epoch": 0.22435897435897437,
      "grad_norm": 1.6484375,
      "learning_rate": 8.96449704142012e-05,
      "loss": 0.1583,
      "step": 910
    },
    {
      "epoch": 0.22485207100591717,
      "grad_norm": 1.359375,
      "learning_rate": 8.984220907297832e-05,
      "loss": 0.1325,
      "step": 912
    },
    {
      "epoch": 0.22534516765285997,
      "grad_norm": 1.2734375,
      "learning_rate": 9.003944773175543e-05,
      "loss": 0.1554,
      "step": 914
    },
    {
      "epoch": 0.22583826429980275,
      "grad_norm": 1.484375,
      "learning_rate": 9.023668639053254e-05,
      "loss": 0.1487,
      "step": 916
    },
    {
      "epoch": 0.22633136094674555,
      "grad_norm": 2.109375,
      "learning_rate": 9.043392504930966e-05,
      "loss": 0.1478,
      "step": 918
    },
    {
      "epoch": 0.22682445759368836,
      "grad_norm": 1.703125,
      "learning_rate": 9.063116370808678e-05,
      "loss": 0.1496,
      "step": 920
    },
    {
      "epoch": 0.22731755424063116,
      "grad_norm": 1.375,
      "learning_rate": 9.082840236686391e-05,
      "loss": 0.1467,
      "step": 922
    },
    {
      "epoch": 0.22781065088757396,
      "grad_norm": 1.734375,
      "learning_rate": 9.102564102564103e-05,
      "loss": 0.1491,
      "step": 924
    },
    {
      "epoch": 0.22830374753451677,
      "grad_norm": 2.359375,
      "learning_rate": 9.122287968441815e-05,
      "loss": 0.1399,
      "step": 926
    },
    {
      "epoch": 0.22879684418145957,
      "grad_norm": 1.8984375,
      "learning_rate": 9.142011834319527e-05,
      "loss": 0.1434,
      "step": 928
    },
    {
      "epoch": 0.22928994082840237,
      "grad_norm": 1.40625,
      "learning_rate": 9.161735700197239e-05,
      "loss": 0.1514,
      "step": 930
    },
    {
      "epoch": 0.22978303747534518,
      "grad_norm": 1.203125,
      "learning_rate": 9.181459566074951e-05,
      "loss": 0.1448,
      "step": 932
    },
    {
      "epoch": 0.23027613412228798,
      "grad_norm": 1.2578125,
      "learning_rate": 9.201183431952663e-05,
      "loss": 0.153,
      "step": 934
    },
    {
      "epoch": 0.23076923076923078,
      "grad_norm": 1.421875,
      "learning_rate": 9.220907297830376e-05,
      "loss": 0.1439,
      "step": 936
    },
    {
      "epoch": 0.23126232741617356,
      "grad_norm": 1.4921875,
      "learning_rate": 9.240631163708087e-05,
      "loss": 0.1427,
      "step": 938
    },
    {
      "epoch": 0.23175542406311636,
      "grad_norm": 1.4765625,
      "learning_rate": 9.260355029585799e-05,
      "loss": 0.1484,
      "step": 940
    },
    {
      "epoch": 0.23224852071005916,
      "grad_norm": 1.2578125,
      "learning_rate": 9.280078895463511e-05,
      "loss": 0.1396,
      "step": 942
    },
    {
      "epoch": 0.23274161735700197,
      "grad_norm": 1.6640625,
      "learning_rate": 9.299802761341223e-05,
      "loss": 0.1492,
      "step": 944
    },
    {
      "epoch": 0.23323471400394477,
      "grad_norm": 1.1015625,
      "learning_rate": 9.319526627218935e-05,
      "loss": 0.1434,
      "step": 946
    },
    {
      "epoch": 0.23372781065088757,
      "grad_norm": 1.6484375,
      "learning_rate": 9.339250493096647e-05,
      "loss": 0.134,
      "step": 948
    },
    {
      "epoch": 0.23422090729783038,
      "grad_norm": 1.421875,
      "learning_rate": 9.35897435897436e-05,
      "loss": 0.144,
      "step": 950
    },
    {
      "epoch": 0.23471400394477318,
      "grad_norm": 1.3046875,
      "learning_rate": 9.378698224852072e-05,
      "loss": 0.1412,
      "step": 952
    },
    {
      "epoch": 0.23520710059171598,
      "grad_norm": 1.3828125,
      "learning_rate": 9.398422090729784e-05,
      "loss": 0.1347,
      "step": 954
    },
    {
      "epoch": 0.2357001972386588,
      "grad_norm": 1.7734375,
      "learning_rate": 9.418145956607496e-05,
      "loss": 0.1434,
      "step": 956
    },
    {
      "epoch": 0.2361932938856016,
      "grad_norm": 1.2734375,
      "learning_rate": 9.437869822485208e-05,
      "loss": 0.1425,
      "step": 958
    },
    {
      "epoch": 0.23668639053254437,
      "grad_norm": 1.28125,
      "learning_rate": 9.457593688362918e-05,
      "loss": 0.1494,
      "step": 960
    },
    {
      "epoch": 0.23717948717948717,
      "grad_norm": 1.6484375,
      "learning_rate": 9.477317554240632e-05,
      "loss": 0.1423,
      "step": 962
    },
    {
      "epoch": 0.23767258382642997,
      "grad_norm": 1.578125,
      "learning_rate": 9.497041420118344e-05,
      "loss": 0.1441,
      "step": 964
    },
    {
      "epoch": 0.23816568047337278,
      "grad_norm": 1.2890625,
      "learning_rate": 9.516765285996056e-05,
      "loss": 0.1389,
      "step": 966
    },
    {
      "epoch": 0.23865877712031558,
      "grad_norm": 1.2421875,
      "learning_rate": 9.536489151873768e-05,
      "loss": 0.1483,
      "step": 968
    },
    {
      "epoch": 0.23915187376725838,
      "grad_norm": 1.125,
      "learning_rate": 9.55621301775148e-05,
      "loss": 0.1383,
      "step": 970
    },
    {
      "epoch": 0.23964497041420119,
      "grad_norm": 1.1484375,
      "learning_rate": 9.575936883629191e-05,
      "loss": 0.14,
      "step": 972
    },
    {
      "epoch": 0.240138067061144,
      "grad_norm": 1.5078125,
      "learning_rate": 9.595660749506903e-05,
      "loss": 0.1458,
      "step": 974
    },
    {
      "epoch": 0.2406311637080868,
      "grad_norm": 1.1640625,
      "learning_rate": 9.615384615384617e-05,
      "loss": 0.1371,
      "step": 976
    },
    {
      "epoch": 0.2411242603550296,
      "grad_norm": 1.1484375,
      "learning_rate": 9.635108481262329e-05,
      "loss": 0.1391,
      "step": 978
    },
    {
      "epoch": 0.2416173570019724,
      "grad_norm": 1.34375,
      "learning_rate": 9.65483234714004e-05,
      "loss": 0.1414,
      "step": 980
    },
    {
      "epoch": 0.24211045364891517,
      "grad_norm": 1.4453125,
      "learning_rate": 9.674556213017751e-05,
      "loss": 0.1407,
      "step": 982
    },
    {
      "epoch": 0.24260355029585798,
      "grad_norm": 1.5078125,
      "learning_rate": 9.694280078895463e-05,
      "loss": 0.1461,
      "step": 984
    },
    {
      "epoch": 0.24309664694280078,
      "grad_norm": 1.3359375,
      "learning_rate": 9.714003944773175e-05,
      "loss": 0.1351,
      "step": 986
    },
    {
      "epoch": 0.24358974358974358,
      "grad_norm": 1.2109375,
      "learning_rate": 9.733727810650888e-05,
      "loss": 0.1358,
      "step": 988
    },
    {
      "epoch": 0.2440828402366864,
      "grad_norm": 1.0859375,
      "learning_rate": 9.7534516765286e-05,
      "loss": 0.1483,
      "step": 990
    },
    {
      "epoch": 0.2445759368836292,
      "grad_norm": 1.171875,
      "learning_rate": 9.773175542406312e-05,
      "loss": 0.1372,
      "step": 992
    },
    {
      "epoch": 0.245069033530572,
      "grad_norm": 1.625,
      "learning_rate": 9.792899408284024e-05,
      "loss": 0.1424,
      "step": 994
    },
    {
      "epoch": 0.2455621301775148,
      "grad_norm": 1.6640625,
      "learning_rate": 9.812623274161736e-05,
      "loss": 0.1356,
      "step": 996
    },
    {
      "epoch": 0.2460552268244576,
      "grad_norm": 1.0703125,
      "learning_rate": 9.832347140039448e-05,
      "loss": 0.1405,
      "step": 998
    },
    {
      "epoch": 0.2465483234714004,
      "grad_norm": 1.703125,
      "learning_rate": 9.85207100591716e-05,
      "loss": 0.1285,
      "step": 1000
    },
    {
      "epoch": 0.2470414201183432,
      "grad_norm": 1.390625,
      "learning_rate": 9.871794871794872e-05,
      "loss": 0.1395,
      "step": 1002
    },
    {
      "epoch": 0.24753451676528598,
      "grad_norm": 1.6484375,
      "learning_rate": 9.891518737672584e-05,
      "loss": 0.1501,
      "step": 1004
    },
    {
      "epoch": 0.2480276134122288,
      "grad_norm": 1.765625,
      "learning_rate": 9.911242603550296e-05,
      "loss": 0.133,
      "step": 1006
    },
    {
      "epoch": 0.2485207100591716,
      "grad_norm": 1.90625,
      "learning_rate": 9.930966469428008e-05,
      "loss": 0.1452,
      "step": 1008
    },
    {
      "epoch": 0.2490138067061144,
      "grad_norm": 1.2265625,
      "learning_rate": 9.95069033530572e-05,
      "loss": 0.1403,
      "step": 1010
    },
    {
      "epoch": 0.2495069033530572,
      "grad_norm": 1.5390625,
      "learning_rate": 9.970414201183432e-05,
      "loss": 0.1481,
      "step": 1012
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1875,
      "learning_rate": 9.990138067061144e-05,
      "loss": 0.1352,
      "step": 1014
    },
    {
      "epoch": 0.2504930966469428,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00010009861932938856,
      "loss": 0.1382,
      "step": 1016
    },
    {
      "epoch": 0.2509861932938856,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00010029585798816568,
      "loss": 0.1354,
      "step": 1018
    },
    {
      "epoch": 0.2514792899408284,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001004930966469428,
      "loss": 0.1311,
      "step": 1020
    },
    {
      "epoch": 0.2519723865877712,
      "grad_norm": 1.1875,
      "learning_rate": 0.00010069033530571992,
      "loss": 0.1415,
      "step": 1022
    },
    {
      "epoch": 0.252465483234714,
      "grad_norm": 2.3125,
      "learning_rate": 0.00010088757396449704,
      "loss": 0.135,
      "step": 1024
    },
    {
      "epoch": 0.2529585798816568,
      "grad_norm": 1.59375,
      "learning_rate": 0.00010108481262327416,
      "loss": 0.1361,
      "step": 1026
    },
    {
      "epoch": 0.2534516765285996,
      "grad_norm": 1.75,
      "learning_rate": 0.00010128205128205129,
      "loss": 0.1237,
      "step": 1028
    },
    {
      "epoch": 0.2539447731755424,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00010147928994082841,
      "loss": 0.138,
      "step": 1030
    },
    {
      "epoch": 0.25443786982248523,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00010167652859960553,
      "loss": 0.134,
      "step": 1032
    },
    {
      "epoch": 0.25493096646942803,
      "grad_norm": 1.296875,
      "learning_rate": 0.00010187376725838265,
      "loss": 0.1373,
      "step": 1034
    },
    {
      "epoch": 0.25542406311637084,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00010207100591715977,
      "loss": 0.1427,
      "step": 1036
    },
    {
      "epoch": 0.2559171597633136,
      "grad_norm": 1.015625,
      "learning_rate": 0.00010226824457593689,
      "loss": 0.1279,
      "step": 1038
    },
    {
      "epoch": 0.2564102564102564,
      "grad_norm": 1.0546875,
      "learning_rate": 0.000102465483234714,
      "loss": 0.1439,
      "step": 1040
    },
    {
      "epoch": 0.2569033530571992,
      "grad_norm": 1.4375,
      "learning_rate": 0.00010266272189349113,
      "loss": 0.1433,
      "step": 1042
    },
    {
      "epoch": 0.257396449704142,
      "grad_norm": 1.40625,
      "learning_rate": 0.00010285996055226826,
      "loss": 0.1435,
      "step": 1044
    },
    {
      "epoch": 0.2578895463510848,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00010305719921104538,
      "loss": 0.1434,
      "step": 1046
    },
    {
      "epoch": 0.2583826429980276,
      "grad_norm": 1.5546875,
      "learning_rate": 0.0001032544378698225,
      "loss": 0.1415,
      "step": 1048
    },
    {
      "epoch": 0.2588757396449704,
      "grad_norm": 1.546875,
      "learning_rate": 0.00010345167652859962,
      "loss": 0.1287,
      "step": 1050
    },
    {
      "epoch": 0.2593688362919132,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00010364891518737674,
      "loss": 0.1401,
      "step": 1052
    },
    {
      "epoch": 0.259861932938856,
      "grad_norm": 1.546875,
      "learning_rate": 0.00010384615384615386,
      "loss": 0.1298,
      "step": 1054
    },
    {
      "epoch": 0.2603550295857988,
      "grad_norm": 1.265625,
      "learning_rate": 0.00010404339250493098,
      "loss": 0.1391,
      "step": 1056
    },
    {
      "epoch": 0.2608481262327416,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010424063116370811,
      "loss": 0.1378,
      "step": 1058
    },
    {
      "epoch": 0.2613412228796844,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001044378698224852,
      "loss": 0.1277,
      "step": 1060
    },
    {
      "epoch": 0.2618343195266272,
      "grad_norm": 1.484375,
      "learning_rate": 0.00010463510848126232,
      "loss": 0.1371,
      "step": 1062
    },
    {
      "epoch": 0.26232741617357,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00010483234714003944,
      "loss": 0.138,
      "step": 1064
    },
    {
      "epoch": 0.26282051282051283,
      "grad_norm": 1.5625,
      "learning_rate": 0.00010502958579881656,
      "loss": 0.134,
      "step": 1066
    },
    {
      "epoch": 0.26331360946745563,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00010522682445759369,
      "loss": 0.1365,
      "step": 1068
    },
    {
      "epoch": 0.26380670611439844,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00010542406311637081,
      "loss": 0.1419,
      "step": 1070
    },
    {
      "epoch": 0.26429980276134124,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00010562130177514793,
      "loss": 0.1375,
      "step": 1072
    },
    {
      "epoch": 0.26479289940828404,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00010581854043392505,
      "loss": 0.1412,
      "step": 1074
    },
    {
      "epoch": 0.26528599605522685,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00010601577909270217,
      "loss": 0.1358,
      "step": 1076
    },
    {
      "epoch": 0.26577909270216965,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010621301775147929,
      "loss": 0.1412,
      "step": 1078
    },
    {
      "epoch": 0.26627218934911245,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00010641025641025641,
      "loss": 0.1411,
      "step": 1080
    },
    {
      "epoch": 0.2667652859960552,
      "grad_norm": 1.359375,
      "learning_rate": 0.00010660749506903354,
      "loss": 0.1345,
      "step": 1082
    },
    {
      "epoch": 0.267258382642998,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00010680473372781066,
      "loss": 0.1417,
      "step": 1084
    },
    {
      "epoch": 0.2677514792899408,
      "grad_norm": 1.671875,
      "learning_rate": 0.00010700197238658778,
      "loss": 0.1371,
      "step": 1086
    },
    {
      "epoch": 0.2682445759368836,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001071992110453649,
      "loss": 0.1473,
      "step": 1088
    },
    {
      "epoch": 0.2687376725838264,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00010739644970414202,
      "loss": 0.131,
      "step": 1090
    },
    {
      "epoch": 0.2692307692307692,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00010759368836291914,
      "loss": 0.1346,
      "step": 1092
    },
    {
      "epoch": 0.269723865877712,
      "grad_norm": 1.390625,
      "learning_rate": 0.00010779092702169626,
      "loss": 0.1338,
      "step": 1094
    },
    {
      "epoch": 0.2702169625246548,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00010798816568047338,
      "loss": 0.1239,
      "step": 1096
    },
    {
      "epoch": 0.27071005917159763,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00010818540433925051,
      "loss": 0.1402,
      "step": 1098
    },
    {
      "epoch": 0.27120315581854043,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00010838264299802763,
      "loss": 0.1334,
      "step": 1100
    },
    {
      "epoch": 0.27169625246548323,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010857988165680475,
      "loss": 0.1224,
      "step": 1102
    },
    {
      "epoch": 0.27218934911242604,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00010877712031558184,
      "loss": 0.1441,
      "step": 1104
    },
    {
      "epoch": 0.27268244575936884,
      "grad_norm": 0.953125,
      "learning_rate": 0.00010897435897435896,
      "loss": 0.1335,
      "step": 1106
    },
    {
      "epoch": 0.27317554240631164,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001091715976331361,
      "loss": 0.1385,
      "step": 1108
    },
    {
      "epoch": 0.27366863905325445,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00010936883629191322,
      "loss": 0.1362,
      "step": 1110
    },
    {
      "epoch": 0.27416173570019725,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00010956607495069034,
      "loss": 0.1331,
      "step": 1112
    },
    {
      "epoch": 0.27465483234714005,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00010976331360946746,
      "loss": 0.1225,
      "step": 1114
    },
    {
      "epoch": 0.27514792899408286,
      "grad_norm": 1.078125,
      "learning_rate": 0.00010996055226824457,
      "loss": 0.1287,
      "step": 1116
    },
    {
      "epoch": 0.27564102564102566,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001101577909270217,
      "loss": 0.1383,
      "step": 1118
    },
    {
      "epoch": 0.27613412228796846,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00011035502958579881,
      "loss": 0.1304,
      "step": 1120
    },
    {
      "epoch": 0.27662721893491127,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00011055226824457595,
      "loss": 0.1328,
      "step": 1122
    },
    {
      "epoch": 0.27712031558185407,
      "grad_norm": 1.984375,
      "learning_rate": 0.00011074950690335307,
      "loss": 0.1332,
      "step": 1124
    },
    {
      "epoch": 0.2776134122287968,
      "grad_norm": 1.703125,
      "learning_rate": 0.00011094674556213019,
      "loss": 0.1343,
      "step": 1126
    },
    {
      "epoch": 0.2781065088757396,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001111439842209073,
      "loss": 0.1338,
      "step": 1128
    },
    {
      "epoch": 0.2785996055226824,
      "grad_norm": 1.625,
      "learning_rate": 0.00011134122287968442,
      "loss": 0.122,
      "step": 1130
    },
    {
      "epoch": 0.27909270216962523,
      "grad_norm": 1.390625,
      "learning_rate": 0.00011153846153846154,
      "loss": 0.1247,
      "step": 1132
    },
    {
      "epoch": 0.27958579881656803,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00011173570019723866,
      "loss": 0.1259,
      "step": 1134
    },
    {
      "epoch": 0.28007889546351084,
      "grad_norm": 0.984375,
      "learning_rate": 0.00011193293885601578,
      "loss": 0.1254,
      "step": 1136
    },
    {
      "epoch": 0.28057199211045364,
      "grad_norm": 1.390625,
      "learning_rate": 0.00011213017751479292,
      "loss": 0.1319,
      "step": 1138
    },
    {
      "epoch": 0.28106508875739644,
      "grad_norm": 1.078125,
      "learning_rate": 0.00011232741617357004,
      "loss": 0.1331,
      "step": 1140
    },
    {
      "epoch": 0.28155818540433925,
      "grad_norm": 1.421875,
      "learning_rate": 0.00011252465483234716,
      "loss": 0.1289,
      "step": 1142
    },
    {
      "epoch": 0.28205128205128205,
      "grad_norm": 2.203125,
      "learning_rate": 0.00011272189349112428,
      "loss": 0.1333,
      "step": 1144
    },
    {
      "epoch": 0.28254437869822485,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001129191321499014,
      "loss": 0.1409,
      "step": 1146
    },
    {
      "epoch": 0.28303747534516766,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001131163708086785,
      "loss": 0.1285,
      "step": 1148
    },
    {
      "epoch": 0.28353057199211046,
      "grad_norm": 1.40625,
      "learning_rate": 0.00011331360946745562,
      "loss": 0.1388,
      "step": 1150
    },
    {
      "epoch": 0.28402366863905326,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00011351084812623274,
      "loss": 0.1263,
      "step": 1152
    },
    {
      "epoch": 0.28451676528599606,
      "grad_norm": 1.265625,
      "learning_rate": 0.00011370808678500986,
      "loss": 0.1346,
      "step": 1154
    },
    {
      "epoch": 0.28500986193293887,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00011390532544378698,
      "loss": 0.1284,
      "step": 1156
    },
    {
      "epoch": 0.28550295857988167,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001141025641025641,
      "loss": 0.1356,
      "step": 1158
    },
    {
      "epoch": 0.2859960552268245,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00011429980276134122,
      "loss": 0.1199,
      "step": 1160
    },
    {
      "epoch": 0.2864891518737673,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00011449704142011835,
      "loss": 0.1353,
      "step": 1162
    },
    {
      "epoch": 0.2869822485207101,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00011469428007889547,
      "loss": 0.1319,
      "step": 1164
    },
    {
      "epoch": 0.2874753451676529,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011489151873767259,
      "loss": 0.1319,
      "step": 1166
    },
    {
      "epoch": 0.2879684418145957,
      "grad_norm": 1.203125,
      "learning_rate": 0.00011508875739644971,
      "loss": 0.1363,
      "step": 1168
    },
    {
      "epoch": 0.28846153846153844,
      "grad_norm": 1.4765625,
      "learning_rate": 0.00011528599605522683,
      "loss": 0.1276,
      "step": 1170
    },
    {
      "epoch": 0.28895463510848124,
      "grad_norm": 1.078125,
      "learning_rate": 0.00011548323471400395,
      "loss": 0.1278,
      "step": 1172
    },
    {
      "epoch": 0.28944773175542404,
      "grad_norm": 1.5625,
      "learning_rate": 0.00011568047337278107,
      "loss": 0.1265,
      "step": 1174
    },
    {
      "epoch": 0.28994082840236685,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001158777120315582,
      "loss": 0.1336,
      "step": 1176
    },
    {
      "epoch": 0.29043392504930965,
      "grad_norm": 1.359375,
      "learning_rate": 0.00011607495069033532,
      "loss": 0.1248,
      "step": 1178
    },
    {
      "epoch": 0.29092702169625245,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00011627218934911244,
      "loss": 0.1286,
      "step": 1180
    },
    {
      "epoch": 0.29142011834319526,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00011646942800788956,
      "loss": 0.1299,
      "step": 1182
    },
    {
      "epoch": 0.29191321499013806,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00011666666666666668,
      "loss": 0.1266,
      "step": 1184
    },
    {
      "epoch": 0.29240631163708086,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001168639053254438,
      "loss": 0.1235,
      "step": 1186
    },
    {
      "epoch": 0.29289940828402367,
      "grad_norm": 1.21875,
      "learning_rate": 0.00011706114398422092,
      "loss": 0.1241,
      "step": 1188
    },
    {
      "epoch": 0.29339250493096647,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00011725838264299804,
      "loss": 0.1312,
      "step": 1190
    },
    {
      "epoch": 0.2938856015779093,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00011745562130177514,
      "loss": 0.1195,
      "step": 1192
    },
    {
      "epoch": 0.2943786982248521,
      "grad_norm": 1.5,
      "learning_rate": 0.00011765285996055226,
      "loss": 0.1281,
      "step": 1194
    },
    {
      "epoch": 0.2948717948717949,
      "grad_norm": 0.984375,
      "learning_rate": 0.00011785009861932938,
      "loss": 0.1321,
      "step": 1196
    },
    {
      "epoch": 0.2953648915187377,
      "grad_norm": 1.453125,
      "learning_rate": 0.0001180473372781065,
      "loss": 0.1341,
      "step": 1198
    },
    {
      "epoch": 0.2958579881656805,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00011824457593688362,
      "loss": 0.1284,
      "step": 1200
    },
    {
      "epoch": 0.2963510848126233,
      "grad_norm": 2.0,
      "learning_rate": 0.00011844181459566075,
      "loss": 0.1255,
      "step": 1202
    },
    {
      "epoch": 0.2968441814595661,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00011863905325443787,
      "loss": 0.1295,
      "step": 1204
    },
    {
      "epoch": 0.2973372781065089,
      "grad_norm": 2.3125,
      "learning_rate": 0.000118836291913215,
      "loss": 0.1218,
      "step": 1206
    },
    {
      "epoch": 0.2978303747534517,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00011903353057199211,
      "loss": 0.1285,
      "step": 1208
    },
    {
      "epoch": 0.2983234714003945,
      "grad_norm": 1.0,
      "learning_rate": 0.00011923076923076923,
      "loss": 0.1199,
      "step": 1210
    },
    {
      "epoch": 0.2988165680473373,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00011942800788954635,
      "loss": 0.1196,
      "step": 1212
    },
    {
      "epoch": 0.29930966469428005,
      "grad_norm": 1.921875,
      "learning_rate": 0.00011962524654832347,
      "loss": 0.1321,
      "step": 1214
    },
    {
      "epoch": 0.29980276134122286,
      "grad_norm": 1.71875,
      "learning_rate": 0.0001198224852071006,
      "loss": 0.1291,
      "step": 1216
    },
    {
      "epoch": 0.30029585798816566,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00012001972386587772,
      "loss": 0.1141,
      "step": 1218
    },
    {
      "epoch": 0.30078895463510846,
      "grad_norm": 1.25,
      "learning_rate": 0.00012021696252465484,
      "loss": 0.1292,
      "step": 1220
    },
    {
      "epoch": 0.30128205128205127,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00012041420118343196,
      "loss": 0.1319,
      "step": 1222
    },
    {
      "epoch": 0.30177514792899407,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00012061143984220908,
      "loss": 0.1188,
      "step": 1224
    },
    {
      "epoch": 0.3022682445759369,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001208086785009862,
      "loss": 0.1125,
      "step": 1226
    },
    {
      "epoch": 0.3027613412228797,
      "grad_norm": 0.8125,
      "learning_rate": 0.00012100591715976332,
      "loss": 0.1272,
      "step": 1228
    },
    {
      "epoch": 0.3032544378698225,
      "grad_norm": 1.6015625,
      "learning_rate": 0.00012120315581854044,
      "loss": 0.1318,
      "step": 1230
    },
    {
      "epoch": 0.3037475345167653,
      "grad_norm": 1.5,
      "learning_rate": 0.00012140039447731757,
      "loss": 0.1342,
      "step": 1232
    },
    {
      "epoch": 0.3042406311637081,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001215976331360947,
      "loss": 0.1276,
      "step": 1234
    },
    {
      "epoch": 0.3047337278106509,
      "grad_norm": 1.703125,
      "learning_rate": 0.00012179487179487179,
      "loss": 0.1244,
      "step": 1236
    },
    {
      "epoch": 0.3052268244575937,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001219921104536489,
      "loss": 0.1304,
      "step": 1238
    },
    {
      "epoch": 0.3057199211045365,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012218934911242604,
      "loss": 0.1272,
      "step": 1240
    },
    {
      "epoch": 0.3062130177514793,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00012238658777120315,
      "loss": 0.1243,
      "step": 1242
    },
    {
      "epoch": 0.3067061143984221,
      "grad_norm": 1.0625,
      "learning_rate": 0.00012258382642998028,
      "loss": 0.1207,
      "step": 1244
    },
    {
      "epoch": 0.3071992110453649,
      "grad_norm": 1.28125,
      "learning_rate": 0.00012278106508875738,
      "loss": 0.1261,
      "step": 1246
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00012297830374753452,
      "loss": 0.1288,
      "step": 1248
    },
    {
      "epoch": 0.3081854043392505,
      "grad_norm": 0.84375,
      "learning_rate": 0.00012317554240631165,
      "loss": 0.1266,
      "step": 1250
    },
    {
      "epoch": 0.3086785009861933,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00012337278106508876,
      "loss": 0.122,
      "step": 1252
    },
    {
      "epoch": 0.3091715976331361,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0001235700197238659,
      "loss": 0.1294,
      "step": 1254
    },
    {
      "epoch": 0.3096646942800789,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000123767258382643,
      "loss": 0.1355,
      "step": 1256
    },
    {
      "epoch": 0.31015779092702167,
      "grad_norm": 1.203125,
      "learning_rate": 0.00012396449704142013,
      "loss": 0.1309,
      "step": 1258
    },
    {
      "epoch": 0.3106508875739645,
      "grad_norm": 1.6171875,
      "learning_rate": 0.00012416173570019723,
      "loss": 0.1321,
      "step": 1260
    },
    {
      "epoch": 0.3111439842209073,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00012435897435897437,
      "loss": 0.1411,
      "step": 1262
    },
    {
      "epoch": 0.3116370808678501,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001245562130177515,
      "loss": 0.136,
      "step": 1264
    },
    {
      "epoch": 0.3121301775147929,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001247534516765286,
      "loss": 0.1326,
      "step": 1266
    },
    {
      "epoch": 0.3126232741617357,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00012495069033530574,
      "loss": 0.1317,
      "step": 1268
    },
    {
      "epoch": 0.3131163708086785,
      "grad_norm": 1.5,
      "learning_rate": 0.00012514792899408285,
      "loss": 0.1309,
      "step": 1270
    },
    {
      "epoch": 0.3136094674556213,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00012534516765285998,
      "loss": 0.1244,
      "step": 1272
    },
    {
      "epoch": 0.3141025641025641,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00012554240631163708,
      "loss": 0.1306,
      "step": 1274
    },
    {
      "epoch": 0.3145956607495069,
      "grad_norm": 1.546875,
      "learning_rate": 0.00012573964497041422,
      "loss": 0.1381,
      "step": 1276
    },
    {
      "epoch": 0.3150887573964497,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00012593688362919135,
      "loss": 0.1394,
      "step": 1278
    },
    {
      "epoch": 0.3155818540433925,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012613412228796843,
      "loss": 0.129,
      "step": 1280
    },
    {
      "epoch": 0.3160749506903353,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00012633136094674556,
      "loss": 0.1218,
      "step": 1282
    },
    {
      "epoch": 0.3165680473372781,
      "grad_norm": 1.078125,
      "learning_rate": 0.00012652859960552267,
      "loss": 0.128,
      "step": 1284
    },
    {
      "epoch": 0.3170611439842209,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001267258382642998,
      "loss": 0.1297,
      "step": 1286
    },
    {
      "epoch": 0.3175542406311637,
      "grad_norm": 1.40625,
      "learning_rate": 0.00012692307692307693,
      "loss": 0.123,
      "step": 1288
    },
    {
      "epoch": 0.3180473372781065,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00012712031558185404,
      "loss": 0.1238,
      "step": 1290
    },
    {
      "epoch": 0.3185404339250493,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00012731755424063117,
      "loss": 0.1181,
      "step": 1292
    },
    {
      "epoch": 0.31903353057199213,
      "grad_norm": 1.7265625,
      "learning_rate": 0.00012751479289940828,
      "loss": 0.1305,
      "step": 1294
    },
    {
      "epoch": 0.31952662721893493,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001277120315581854,
      "loss": 0.1229,
      "step": 1296
    },
    {
      "epoch": 0.32001972386587774,
      "grad_norm": 1.9296875,
      "learning_rate": 0.00012790927021696252,
      "loss": 0.132,
      "step": 1298
    },
    {
      "epoch": 0.32051282051282054,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00012810650887573965,
      "loss": 0.132,
      "step": 1300
    },
    {
      "epoch": 0.3210059171597633,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00012830374753451679,
      "loss": 0.1248,
      "step": 1302
    },
    {
      "epoch": 0.3214990138067061,
      "grad_norm": 1.4296875,
      "learning_rate": 0.0001285009861932939,
      "loss": 0.1265,
      "step": 1304
    },
    {
      "epoch": 0.3219921104536489,
      "grad_norm": 1.5390625,
      "learning_rate": 0.00012869822485207102,
      "loss": 0.1222,
      "step": 1306
    },
    {
      "epoch": 0.3224852071005917,
      "grad_norm": 1.390625,
      "learning_rate": 0.00012889546351084813,
      "loss": 0.1249,
      "step": 1308
    },
    {
      "epoch": 0.3229783037475345,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00012909270216962526,
      "loss": 0.1318,
      "step": 1310
    },
    {
      "epoch": 0.3234714003944773,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00012928994082840237,
      "loss": 0.1277,
      "step": 1312
    },
    {
      "epoch": 0.3239644970414201,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001294871794871795,
      "loss": 0.1192,
      "step": 1314
    },
    {
      "epoch": 0.3244575936883629,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001296844181459566,
      "loss": 0.1196,
      "step": 1316
    },
    {
      "epoch": 0.3249506903353057,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00012988165680473374,
      "loss": 0.1317,
      "step": 1318
    },
    {
      "epoch": 0.3254437869822485,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00013007889546351087,
      "loss": 0.1258,
      "step": 1320
    },
    {
      "epoch": 0.3259368836291913,
      "grad_norm": 1.09375,
      "learning_rate": 0.00013027613412228798,
      "loss": 0.1257,
      "step": 1322
    },
    {
      "epoch": 0.3264299802761341,
      "grad_norm": 1.328125,
      "learning_rate": 0.00013047337278106509,
      "loss": 0.1321,
      "step": 1324
    },
    {
      "epoch": 0.3269230769230769,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001306706114398422,
      "loss": 0.127,
      "step": 1326
    },
    {
      "epoch": 0.32741617357001973,
      "grad_norm": 1.046875,
      "learning_rate": 0.00013086785009861933,
      "loss": 0.1225,
      "step": 1328
    },
    {
      "epoch": 0.32790927021696253,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00013106508875739646,
      "loss": 0.1222,
      "step": 1330
    },
    {
      "epoch": 0.32840236686390534,
      "grad_norm": 1.34375,
      "learning_rate": 0.00013126232741617356,
      "loss": 0.1253,
      "step": 1332
    },
    {
      "epoch": 0.32889546351084814,
      "grad_norm": 1.75,
      "learning_rate": 0.0001314595660749507,
      "loss": 0.127,
      "step": 1334
    },
    {
      "epoch": 0.32938856015779094,
      "grad_norm": 1.5625,
      "learning_rate": 0.0001316568047337278,
      "loss": 0.1201,
      "step": 1336
    },
    {
      "epoch": 0.32988165680473375,
      "grad_norm": 1.15625,
      "learning_rate": 0.00013185404339250494,
      "loss": 0.1292,
      "step": 1338
    },
    {
      "epoch": 0.33037475345167655,
      "grad_norm": 1.78125,
      "learning_rate": 0.00013205128205128204,
      "loss": 0.1226,
      "step": 1340
    },
    {
      "epoch": 0.33086785009861935,
      "grad_norm": 1.671875,
      "learning_rate": 0.00013224852071005918,
      "loss": 0.1267,
      "step": 1342
    },
    {
      "epoch": 0.33136094674556216,
      "grad_norm": 1.578125,
      "learning_rate": 0.0001324457593688363,
      "loss": 0.1217,
      "step": 1344
    },
    {
      "epoch": 0.3318540433925049,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00013264299802761341,
      "loss": 0.1261,
      "step": 1346
    },
    {
      "epoch": 0.3323471400394477,
      "grad_norm": 1.375,
      "learning_rate": 0.00013284023668639055,
      "loss": 0.1227,
      "step": 1348
    },
    {
      "epoch": 0.3328402366863905,
      "grad_norm": 1.75,
      "learning_rate": 0.00013303747534516765,
      "loss": 0.1206,
      "step": 1350
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001332347140039448,
      "loss": 0.1186,
      "step": 1352
    },
    {
      "epoch": 0.3338264299802761,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001334319526627219,
      "loss": 0.1137,
      "step": 1354
    },
    {
      "epoch": 0.3343195266272189,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00013362919132149903,
      "loss": 0.1135,
      "step": 1356
    },
    {
      "epoch": 0.3348126232741617,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00013382642998027616,
      "loss": 0.1256,
      "step": 1358
    },
    {
      "epoch": 0.33530571992110453,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00013402366863905326,
      "loss": 0.1241,
      "step": 1360
    },
    {
      "epoch": 0.33579881656804733,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001342209072978304,
      "loss": 0.1126,
      "step": 1362
    },
    {
      "epoch": 0.33629191321499013,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001344181459566075,
      "loss": 0.128,
      "step": 1364
    },
    {
      "epoch": 0.33678500986193294,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013461538461538464,
      "loss": 0.1197,
      "step": 1366
    },
    {
      "epoch": 0.33727810650887574,
      "grad_norm": 1.375,
      "learning_rate": 0.00013481262327416174,
      "loss": 0.1326,
      "step": 1368
    },
    {
      "epoch": 0.33777120315581854,
      "grad_norm": 1.09375,
      "learning_rate": 0.00013500986193293885,
      "loss": 0.1232,
      "step": 1370
    },
    {
      "epoch": 0.33826429980276135,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00013520710059171598,
      "loss": 0.1286,
      "step": 1372
    },
    {
      "epoch": 0.33875739644970415,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001354043392504931,
      "loss": 0.1266,
      "step": 1374
    },
    {
      "epoch": 0.33925049309664695,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00013560157790927022,
      "loss": 0.1202,
      "step": 1376
    },
    {
      "epoch": 0.33974358974358976,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00013579881656804733,
      "loss": 0.1291,
      "step": 1378
    },
    {
      "epoch": 0.34023668639053256,
      "grad_norm": 1.0625,
      "learning_rate": 0.00013599605522682446,
      "loss": 0.1225,
      "step": 1380
    },
    {
      "epoch": 0.34072978303747536,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0001361932938856016,
      "loss": 0.1282,
      "step": 1382
    },
    {
      "epoch": 0.34122287968441817,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001363905325443787,
      "loss": 0.134,
      "step": 1384
    },
    {
      "epoch": 0.34171597633136097,
      "grad_norm": 1.21875,
      "learning_rate": 0.00013658777120315583,
      "loss": 0.1168,
      "step": 1386
    },
    {
      "epoch": 0.3422090729783038,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00013678500986193294,
      "loss": 0.124,
      "step": 1388
    },
    {
      "epoch": 0.3427021696252465,
      "grad_norm": 1.203125,
      "learning_rate": 0.00013698224852071007,
      "loss": 0.1212,
      "step": 1390
    },
    {
      "epoch": 0.3431952662721893,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00013717948717948718,
      "loss": 0.1229,
      "step": 1392
    },
    {
      "epoch": 0.34368836291913213,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001373767258382643,
      "loss": 0.1346,
      "step": 1394
    },
    {
      "epoch": 0.34418145956607493,
      "grad_norm": 1.546875,
      "learning_rate": 0.00013757396449704144,
      "loss": 0.1272,
      "step": 1396
    },
    {
      "epoch": 0.34467455621301774,
      "grad_norm": 1.0,
      "learning_rate": 0.00013777120315581855,
      "loss": 0.136,
      "step": 1398
    },
    {
      "epoch": 0.34516765285996054,
      "grad_norm": 1.5078125,
      "learning_rate": 0.00013796844181459568,
      "loss": 0.1259,
      "step": 1400
    },
    {
      "epoch": 0.34566074950690334,
      "grad_norm": 1.3515625,
      "learning_rate": 0.0001381656804733728,
      "loss": 0.129,
      "step": 1402
    },
    {
      "epoch": 0.34615384615384615,
      "grad_norm": 1.21875,
      "learning_rate": 0.00013836291913214992,
      "loss": 0.1246,
      "step": 1404
    },
    {
      "epoch": 0.34664694280078895,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00013856015779092703,
      "loss": 0.1292,
      "step": 1406
    },
    {
      "epoch": 0.34714003944773175,
      "grad_norm": 0.765625,
      "learning_rate": 0.00013875739644970416,
      "loss": 0.1219,
      "step": 1408
    },
    {
      "epoch": 0.34763313609467456,
      "grad_norm": 0.75,
      "learning_rate": 0.00013895463510848127,
      "loss": 0.1288,
      "step": 1410
    },
    {
      "epoch": 0.34812623274161736,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00013915187376725837,
      "loss": 0.111,
      "step": 1412
    },
    {
      "epoch": 0.34861932938856016,
      "grad_norm": 1.125,
      "learning_rate": 0.0001393491124260355,
      "loss": 0.124,
      "step": 1414
    },
    {
      "epoch": 0.34911242603550297,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001395463510848126,
      "loss": 0.1168,
      "step": 1416
    },
    {
      "epoch": 0.34960552268244577,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00013974358974358974,
      "loss": 0.1117,
      "step": 1418
    },
    {
      "epoch": 0.35009861932938857,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00013994082840236685,
      "loss": 0.118,
      "step": 1420
    },
    {
      "epoch": 0.3505917159763314,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014013806706114398,
      "loss": 0.1185,
      "step": 1422
    },
    {
      "epoch": 0.3510848126232742,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00014033530571992112,
      "loss": 0.1269,
      "step": 1424
    },
    {
      "epoch": 0.351577909270217,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014053254437869822,
      "loss": 0.124,
      "step": 1426
    },
    {
      "epoch": 0.3520710059171598,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014072978303747536,
      "loss": 0.1184,
      "step": 1428
    },
    {
      "epoch": 0.3525641025641026,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00014092702169625246,
      "loss": 0.115,
      "step": 1430
    },
    {
      "epoch": 0.3530571992110454,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001411242603550296,
      "loss": 0.125,
      "step": 1432
    },
    {
      "epoch": 0.35355029585798814,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001413214990138067,
      "loss": 0.1193,
      "step": 1434
    },
    {
      "epoch": 0.35404339250493094,
      "grad_norm": 1.796875,
      "learning_rate": 0.00014151873767258383,
      "loss": 0.116,
      "step": 1436
    },
    {
      "epoch": 0.35453648915187375,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014171597633136097,
      "loss": 0.122,
      "step": 1438
    },
    {
      "epoch": 0.35502958579881655,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00014191321499013807,
      "loss": 0.1187,
      "step": 1440
    },
    {
      "epoch": 0.35552268244575935,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001421104536489152,
      "loss": 0.115,
      "step": 1442
    },
    {
      "epoch": 0.35601577909270216,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001423076923076923,
      "loss": 0.1205,
      "step": 1444
    },
    {
      "epoch": 0.35650887573964496,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00014250493096646944,
      "loss": 0.1164,
      "step": 1446
    },
    {
      "epoch": 0.35700197238658776,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00014270216962524655,
      "loss": 0.1188,
      "step": 1448
    },
    {
      "epoch": 0.35749506903353057,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00014289940828402368,
      "loss": 0.1307,
      "step": 1450
    },
    {
      "epoch": 0.35798816568047337,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00014309664694280082,
      "loss": 0.1101,
      "step": 1452
    },
    {
      "epoch": 0.3584812623274162,
      "grad_norm": 1.34375,
      "learning_rate": 0.00014329388560157792,
      "loss": 0.1172,
      "step": 1454
    },
    {
      "epoch": 0.358974358974359,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014349112426035503,
      "loss": 0.1116,
      "step": 1456
    },
    {
      "epoch": 0.3594674556213018,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00014368836291913214,
      "loss": 0.1055,
      "step": 1458
    },
    {
      "epoch": 0.3599605522682446,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014388560157790927,
      "loss": 0.1086,
      "step": 1460
    },
    {
      "epoch": 0.3604536489151874,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001440828402366864,
      "loss": 0.1245,
      "step": 1462
    },
    {
      "epoch": 0.3609467455621302,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001442800788954635,
      "loss": 0.1289,
      "step": 1464
    },
    {
      "epoch": 0.361439842209073,
      "grad_norm": 1.71875,
      "learning_rate": 0.00014447731755424064,
      "loss": 0.1195,
      "step": 1466
    },
    {
      "epoch": 0.3619329388560158,
      "grad_norm": 1.5234375,
      "learning_rate": 0.00014467455621301775,
      "loss": 0.1061,
      "step": 1468
    },
    {
      "epoch": 0.3624260355029586,
      "grad_norm": 2.0,
      "learning_rate": 0.00014487179487179488,
      "loss": 0.122,
      "step": 1470
    },
    {
      "epoch": 0.3629191321499014,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014506903353057199,
      "loss": 0.1175,
      "step": 1472
    },
    {
      "epoch": 0.3634122287968442,
      "grad_norm": 3.640625,
      "learning_rate": 0.00014526627218934912,
      "loss": 0.1194,
      "step": 1474
    },
    {
      "epoch": 0.363905325443787,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014546351084812625,
      "loss": 0.1178,
      "step": 1476
    },
    {
      "epoch": 0.36439842209072976,
      "grad_norm": 3.515625,
      "learning_rate": 0.00014566074950690336,
      "loss": 0.1174,
      "step": 1478
    },
    {
      "epoch": 0.36489151873767256,
      "grad_norm": 1.375,
      "learning_rate": 0.0001458579881656805,
      "loss": 0.1106,
      "step": 1480
    },
    {
      "epoch": 0.36538461538461536,
      "grad_norm": 1.3671875,
      "learning_rate": 0.0001460552268244576,
      "loss": 0.1156,
      "step": 1482
    },
    {
      "epoch": 0.36587771203155817,
      "grad_norm": 1.78125,
      "learning_rate": 0.00014625246548323473,
      "loss": 0.1118,
      "step": 1484
    },
    {
      "epoch": 0.36637080867850097,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00014644970414201184,
      "loss": 0.1216,
      "step": 1486
    },
    {
      "epoch": 0.3668639053254438,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014664694280078897,
      "loss": 0.1094,
      "step": 1488
    },
    {
      "epoch": 0.3673570019723866,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001468441814595661,
      "loss": 0.1076,
      "step": 1490
    },
    {
      "epoch": 0.3678500986193294,
      "grad_norm": 0.875,
      "learning_rate": 0.0001470414201183432,
      "loss": 0.1038,
      "step": 1492
    },
    {
      "epoch": 0.3683431952662722,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014723865877712034,
      "loss": 0.1115,
      "step": 1494
    },
    {
      "epoch": 0.368836291913215,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00014743589743589745,
      "loss": 0.1114,
      "step": 1496
    },
    {
      "epoch": 0.3693293885601578,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014763313609467458,
      "loss": 0.1076,
      "step": 1498
    },
    {
      "epoch": 0.3698224852071006,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00014783037475345169,
      "loss": 0.1154,
      "step": 1500
    },
    {
      "epoch": 0.3703155818540434,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001480276134122288,
      "loss": 0.1143,
      "step": 1502
    },
    {
      "epoch": 0.3708086785009862,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014822485207100592,
      "loss": 0.1142,
      "step": 1504
    },
    {
      "epoch": 0.371301775147929,
      "grad_norm": 0.890625,
      "learning_rate": 0.00014842209072978303,
      "loss": 0.1226,
      "step": 1506
    },
    {
      "epoch": 0.3717948717948718,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014861932938856016,
      "loss": 0.117,
      "step": 1508
    },
    {
      "epoch": 0.3722879684418146,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00014881656804733727,
      "loss": 0.1203,
      "step": 1510
    },
    {
      "epoch": 0.3727810650887574,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001490138067061144,
      "loss": 0.1216,
      "step": 1512
    },
    {
      "epoch": 0.3732741617357002,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001492110453648915,
      "loss": 0.1219,
      "step": 1514
    },
    {
      "epoch": 0.373767258382643,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00014940828402366864,
      "loss": 0.1226,
      "step": 1516
    },
    {
      "epoch": 0.3742603550295858,
      "grad_norm": 1.265625,
      "learning_rate": 0.00014960552268244577,
      "loss": 0.1176,
      "step": 1518
    },
    {
      "epoch": 0.3747534516765286,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014980276134122288,
      "loss": 0.1111,
      "step": 1520
    },
    {
      "epoch": 0.3752465483234714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.1142,
      "step": 1522
    },
    {
      "epoch": 0.3757396449704142,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015019723865877712,
      "loss": 0.1113,
      "step": 1524
    },
    {
      "epoch": 0.376232741617357,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015039447731755425,
      "loss": 0.12,
      "step": 1526
    },
    {
      "epoch": 0.3767258382642998,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00015059171597633136,
      "loss": 0.1117,
      "step": 1528
    },
    {
      "epoch": 0.3772189349112426,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0001507889546351085,
      "loss": 0.1098,
      "step": 1530
    },
    {
      "epoch": 0.3777120315581854,
      "grad_norm": 1.140625,
      "learning_rate": 0.00015098619329388563,
      "loss": 0.1061,
      "step": 1532
    },
    {
      "epoch": 0.3782051282051282,
      "grad_norm": 1.375,
      "learning_rate": 0.00015118343195266273,
      "loss": 0.1158,
      "step": 1534
    },
    {
      "epoch": 0.378698224852071,
      "grad_norm": 1.40625,
      "learning_rate": 0.00015138067061143986,
      "loss": 0.1166,
      "step": 1536
    },
    {
      "epoch": 0.3791913214990138,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015157790927021697,
      "loss": 0.1183,
      "step": 1538
    },
    {
      "epoch": 0.3796844181459566,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001517751479289941,
      "loss": 0.1123,
      "step": 1540
    },
    {
      "epoch": 0.3801775147928994,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001519723865877712,
      "loss": 0.1142,
      "step": 1542
    },
    {
      "epoch": 0.3806706114398422,
      "grad_norm": 1.21875,
      "learning_rate": 0.00015216962524654834,
      "loss": 0.1165,
      "step": 1544
    },
    {
      "epoch": 0.381163708086785,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00015236686390532545,
      "loss": 0.1139,
      "step": 1546
    },
    {
      "epoch": 0.3816568047337278,
      "grad_norm": 1.296875,
      "learning_rate": 0.00015256410256410255,
      "loss": 0.1147,
      "step": 1548
    },
    {
      "epoch": 0.3821499013806706,
      "grad_norm": 1.421875,
      "learning_rate": 0.0001527613412228797,
      "loss": 0.1063,
      "step": 1550
    },
    {
      "epoch": 0.3826429980276134,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001529585798816568,
      "loss": 0.1122,
      "step": 1552
    },
    {
      "epoch": 0.3831360946745562,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015315581854043393,
      "loss": 0.1069,
      "step": 1554
    },
    {
      "epoch": 0.38362919132149903,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015335305719921106,
      "loss": 0.1088,
      "step": 1556
    },
    {
      "epoch": 0.38412228796844183,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00015355029585798817,
      "loss": 0.117,
      "step": 1558
    },
    {
      "epoch": 0.38461538461538464,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001537475345167653,
      "loss": 0.1204,
      "step": 1560
    },
    {
      "epoch": 0.38510848126232744,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001539447731755424,
      "loss": 0.1173,
      "step": 1562
    },
    {
      "epoch": 0.38560157790927024,
      "grad_norm": 1.03125,
      "learning_rate": 0.00015414201183431954,
      "loss": 0.1141,
      "step": 1564
    },
    {
      "epoch": 0.386094674556213,
      "grad_norm": 0.75,
      "learning_rate": 0.00015433925049309664,
      "loss": 0.1052,
      "step": 1566
    },
    {
      "epoch": 0.3865877712031558,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015453648915187378,
      "loss": 0.1028,
      "step": 1568
    },
    {
      "epoch": 0.3870808678500986,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001547337278106509,
      "loss": 0.1164,
      "step": 1570
    },
    {
      "epoch": 0.3875739644970414,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015493096646942802,
      "loss": 0.1104,
      "step": 1572
    },
    {
      "epoch": 0.3880670611439842,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015512820512820515,
      "loss": 0.1204,
      "step": 1574
    },
    {
      "epoch": 0.388560157790927,
      "grad_norm": 0.984375,
      "learning_rate": 0.00015532544378698225,
      "loss": 0.1249,
      "step": 1576
    },
    {
      "epoch": 0.3890532544378698,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001555226824457594,
      "loss": 0.1194,
      "step": 1578
    },
    {
      "epoch": 0.3895463510848126,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001557199211045365,
      "loss": 0.111,
      "step": 1580
    },
    {
      "epoch": 0.3900394477317554,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015591715976331363,
      "loss": 0.1095,
      "step": 1582
    },
    {
      "epoch": 0.3905325443786982,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015611439842209076,
      "loss": 0.1218,
      "step": 1584
    },
    {
      "epoch": 0.391025641025641,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015631163708086787,
      "loss": 0.1131,
      "step": 1586
    },
    {
      "epoch": 0.3915187376725838,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000156508875739645,
      "loss": 0.1128,
      "step": 1588
    },
    {
      "epoch": 0.39201183431952663,
      "grad_norm": 1.03125,
      "learning_rate": 0.00015670611439842208,
      "loss": 0.1062,
      "step": 1590
    },
    {
      "epoch": 0.39250493096646943,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001569033530571992,
      "loss": 0.1089,
      "step": 1592
    },
    {
      "epoch": 0.39299802761341224,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015710059171597634,
      "loss": 0.1064,
      "step": 1594
    },
    {
      "epoch": 0.39349112426035504,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015729783037475345,
      "loss": 0.1149,
      "step": 1596
    },
    {
      "epoch": 0.39398422090729784,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00015749506903353058,
      "loss": 0.1179,
      "step": 1598
    },
    {
      "epoch": 0.39447731755424065,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001576923076923077,
      "loss": 0.1149,
      "step": 1600
    },
    {
      "epoch": 0.39497041420118345,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015788954635108482,
      "loss": 0.1044,
      "step": 1602
    },
    {
      "epoch": 0.39546351084812625,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015808678500986193,
      "loss": 0.1062,
      "step": 1604
    },
    {
      "epoch": 0.39595660749506906,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00015828402366863906,
      "loss": 0.1162,
      "step": 1606
    },
    {
      "epoch": 0.39644970414201186,
      "grad_norm": 1.515625,
      "learning_rate": 0.00015848126232741617,
      "loss": 0.1087,
      "step": 1608
    },
    {
      "epoch": 0.3969428007889546,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001586785009861933,
      "loss": 0.1169,
      "step": 1610
    },
    {
      "epoch": 0.3974358974358974,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015887573964497043,
      "loss": 0.112,
      "step": 1612
    },
    {
      "epoch": 0.3979289940828402,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015907297830374754,
      "loss": 0.0958,
      "step": 1614
    },
    {
      "epoch": 0.398422090729783,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00015927021696252467,
      "loss": 0.1214,
      "step": 1616
    },
    {
      "epoch": 0.3989151873767258,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015946745562130178,
      "loss": 0.1087,
      "step": 1618
    },
    {
      "epoch": 0.3994082840236686,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001596646942800789,
      "loss": 0.107,
      "step": 1620
    },
    {
      "epoch": 0.39990138067061143,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015986193293885602,
      "loss": 0.1051,
      "step": 1622
    },
    {
      "epoch": 0.40039447731755423,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016005917159763315,
      "loss": 0.1053,
      "step": 1624
    },
    {
      "epoch": 0.40088757396449703,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016025641025641028,
      "loss": 0.1106,
      "step": 1626
    },
    {
      "epoch": 0.40138067061143984,
      "grad_norm": 1.4453125,
      "learning_rate": 0.0001604536489151874,
      "loss": 0.1124,
      "step": 1628
    },
    {
      "epoch": 0.40187376725838264,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00016065088757396452,
      "loss": 0.1169,
      "step": 1630
    },
    {
      "epoch": 0.40236686390532544,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016084812623274163,
      "loss": 0.1133,
      "step": 1632
    },
    {
      "epoch": 0.40285996055226825,
      "grad_norm": 0.859375,
      "learning_rate": 0.00016104536489151873,
      "loss": 0.1054,
      "step": 1634
    },
    {
      "epoch": 0.40335305719921105,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00016124260355029587,
      "loss": 0.1069,
      "step": 1636
    },
    {
      "epoch": 0.40384615384615385,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00016143984220907297,
      "loss": 0.1104,
      "step": 1638
    },
    {
      "epoch": 0.40433925049309666,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001616370808678501,
      "loss": 0.1023,
      "step": 1640
    },
    {
      "epoch": 0.40483234714003946,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0001618343195266272,
      "loss": 0.1058,
      "step": 1642
    },
    {
      "epoch": 0.40532544378698226,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00016203155818540435,
      "loss": 0.1085,
      "step": 1644
    },
    {
      "epoch": 0.40581854043392507,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016222879684418145,
      "loss": 0.1058,
      "step": 1646
    },
    {
      "epoch": 0.40631163708086787,
      "grad_norm": 1.53125,
      "learning_rate": 0.00016242603550295858,
      "loss": 0.1099,
      "step": 1648
    },
    {
      "epoch": 0.4068047337278107,
      "grad_norm": 1.6640625,
      "learning_rate": 0.00016262327416173572,
      "loss": 0.1087,
      "step": 1650
    },
    {
      "epoch": 0.4072978303747535,
      "grad_norm": 1.328125,
      "learning_rate": 0.00016282051282051282,
      "loss": 0.108,
      "step": 1652
    },
    {
      "epoch": 0.4077909270216962,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016301775147928996,
      "loss": 0.1091,
      "step": 1654
    },
    {
      "epoch": 0.40828402366863903,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00016321499013806706,
      "loss": 0.0988,
      "step": 1656
    },
    {
      "epoch": 0.40877712031558183,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001634122287968442,
      "loss": 0.1052,
      "step": 1658
    },
    {
      "epoch": 0.40927021696252464,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001636094674556213,
      "loss": 0.1074,
      "step": 1660
    },
    {
      "epoch": 0.40976331360946744,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00016380670611439843,
      "loss": 0.1022,
      "step": 1662
    },
    {
      "epoch": 0.41025641025641024,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016400394477317557,
      "loss": 0.1096,
      "step": 1664
    },
    {
      "epoch": 0.41074950690335305,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016420118343195267,
      "loss": 0.103,
      "step": 1666
    },
    {
      "epoch": 0.41124260355029585,
      "grad_norm": 0.875,
      "learning_rate": 0.0001643984220907298,
      "loss": 0.0972,
      "step": 1668
    },
    {
      "epoch": 0.41173570019723865,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001645956607495069,
      "loss": 0.1001,
      "step": 1670
    },
    {
      "epoch": 0.41222879684418146,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00016479289940828405,
      "loss": 0.1043,
      "step": 1672
    },
    {
      "epoch": 0.41272189349112426,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00016499013806706115,
      "loss": 0.1121,
      "step": 1674
    },
    {
      "epoch": 0.41321499013806706,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00016518737672583829,
      "loss": 0.1105,
      "step": 1676
    },
    {
      "epoch": 0.41370808678500987,
      "grad_norm": 0.75,
      "learning_rate": 0.0001653846153846154,
      "loss": 0.1062,
      "step": 1678
    },
    {
      "epoch": 0.41420118343195267,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001655818540433925,
      "loss": 0.1064,
      "step": 1680
    },
    {
      "epoch": 0.41469428007889547,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00016577909270216963,
      "loss": 0.1155,
      "step": 1682
    },
    {
      "epoch": 0.4151873767258383,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00016597633136094674,
      "loss": 0.1089,
      "step": 1684
    },
    {
      "epoch": 0.4156804733727811,
      "grad_norm": 0.796875,
      "learning_rate": 0.00016617357001972387,
      "loss": 0.1021,
      "step": 1686
    },
    {
      "epoch": 0.4161735700197239,
      "grad_norm": 0.8125,
      "learning_rate": 0.000166370808678501,
      "loss": 0.105,
      "step": 1688
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.75,
      "learning_rate": 0.0001665680473372781,
      "loss": 0.1018,
      "step": 1690
    },
    {
      "epoch": 0.4171597633136095,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016676528599605524,
      "loss": 0.1072,
      "step": 1692
    },
    {
      "epoch": 0.4176528599605523,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00016696252465483235,
      "loss": 0.1011,
      "step": 1694
    },
    {
      "epoch": 0.4181459566074951,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016715976331360948,
      "loss": 0.1107,
      "step": 1696
    },
    {
      "epoch": 0.41863905325443784,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016735700197238659,
      "loss": 0.0995,
      "step": 1698
    },
    {
      "epoch": 0.41913214990138065,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016755424063116372,
      "loss": 0.0969,
      "step": 1700
    },
    {
      "epoch": 0.41962524654832345,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016775147928994083,
      "loss": 0.1057,
      "step": 1702
    },
    {
      "epoch": 0.42011834319526625,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016794871794871796,
      "loss": 0.0933,
      "step": 1704
    },
    {
      "epoch": 0.42061143984220906,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001681459566074951,
      "loss": 0.1105,
      "step": 1706
    },
    {
      "epoch": 0.42110453648915186,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001683431952662722,
      "loss": 0.1093,
      "step": 1708
    },
    {
      "epoch": 0.42159763313609466,
      "grad_norm": 1.4609375,
      "learning_rate": 0.00016854043392504933,
      "loss": 0.1084,
      "step": 1710
    },
    {
      "epoch": 0.42209072978303747,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00016873767258382644,
      "loss": 0.1085,
      "step": 1712
    },
    {
      "epoch": 0.42258382642998027,
      "grad_norm": 1.1875,
      "learning_rate": 0.00016893491124260357,
      "loss": 0.1029,
      "step": 1714
    },
    {
      "epoch": 0.4230769230769231,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016913214990138068,
      "loss": 0.1156,
      "step": 1716
    },
    {
      "epoch": 0.4235700197238659,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001693293885601578,
      "loss": 0.1067,
      "step": 1718
    },
    {
      "epoch": 0.4240631163708087,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016952662721893494,
      "loss": 0.1072,
      "step": 1720
    },
    {
      "epoch": 0.4245562130177515,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016972386587771202,
      "loss": 0.1069,
      "step": 1722
    },
    {
      "epoch": 0.4250493096646943,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016992110453648915,
      "loss": 0.0973,
      "step": 1724
    },
    {
      "epoch": 0.4255424063116371,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017011834319526626,
      "loss": 0.1111,
      "step": 1726
    },
    {
      "epoch": 0.4260355029585799,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001703155818540434,
      "loss": 0.1096,
      "step": 1728
    },
    {
      "epoch": 0.4265285996055227,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017051282051282053,
      "loss": 0.108,
      "step": 1730
    },
    {
      "epoch": 0.4270216962524655,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017071005917159763,
      "loss": 0.107,
      "step": 1732
    },
    {
      "epoch": 0.4275147928994083,
      "grad_norm": 1.125,
      "learning_rate": 0.00017090729783037476,
      "loss": 0.1051,
      "step": 1734
    },
    {
      "epoch": 0.4280078895463511,
      "grad_norm": 0.875,
      "learning_rate": 0.00017110453648915187,
      "loss": 0.1009,
      "step": 1736
    },
    {
      "epoch": 0.4285009861932939,
      "grad_norm": 1.140625,
      "learning_rate": 0.000171301775147929,
      "loss": 0.1054,
      "step": 1738
    },
    {
      "epoch": 0.4289940828402367,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001714990138067061,
      "loss": 0.1056,
      "step": 1740
    },
    {
      "epoch": 0.42948717948717946,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017169625246548324,
      "loss": 0.1049,
      "step": 1742
    },
    {
      "epoch": 0.42998027613412226,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017189349112426038,
      "loss": 0.1014,
      "step": 1744
    },
    {
      "epoch": 0.43047337278106507,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017209072978303748,
      "loss": 0.0971,
      "step": 1746
    },
    {
      "epoch": 0.43096646942800787,
      "grad_norm": 1.4921875,
      "learning_rate": 0.00017228796844181461,
      "loss": 0.1064,
      "step": 1748
    },
    {
      "epoch": 0.4314595660749507,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017248520710059172,
      "loss": 0.1022,
      "step": 1750
    },
    {
      "epoch": 0.4319526627218935,
      "grad_norm": 0.609375,
      "learning_rate": 0.00017268244575936885,
      "loss": 0.0964,
      "step": 1752
    },
    {
      "epoch": 0.4324457593688363,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00017287968441814596,
      "loss": 0.1065,
      "step": 1754
    },
    {
      "epoch": 0.4329388560157791,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001730769230769231,
      "loss": 0.097,
      "step": 1756
    },
    {
      "epoch": 0.4334319526627219,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00017327416173570023,
      "loss": 0.1049,
      "step": 1758
    },
    {
      "epoch": 0.4339250493096647,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017347140039447733,
      "loss": 0.1005,
      "step": 1760
    },
    {
      "epoch": 0.4344181459566075,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017366863905325447,
      "loss": 0.1073,
      "step": 1762
    },
    {
      "epoch": 0.4349112426035503,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017386587771203157,
      "loss": 0.1075,
      "step": 1764
    },
    {
      "epoch": 0.4354043392504931,
      "grad_norm": 0.671875,
      "learning_rate": 0.00017406311637080868,
      "loss": 0.1033,
      "step": 1766
    },
    {
      "epoch": 0.4358974358974359,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001742603550295858,
      "loss": 0.105,
      "step": 1768
    },
    {
      "epoch": 0.4363905325443787,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017445759368836292,
      "loss": 0.1109,
      "step": 1770
    },
    {
      "epoch": 0.4368836291913215,
      "grad_norm": 1.609375,
      "learning_rate": 0.00017465483234714005,
      "loss": 0.1024,
      "step": 1772
    },
    {
      "epoch": 0.4373767258382643,
      "grad_norm": 1.578125,
      "learning_rate": 0.00017485207100591716,
      "loss": 0.0935,
      "step": 1774
    },
    {
      "epoch": 0.4378698224852071,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001750493096646943,
      "loss": 0.0939,
      "step": 1776
    },
    {
      "epoch": 0.4383629191321499,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001752465483234714,
      "loss": 0.0986,
      "step": 1778
    },
    {
      "epoch": 0.4388560157790927,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017544378698224853,
      "loss": 0.1001,
      "step": 1780
    },
    {
      "epoch": 0.4393491124260355,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017564102564102566,
      "loss": 0.1015,
      "step": 1782
    },
    {
      "epoch": 0.43984220907297833,
      "grad_norm": 1.28125,
      "learning_rate": 0.00017583826429980277,
      "loss": 0.0988,
      "step": 1784
    },
    {
      "epoch": 0.4403353057199211,
      "grad_norm": 1.3828125,
      "learning_rate": 0.0001760355029585799,
      "loss": 0.1007,
      "step": 1786
    },
    {
      "epoch": 0.4408284023668639,
      "grad_norm": 0.859375,
      "learning_rate": 0.000176232741617357,
      "loss": 0.1039,
      "step": 1788
    },
    {
      "epoch": 0.4413214990138067,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00017642998027613414,
      "loss": 0.0953,
      "step": 1790
    },
    {
      "epoch": 0.4418145956607495,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00017662721893491124,
      "loss": 0.0984,
      "step": 1792
    },
    {
      "epoch": 0.4423076923076923,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017682445759368838,
      "loss": 0.0975,
      "step": 1794
    },
    {
      "epoch": 0.4428007889546351,
      "grad_norm": 0.65625,
      "learning_rate": 0.00017702169625246548,
      "loss": 0.0921,
      "step": 1796
    },
    {
      "epoch": 0.4432938856015779,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017721893491124262,
      "loss": 0.0988,
      "step": 1798
    },
    {
      "epoch": 0.4437869822485207,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00017741617357001975,
      "loss": 0.0926,
      "step": 1800
    },
    {
      "epoch": 0.4442800788954635,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00017761341222879686,
      "loss": 0.1028,
      "step": 1802
    },
    {
      "epoch": 0.4447731755424063,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000177810650887574,
      "loss": 0.0946,
      "step": 1804
    },
    {
      "epoch": 0.4452662721893491,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001780078895463511,
      "loss": 0.1017,
      "step": 1806
    },
    {
      "epoch": 0.4457593688362919,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017820512820512823,
      "loss": 0.1024,
      "step": 1808
    },
    {
      "epoch": 0.4462524654832347,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017840236686390533,
      "loss": 0.0991,
      "step": 1810
    },
    {
      "epoch": 0.4467455621301775,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00017859960552268244,
      "loss": 0.0932,
      "step": 1812
    },
    {
      "epoch": 0.4472386587771203,
      "grad_norm": 1.21875,
      "learning_rate": 0.00017879684418145957,
      "loss": 0.111,
      "step": 1814
    },
    {
      "epoch": 0.4477317554240631,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00017899408284023668,
      "loss": 0.0943,
      "step": 1816
    },
    {
      "epoch": 0.44822485207100593,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001791913214990138,
      "loss": 0.0949,
      "step": 1818
    },
    {
      "epoch": 0.44871794871794873,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017938856015779092,
      "loss": 0.095,
      "step": 1820
    },
    {
      "epoch": 0.44921104536489154,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017958579881656805,
      "loss": 0.1036,
      "step": 1822
    },
    {
      "epoch": 0.44970414201183434,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017978303747534518,
      "loss": 0.0985,
      "step": 1824
    },
    {
      "epoch": 0.45019723865877714,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001799802761341223,
      "loss": 0.1005,
      "step": 1826
    },
    {
      "epoch": 0.45069033530571995,
      "grad_norm": 0.828125,
      "learning_rate": 0.00018017751479289942,
      "loss": 0.1037,
      "step": 1828
    },
    {
      "epoch": 0.4511834319526627,
      "grad_norm": 0.71875,
      "learning_rate": 0.00018037475345167653,
      "loss": 0.0914,
      "step": 1830
    },
    {
      "epoch": 0.4516765285996055,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018057199211045366,
      "loss": 0.1018,
      "step": 1832
    },
    {
      "epoch": 0.4521696252465483,
      "grad_norm": 1.40625,
      "learning_rate": 0.00018076923076923077,
      "loss": 0.1051,
      "step": 1834
    },
    {
      "epoch": 0.4526627218934911,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001809664694280079,
      "loss": 0.1053,
      "step": 1836
    },
    {
      "epoch": 0.4531558185404339,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018116370808678503,
      "loss": 0.0942,
      "step": 1838
    },
    {
      "epoch": 0.4536489151873767,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00018136094674556214,
      "loss": 0.0964,
      "step": 1840
    },
    {
      "epoch": 0.4541420118343195,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00018155818540433927,
      "loss": 0.1001,
      "step": 1842
    },
    {
      "epoch": 0.4546351084812623,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018175542406311638,
      "loss": 0.1023,
      "step": 1844
    },
    {
      "epoch": 0.4551282051282051,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001819526627218935,
      "loss": 0.0977,
      "step": 1846
    },
    {
      "epoch": 0.4556213017751479,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00018214990138067062,
      "loss": 0.0992,
      "step": 1848
    },
    {
      "epoch": 0.4561143984220907,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018234714003944775,
      "loss": 0.0949,
      "step": 1850
    },
    {
      "epoch": 0.45660749506903353,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00018254437869822488,
      "loss": 0.0998,
      "step": 1852
    },
    {
      "epoch": 0.45710059171597633,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018274161735700196,
      "loss": 0.1033,
      "step": 1854
    },
    {
      "epoch": 0.45759368836291914,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001829388560157791,
      "loss": 0.1019,
      "step": 1856
    },
    {
      "epoch": 0.45808678500986194,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001831360946745562,
      "loss": 0.1061,
      "step": 1858
    },
    {
      "epoch": 0.45857988165680474,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018333333333333334,
      "loss": 0.0965,
      "step": 1860
    },
    {
      "epoch": 0.45907297830374755,
      "grad_norm": 1.28125,
      "learning_rate": 0.00018353057199211047,
      "loss": 0.0915,
      "step": 1862
    },
    {
      "epoch": 0.45956607495069035,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018372781065088757,
      "loss": 0.1025,
      "step": 1864
    },
    {
      "epoch": 0.46005917159763315,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001839250493096647,
      "loss": 0.0973,
      "step": 1866
    },
    {
      "epoch": 0.46055226824457596,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001841222879684418,
      "loss": 0.0975,
      "step": 1868
    },
    {
      "epoch": 0.46104536489151876,
      "grad_norm": 1.125,
      "learning_rate": 0.00018431952662721895,
      "loss": 0.0943,
      "step": 1870
    },
    {
      "epoch": 0.46153846153846156,
      "grad_norm": 1.125,
      "learning_rate": 0.00018451676528599605,
      "loss": 0.1008,
      "step": 1872
    },
    {
      "epoch": 0.4620315581854043,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018471400394477319,
      "loss": 0.0985,
      "step": 1874
    },
    {
      "epoch": 0.4625246548323471,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018491124260355032,
      "loss": 0.0998,
      "step": 1876
    },
    {
      "epoch": 0.4630177514792899,
      "grad_norm": 0.765625,
      "learning_rate": 0.00018510848126232742,
      "loss": 0.0986,
      "step": 1878
    },
    {
      "epoch": 0.4635108481262327,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018530571992110456,
      "loss": 0.1121,
      "step": 1880
    },
    {
      "epoch": 0.4640039447731755,
      "grad_norm": 1.3125,
      "learning_rate": 0.00018550295857988166,
      "loss": 0.0962,
      "step": 1882
    },
    {
      "epoch": 0.46449704142011833,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001857001972386588,
      "loss": 0.1077,
      "step": 1884
    },
    {
      "epoch": 0.46499013806706113,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001858974358974359,
      "loss": 0.0981,
      "step": 1886
    },
    {
      "epoch": 0.46548323471400394,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00018609467455621304,
      "loss": 0.0962,
      "step": 1888
    },
    {
      "epoch": 0.46597633136094674,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018629191321499014,
      "loss": 0.0977,
      "step": 1890
    },
    {
      "epoch": 0.46646942800788954,
      "grad_norm": 1.515625,
      "learning_rate": 0.00018648915187376727,
      "loss": 0.1,
      "step": 1892
    },
    {
      "epoch": 0.46696252465483234,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001866863905325444,
      "loss": 0.0936,
      "step": 1894
    },
    {
      "epoch": 0.46745562130177515,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018688362919132151,
      "loss": 0.0998,
      "step": 1896
    },
    {
      "epoch": 0.46794871794871795,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00018708086785009862,
      "loss": 0.1021,
      "step": 1898
    },
    {
      "epoch": 0.46844181459566075,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018727810650887573,
      "loss": 0.1037,
      "step": 1900
    },
    {
      "epoch": 0.46893491124260356,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018747534516765286,
      "loss": 0.0964,
      "step": 1902
    },
    {
      "epoch": 0.46942800788954636,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00018767258382643,
      "loss": 0.0954,
      "step": 1904
    },
    {
      "epoch": 0.46992110453648916,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001878698224852071,
      "loss": 0.099,
      "step": 1906
    },
    {
      "epoch": 0.47041420118343197,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018806706114398423,
      "loss": 0.0974,
      "step": 1908
    },
    {
      "epoch": 0.47090729783037477,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018826429980276134,
      "loss": 0.0946,
      "step": 1910
    },
    {
      "epoch": 0.4714003944773176,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018846153846153847,
      "loss": 0.0954,
      "step": 1912
    },
    {
      "epoch": 0.4718934911242604,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018865877712031558,
      "loss": 0.0951,
      "step": 1914
    },
    {
      "epoch": 0.4723865877712032,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001888560157790927,
      "loss": 0.0978,
      "step": 1916
    },
    {
      "epoch": 0.47287968441814593,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00018905325443786984,
      "loss": 0.1012,
      "step": 1918
    },
    {
      "epoch": 0.47337278106508873,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018925049309664695,
      "loss": 0.0894,
      "step": 1920
    },
    {
      "epoch": 0.47386587771203154,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00018944773175542408,
      "loss": 0.0941,
      "step": 1922
    },
    {
      "epoch": 0.47435897435897434,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001896449704142012,
      "loss": 0.0935,
      "step": 1924
    },
    {
      "epoch": 0.47485207100591714,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00018984220907297832,
      "loss": 0.0959,
      "step": 1926
    },
    {
      "epoch": 0.47534516765285995,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019003944773175543,
      "loss": 0.1012,
      "step": 1928
    },
    {
      "epoch": 0.47583826429980275,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019023668639053256,
      "loss": 0.096,
      "step": 1930
    },
    {
      "epoch": 0.47633136094674555,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001904339250493097,
      "loss": 0.0945,
      "step": 1932
    },
    {
      "epoch": 0.47682445759368836,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001906311637080868,
      "loss": 0.0975,
      "step": 1934
    },
    {
      "epoch": 0.47731755424063116,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019082840236686393,
      "loss": 0.0996,
      "step": 1936
    },
    {
      "epoch": 0.47781065088757396,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019102564102564104,
      "loss": 0.0889,
      "step": 1938
    },
    {
      "epoch": 0.47830374753451677,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019122287968441817,
      "loss": 0.1021,
      "step": 1940
    },
    {
      "epoch": 0.47879684418145957,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019142011834319528,
      "loss": 0.0964,
      "step": 1942
    },
    {
      "epoch": 0.47928994082840237,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019161735700197238,
      "loss": 0.0911,
      "step": 1944
    },
    {
      "epoch": 0.4797830374753452,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019181459566074952,
      "loss": 0.092,
      "step": 1946
    },
    {
      "epoch": 0.480276134122288,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019201183431952662,
      "loss": 0.1005,
      "step": 1948
    },
    {
      "epoch": 0.4807692307692308,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019220907297830375,
      "loss": 0.0948,
      "step": 1950
    },
    {
      "epoch": 0.4812623274161736,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019240631163708086,
      "loss": 0.0947,
      "step": 1952
    },
    {
      "epoch": 0.4817554240631164,
      "grad_norm": 0.5390625,
      "learning_rate": 0.000192603550295858,
      "loss": 0.093,
      "step": 1954
    },
    {
      "epoch": 0.4822485207100592,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019280078895463513,
      "loss": 0.0904,
      "step": 1956
    },
    {
      "epoch": 0.482741617357002,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019299802761341223,
      "loss": 0.0892,
      "step": 1958
    },
    {
      "epoch": 0.4832347140039448,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019319526627218937,
      "loss": 0.098,
      "step": 1960
    },
    {
      "epoch": 0.48372781065088755,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019339250493096647,
      "loss": 0.0929,
      "step": 1962
    },
    {
      "epoch": 0.48422090729783035,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001935897435897436,
      "loss": 0.0985,
      "step": 1964
    },
    {
      "epoch": 0.48471400394477315,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001937869822485207,
      "loss": 0.0947,
      "step": 1966
    },
    {
      "epoch": 0.48520710059171596,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019398422090729784,
      "loss": 0.1011,
      "step": 1968
    },
    {
      "epoch": 0.48570019723865876,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019418145956607498,
      "loss": 0.0946,
      "step": 1970
    },
    {
      "epoch": 0.48619329388560156,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019437869822485208,
      "loss": 0.0891,
      "step": 1972
    },
    {
      "epoch": 0.48668639053254437,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019457593688362922,
      "loss": 0.0955,
      "step": 1974
    },
    {
      "epoch": 0.48717948717948717,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019477317554240632,
      "loss": 0.1009,
      "step": 1976
    },
    {
      "epoch": 0.48767258382643,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019497041420118345,
      "loss": 0.0999,
      "step": 1978
    },
    {
      "epoch": 0.4881656804733728,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019516765285996056,
      "loss": 0.092,
      "step": 1980
    },
    {
      "epoch": 0.4886587771203156,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001953648915187377,
      "loss": 0.0983,
      "step": 1982
    },
    {
      "epoch": 0.4891518737672584,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001955621301775148,
      "loss": 0.0983,
      "step": 1984
    },
    {
      "epoch": 0.4896449704142012,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001957593688362919,
      "loss": 0.1015,
      "step": 1986
    },
    {
      "epoch": 0.490138067061144,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019595660749506904,
      "loss": 0.0915,
      "step": 1988
    },
    {
      "epoch": 0.4906311637080868,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019615384615384615,
      "loss": 0.0947,
      "step": 1990
    },
    {
      "epoch": 0.4911242603550296,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019635108481262328,
      "loss": 0.0877,
      "step": 1992
    },
    {
      "epoch": 0.4916173570019724,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019654832347140038,
      "loss": 0.0931,
      "step": 1994
    },
    {
      "epoch": 0.4921104536489152,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019674556213017752,
      "loss": 0.0988,
      "step": 1996
    },
    {
      "epoch": 0.492603550295858,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019694280078895465,
      "loss": 0.0988,
      "step": 1998
    },
    {
      "epoch": 0.4930966469428008,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019714003944773176,
      "loss": 0.098,
      "step": 2000
    },
    {
      "epoch": 0.4935897435897436,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001973372781065089,
      "loss": 0.097,
      "step": 2002
    },
    {
      "epoch": 0.4940828402366864,
      "grad_norm": 0.734375,
      "learning_rate": 0.000197534516765286,
      "loss": 0.0859,
      "step": 2004
    },
    {
      "epoch": 0.49457593688362916,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019773175542406313,
      "loss": 0.0976,
      "step": 2006
    },
    {
      "epoch": 0.49506903353057197,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019792899408284023,
      "loss": 0.0956,
      "step": 2008
    },
    {
      "epoch": 0.49556213017751477,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019812623274161737,
      "loss": 0.1056,
      "step": 2010
    },
    {
      "epoch": 0.4960552268244576,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001983234714003945,
      "loss": 0.0916,
      "step": 2012
    },
    {
      "epoch": 0.4965483234714004,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001985207100591716,
      "loss": 0.1042,
      "step": 2014
    },
    {
      "epoch": 0.4970414201183432,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019871794871794874,
      "loss": 0.0963,
      "step": 2016
    },
    {
      "epoch": 0.497534516765286,
      "grad_norm": 1.0,
      "learning_rate": 0.00019891518737672585,
      "loss": 0.0916,
      "step": 2018
    },
    {
      "epoch": 0.4980276134122288,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019911242603550298,
      "loss": 0.0962,
      "step": 2020
    },
    {
      "epoch": 0.4985207100591716,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019930966469428008,
      "loss": 0.0963,
      "step": 2022
    },
    {
      "epoch": 0.4990138067061144,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019950690335305722,
      "loss": 0.0934,
      "step": 2024
    },
    {
      "epoch": 0.4995069033530572,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019970414201183435,
      "loss": 0.1109,
      "step": 2026
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019990138067061146,
      "loss": 0.1064,
      "step": 2028
    },
    {
      "epoch": 0.5004930966469427,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00019999999851868022,
      "loss": 0.0915,
      "step": 2030
    },
    {
      "epoch": 0.5009861932938856,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019999998666812206,
      "loss": 0.0898,
      "step": 2032
    },
    {
      "epoch": 0.5014792899408284,
      "grad_norm": 1.0,
      "learning_rate": 0.00019999996296700715,
      "loss": 0.0973,
      "step": 2034
    },
    {
      "epoch": 0.5019723865877712,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001999999274153383,
      "loss": 0.099,
      "step": 2036
    },
    {
      "epoch": 0.502465483234714,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019999988001311977,
      "loss": 0.0948,
      "step": 2038
    },
    {
      "epoch": 0.5029585798816568,
      "grad_norm": 0.875,
      "learning_rate": 0.00019999982076035705,
      "loss": 0.095,
      "step": 2040
    },
    {
      "epoch": 0.5034516765285996,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001999997496570573,
      "loss": 0.0993,
      "step": 2042
    },
    {
      "epoch": 0.5039447731755424,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019999966670322886,
      "loss": 0.0909,
      "step": 2044
    },
    {
      "epoch": 0.5044378698224852,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001999995718988816,
      "loss": 0.0937,
      "step": 2046
    },
    {
      "epoch": 0.504930966469428,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019999946524402678,
      "loss": 0.0921,
      "step": 2048
    },
    {
      "epoch": 0.5054240631163708,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019999934673867696,
      "loss": 0.0898,
      "step": 2050
    },
    {
      "epoch": 0.5059171597633136,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019999921638284623,
      "loss": 0.0971,
      "step": 2052
    },
    {
      "epoch": 0.5064102564102564,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019999907417655004,
      "loss": 0.0966,
      "step": 2054
    },
    {
      "epoch": 0.5069033530571992,
      "grad_norm": 1.34375,
      "learning_rate": 0.00019999892011980522,
      "loss": 0.094,
      "step": 2056
    },
    {
      "epoch": 0.507396449704142,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019999875421263007,
      "loss": 0.0972,
      "step": 2058
    },
    {
      "epoch": 0.5078895463510849,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019999857645504424,
      "loss": 0.0895,
      "step": 2060
    },
    {
      "epoch": 0.5083826429980276,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019999838684706873,
      "loss": 0.0905,
      "step": 2062
    },
    {
      "epoch": 0.5088757396449705,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001999981853887261,
      "loss": 0.092,
      "step": 2064
    },
    {
      "epoch": 0.5093688362919132,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019999797208004015,
      "loss": 0.0923,
      "step": 2066
    },
    {
      "epoch": 0.5098619329388561,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019999774692103621,
      "loss": 0.0854,
      "step": 2068
    },
    {
      "epoch": 0.5103550295857988,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019999750991174097,
      "loss": 0.0936,
      "step": 2070
    },
    {
      "epoch": 0.5108481262327417,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019999726105218244,
      "loss": 0.0993,
      "step": 2072
    },
    {
      "epoch": 0.5113412228796844,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001999970003423902,
      "loss": 0.0951,
      "step": 2074
    },
    {
      "epoch": 0.5118343195266272,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019999672778239506,
      "loss": 0.0932,
      "step": 2076
    },
    {
      "epoch": 0.51232741617357,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001999964433722294,
      "loss": 0.0897,
      "step": 2078
    },
    {
      "epoch": 0.5128205128205128,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001999961471119269,
      "loss": 0.0935,
      "step": 2080
    },
    {
      "epoch": 0.5133136094674556,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019999583900152265,
      "loss": 0.0988,
      "step": 2082
    },
    {
      "epoch": 0.5138067061143984,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019999551904105318,
      "loss": 0.0975,
      "step": 2084
    },
    {
      "epoch": 0.5142998027613412,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001999951872305564,
      "loss": 0.0878,
      "step": 2086
    },
    {
      "epoch": 0.514792899408284,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019999484357007162,
      "loss": 0.0938,
      "step": 2088
    },
    {
      "epoch": 0.5152859960552268,
      "grad_norm": 0.625,
      "learning_rate": 0.00019999448805963957,
      "loss": 0.093,
      "step": 2090
    },
    {
      "epoch": 0.5157790927021696,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001999941206993024,
      "loss": 0.0904,
      "step": 2092
    },
    {
      "epoch": 0.5162721893491125,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019999374148910363,
      "loss": 0.0945,
      "step": 2094
    },
    {
      "epoch": 0.5167652859960552,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019999335042908823,
      "loss": 0.0949,
      "step": 2096
    },
    {
      "epoch": 0.5172583826429981,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019999294751930249,
      "loss": 0.0929,
      "step": 2098
    },
    {
      "epoch": 0.5177514792899408,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019999253275979414,
      "loss": 0.0937,
      "step": 2100
    },
    {
      "epoch": 0.5182445759368837,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019999210615061243,
      "loss": 0.0948,
      "step": 2102
    },
    {
      "epoch": 0.5187376725838264,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019999166769180783,
      "loss": 0.086,
      "step": 2104
    },
    {
      "epoch": 0.5192307692307693,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001999912173834323,
      "loss": 0.0921,
      "step": 2106
    },
    {
      "epoch": 0.519723865877712,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001999907552255393,
      "loss": 0.0962,
      "step": 2108
    },
    {
      "epoch": 0.5202169625246549,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001999902812181835,
      "loss": 0.0926,
      "step": 2110
    },
    {
      "epoch": 0.5207100591715976,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019998979536142106,
      "loss": 0.094,
      "step": 2112
    },
    {
      "epoch": 0.5212031558185405,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019998929765530963,
      "loss": 0.0861,
      "step": 2114
    },
    {
      "epoch": 0.5216962524654832,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019998878809990815,
      "loss": 0.0889,
      "step": 2116
    },
    {
      "epoch": 0.522189349112426,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019998826669527704,
      "loss": 0.0878,
      "step": 2118
    },
    {
      "epoch": 0.5226824457593688,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019998773344147806,
      "loss": 0.0948,
      "step": 2120
    },
    {
      "epoch": 0.5231755424063116,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019998718833857436,
      "loss": 0.0876,
      "step": 2122
    },
    {
      "epoch": 0.5236686390532544,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019998663138663062,
      "loss": 0.0907,
      "step": 2124
    },
    {
      "epoch": 0.5241617357001972,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001999860625857128,
      "loss": 0.0962,
      "step": 2126
    },
    {
      "epoch": 0.52465483234714,
      "grad_norm": 1.4296875,
      "learning_rate": 0.00019998548193588832,
      "loss": 0.0937,
      "step": 2128
    },
    {
      "epoch": 0.5251479289940828,
      "grad_norm": 0.625,
      "learning_rate": 0.000199984889437226,
      "loss": 0.0892,
      "step": 2130
    },
    {
      "epoch": 0.5256410256410257,
      "grad_norm": 0.96484375,
      "learning_rate": 0.000199984285089796,
      "loss": 0.0952,
      "step": 2132
    },
    {
      "epoch": 0.5261341222879684,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019998366889366999,
      "loss": 0.0915,
      "step": 2134
    },
    {
      "epoch": 0.5266272189349113,
      "grad_norm": 0.921875,
      "learning_rate": 0.000199983040848921,
      "loss": 0.0932,
      "step": 2136
    },
    {
      "epoch": 0.527120315581854,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001999824009556234,
      "loss": 0.0945,
      "step": 2138
    },
    {
      "epoch": 0.5276134122287969,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019998174921385308,
      "loss": 0.0849,
      "step": 2140
    },
    {
      "epoch": 0.5281065088757396,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019998108562368727,
      "loss": 0.0874,
      "step": 2142
    },
    {
      "epoch": 0.5285996055226825,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019998041018520453,
      "loss": 0.0911,
      "step": 2144
    },
    {
      "epoch": 0.5290927021696252,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019997972289848503,
      "loss": 0.093,
      "step": 2146
    },
    {
      "epoch": 0.5295857988165681,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019997902376361012,
      "loss": 0.0845,
      "step": 2148
    },
    {
      "epoch": 0.5300788954635108,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001999783127806627,
      "loss": 0.0938,
      "step": 2150
    },
    {
      "epoch": 0.5305719921104537,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019997758994972698,
      "loss": 0.0968,
      "step": 2152
    },
    {
      "epoch": 0.5310650887573964,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019997685527088865,
      "loss": 0.0914,
      "step": 2154
    },
    {
      "epoch": 0.5315581854043393,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001999761087442348,
      "loss": 0.0988,
      "step": 2156
    },
    {
      "epoch": 0.532051282051282,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001999753503698538,
      "loss": 0.0863,
      "step": 2158
    },
    {
      "epoch": 0.5325443786982249,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019997458014783568,
      "loss": 0.0873,
      "step": 2160
    },
    {
      "epoch": 0.5330374753451677,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019997379807827156,
      "loss": 0.0884,
      "step": 2162
    },
    {
      "epoch": 0.5335305719921104,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019997300416125423,
      "loss": 0.0893,
      "step": 2164
    },
    {
      "epoch": 0.5340236686390533,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001999721983968777,
      "loss": 0.0854,
      "step": 2166
    },
    {
      "epoch": 0.534516765285996,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019997138078523753,
      "loss": 0.0848,
      "step": 2168
    },
    {
      "epoch": 0.5350098619329389,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001999705513264305,
      "loss": 0.0804,
      "step": 2170
    },
    {
      "epoch": 0.5355029585798816,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019996971002055503,
      "loss": 0.094,
      "step": 2172
    },
    {
      "epoch": 0.5359960552268245,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019996885686771073,
      "loss": 0.0887,
      "step": 2174
    },
    {
      "epoch": 0.5364891518737672,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019996799186799876,
      "loss": 0.0894,
      "step": 2176
    },
    {
      "epoch": 0.5369822485207101,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019996711502152158,
      "loss": 0.0846,
      "step": 2178
    },
    {
      "epoch": 0.5374753451676528,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019996622632838316,
      "loss": 0.0841,
      "step": 2180
    },
    {
      "epoch": 0.5379684418145957,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019996532578868875,
      "loss": 0.0875,
      "step": 2182
    },
    {
      "epoch": 0.5384615384615384,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001999644134025451,
      "loss": 0.0869,
      "step": 2184
    },
    {
      "epoch": 0.5389546351084813,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019996348917006036,
      "loss": 0.0937,
      "step": 2186
    },
    {
      "epoch": 0.539447731755424,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000199962553091344,
      "loss": 0.0948,
      "step": 2188
    },
    {
      "epoch": 0.5399408284023669,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019996160516650696,
      "loss": 0.0935,
      "step": 2190
    },
    {
      "epoch": 0.5404339250493096,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019996064539566164,
      "loss": 0.0864,
      "step": 2192
    },
    {
      "epoch": 0.5409270216962525,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019995967377892172,
      "loss": 0.0925,
      "step": 2194
    },
    {
      "epoch": 0.5414201183431953,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019995869031640235,
      "loss": 0.0866,
      "step": 2196
    },
    {
      "epoch": 0.5419132149901381,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001999576950082201,
      "loss": 0.0842,
      "step": 2198
    },
    {
      "epoch": 0.5424063116370809,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019995668785449284,
      "loss": 0.0914,
      "step": 2200
    },
    {
      "epoch": 0.5428994082840237,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019995566885534002,
      "loss": 0.0923,
      "step": 2202
    },
    {
      "epoch": 0.5433925049309665,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019995463801088237,
      "loss": 0.0869,
      "step": 2204
    },
    {
      "epoch": 0.5438856015779092,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000199953595321242,
      "loss": 0.0831,
      "step": 2206
    },
    {
      "epoch": 0.5443786982248521,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019995254078654257,
      "loss": 0.0842,
      "step": 2208
    },
    {
      "epoch": 0.5448717948717948,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019995147440690896,
      "loss": 0.0827,
      "step": 2210
    },
    {
      "epoch": 0.5453648915187377,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001999503961824676,
      "loss": 0.0916,
      "step": 2212
    },
    {
      "epoch": 0.5458579881656804,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019994930611334623,
      "loss": 0.0782,
      "step": 2214
    },
    {
      "epoch": 0.5463510848126233,
      "grad_norm": 0.91796875,
      "learning_rate": 0.000199948204199674,
      "loss": 0.0906,
      "step": 2216
    },
    {
      "epoch": 0.546844181459566,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019994709044158157,
      "loss": 0.0901,
      "step": 2218
    },
    {
      "epoch": 0.5473372781065089,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001999459648392009,
      "loss": 0.0936,
      "step": 2220
    },
    {
      "epoch": 0.5478303747534516,
      "grad_norm": 0.875,
      "learning_rate": 0.00019994482739266534,
      "loss": 0.0864,
      "step": 2222
    },
    {
      "epoch": 0.5483234714003945,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019994367810210973,
      "loss": 0.0926,
      "step": 2224
    },
    {
      "epoch": 0.5488165680473372,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019994251696767027,
      "loss": 0.0924,
      "step": 2226
    },
    {
      "epoch": 0.5493096646942801,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001999413439894845,
      "loss": 0.0862,
      "step": 2228
    },
    {
      "epoch": 0.5498027613412229,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001999401591676915,
      "loss": 0.0905,
      "step": 2230
    },
    {
      "epoch": 0.5502958579881657,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019993896250243163,
      "loss": 0.0926,
      "step": 2232
    },
    {
      "epoch": 0.5507889546351085,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001999377539938467,
      "loss": 0.0888,
      "step": 2234
    },
    {
      "epoch": 0.5512820512820513,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019993653364207997,
      "loss": 0.0802,
      "step": 2236
    },
    {
      "epoch": 0.5517751479289941,
      "grad_norm": 1.1015625,
      "learning_rate": 0.000199935301447276,
      "loss": 0.0889,
      "step": 2238
    },
    {
      "epoch": 0.5522682445759369,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019993405740958086,
      "loss": 0.0935,
      "step": 2240
    },
    {
      "epoch": 0.5527613412228797,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019993280152914196,
      "loss": 0.0866,
      "step": 2242
    },
    {
      "epoch": 0.5532544378698225,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019993153380610813,
      "loss": 0.0923,
      "step": 2244
    },
    {
      "epoch": 0.5537475345167653,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001999302542406296,
      "loss": 0.0838,
      "step": 2246
    },
    {
      "epoch": 0.5542406311637081,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019992896283285797,
      "loss": 0.0841,
      "step": 2248
    },
    {
      "epoch": 0.5547337278106509,
      "grad_norm": 1.125,
      "learning_rate": 0.00019992765958294634,
      "loss": 0.0831,
      "step": 2250
    },
    {
      "epoch": 0.5552268244575936,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019992634449104911,
      "loss": 0.0864,
      "step": 2252
    },
    {
      "epoch": 0.5557199211045365,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019992501755732215,
      "loss": 0.0857,
      "step": 2254
    },
    {
      "epoch": 0.5562130177514792,
      "grad_norm": 1.46875,
      "learning_rate": 0.0001999236787819227,
      "loss": 0.0845,
      "step": 2256
    },
    {
      "epoch": 0.5567061143984221,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001999223281650094,
      "loss": 0.0922,
      "step": 2258
    },
    {
      "epoch": 0.5571992110453649,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019992096570674235,
      "loss": 0.086,
      "step": 2260
    },
    {
      "epoch": 0.5576923076923077,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019991959140728296,
      "loss": 0.0849,
      "step": 2262
    },
    {
      "epoch": 0.5581854043392505,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001999182052667941,
      "loss": 0.0885,
      "step": 2264
    },
    {
      "epoch": 0.5586785009861933,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019991680728544004,
      "loss": 0.0868,
      "step": 2266
    },
    {
      "epoch": 0.5591715976331361,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019991539746338645,
      "loss": 0.0902,
      "step": 2268
    },
    {
      "epoch": 0.5596646942800789,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019991397580080045,
      "loss": 0.0829,
      "step": 2270
    },
    {
      "epoch": 0.5601577909270217,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001999125422978504,
      "loss": 0.0839,
      "step": 2272
    },
    {
      "epoch": 0.5606508875739645,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001999110969547063,
      "loss": 0.0927,
      "step": 2274
    },
    {
      "epoch": 0.5611439842209073,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019990963977153936,
      "loss": 0.0916,
      "step": 2276
    },
    {
      "epoch": 0.5616370808678501,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019990817074852228,
      "loss": 0.0853,
      "step": 2278
    },
    {
      "epoch": 0.5621301775147929,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019990668988582915,
      "loss": 0.0868,
      "step": 2280
    },
    {
      "epoch": 0.5626232741617357,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019990519718363547,
      "loss": 0.0931,
      "step": 2282
    },
    {
      "epoch": 0.5631163708086785,
      "grad_norm": 1.6015625,
      "learning_rate": 0.0001999036926421181,
      "loss": 0.0897,
      "step": 2284
    },
    {
      "epoch": 0.5636094674556213,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001999021762614554,
      "loss": 0.0776,
      "step": 2286
    },
    {
      "epoch": 0.5641025641025641,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019990064804182702,
      "loss": 0.0831,
      "step": 2288
    },
    {
      "epoch": 0.564595660749507,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019989910798341406,
      "loss": 0.0853,
      "step": 2290
    },
    {
      "epoch": 0.5650887573964497,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019989755608639903,
      "loss": 0.087,
      "step": 2292
    },
    {
      "epoch": 0.5655818540433925,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019989599235096587,
      "loss": 0.0811,
      "step": 2294
    },
    {
      "epoch": 0.5660749506903353,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019989441677729984,
      "loss": 0.0813,
      "step": 2296
    },
    {
      "epoch": 0.5665680473372781,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019989282936558773,
      "loss": 0.0825,
      "step": 2298
    },
    {
      "epoch": 0.5670611439842209,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019989123011601758,
      "loss": 0.0858,
      "step": 2300
    },
    {
      "epoch": 0.5675542406311637,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019988961902877894,
      "loss": 0.0892,
      "step": 2302
    },
    {
      "epoch": 0.5680473372781065,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019988799610406272,
      "loss": 0.0863,
      "step": 2304
    },
    {
      "epoch": 0.5685404339250493,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019988636134206128,
      "loss": 0.0853,
      "step": 2306
    },
    {
      "epoch": 0.5690335305719921,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019988471474296831,
      "loss": 0.0846,
      "step": 2308
    },
    {
      "epoch": 0.5695266272189349,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000199883056306979,
      "loss": 0.0869,
      "step": 2310
    },
    {
      "epoch": 0.5700197238658777,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001998813860342898,
      "loss": 0.0916,
      "step": 2312
    },
    {
      "epoch": 0.5705128205128205,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001998797039250987,
      "loss": 0.0888,
      "step": 2314
    },
    {
      "epoch": 0.5710059171597633,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019987800997960502,
      "loss": 0.0941,
      "step": 2316
    },
    {
      "epoch": 0.5714990138067061,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019987630419800954,
      "loss": 0.0955,
      "step": 2318
    },
    {
      "epoch": 0.571992110453649,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019987458658051434,
      "loss": 0.082,
      "step": 2320
    },
    {
      "epoch": 0.5724852071005917,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000199872857127323,
      "loss": 0.0864,
      "step": 2322
    },
    {
      "epoch": 0.5729783037475346,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001998711158386405,
      "loss": 0.0893,
      "step": 2324
    },
    {
      "epoch": 0.5734714003944773,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019986936271467314,
      "loss": 0.0886,
      "step": 2326
    },
    {
      "epoch": 0.5739644970414202,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019986759775562872,
      "loss": 0.0911,
      "step": 2328
    },
    {
      "epoch": 0.5744575936883629,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019986582096171636,
      "loss": 0.0916,
      "step": 2330
    },
    {
      "epoch": 0.5749506903353058,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019986403233314667,
      "loss": 0.088,
      "step": 2332
    },
    {
      "epoch": 0.5754437869822485,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00019986223187013153,
      "loss": 0.0778,
      "step": 2334
    },
    {
      "epoch": 0.5759368836291914,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019986041957288438,
      "loss": 0.0883,
      "step": 2336
    },
    {
      "epoch": 0.5764299802761341,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019985859544161994,
      "loss": 0.0875,
      "step": 2338
    },
    {
      "epoch": 0.5769230769230769,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019985675947655442,
      "loss": 0.0936,
      "step": 2340
    },
    {
      "epoch": 0.5774161735700197,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019985491167790538,
      "loss": 0.0814,
      "step": 2342
    },
    {
      "epoch": 0.5779092702169625,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019985305204589177,
      "loss": 0.0912,
      "step": 2344
    },
    {
      "epoch": 0.5784023668639053,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000199851180580734,
      "loss": 0.0939,
      "step": 2346
    },
    {
      "epoch": 0.5788954635108481,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019984929728265383,
      "loss": 0.0924,
      "step": 2348
    },
    {
      "epoch": 0.5793885601577909,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019984740215187445,
      "loss": 0.086,
      "step": 2350
    },
    {
      "epoch": 0.5798816568047337,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019984549518862042,
      "loss": 0.0842,
      "step": 2352
    },
    {
      "epoch": 0.5803747534516766,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019984357639311778,
      "loss": 0.089,
      "step": 2354
    },
    {
      "epoch": 0.5808678500986193,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019984164576559383,
      "loss": 0.0846,
      "step": 2356
    },
    {
      "epoch": 0.5813609467455622,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019983970330627747,
      "loss": 0.0867,
      "step": 2358
    },
    {
      "epoch": 0.5818540433925049,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019983774901539877,
      "loss": 0.0819,
      "step": 2360
    },
    {
      "epoch": 0.5823471400394478,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019983578289318942,
      "loss": 0.0876,
      "step": 2362
    },
    {
      "epoch": 0.5828402366863905,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019983380493988242,
      "loss": 0.0841,
      "step": 2364
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001998318151557121,
      "loss": 0.0822,
      "step": 2366
    },
    {
      "epoch": 0.5838264299802761,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019982981354091427,
      "loss": 0.0802,
      "step": 2368
    },
    {
      "epoch": 0.584319526627219,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019982780009572619,
      "loss": 0.0909,
      "step": 2370
    },
    {
      "epoch": 0.5848126232741617,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00019982577482038645,
      "loss": 0.0904,
      "step": 2372
    },
    {
      "epoch": 0.5853057199211046,
      "grad_norm": 0.59375,
      "learning_rate": 0.000199823737715135,
      "loss": 0.0771,
      "step": 2374
    },
    {
      "epoch": 0.5857988165680473,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001998216887802133,
      "loss": 0.0875,
      "step": 2376
    },
    {
      "epoch": 0.5862919132149902,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019981962801586414,
      "loss": 0.0924,
      "step": 2378
    },
    {
      "epoch": 0.5867850098619329,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019981755542233177,
      "loss": 0.0929,
      "step": 2380
    },
    {
      "epoch": 0.5872781065088757,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019981547099986172,
      "loss": 0.08,
      "step": 2382
    },
    {
      "epoch": 0.5877712031558185,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001998133747487011,
      "loss": 0.0841,
      "step": 2384
    },
    {
      "epoch": 0.5882642998027613,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019981126666909828,
      "loss": 0.0886,
      "step": 2386
    },
    {
      "epoch": 0.5887573964497042,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019980914676130307,
      "loss": 0.0778,
      "step": 2388
    },
    {
      "epoch": 0.5892504930966469,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001998070150255667,
      "loss": 0.0902,
      "step": 2390
    },
    {
      "epoch": 0.5897435897435898,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019980487146214183,
      "loss": 0.0862,
      "step": 2392
    },
    {
      "epoch": 0.5902366863905325,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019980271607128244,
      "loss": 0.09,
      "step": 2394
    },
    {
      "epoch": 0.5907297830374754,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019980054885324395,
      "loss": 0.0829,
      "step": 2396
    },
    {
      "epoch": 0.5912228796844181,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019979836980828323,
      "loss": 0.0773,
      "step": 2398
    },
    {
      "epoch": 0.591715976331361,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019979617893665848,
      "loss": 0.0871,
      "step": 2400
    },
    {
      "epoch": 0.5922090729783037,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019979397623862933,
      "loss": 0.09,
      "step": 2402
    },
    {
      "epoch": 0.5927021696252466,
      "grad_norm": 0.625,
      "learning_rate": 0.0001997917617144568,
      "loss": 0.0822,
      "step": 2404
    },
    {
      "epoch": 0.5931952662721893,
      "grad_norm": 0.875,
      "learning_rate": 0.00019978953536440336,
      "loss": 0.0907,
      "step": 2406
    },
    {
      "epoch": 0.5936883629191322,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019978729718873281,
      "loss": 0.0882,
      "step": 2408
    },
    {
      "epoch": 0.5941814595660749,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019978504718771043,
      "loss": 0.084,
      "step": 2410
    },
    {
      "epoch": 0.5946745562130178,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001997827853616028,
      "loss": 0.0812,
      "step": 2412
    },
    {
      "epoch": 0.5951676528599605,
      "grad_norm": 1.296875,
      "learning_rate": 0.00019978051171067805,
      "loss": 0.0883,
      "step": 2414
    },
    {
      "epoch": 0.5956607495069034,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019977822623520547,
      "loss": 0.0898,
      "step": 2416
    },
    {
      "epoch": 0.5961538461538461,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019977592893545608,
      "loss": 0.0817,
      "step": 2418
    },
    {
      "epoch": 0.596646942800789,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019977361981170195,
      "loss": 0.0907,
      "step": 2420
    },
    {
      "epoch": 0.5971400394477318,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019977129886421686,
      "loss": 0.0854,
      "step": 2422
    },
    {
      "epoch": 0.5976331360946746,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001997689660932758,
      "loss": 0.084,
      "step": 2424
    },
    {
      "epoch": 0.5981262327416174,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019976662149915518,
      "loss": 0.0796,
      "step": 2426
    },
    {
      "epoch": 0.5986193293885601,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019976426508213293,
      "loss": 0.0814,
      "step": 2428
    },
    {
      "epoch": 0.599112426035503,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019976189684248825,
      "loss": 0.0837,
      "step": 2430
    },
    {
      "epoch": 0.5996055226824457,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001997595167805018,
      "loss": 0.0807,
      "step": 2432
    },
    {
      "epoch": 0.6000986193293886,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001997571248964556,
      "loss": 0.0884,
      "step": 2434
    },
    {
      "epoch": 0.6005917159763313,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019975472119063313,
      "loss": 0.0856,
      "step": 2436
    },
    {
      "epoch": 0.6010848126232742,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019975230566331926,
      "loss": 0.0865,
      "step": 2438
    },
    {
      "epoch": 0.6015779092702169,
      "grad_norm": 1.25,
      "learning_rate": 0.00019974987831480022,
      "loss": 0.0864,
      "step": 2440
    },
    {
      "epoch": 0.6020710059171598,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00019974743914536364,
      "loss": 0.0879,
      "step": 2442
    },
    {
      "epoch": 0.6025641025641025,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019974498815529863,
      "loss": 0.0873,
      "step": 2444
    },
    {
      "epoch": 0.6030571992110454,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001997425253448956,
      "loss": 0.0895,
      "step": 2446
    },
    {
      "epoch": 0.6035502958579881,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019974005071444646,
      "loss": 0.0858,
      "step": 2448
    },
    {
      "epoch": 0.604043392504931,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019973756426424442,
      "loss": 0.0848,
      "step": 2450
    },
    {
      "epoch": 0.6045364891518737,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001997350659945841,
      "loss": 0.0822,
      "step": 2452
    },
    {
      "epoch": 0.6050295857988166,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019973255590576168,
      "loss": 0.0827,
      "step": 2454
    },
    {
      "epoch": 0.6055226824457594,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001997300339980745,
      "loss": 0.0863,
      "step": 2456
    },
    {
      "epoch": 0.6060157790927022,
      "grad_norm": 0.625,
      "learning_rate": 0.00019972750027182152,
      "loss": 0.0924,
      "step": 2458
    },
    {
      "epoch": 0.606508875739645,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019972495472730294,
      "loss": 0.0838,
      "step": 2460
    },
    {
      "epoch": 0.6070019723865878,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019972239736482038,
      "loss": 0.0857,
      "step": 2462
    },
    {
      "epoch": 0.6074950690335306,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000199719828184677,
      "loss": 0.0835,
      "step": 2464
    },
    {
      "epoch": 0.6079881656804734,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001997172471871772,
      "loss": 0.0854,
      "step": 2466
    },
    {
      "epoch": 0.6084812623274162,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019971465437262686,
      "loss": 0.0832,
      "step": 2468
    },
    {
      "epoch": 0.6089743589743589,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019971204974133326,
      "loss": 0.0793,
      "step": 2470
    },
    {
      "epoch": 0.6094674556213018,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000199709433293605,
      "loss": 0.0868,
      "step": 2472
    },
    {
      "epoch": 0.6099605522682445,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019970680502975224,
      "loss": 0.0853,
      "step": 2474
    },
    {
      "epoch": 0.6104536489151874,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019970416495008637,
      "loss": 0.0878,
      "step": 2476
    },
    {
      "epoch": 0.6109467455621301,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001997015130549203,
      "loss": 0.0744,
      "step": 2478
    },
    {
      "epoch": 0.611439842209073,
      "grad_norm": 0.625,
      "learning_rate": 0.00019969884934456823,
      "loss": 0.085,
      "step": 2480
    },
    {
      "epoch": 0.6119329388560157,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019969617381934587,
      "loss": 0.0876,
      "step": 2482
    },
    {
      "epoch": 0.6124260355029586,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001996934864795703,
      "loss": 0.0836,
      "step": 2484
    },
    {
      "epoch": 0.6129191321499013,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019969078732555998,
      "loss": 0.0823,
      "step": 2486
    },
    {
      "epoch": 0.6134122287968442,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001996880763576347,
      "loss": 0.0812,
      "step": 2488
    },
    {
      "epoch": 0.613905325443787,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019968535357611585,
      "loss": 0.0789,
      "step": 2490
    },
    {
      "epoch": 0.6143984220907298,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019968261898132598,
      "loss": 0.0906,
      "step": 2492
    },
    {
      "epoch": 0.6148915187376726,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019967987257358925,
      "loss": 0.0843,
      "step": 2494
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019967711435323103,
      "loss": 0.0852,
      "step": 2496
    },
    {
      "epoch": 0.6158777120315582,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001996743443205783,
      "loss": 0.0789,
      "step": 2498
    },
    {
      "epoch": 0.616370808678501,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001996715624759592,
      "loss": 0.0873,
      "step": 2500
    },
    {
      "epoch": 0.6168639053254438,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019966876881970347,
      "loss": 0.0807,
      "step": 2502
    },
    {
      "epoch": 0.6173570019723866,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019966596335214216,
      "loss": 0.0819,
      "step": 2504
    },
    {
      "epoch": 0.6178500986193294,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019966314607360773,
      "loss": 0.085,
      "step": 2506
    },
    {
      "epoch": 0.6183431952662722,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019966031698443403,
      "loss": 0.0888,
      "step": 2508
    },
    {
      "epoch": 0.618836291913215,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019965747608495632,
      "loss": 0.0802,
      "step": 2510
    },
    {
      "epoch": 0.6193293885601578,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001996546233755113,
      "loss": 0.0752,
      "step": 2512
    },
    {
      "epoch": 0.6198224852071006,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019965175885643704,
      "loss": 0.0799,
      "step": 2514
    },
    {
      "epoch": 0.6203155818540433,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019964888252807294,
      "loss": 0.0829,
      "step": 2516
    },
    {
      "epoch": 0.6208086785009862,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001996459943907599,
      "loss": 0.088,
      "step": 2518
    },
    {
      "epoch": 0.621301775147929,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019964309444484016,
      "loss": 0.0804,
      "step": 2520
    },
    {
      "epoch": 0.6217948717948718,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001996401826906574,
      "loss": 0.0932,
      "step": 2522
    },
    {
      "epoch": 0.6222879684418146,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019963725912855665,
      "loss": 0.0843,
      "step": 2524
    },
    {
      "epoch": 0.6227810650887574,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019963432375888441,
      "loss": 0.0828,
      "step": 2526
    },
    {
      "epoch": 0.6232741617357002,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019963137658198852,
      "loss": 0.0776,
      "step": 2528
    },
    {
      "epoch": 0.623767258382643,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019962841759821825,
      "loss": 0.0827,
      "step": 2530
    },
    {
      "epoch": 0.6242603550295858,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019962544680792422,
      "loss": 0.0854,
      "step": 2532
    },
    {
      "epoch": 0.6247534516765286,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001996224642114585,
      "loss": 0.0788,
      "step": 2534
    },
    {
      "epoch": 0.6252465483234714,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019961946980917456,
      "loss": 0.0778,
      "step": 2536
    },
    {
      "epoch": 0.6257396449704142,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019961646360142725,
      "loss": 0.0859,
      "step": 2538
    },
    {
      "epoch": 0.626232741617357,
      "grad_norm": 0.875,
      "learning_rate": 0.0001996134455885728,
      "loss": 0.0843,
      "step": 2540
    },
    {
      "epoch": 0.6267258382642998,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001996104157709689,
      "loss": 0.0834,
      "step": 2542
    },
    {
      "epoch": 0.6272189349112426,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019960737414897458,
      "loss": 0.0819,
      "step": 2544
    },
    {
      "epoch": 0.6277120315581854,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019960432072295024,
      "loss": 0.0799,
      "step": 2546
    },
    {
      "epoch": 0.6282051282051282,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001996012554932578,
      "loss": 0.0848,
      "step": 2548
    },
    {
      "epoch": 0.628698224852071,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001995981784602605,
      "loss": 0.0874,
      "step": 2550
    },
    {
      "epoch": 0.6291913214990138,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019959508962432297,
      "loss": 0.0879,
      "step": 2552
    },
    {
      "epoch": 0.6296844181459567,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019959198898581124,
      "loss": 0.0874,
      "step": 2554
    },
    {
      "epoch": 0.6301775147928994,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019958887654509275,
      "loss": 0.0798,
      "step": 2556
    },
    {
      "epoch": 0.6306706114398422,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00019958575230253635,
      "loss": 0.079,
      "step": 2558
    },
    {
      "epoch": 0.631163708086785,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019958261625851233,
      "loss": 0.0796,
      "step": 2560
    },
    {
      "epoch": 0.6316568047337278,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019957946841339227,
      "loss": 0.0891,
      "step": 2562
    },
    {
      "epoch": 0.6321499013806706,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001995763087675492,
      "loss": 0.0797,
      "step": 2564
    },
    {
      "epoch": 0.6326429980276134,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019957313732135758,
      "loss": 0.0831,
      "step": 2566
    },
    {
      "epoch": 0.6331360946745562,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019956995407519328,
      "loss": 0.0806,
      "step": 2568
    },
    {
      "epoch": 0.633629191321499,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019956675902943344,
      "loss": 0.0754,
      "step": 2570
    },
    {
      "epoch": 0.6341222879684418,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019956355218445678,
      "loss": 0.0818,
      "step": 2572
    },
    {
      "epoch": 0.6346153846153846,
      "grad_norm": 0.5,
      "learning_rate": 0.00019956033354064329,
      "loss": 0.0854,
      "step": 2574
    },
    {
      "epoch": 0.6351084812623274,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001995571030983744,
      "loss": 0.0803,
      "step": 2576
    },
    {
      "epoch": 0.6356015779092702,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019955386085803293,
      "loss": 0.0823,
      "step": 2578
    },
    {
      "epoch": 0.636094674556213,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019955060682000313,
      "loss": 0.0848,
      "step": 2580
    },
    {
      "epoch": 0.6365877712031558,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019954734098467057,
      "loss": 0.0788,
      "step": 2582
    },
    {
      "epoch": 0.6370808678500987,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00019954406335242233,
      "loss": 0.0808,
      "step": 2584
    },
    {
      "epoch": 0.6375739644970414,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001995407739236468,
      "loss": 0.0861,
      "step": 2586
    },
    {
      "epoch": 0.6380670611439843,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001995374726987338,
      "loss": 0.0813,
      "step": 2588
    },
    {
      "epoch": 0.638560157790927,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019953415967807452,
      "loss": 0.0866,
      "step": 2590
    },
    {
      "epoch": 0.6390532544378699,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001995308348620616,
      "loss": 0.0847,
      "step": 2592
    },
    {
      "epoch": 0.6395463510848126,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019952749825108903,
      "loss": 0.081,
      "step": 2594
    },
    {
      "epoch": 0.6400394477317555,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019952414984555222,
      "loss": 0.0782,
      "step": 2596
    },
    {
      "epoch": 0.6405325443786982,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019952078964584802,
      "loss": 0.0794,
      "step": 2598
    },
    {
      "epoch": 0.6410256410256411,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019951741765237454,
      "loss": 0.0746,
      "step": 2600
    },
    {
      "epoch": 0.6415187376725838,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019951403386553147,
      "loss": 0.0818,
      "step": 2602
    },
    {
      "epoch": 0.6420118343195266,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001995106382857198,
      "loss": 0.0809,
      "step": 2604
    },
    {
      "epoch": 0.6425049309664694,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019950723091334186,
      "loss": 0.0822,
      "step": 2606
    },
    {
      "epoch": 0.6429980276134122,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019950381174880147,
      "loss": 0.0835,
      "step": 2608
    },
    {
      "epoch": 0.643491124260355,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019950038079250384,
      "loss": 0.0778,
      "step": 2610
    },
    {
      "epoch": 0.6439842209072978,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019949693804485554,
      "loss": 0.0785,
      "step": 2612
    },
    {
      "epoch": 0.6444773175542406,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019949348350626456,
      "loss": 0.0837,
      "step": 2614
    },
    {
      "epoch": 0.6449704142011834,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001994900171771403,
      "loss": 0.0765,
      "step": 2616
    },
    {
      "epoch": 0.6454635108481263,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019948653905789353,
      "loss": 0.0724,
      "step": 2618
    },
    {
      "epoch": 0.645956607495069,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019948304914893639,
      "loss": 0.0758,
      "step": 2620
    },
    {
      "epoch": 0.6464497041420119,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019947954745068252,
      "loss": 0.0789,
      "step": 2622
    },
    {
      "epoch": 0.6469428007889546,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019947603396354682,
      "loss": 0.0834,
      "step": 2624
    },
    {
      "epoch": 0.6474358974358975,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001994725086879457,
      "loss": 0.0808,
      "step": 2626
    },
    {
      "epoch": 0.6479289940828402,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019946897162429695,
      "loss": 0.0804,
      "step": 2628
    },
    {
      "epoch": 0.6484220907297831,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019946542277301968,
      "loss": 0.0808,
      "step": 2630
    },
    {
      "epoch": 0.6489151873767258,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019946186213453445,
      "loss": 0.0754,
      "step": 2632
    },
    {
      "epoch": 0.6494082840236687,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019945828970926327,
      "loss": 0.0828,
      "step": 2634
    },
    {
      "epoch": 0.6499013806706114,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019945470549762942,
      "loss": 0.0869,
      "step": 2636
    },
    {
      "epoch": 0.6503944773175543,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001994511095000577,
      "loss": 0.0782,
      "step": 2638
    },
    {
      "epoch": 0.650887573964497,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019944750171697423,
      "loss": 0.0807,
      "step": 2640
    },
    {
      "epoch": 0.6513806706114399,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019944388214880653,
      "loss": 0.0821,
      "step": 2642
    },
    {
      "epoch": 0.6518737672583826,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00019944025079598362,
      "loss": 0.0837,
      "step": 2644
    },
    {
      "epoch": 0.6523668639053254,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001994366076589358,
      "loss": 0.0821,
      "step": 2646
    },
    {
      "epoch": 0.6528599605522682,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019943295273809475,
      "loss": 0.0886,
      "step": 2648
    },
    {
      "epoch": 0.653353057199211,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019942928603389366,
      "loss": 0.0819,
      "step": 2650
    },
    {
      "epoch": 0.6538461538461539,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019942560754676703,
      "loss": 0.0827,
      "step": 2652
    },
    {
      "epoch": 0.6543392504930966,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019942191727715075,
      "loss": 0.0773,
      "step": 2654
    },
    {
      "epoch": 0.6548323471400395,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019941821522548222,
      "loss": 0.0818,
      "step": 2656
    },
    {
      "epoch": 0.6553254437869822,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001994145013922001,
      "loss": 0.0822,
      "step": 2658
    },
    {
      "epoch": 0.6558185404339251,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001994107757777445,
      "loss": 0.0815,
      "step": 2660
    },
    {
      "epoch": 0.6563116370808678,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019940703838255692,
      "loss": 0.0807,
      "step": 2662
    },
    {
      "epoch": 0.6568047337278107,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019940328920708028,
      "loss": 0.0856,
      "step": 2664
    },
    {
      "epoch": 0.6572978303747534,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019939952825175888,
      "loss": 0.0797,
      "step": 2666
    },
    {
      "epoch": 0.6577909270216963,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001993957555170384,
      "loss": 0.0778,
      "step": 2668
    },
    {
      "epoch": 0.658284023668639,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00019939197100336592,
      "loss": 0.0759,
      "step": 2670
    },
    {
      "epoch": 0.6587771203155819,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019938817471118996,
      "loss": 0.0734,
      "step": 2672
    },
    {
      "epoch": 0.6592702169625246,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001993843666409604,
      "loss": 0.0833,
      "step": 2674
    },
    {
      "epoch": 0.6597633136094675,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019938054679312844,
      "loss": 0.0792,
      "step": 2676
    },
    {
      "epoch": 0.6602564102564102,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001993767151681469,
      "loss": 0.0792,
      "step": 2678
    },
    {
      "epoch": 0.6607495069033531,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001993728717664697,
      "loss": 0.0856,
      "step": 2680
    },
    {
      "epoch": 0.6612426035502958,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001993690165885524,
      "loss": 0.081,
      "step": 2682
    },
    {
      "epoch": 0.6617357001972387,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019936514963485183,
      "loss": 0.074,
      "step": 2684
    },
    {
      "epoch": 0.6622287968441815,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001993612709058262,
      "loss": 0.0796,
      "step": 2686
    },
    {
      "epoch": 0.6627218934911243,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019935738040193526,
      "loss": 0.0777,
      "step": 2688
    },
    {
      "epoch": 0.6632149901380671,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019935347812363998,
      "loss": 0.0838,
      "step": 2690
    },
    {
      "epoch": 0.6637080867850098,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019934956407140285,
      "loss": 0.0854,
      "step": 2692
    },
    {
      "epoch": 0.6642011834319527,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001993456382456876,
      "loss": 0.0824,
      "step": 2694
    },
    {
      "epoch": 0.6646942800788954,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019934170064695963,
      "loss": 0.08,
      "step": 2696
    },
    {
      "epoch": 0.6651873767258383,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019933775127568547,
      "loss": 0.0877,
      "step": 2698
    },
    {
      "epoch": 0.665680473372781,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019933379013233312,
      "loss": 0.0816,
      "step": 2700
    },
    {
      "epoch": 0.6661735700197239,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019932981721737202,
      "loss": 0.0971,
      "step": 2702
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000199325832531273,
      "loss": 0.075,
      "step": 2704
    },
    {
      "epoch": 0.6671597633136095,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001993218360745083,
      "loss": 0.0827,
      "step": 2706
    },
    {
      "epoch": 0.6676528599605522,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019931782784755146,
      "loss": 0.0743,
      "step": 2708
    },
    {
      "epoch": 0.6681459566074951,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001993138078508775,
      "loss": 0.0861,
      "step": 2710
    },
    {
      "epoch": 0.6686390532544378,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001993097760849628,
      "loss": 0.0726,
      "step": 2712
    },
    {
      "epoch": 0.6691321499013807,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019930573255028518,
      "loss": 0.0796,
      "step": 2714
    },
    {
      "epoch": 0.6696252465483234,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019930167724732377,
      "loss": 0.0779,
      "step": 2716
    },
    {
      "epoch": 0.6701183431952663,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001992976101765592,
      "loss": 0.0779,
      "step": 2718
    },
    {
      "epoch": 0.6706114398422091,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019929353133847343,
      "loss": 0.0753,
      "step": 2720
    },
    {
      "epoch": 0.6711045364891519,
      "grad_norm": 1.0,
      "learning_rate": 0.00019928944073354982,
      "loss": 0.0921,
      "step": 2722
    },
    {
      "epoch": 0.6715976331360947,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019928533836227307,
      "loss": 0.0875,
      "step": 2724
    },
    {
      "epoch": 0.6720907297830375,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001992812242251294,
      "loss": 0.0768,
      "step": 2726
    },
    {
      "epoch": 0.6725838264299803,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001992770983226064,
      "loss": 0.084,
      "step": 2728
    },
    {
      "epoch": 0.6730769230769231,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019927296065519289,
      "loss": 0.0807,
      "step": 2730
    },
    {
      "epoch": 0.6735700197238659,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0001992688112233793,
      "loss": 0.0761,
      "step": 2732
    },
    {
      "epoch": 0.6740631163708086,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019926465002765732,
      "loss": 0.0834,
      "step": 2734
    },
    {
      "epoch": 0.6745562130177515,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019926047706852012,
      "loss": 0.0825,
      "step": 2736
    },
    {
      "epoch": 0.6750493096646942,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019925629234646217,
      "loss": 0.0822,
      "step": 2738
    },
    {
      "epoch": 0.6755424063116371,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001992520958619794,
      "loss": 0.0788,
      "step": 2740
    },
    {
      "epoch": 0.6760355029585798,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001992478876155691,
      "loss": 0.0808,
      "step": 2742
    },
    {
      "epoch": 0.6765285996055227,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019924366760773002,
      "loss": 0.072,
      "step": 2744
    },
    {
      "epoch": 0.6770216962524654,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001992394358389622,
      "loss": 0.0786,
      "step": 2746
    },
    {
      "epoch": 0.6775147928994083,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019923519230976715,
      "loss": 0.0833,
      "step": 2748
    },
    {
      "epoch": 0.678007889546351,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00019923093702064776,
      "loss": 0.0807,
      "step": 2750
    },
    {
      "epoch": 0.6785009861932939,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00019922666997210833,
      "loss": 0.0818,
      "step": 2752
    },
    {
      "epoch": 0.6789940828402367,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019922239116465447,
      "loss": 0.0845,
      "step": 2754
    },
    {
      "epoch": 0.6794871794871795,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019921810059879324,
      "loss": 0.0784,
      "step": 2756
    },
    {
      "epoch": 0.6799802761341223,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019921379827503316,
      "loss": 0.0837,
      "step": 2758
    },
    {
      "epoch": 0.6804733727810651,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019920948419388404,
      "loss": 0.0927,
      "step": 2760
    },
    {
      "epoch": 0.6809664694280079,
      "grad_norm": 0.453125,
      "learning_rate": 0.00019920515835585714,
      "loss": 0.0786,
      "step": 2762
    },
    {
      "epoch": 0.6814595660749507,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019920082076146508,
      "loss": 0.083,
      "step": 2764
    },
    {
      "epoch": 0.6819526627218935,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019919647141122186,
      "loss": 0.0798,
      "step": 2766
    },
    {
      "epoch": 0.6824457593688363,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019919211030564298,
      "loss": 0.0786,
      "step": 2768
    },
    {
      "epoch": 0.6829388560157791,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001991877374452452,
      "loss": 0.0792,
      "step": 2770
    },
    {
      "epoch": 0.6834319526627219,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019918335283054673,
      "loss": 0.0769,
      "step": 2772
    },
    {
      "epoch": 0.6839250493096647,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019917895646206715,
      "loss": 0.0726,
      "step": 2774
    },
    {
      "epoch": 0.6844181459566075,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001991745483403275,
      "loss": 0.0834,
      "step": 2776
    },
    {
      "epoch": 0.6849112426035503,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001991701284658502,
      "loss": 0.0736,
      "step": 2778
    },
    {
      "epoch": 0.685404339250493,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019916569683915895,
      "loss": 0.076,
      "step": 2780
    },
    {
      "epoch": 0.6858974358974359,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019916125346077898,
      "loss": 0.0776,
      "step": 2782
    },
    {
      "epoch": 0.6863905325443787,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019915679833123676,
      "loss": 0.0731,
      "step": 2784
    },
    {
      "epoch": 0.6868836291913215,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019915233145106038,
      "loss": 0.0791,
      "step": 2786
    },
    {
      "epoch": 0.6873767258382643,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001991478528207791,
      "loss": 0.0797,
      "step": 2788
    },
    {
      "epoch": 0.6878698224852071,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001991433624409237,
      "loss": 0.0789,
      "step": 2790
    },
    {
      "epoch": 0.6883629191321499,
      "grad_norm": 0.625,
      "learning_rate": 0.0001991388603120263,
      "loss": 0.0784,
      "step": 2792
    },
    {
      "epoch": 0.6888560157790927,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019913434643462045,
      "loss": 0.0792,
      "step": 2794
    },
    {
      "epoch": 0.6893491124260355,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019912982080924103,
      "loss": 0.0745,
      "step": 2796
    },
    {
      "epoch": 0.6898422090729783,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019912528343642437,
      "loss": 0.082,
      "step": 2798
    },
    {
      "epoch": 0.6903353057199211,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019912073431670822,
      "loss": 0.0706,
      "step": 2800
    },
    {
      "epoch": 0.6908284023668639,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001991161734506316,
      "loss": 0.0774,
      "step": 2802
    },
    {
      "epoch": 0.6913214990138067,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019911160083873505,
      "loss": 0.0744,
      "step": 2804
    },
    {
      "epoch": 0.6918145956607495,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001991070164815604,
      "loss": 0.0797,
      "step": 2806
    },
    {
      "epoch": 0.6923076923076923,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019910242037965098,
      "loss": 0.0755,
      "step": 2808
    },
    {
      "epoch": 0.6928007889546351,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019909781253355142,
      "loss": 0.0758,
      "step": 2810
    },
    {
      "epoch": 0.6932938856015779,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019909319294380778,
      "loss": 0.088,
      "step": 2812
    },
    {
      "epoch": 0.6937869822485208,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001990885616109675,
      "loss": 0.0805,
      "step": 2814
    },
    {
      "epoch": 0.6942800788954635,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019908391853557946,
      "loss": 0.0784,
      "step": 2816
    },
    {
      "epoch": 0.6947731755424064,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019907926371819383,
      "loss": 0.0873,
      "step": 2818
    },
    {
      "epoch": 0.6952662721893491,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019907459715936228,
      "loss": 0.0805,
      "step": 2820
    },
    {
      "epoch": 0.6957593688362919,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001990699188596378,
      "loss": 0.0784,
      "step": 2822
    },
    {
      "epoch": 0.6962524654832347,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001990652288195748,
      "loss": 0.0779,
      "step": 2824
    },
    {
      "epoch": 0.6967455621301775,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001990605270397291,
      "loss": 0.0843,
      "step": 2826
    },
    {
      "epoch": 0.6972386587771203,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001990558135206578,
      "loss": 0.0756,
      "step": 2828
    },
    {
      "epoch": 0.6977317554240631,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001990510882629196,
      "loss": 0.0751,
      "step": 2830
    },
    {
      "epoch": 0.6982248520710059,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019904635126707438,
      "loss": 0.0788,
      "step": 2832
    },
    {
      "epoch": 0.6987179487179487,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019904160253368354,
      "loss": 0.0777,
      "step": 2834
    },
    {
      "epoch": 0.6992110453648915,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00019903684206330982,
      "loss": 0.0751,
      "step": 2836
    },
    {
      "epoch": 0.6997041420118343,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00019903206985651738,
      "loss": 0.0772,
      "step": 2838
    },
    {
      "epoch": 0.7001972386587771,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001990272859138717,
      "loss": 0.0793,
      "step": 2840
    },
    {
      "epoch": 0.7006903353057199,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001990224902359398,
      "loss": 0.0829,
      "step": 2842
    },
    {
      "epoch": 0.7011834319526628,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019901768282328986,
      "loss": 0.0804,
      "step": 2844
    },
    {
      "epoch": 0.7016765285996055,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019901286367649172,
      "loss": 0.079,
      "step": 2846
    },
    {
      "epoch": 0.7021696252465484,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001990080327961164,
      "loss": 0.0794,
      "step": 2848
    },
    {
      "epoch": 0.7026627218934911,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001990031901827364,
      "loss": 0.0822,
      "step": 2850
    },
    {
      "epoch": 0.703155818540434,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019899833583692563,
      "loss": 0.0777,
      "step": 2852
    },
    {
      "epoch": 0.7036489151873767,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001989934697592593,
      "loss": 0.0764,
      "step": 2854
    },
    {
      "epoch": 0.7041420118343196,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019898859195031408,
      "loss": 0.0804,
      "step": 2856
    },
    {
      "epoch": 0.7046351084812623,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019898370241066806,
      "loss": 0.0779,
      "step": 2858
    },
    {
      "epoch": 0.7051282051282052,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019897880114090065,
      "loss": 0.0744,
      "step": 2860
    },
    {
      "epoch": 0.7056213017751479,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019897388814159265,
      "loss": 0.0717,
      "step": 2862
    },
    {
      "epoch": 0.7061143984220908,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019896896341332633,
      "loss": 0.0733,
      "step": 2864
    },
    {
      "epoch": 0.7066074950690335,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019896402695668527,
      "loss": 0.0765,
      "step": 2866
    },
    {
      "epoch": 0.7071005917159763,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019895907877225447,
      "loss": 0.0752,
      "step": 2868
    },
    {
      "epoch": 0.7075936883629191,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019895411886062031,
      "loss": 0.0739,
      "step": 2870
    },
    {
      "epoch": 0.7080867850098619,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001989491472223706,
      "loss": 0.0721,
      "step": 2872
    },
    {
      "epoch": 0.7085798816568047,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019894416385809444,
      "loss": 0.078,
      "step": 2874
    },
    {
      "epoch": 0.7090729783037475,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001989391687683825,
      "loss": 0.0861,
      "step": 2876
    },
    {
      "epoch": 0.7095660749506904,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001989341619538266,
      "loss": 0.0793,
      "step": 2878
    },
    {
      "epoch": 0.7100591715976331,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019892914341502013,
      "loss": 0.0785,
      "step": 2880
    },
    {
      "epoch": 0.710552268244576,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019892411315255783,
      "loss": 0.081,
      "step": 2882
    },
    {
      "epoch": 0.7110453648915187,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019891907116703582,
      "loss": 0.0845,
      "step": 2884
    },
    {
      "epoch": 0.7115384615384616,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00019891401745905152,
      "loss": 0.0762,
      "step": 2886
    },
    {
      "epoch": 0.7120315581854043,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019890895202920395,
      "loss": 0.0778,
      "step": 2888
    },
    {
      "epoch": 0.7125246548323472,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001989038748780933,
      "loss": 0.0727,
      "step": 2890
    },
    {
      "epoch": 0.7130177514792899,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019889878600632126,
      "loss": 0.0844,
      "step": 2892
    },
    {
      "epoch": 0.7135108481262328,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001988936854144909,
      "loss": 0.0816,
      "step": 2894
    },
    {
      "epoch": 0.7140039447731755,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001988885731032067,
      "loss": 0.0808,
      "step": 2896
    },
    {
      "epoch": 0.7144970414201184,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019888344907307442,
      "loss": 0.0813,
      "step": 2898
    },
    {
      "epoch": 0.7149901380670611,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019887831332470138,
      "loss": 0.0823,
      "step": 2900
    },
    {
      "epoch": 0.715483234714004,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001988731658586961,
      "loss": 0.081,
      "step": 2902
    },
    {
      "epoch": 0.7159763313609467,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019886800667566864,
      "loss": 0.0805,
      "step": 2904
    },
    {
      "epoch": 0.7164694280078896,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001988628357762304,
      "loss": 0.0791,
      "step": 2906
    },
    {
      "epoch": 0.7169625246548323,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001988576531609941,
      "loss": 0.0666,
      "step": 2908
    },
    {
      "epoch": 0.7174556213017751,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019885245883057403,
      "loss": 0.0778,
      "step": 2910
    },
    {
      "epoch": 0.717948717948718,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019884725278558558,
      "loss": 0.08,
      "step": 2912
    },
    {
      "epoch": 0.7184418145956607,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019884203502664583,
      "loss": 0.0796,
      "step": 2914
    },
    {
      "epoch": 0.7189349112426036,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019883680555437305,
      "loss": 0.0781,
      "step": 2916
    },
    {
      "epoch": 0.7194280078895463,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019883156436938698,
      "loss": 0.0781,
      "step": 2918
    },
    {
      "epoch": 0.7199211045364892,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019882631147230876,
      "loss": 0.0787,
      "step": 2920
    },
    {
      "epoch": 0.7204142011834319,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001988210468637608,
      "loss": 0.0884,
      "step": 2922
    },
    {
      "epoch": 0.7209072978303748,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019881577054436705,
      "loss": 0.076,
      "step": 2924
    },
    {
      "epoch": 0.7214003944773175,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001988104825147528,
      "loss": 0.0768,
      "step": 2926
    },
    {
      "epoch": 0.7218934911242604,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019880518277554468,
      "loss": 0.0783,
      "step": 2928
    },
    {
      "epoch": 0.7223865877712031,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019879987132737073,
      "loss": 0.0774,
      "step": 2930
    },
    {
      "epoch": 0.722879684418146,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001987945481708604,
      "loss": 0.0725,
      "step": 2932
    },
    {
      "epoch": 0.7233727810650887,
      "grad_norm": 0.4375,
      "learning_rate": 0.00019878921330664453,
      "loss": 0.0828,
      "step": 2934
    },
    {
      "epoch": 0.7238658777120316,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001987838667353553,
      "loss": 0.0778,
      "step": 2936
    },
    {
      "epoch": 0.7243589743589743,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019877850845762633,
      "loss": 0.075,
      "step": 2938
    },
    {
      "epoch": 0.7248520710059172,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019877313847409261,
      "loss": 0.083,
      "step": 2940
    },
    {
      "epoch": 0.72534516765286,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001987677567853905,
      "loss": 0.0787,
      "step": 2942
    },
    {
      "epoch": 0.7258382642998028,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019876236339215773,
      "loss": 0.0824,
      "step": 2944
    },
    {
      "epoch": 0.7263313609467456,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001987569582950335,
      "loss": 0.0777,
      "step": 2946
    },
    {
      "epoch": 0.7268244575936884,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019875154149465833,
      "loss": 0.0786,
      "step": 2948
    },
    {
      "epoch": 0.7273175542406312,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019874611299167412,
      "loss": 0.0736,
      "step": 2950
    },
    {
      "epoch": 0.727810650887574,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001987406727867242,
      "loss": 0.0754,
      "step": 2952
    },
    {
      "epoch": 0.7283037475345168,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001987352208804533,
      "loss": 0.0764,
      "step": 2954
    },
    {
      "epoch": 0.7287968441814595,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001987297572735074,
      "loss": 0.0795,
      "step": 2956
    },
    {
      "epoch": 0.7292899408284024,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00019872428196653403,
      "loss": 0.0767,
      "step": 2958
    },
    {
      "epoch": 0.7297830374753451,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019871879496018206,
      "loss": 0.0712,
      "step": 2960
    },
    {
      "epoch": 0.730276134122288,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019871329625510172,
      "loss": 0.0753,
      "step": 2962
    },
    {
      "epoch": 0.7307692307692307,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019870778585194461,
      "loss": 0.0739,
      "step": 2964
    },
    {
      "epoch": 0.7312623274161736,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019870226375136378,
      "loss": 0.0756,
      "step": 2966
    },
    {
      "epoch": 0.7317554240631163,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001986967299540136,
      "loss": 0.0747,
      "step": 2968
    },
    {
      "epoch": 0.7322485207100592,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019869118446054986,
      "loss": 0.0756,
      "step": 2970
    },
    {
      "epoch": 0.7327416173570019,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019868562727162975,
      "loss": 0.081,
      "step": 2972
    },
    {
      "epoch": 0.7332347140039448,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019868005838791184,
      "loss": 0.0706,
      "step": 2974
    },
    {
      "epoch": 0.7337278106508875,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019867447781005604,
      "loss": 0.0789,
      "step": 2976
    },
    {
      "epoch": 0.7342209072978304,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019866888553872366,
      "loss": 0.0795,
      "step": 2978
    },
    {
      "epoch": 0.7347140039447732,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019866328157457748,
      "loss": 0.0777,
      "step": 2980
    },
    {
      "epoch": 0.735207100591716,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019865766591828157,
      "loss": 0.0812,
      "step": 2982
    },
    {
      "epoch": 0.7357001972386588,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001986520385705014,
      "loss": 0.0742,
      "step": 2984
    },
    {
      "epoch": 0.7361932938856016,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019864639953190387,
      "loss": 0.075,
      "step": 2986
    },
    {
      "epoch": 0.7366863905325444,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001986407488031572,
      "loss": 0.0854,
      "step": 2988
    },
    {
      "epoch": 0.7371794871794872,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001986350863849311,
      "loss": 0.0714,
      "step": 2990
    },
    {
      "epoch": 0.73767258382643,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019862941227789653,
      "loss": 0.0722,
      "step": 2992
    },
    {
      "epoch": 0.7381656804733728,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001986237264827259,
      "loss": 0.0778,
      "step": 2994
    },
    {
      "epoch": 0.7386587771203156,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019861802900009309,
      "loss": 0.0694,
      "step": 2996
    },
    {
      "epoch": 0.7391518737672583,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001986123198306732,
      "loss": 0.0775,
      "step": 2998
    },
    {
      "epoch": 0.7396449704142012,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019860659897514286,
      "loss": 0.0784,
      "step": 3000
    },
    {
      "epoch": 0.7401380670611439,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019860086643417994,
      "loss": 0.076,
      "step": 3002
    },
    {
      "epoch": 0.7406311637080868,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019859512220846387,
      "loss": 0.0745,
      "step": 3004
    },
    {
      "epoch": 0.7411242603550295,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001985893662986753,
      "loss": 0.0803,
      "step": 3006
    },
    {
      "epoch": 0.7416173570019724,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001985835987054964,
      "loss": 0.081,
      "step": 3008
    },
    {
      "epoch": 0.7421104536489151,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019857781942961064,
      "loss": 0.0872,
      "step": 3010
    },
    {
      "epoch": 0.742603550295858,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019857202847170288,
      "loss": 0.0729,
      "step": 3012
    },
    {
      "epoch": 0.7430966469428008,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019856622583245938,
      "loss": 0.0741,
      "step": 3014
    },
    {
      "epoch": 0.7435897435897436,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001985604115125678,
      "loss": 0.0802,
      "step": 3016
    },
    {
      "epoch": 0.7440828402366864,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001985545855127172,
      "loss": 0.0777,
      "step": 3018
    },
    {
      "epoch": 0.7445759368836292,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001985487478335979,
      "loss": 0.0765,
      "step": 3020
    },
    {
      "epoch": 0.745069033530572,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019854289847590176,
      "loss": 0.069,
      "step": 3022
    },
    {
      "epoch": 0.7455621301775148,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019853703744032202,
      "loss": 0.0711,
      "step": 3024
    },
    {
      "epoch": 0.7460552268244576,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019853116472755313,
      "loss": 0.079,
      "step": 3026
    },
    {
      "epoch": 0.7465483234714004,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001985252803382911,
      "loss": 0.0663,
      "step": 3028
    },
    {
      "epoch": 0.7470414201183432,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019851938427323327,
      "loss": 0.0762,
      "step": 3030
    },
    {
      "epoch": 0.747534516765286,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001985134765330783,
      "loss": 0.0746,
      "step": 3032
    },
    {
      "epoch": 0.7480276134122288,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001985075571185264,
      "loss": 0.0771,
      "step": 3034
    },
    {
      "epoch": 0.7485207100591716,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019850162603027896,
      "loss": 0.0779,
      "step": 3036
    },
    {
      "epoch": 0.7490138067061144,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019849568326903884,
      "loss": 0.0713,
      "step": 3038
    },
    {
      "epoch": 0.7495069033530573,
      "grad_norm": 0.453125,
      "learning_rate": 0.00019848972883551035,
      "loss": 0.0774,
      "step": 3040
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019848376273039908,
      "loss": 0.0802,
      "step": 3042
    },
    {
      "epoch": 0.7504930966469427,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019847778495441206,
      "loss": 0.077,
      "step": 3044
    },
    {
      "epoch": 0.7509861932938856,
      "grad_norm": 0.4375,
      "learning_rate": 0.00019847179550825774,
      "loss": 0.0797,
      "step": 3046
    },
    {
      "epoch": 0.7514792899408284,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019846579439264584,
      "loss": 0.0758,
      "step": 3048
    },
    {
      "epoch": 0.7519723865877712,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001984597816082875,
      "loss": 0.0828,
      "step": 3050
    },
    {
      "epoch": 0.752465483234714,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019845375715589535,
      "loss": 0.0751,
      "step": 3052
    },
    {
      "epoch": 0.7529585798816568,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019844772103618327,
      "loss": 0.0802,
      "step": 3054
    },
    {
      "epoch": 0.7534516765285996,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019844167324986657,
      "loss": 0.0705,
      "step": 3056
    },
    {
      "epoch": 0.7539447731755424,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000198435613797662,
      "loss": 0.0727,
      "step": 3058
    },
    {
      "epoch": 0.7544378698224852,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019842954268028758,
      "loss": 0.0735,
      "step": 3060
    },
    {
      "epoch": 0.754930966469428,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019842345989846277,
      "loss": 0.0762,
      "step": 3062
    },
    {
      "epoch": 0.7554240631163708,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019841736545290846,
      "loss": 0.0781,
      "step": 3064
    },
    {
      "epoch": 0.7559171597633136,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019841125934434684,
      "loss": 0.0792,
      "step": 3066
    },
    {
      "epoch": 0.7564102564102564,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019840514157350153,
      "loss": 0.08,
      "step": 3068
    },
    {
      "epoch": 0.7569033530571992,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019839901214109754,
      "loss": 0.0717,
      "step": 3070
    },
    {
      "epoch": 0.757396449704142,
      "grad_norm": 0.5,
      "learning_rate": 0.0001983928710478612,
      "loss": 0.0808,
      "step": 3072
    },
    {
      "epoch": 0.7578895463510849,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019838671829452032,
      "loss": 0.0788,
      "step": 3074
    },
    {
      "epoch": 0.7583826429980276,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019838055388180396,
      "loss": 0.0773,
      "step": 3076
    },
    {
      "epoch": 0.7588757396449705,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019837437781044268,
      "loss": 0.0678,
      "step": 3078
    },
    {
      "epoch": 0.7593688362919132,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001983681900811684,
      "loss": 0.0684,
      "step": 3080
    },
    {
      "epoch": 0.7598619329388561,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019836199069471437,
      "loss": 0.0761,
      "step": 3082
    },
    {
      "epoch": 0.7603550295857988,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019835577965181527,
      "loss": 0.075,
      "step": 3084
    },
    {
      "epoch": 0.7608481262327417,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019834955695320709,
      "loss": 0.0764,
      "step": 3086
    },
    {
      "epoch": 0.7613412228796844,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019834332259962734,
      "loss": 0.0756,
      "step": 3088
    },
    {
      "epoch": 0.7618343195266272,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019833707659181477,
      "loss": 0.0708,
      "step": 3090
    },
    {
      "epoch": 0.76232741617357,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00019833081893050957,
      "loss": 0.0788,
      "step": 3092
    },
    {
      "epoch": 0.7628205128205128,
      "grad_norm": 0.5,
      "learning_rate": 0.00019832454961645334,
      "loss": 0.0746,
      "step": 3094
    },
    {
      "epoch": 0.7633136094674556,
      "grad_norm": 0.609375,
      "learning_rate": 0.000198318268650389,
      "loss": 0.0777,
      "step": 3096
    },
    {
      "epoch": 0.7638067061143984,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00019831197603306086,
      "loss": 0.0707,
      "step": 3098
    },
    {
      "epoch": 0.7642998027613412,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001983056717652147,
      "loss": 0.0781,
      "step": 3100
    },
    {
      "epoch": 0.764792899408284,
      "grad_norm": 1.0,
      "learning_rate": 0.00019829935584759754,
      "loss": 0.0767,
      "step": 3102
    },
    {
      "epoch": 0.7652859960552268,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001982930282809579,
      "loss": 0.0676,
      "step": 3104
    },
    {
      "epoch": 0.7657790927021696,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019828668906604556,
      "loss": 0.078,
      "step": 3106
    },
    {
      "epoch": 0.7662721893491125,
      "grad_norm": 0.5,
      "learning_rate": 0.00019828033820361185,
      "loss": 0.0797,
      "step": 3108
    },
    {
      "epoch": 0.7667652859960552,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019827397569440933,
      "loss": 0.066,
      "step": 3110
    },
    {
      "epoch": 0.7672583826429981,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00019826760153919198,
      "loss": 0.0745,
      "step": 3112
    },
    {
      "epoch": 0.7677514792899408,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001982612157387152,
      "loss": 0.076,
      "step": 3114
    },
    {
      "epoch": 0.7682445759368837,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019825481829373575,
      "loss": 0.0718,
      "step": 3116
    },
    {
      "epoch": 0.7687376725838264,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019824840920501174,
      "loss": 0.0726,
      "step": 3118
    },
    {
      "epoch": 0.7692307692307693,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019824198847330266,
      "loss": 0.0697,
      "step": 3120
    },
    {
      "epoch": 0.769723865877712,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001982355560993695,
      "loss": 0.0788,
      "step": 3122
    },
    {
      "epoch": 0.7702169625246549,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019822911208397441,
      "loss": 0.0712,
      "step": 3124
    },
    {
      "epoch": 0.7707100591715976,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019822265642788111,
      "loss": 0.0706,
      "step": 3126
    },
    {
      "epoch": 0.7712031558185405,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019821618913185463,
      "loss": 0.0752,
      "step": 3128
    },
    {
      "epoch": 0.7716962524654832,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019820971019666135,
      "loss": 0.0776,
      "step": 3130
    },
    {
      "epoch": 0.772189349112426,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019820321962306907,
      "loss": 0.0778,
      "step": 3132
    },
    {
      "epoch": 0.7726824457593688,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000198196717411847,
      "loss": 0.0741,
      "step": 3134
    },
    {
      "epoch": 0.7731755424063116,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019819020356376564,
      "loss": 0.0738,
      "step": 3136
    },
    {
      "epoch": 0.7736686390532544,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019818367807959692,
      "loss": 0.0742,
      "step": 3138
    },
    {
      "epoch": 0.7741617357001972,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019817714096011417,
      "loss": 0.0846,
      "step": 3140
    },
    {
      "epoch": 0.77465483234714,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001981705922060921,
      "loss": 0.0813,
      "step": 3142
    },
    {
      "epoch": 0.7751479289940828,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019816403181830667,
      "loss": 0.0763,
      "step": 3144
    },
    {
      "epoch": 0.7756410256410257,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019815745979753543,
      "loss": 0.0786,
      "step": 3146
    },
    {
      "epoch": 0.7761341222879684,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019815087614455718,
      "loss": 0.0764,
      "step": 3148
    },
    {
      "epoch": 0.7766272189349113,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001981442808601521,
      "loss": 0.0768,
      "step": 3150
    },
    {
      "epoch": 0.777120315581854,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019813767394510174,
      "loss": 0.075,
      "step": 3152
    },
    {
      "epoch": 0.7776134122287969,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019813105540018912,
      "loss": 0.0782,
      "step": 3154
    },
    {
      "epoch": 0.7781065088757396,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001981244252261985,
      "loss": 0.0763,
      "step": 3156
    },
    {
      "epoch": 0.7785996055226825,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00019811778342391568,
      "loss": 0.0757,
      "step": 3158
    },
    {
      "epoch": 0.7790927021696252,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019811112999412766,
      "loss": 0.0736,
      "step": 3160
    },
    {
      "epoch": 0.7795857988165681,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019810446493762301,
      "loss": 0.0772,
      "step": 3162
    },
    {
      "epoch": 0.7800788954635108,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019809778825519148,
      "loss": 0.074,
      "step": 3164
    },
    {
      "epoch": 0.7805719921104537,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019809109994762433,
      "loss": 0.0744,
      "step": 3166
    },
    {
      "epoch": 0.7810650887573964,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00019808440001571417,
      "loss": 0.0724,
      "step": 3168
    },
    {
      "epoch": 0.7815581854043393,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000198077688460255,
      "loss": 0.0761,
      "step": 3170
    },
    {
      "epoch": 0.782051282051282,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019807096528204213,
      "loss": 0.0778,
      "step": 3172
    },
    {
      "epoch": 0.7825443786982249,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019806423048187232,
      "loss": 0.0788,
      "step": 3174
    },
    {
      "epoch": 0.7830374753451677,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019805748406054367,
      "loss": 0.0699,
      "step": 3176
    },
    {
      "epoch": 0.7835305719921104,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019805072601885567,
      "loss": 0.0826,
      "step": 3178
    },
    {
      "epoch": 0.7840236686390533,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019804395635760922,
      "loss": 0.0694,
      "step": 3180
    },
    {
      "epoch": 0.784516765285996,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001980371750776065,
      "loss": 0.0718,
      "step": 3182
    },
    {
      "epoch": 0.7850098619329389,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001980303821796512,
      "loss": 0.0762,
      "step": 3184
    },
    {
      "epoch": 0.7855029585798816,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019802357766454827,
      "loss": 0.0686,
      "step": 3186
    },
    {
      "epoch": 0.7859960552268245,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019801676153310407,
      "loss": 0.0703,
      "step": 3188
    },
    {
      "epoch": 0.7864891518737672,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001980099337861264,
      "loss": 0.0731,
      "step": 3190
    },
    {
      "epoch": 0.7869822485207101,
      "grad_norm": 0.5,
      "learning_rate": 0.00019800309442442434,
      "loss": 0.0741,
      "step": 3192
    },
    {
      "epoch": 0.7874753451676528,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00019799624344880843,
      "loss": 0.0781,
      "step": 3194
    },
    {
      "epoch": 0.7879684418145957,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001979893808600905,
      "loss": 0.0795,
      "step": 3196
    },
    {
      "epoch": 0.7884615384615384,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019798250665908386,
      "loss": 0.0816,
      "step": 3198
    },
    {
      "epoch": 0.7889546351084813,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019797562084660313,
      "loss": 0.0738,
      "step": 3200
    },
    {
      "epoch": 0.789447731755424,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019796872342346425,
      "loss": 0.0757,
      "step": 3202
    },
    {
      "epoch": 0.7899408284023669,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001979618143904847,
      "loss": 0.078,
      "step": 3204
    },
    {
      "epoch": 0.7904339250493096,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001979548937484832,
      "loss": 0.072,
      "step": 3206
    },
    {
      "epoch": 0.7909270216962525,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019794796149827986,
      "loss": 0.0769,
      "step": 3208
    },
    {
      "epoch": 0.7914201183431953,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001979410176406962,
      "loss": 0.0769,
      "step": 3210
    },
    {
      "epoch": 0.7919132149901381,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019793406217655517,
      "loss": 0.0769,
      "step": 3212
    },
    {
      "epoch": 0.7924063116370809,
      "grad_norm": 0.75,
      "learning_rate": 0.00019792709510668094,
      "loss": 0.0731,
      "step": 3214
    },
    {
      "epoch": 0.7928994082840237,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019792011643189917,
      "loss": 0.0721,
      "step": 3216
    },
    {
      "epoch": 0.7933925049309665,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001979131261530369,
      "loss": 0.0725,
      "step": 3218
    },
    {
      "epoch": 0.7938856015779092,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019790612427092254,
      "loss": 0.0743,
      "step": 3220
    },
    {
      "epoch": 0.7943786982248521,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019789911078638576,
      "loss": 0.0719,
      "step": 3222
    },
    {
      "epoch": 0.7948717948717948,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019789208570025779,
      "loss": 0.0788,
      "step": 3224
    },
    {
      "epoch": 0.7953648915187377,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001978850490133711,
      "loss": 0.0746,
      "step": 3226
    },
    {
      "epoch": 0.7958579881656804,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019787800072655956,
      "loss": 0.0774,
      "step": 3228
    },
    {
      "epoch": 0.7963510848126233,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001978709408406585,
      "loss": 0.0777,
      "step": 3230
    },
    {
      "epoch": 0.796844181459566,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019786386935650448,
      "loss": 0.0771,
      "step": 3232
    },
    {
      "epoch": 0.7973372781065089,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001978567862749355,
      "loss": 0.0764,
      "step": 3234
    },
    {
      "epoch": 0.7978303747534516,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019784969159679106,
      "loss": 0.0775,
      "step": 3236
    },
    {
      "epoch": 0.7983234714003945,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001978425853229118,
      "loss": 0.0782,
      "step": 3238
    },
    {
      "epoch": 0.7988165680473372,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001978354674541399,
      "loss": 0.0734,
      "step": 3240
    },
    {
      "epoch": 0.7993096646942801,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019782833799131887,
      "loss": 0.0768,
      "step": 3242
    },
    {
      "epoch": 0.7998027613412229,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019782119693529358,
      "loss": 0.0724,
      "step": 3244
    },
    {
      "epoch": 0.8002958579881657,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019781404428691028,
      "loss": 0.0763,
      "step": 3246
    },
    {
      "epoch": 0.8007889546351085,
      "grad_norm": 0.75,
      "learning_rate": 0.00019780688004701662,
      "loss": 0.0803,
      "step": 3248
    },
    {
      "epoch": 0.8012820512820513,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001977997042164616,
      "loss": 0.08,
      "step": 3250
    },
    {
      "epoch": 0.8017751479289941,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019779251679609558,
      "loss": 0.0739,
      "step": 3252
    },
    {
      "epoch": 0.8022682445759369,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001977853177867703,
      "loss": 0.073,
      "step": 3254
    },
    {
      "epoch": 0.8027613412228797,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00019777810718933888,
      "loss": 0.0704,
      "step": 3256
    },
    {
      "epoch": 0.8032544378698225,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019777088500465588,
      "loss": 0.08,
      "step": 3258
    },
    {
      "epoch": 0.8037475345167653,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019776365123357713,
      "loss": 0.0704,
      "step": 3260
    },
    {
      "epoch": 0.8042406311637081,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019775640587695983,
      "loss": 0.0729,
      "step": 3262
    },
    {
      "epoch": 0.8047337278106509,
      "grad_norm": 0.625,
      "learning_rate": 0.00019774914893566265,
      "loss": 0.0691,
      "step": 3264
    },
    {
      "epoch": 0.8052268244575936,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00019774188041054555,
      "loss": 0.0759,
      "step": 3266
    },
    {
      "epoch": 0.8057199211045365,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001977346003024699,
      "loss": 0.0804,
      "step": 3268
    },
    {
      "epoch": 0.8062130177514792,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019772730861229848,
      "loss": 0.0771,
      "step": 3270
    },
    {
      "epoch": 0.8067061143984221,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019772000534089529,
      "loss": 0.0768,
      "step": 3272
    },
    {
      "epoch": 0.8071992110453649,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019771269048912587,
      "loss": 0.0743,
      "step": 3274
    },
    {
      "epoch": 0.8076923076923077,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001977053640578571,
      "loss": 0.078,
      "step": 3276
    },
    {
      "epoch": 0.8081854043392505,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019769802604795717,
      "loss": 0.0707,
      "step": 3278
    },
    {
      "epoch": 0.8086785009861933,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019769067646029567,
      "loss": 0.0739,
      "step": 3280
    },
    {
      "epoch": 0.8091715976331361,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019768331529574357,
      "loss": 0.0757,
      "step": 3282
    },
    {
      "epoch": 0.8096646942800789,
      "grad_norm": 0.875,
      "learning_rate": 0.0001976759425551732,
      "loss": 0.0747,
      "step": 3284
    },
    {
      "epoch": 0.8101577909270217,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001976685582394583,
      "loss": 0.0719,
      "step": 3286
    },
    {
      "epoch": 0.8106508875739645,
      "grad_norm": 0.5,
      "learning_rate": 0.00019766116234947397,
      "loss": 0.0763,
      "step": 3288
    },
    {
      "epoch": 0.8111439842209073,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019765375488609658,
      "loss": 0.0731,
      "step": 3290
    },
    {
      "epoch": 0.8116370808678501,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019764633585020404,
      "loss": 0.0733,
      "step": 3292
    },
    {
      "epoch": 0.8121301775147929,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001976389052426755,
      "loss": 0.0719,
      "step": 3294
    },
    {
      "epoch": 0.8126232741617357,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019763146306439152,
      "loss": 0.0692,
      "step": 3296
    },
    {
      "epoch": 0.8131163708086785,
      "grad_norm": 0.4375,
      "learning_rate": 0.00019762400931623412,
      "loss": 0.0745,
      "step": 3298
    },
    {
      "epoch": 0.8136094674556213,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019761654399908651,
      "loss": 0.0731,
      "step": 3300
    },
    {
      "epoch": 0.8141025641025641,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019760906711383344,
      "loss": 0.0732,
      "step": 3302
    },
    {
      "epoch": 0.814595660749507,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019760157866136093,
      "loss": 0.0789,
      "step": 3304
    },
    {
      "epoch": 0.8150887573964497,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019759407864255638,
      "loss": 0.0796,
      "step": 3306
    },
    {
      "epoch": 0.8155818540433925,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019758656705830865,
      "loss": 0.0696,
      "step": 3308
    },
    {
      "epoch": 0.8160749506903353,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019757904390950787,
      "loss": 0.0791,
      "step": 3310
    },
    {
      "epoch": 0.8165680473372781,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019757150919704554,
      "loss": 0.0745,
      "step": 3312
    },
    {
      "epoch": 0.8170611439842209,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019756396292181462,
      "loss": 0.0726,
      "step": 3314
    },
    {
      "epoch": 0.8175542406311637,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019755640508470942,
      "loss": 0.0748,
      "step": 3316
    },
    {
      "epoch": 0.8180473372781065,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019754883568662545,
      "loss": 0.0776,
      "step": 3318
    },
    {
      "epoch": 0.8185404339250493,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019754125472845987,
      "loss": 0.0742,
      "step": 3320
    },
    {
      "epoch": 0.8190335305719921,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019753366221111096,
      "loss": 0.077,
      "step": 3322
    },
    {
      "epoch": 0.8195266272189349,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019752605813547853,
      "loss": 0.0729,
      "step": 3324
    },
    {
      "epoch": 0.8200197238658777,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001975184425024637,
      "loss": 0.0765,
      "step": 3326
    },
    {
      "epoch": 0.8205128205128205,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019751081531296896,
      "loss": 0.0717,
      "step": 3328
    },
    {
      "epoch": 0.8210059171597633,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019750317656789818,
      "loss": 0.0828,
      "step": 3330
    },
    {
      "epoch": 0.8214990138067061,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019749552626815658,
      "loss": 0.0685,
      "step": 3332
    },
    {
      "epoch": 0.821992110453649,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001974878644146508,
      "loss": 0.0774,
      "step": 3334
    },
    {
      "epoch": 0.8224852071005917,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019748019100828876,
      "loss": 0.0783,
      "step": 3336
    },
    {
      "epoch": 0.8229783037475346,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019747250604997983,
      "loss": 0.0708,
      "step": 3338
    },
    {
      "epoch": 0.8234714003944773,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001974648095406347,
      "loss": 0.0743,
      "step": 3340
    },
    {
      "epoch": 0.8239644970414202,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019745710148116546,
      "loss": 0.0735,
      "step": 3342
    },
    {
      "epoch": 0.8244575936883629,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001974493818724856,
      "loss": 0.0777,
      "step": 3344
    },
    {
      "epoch": 0.8249506903353058,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019744165071550987,
      "loss": 0.0802,
      "step": 3346
    },
    {
      "epoch": 0.8254437869822485,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001974339080111545,
      "loss": 0.0743,
      "step": 3348
    },
    {
      "epoch": 0.8259368836291914,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019742615376033707,
      "loss": 0.0759,
      "step": 3350
    },
    {
      "epoch": 0.8264299802761341,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001974183879639764,
      "loss": 0.0779,
      "step": 3352
    },
    {
      "epoch": 0.8269230769230769,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019741061062299284,
      "loss": 0.0713,
      "step": 3354
    },
    {
      "epoch": 0.8274161735700197,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019740282173830807,
      "loss": 0.0666,
      "step": 3356
    },
    {
      "epoch": 0.8279092702169625,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00019739502131084513,
      "loss": 0.0695,
      "step": 3358
    },
    {
      "epoch": 0.8284023668639053,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001973872093415283,
      "loss": 0.0704,
      "step": 3360
    },
    {
      "epoch": 0.8288954635108481,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001973793858312835,
      "loss": 0.0806,
      "step": 3362
    },
    {
      "epoch": 0.8293885601577909,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019737155078103778,
      "loss": 0.0735,
      "step": 3364
    },
    {
      "epoch": 0.8298816568047337,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001973637041917196,
      "loss": 0.0696,
      "step": 3366
    },
    {
      "epoch": 0.8303747534516766,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019735584606425888,
      "loss": 0.0763,
      "step": 3368
    },
    {
      "epoch": 0.8308678500986193,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001973479763995868,
      "loss": 0.0743,
      "step": 3370
    },
    {
      "epoch": 0.8313609467455622,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019734009519863603,
      "loss": 0.0751,
      "step": 3372
    },
    {
      "epoch": 0.8318540433925049,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00019733220246234053,
      "loss": 0.0688,
      "step": 3374
    },
    {
      "epoch": 0.8323471400394478,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019732429819163555,
      "loss": 0.0832,
      "step": 3376
    },
    {
      "epoch": 0.8328402366863905,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019731638238745789,
      "loss": 0.0688,
      "step": 3378
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019730845505074553,
      "loss": 0.0731,
      "step": 3380
    },
    {
      "epoch": 0.8338264299802761,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000197300516182438,
      "loss": 0.0754,
      "step": 3382
    },
    {
      "epoch": 0.834319526627219,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019729256578347599,
      "loss": 0.0721,
      "step": 3384
    },
    {
      "epoch": 0.8348126232741617,
      "grad_norm": 0.5,
      "learning_rate": 0.00019728460385480174,
      "loss": 0.0757,
      "step": 3386
    },
    {
      "epoch": 0.8353057199211046,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019727663039735877,
      "loss": 0.08,
      "step": 3388
    },
    {
      "epoch": 0.8357988165680473,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019726864541209197,
      "loss": 0.0768,
      "step": 3390
    },
    {
      "epoch": 0.8362919132149902,
      "grad_norm": 0.359375,
      "learning_rate": 0.00019726064889994763,
      "loss": 0.0755,
      "step": 3392
    },
    {
      "epoch": 0.8367850098619329,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019725264086187334,
      "loss": 0.0778,
      "step": 3394
    },
    {
      "epoch": 0.8372781065088757,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019724462129881812,
      "loss": 0.0756,
      "step": 3396
    },
    {
      "epoch": 0.8377712031558185,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019723659021173235,
      "loss": 0.0712,
      "step": 3398
    },
    {
      "epoch": 0.8382642998027613,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001972285476015677,
      "loss": 0.0799,
      "step": 3400
    },
    {
      "epoch": 0.8387573964497042,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019722049346927736,
      "loss": 0.0737,
      "step": 3402
    },
    {
      "epoch": 0.8392504930966469,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001972124278158157,
      "loss": 0.0777,
      "step": 3404
    },
    {
      "epoch": 0.8397435897435898,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001972043506421386,
      "loss": 0.0757,
      "step": 3406
    },
    {
      "epoch": 0.8402366863905325,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001971962619492032,
      "loss": 0.0716,
      "step": 3408
    },
    {
      "epoch": 0.8407297830374754,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001971881617379681,
      "loss": 0.0715,
      "step": 3410
    },
    {
      "epoch": 0.8412228796844181,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001971800500093932,
      "loss": 0.0786,
      "step": 3412
    },
    {
      "epoch": 0.841715976331361,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001971719267644398,
      "loss": 0.0712,
      "step": 3414
    },
    {
      "epoch": 0.8422090729783037,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019716379200407054,
      "loss": 0.0729,
      "step": 3416
    },
    {
      "epoch": 0.8427021696252466,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019715564572924941,
      "loss": 0.0804,
      "step": 3418
    },
    {
      "epoch": 0.8431952662721893,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00019714748794094186,
      "loss": 0.0728,
      "step": 3420
    },
    {
      "epoch": 0.8436883629191322,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00019713931864011452,
      "loss": 0.0727,
      "step": 3422
    },
    {
      "epoch": 0.8441814595660749,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019713113782773563,
      "loss": 0.0698,
      "step": 3424
    },
    {
      "epoch": 0.8446745562130178,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019712294550477456,
      "loss": 0.0732,
      "step": 3426
    },
    {
      "epoch": 0.8451676528599605,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001971147416722022,
      "loss": 0.0696,
      "step": 3428
    },
    {
      "epoch": 0.8456607495069034,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001971065263309907,
      "loss": 0.0728,
      "step": 3430
    },
    {
      "epoch": 0.8461538461538461,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00019709829948211368,
      "loss": 0.0702,
      "step": 3432
    },
    {
      "epoch": 0.846646942800789,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019709006112654604,
      "loss": 0.0766,
      "step": 3434
    },
    {
      "epoch": 0.8471400394477318,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001970818112652641,
      "loss": 0.0717,
      "step": 3436
    },
    {
      "epoch": 0.8476331360946746,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019707354989924547,
      "loss": 0.0686,
      "step": 3438
    },
    {
      "epoch": 0.8481262327416174,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00019706527702946917,
      "loss": 0.0753,
      "step": 3440
    },
    {
      "epoch": 0.8486193293885601,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019705699265691563,
      "loss": 0.0719,
      "step": 3442
    },
    {
      "epoch": 0.849112426035503,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019704869678256656,
      "loss": 0.0755,
      "step": 3444
    },
    {
      "epoch": 0.8496055226824457,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019704038940740505,
      "loss": 0.0732,
      "step": 3446
    },
    {
      "epoch": 0.8500986193293886,
      "grad_norm": 0.40625,
      "learning_rate": 0.00019703207053241558,
      "loss": 0.072,
      "step": 3448
    },
    {
      "epoch": 0.8505917159763313,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00019702374015858403,
      "loss": 0.0702,
      "step": 3450
    },
    {
      "epoch": 0.8510848126232742,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019701539828689753,
      "loss": 0.0747,
      "step": 3452
    },
    {
      "epoch": 0.8515779092702169,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001970070449183447,
      "loss": 0.0798,
      "step": 3454
    },
    {
      "epoch": 0.8520710059171598,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019699868005391543,
      "loss": 0.0713,
      "step": 3456
    },
    {
      "epoch": 0.8525641025641025,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019699030369460096,
      "loss": 0.0765,
      "step": 3458
    },
    {
      "epoch": 0.8530571992110454,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000196981915841394,
      "loss": 0.0658,
      "step": 3460
    },
    {
      "epoch": 0.8535502958579881,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019697351649528854,
      "loss": 0.0698,
      "step": 3462
    },
    {
      "epoch": 0.854043392504931,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019696510565727994,
      "loss": 0.0796,
      "step": 3464
    },
    {
      "epoch": 0.8545364891518737,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019695668332836495,
      "loss": 0.0836,
      "step": 3466
    },
    {
      "epoch": 0.8550295857988166,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019694824950954165,
      "loss": 0.0745,
      "step": 3468
    },
    {
      "epoch": 0.8555226824457594,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019693980420180945,
      "loss": 0.0778,
      "step": 3470
    },
    {
      "epoch": 0.8560157790927022,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019693134740616925,
      "loss": 0.0807,
      "step": 3472
    },
    {
      "epoch": 0.856508875739645,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001969228791236232,
      "loss": 0.0705,
      "step": 3474
    },
    {
      "epoch": 0.8570019723865878,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001969143993551748,
      "loss": 0.0719,
      "step": 3476
    },
    {
      "epoch": 0.8574950690335306,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019690590810182896,
      "loss": 0.0819,
      "step": 3478
    },
    {
      "epoch": 0.8579881656804734,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019689740536459199,
      "loss": 0.075,
      "step": 3480
    },
    {
      "epoch": 0.8584812623274162,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019688889114447146,
      "loss": 0.0724,
      "step": 3482
    },
    {
      "epoch": 0.8589743589743589,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019688036544247635,
      "loss": 0.0671,
      "step": 3484
    },
    {
      "epoch": 0.8594674556213018,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00019687182825961705,
      "loss": 0.0712,
      "step": 3486
    },
    {
      "epoch": 0.8599605522682445,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019686327959690526,
      "loss": 0.0724,
      "step": 3488
    },
    {
      "epoch": 0.8604536489151874,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00019685471945535399,
      "loss": 0.0738,
      "step": 3490
    },
    {
      "epoch": 0.8609467455621301,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001968461478359777,
      "loss": 0.0752,
      "step": 3492
    },
    {
      "epoch": 0.861439842209073,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00019683756473979217,
      "loss": 0.0636,
      "step": 3494
    },
    {
      "epoch": 0.8619329388560157,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001968289701678146,
      "loss": 0.0711,
      "step": 3496
    },
    {
      "epoch": 0.8624260355029586,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019682036412106335,
      "loss": 0.0714,
      "step": 3498
    },
    {
      "epoch": 0.8629191321499013,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019681174660055845,
      "loss": 0.0765,
      "step": 3500
    },
    {
      "epoch": 0.8634122287968442,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019680311760732102,
      "loss": 0.0661,
      "step": 3502
    },
    {
      "epoch": 0.863905325443787,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019679447714237368,
      "loss": 0.07,
      "step": 3504
    },
    {
      "epoch": 0.8643984220907298,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019678582520674036,
      "loss": 0.0786,
      "step": 3506
    },
    {
      "epoch": 0.8648915187376726,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019677716180144638,
      "loss": 0.0708,
      "step": 3508
    },
    {
      "epoch": 0.8653846153846154,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019676848692751838,
      "loss": 0.072,
      "step": 3510
    },
    {
      "epoch": 0.8658777120315582,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001967598005859844,
      "loss": 0.0761,
      "step": 3512
    },
    {
      "epoch": 0.866370808678501,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001967511027778738,
      "loss": 0.0755,
      "step": 3514
    },
    {
      "epoch": 0.8668639053254438,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019674239350421735,
      "loss": 0.0731,
      "step": 3516
    },
    {
      "epoch": 0.8673570019723866,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001967336727660471,
      "loss": 0.0691,
      "step": 3518
    },
    {
      "epoch": 0.8678500986193294,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019672494056439659,
      "loss": 0.0739,
      "step": 3520
    },
    {
      "epoch": 0.8683431952662722,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019671619690030055,
      "loss": 0.0788,
      "step": 3522
    },
    {
      "epoch": 0.868836291913215,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019670744177479517,
      "loss": 0.0668,
      "step": 3524
    },
    {
      "epoch": 0.8693293885601578,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019669867518891802,
      "loss": 0.0805,
      "step": 3526
    },
    {
      "epoch": 0.8698224852071006,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00019668989714370793,
      "loss": 0.0747,
      "step": 3528
    },
    {
      "epoch": 0.8703155818540433,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019668110764020522,
      "loss": 0.0656,
      "step": 3530
    },
    {
      "epoch": 0.8708086785009862,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019667230667945145,
      "loss": 0.0782,
      "step": 3532
    },
    {
      "epoch": 0.871301775147929,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019666349426248955,
      "loss": 0.0786,
      "step": 3534
    },
    {
      "epoch": 0.8717948717948718,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019665467039036393,
      "loss": 0.0728,
      "step": 3536
    },
    {
      "epoch": 0.8722879684418146,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001966458350641202,
      "loss": 0.0662,
      "step": 3538
    },
    {
      "epoch": 0.8727810650887574,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001966369882848054,
      "loss": 0.0676,
      "step": 3540
    },
    {
      "epoch": 0.8732741617357002,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019662813005346793,
      "loss": 0.0747,
      "step": 3542
    },
    {
      "epoch": 0.873767258382643,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019661926037115756,
      "loss": 0.0726,
      "step": 3544
    },
    {
      "epoch": 0.8742603550295858,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001966103792389254,
      "loss": 0.0711,
      "step": 3546
    },
    {
      "epoch": 0.8747534516765286,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019660148665782387,
      "loss": 0.0693,
      "step": 3548
    },
    {
      "epoch": 0.8752465483234714,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019659258262890683,
      "loss": 0.0761,
      "step": 3550
    },
    {
      "epoch": 0.8757396449704142,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019658366715322946,
      "loss": 0.0786,
      "step": 3552
    },
    {
      "epoch": 0.876232741617357,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019657474023184826,
      "loss": 0.0727,
      "step": 3554
    },
    {
      "epoch": 0.8767258382642998,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019656580186582114,
      "loss": 0.0679,
      "step": 3556
    },
    {
      "epoch": 0.8772189349112426,
      "grad_norm": 0.40625,
      "learning_rate": 0.00019655685205620732,
      "loss": 0.0741,
      "step": 3558
    },
    {
      "epoch": 0.8777120315581854,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001965478908040675,
      "loss": 0.0703,
      "step": 3560
    },
    {
      "epoch": 0.8782051282051282,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001965389181104635,
      "loss": 0.0772,
      "step": 3562
    },
    {
      "epoch": 0.878698224852071,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001965299339764587,
      "loss": 0.0781,
      "step": 3564
    },
    {
      "epoch": 0.8791913214990138,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001965209384031178,
      "loss": 0.0703,
      "step": 3566
    },
    {
      "epoch": 0.8796844181459567,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0001965119313915068,
      "loss": 0.0722,
      "step": 3568
    },
    {
      "epoch": 0.8801775147928994,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019650291294269305,
      "loss": 0.0715,
      "step": 3570
    },
    {
      "epoch": 0.8806706114398422,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001964938830577453,
      "loss": 0.0703,
      "step": 3572
    },
    {
      "epoch": 0.881163708086785,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019648484173773368,
      "loss": 0.0713,
      "step": 3574
    },
    {
      "epoch": 0.8816568047337278,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001964757889837296,
      "loss": 0.0681,
      "step": 3576
    },
    {
      "epoch": 0.8821499013806706,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001964667247968059,
      "loss": 0.0759,
      "step": 3578
    },
    {
      "epoch": 0.8826429980276134,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0001964576491780367,
      "loss": 0.0667,
      "step": 3580
    },
    {
      "epoch": 0.8831360946745562,
      "grad_norm": 0.75,
      "learning_rate": 0.0001964485621284975,
      "loss": 0.0791,
      "step": 3582
    },
    {
      "epoch": 0.883629191321499,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001964394636492652,
      "loss": 0.0707,
      "step": 3584
    },
    {
      "epoch": 0.8841222879684418,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000196430353741418,
      "loss": 0.0668,
      "step": 3586
    },
    {
      "epoch": 0.8846153846153846,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001964212324060355,
      "loss": 0.071,
      "step": 3588
    },
    {
      "epoch": 0.8851084812623274,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019641209964419863,
      "loss": 0.0784,
      "step": 3590
    },
    {
      "epoch": 0.8856015779092702,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001964029554569896,
      "loss": 0.0706,
      "step": 3592
    },
    {
      "epoch": 0.886094674556213,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019639379984549217,
      "loss": 0.0678,
      "step": 3594
    },
    {
      "epoch": 0.8865877712031558,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019638463281079123,
      "loss": 0.0663,
      "step": 3596
    },
    {
      "epoch": 0.8870808678500987,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019637545435397314,
      "loss": 0.0761,
      "step": 3598
    },
    {
      "epoch": 0.8875739644970414,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019636626447612563,
      "loss": 0.0706,
      "step": 3600
    },
    {
      "epoch": 0.8880670611439843,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019635706317833776,
      "loss": 0.0685,
      "step": 3602
    },
    {
      "epoch": 0.888560157790927,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019634785046169988,
      "loss": 0.0738,
      "step": 3604
    },
    {
      "epoch": 0.8890532544378699,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001963386263273038,
      "loss": 0.0713,
      "step": 3606
    },
    {
      "epoch": 0.8895463510848126,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019632939077624263,
      "loss": 0.0717,
      "step": 3608
    },
    {
      "epoch": 0.8900394477317555,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001963201438096108,
      "loss": 0.0755,
      "step": 3610
    },
    {
      "epoch": 0.8905325443786982,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019631088542850417,
      "loss": 0.0741,
      "step": 3612
    },
    {
      "epoch": 0.8910256410256411,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019630161563401985,
      "loss": 0.0702,
      "step": 3614
    },
    {
      "epoch": 0.8915187376725838,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019629233442725643,
      "loss": 0.0685,
      "step": 3616
    },
    {
      "epoch": 0.8920118343195266,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019628304180931375,
      "loss": 0.0747,
      "step": 3618
    },
    {
      "epoch": 0.8925049309664694,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00019627373778129302,
      "loss": 0.074,
      "step": 3620
    },
    {
      "epoch": 0.8929980276134122,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019626442234429683,
      "loss": 0.0707,
      "step": 3622
    },
    {
      "epoch": 0.893491124260355,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019625509549942916,
      "loss": 0.067,
      "step": 3624
    },
    {
      "epoch": 0.8939842209072978,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001962457572477952,
      "loss": 0.0746,
      "step": 3626
    },
    {
      "epoch": 0.8944773175542406,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019623640759050168,
      "loss": 0.0789,
      "step": 3628
    },
    {
      "epoch": 0.8949704142011834,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00019622704652865654,
      "loss": 0.0654,
      "step": 3630
    },
    {
      "epoch": 0.8954635108481263,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019621767406336914,
      "loss": 0.0749,
      "step": 3632
    },
    {
      "epoch": 0.895956607495069,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001962082901957501,
      "loss": 0.0741,
      "step": 3634
    },
    {
      "epoch": 0.8964497041420119,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019619889492691154,
      "loss": 0.0769,
      "step": 3636
    },
    {
      "epoch": 0.8969428007889546,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00019618948825796684,
      "loss": 0.0718,
      "step": 3638
    },
    {
      "epoch": 0.8974358974358975,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00019618007019003072,
      "loss": 0.0666,
      "step": 3640
    },
    {
      "epoch": 0.8979289940828402,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001961706407242193,
      "loss": 0.0696,
      "step": 3642
    },
    {
      "epoch": 0.8984220907297831,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019616119986164997,
      "loss": 0.0733,
      "step": 3644
    },
    {
      "epoch": 0.8989151873767258,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00019615174760344157,
      "loss": 0.071,
      "step": 3646
    },
    {
      "epoch": 0.8994082840236687,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019614228395071424,
      "loss": 0.0789,
      "step": 3648
    },
    {
      "epoch": 0.8999013806706114,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019613280890458948,
      "loss": 0.0737,
      "step": 3650
    },
    {
      "epoch": 0.9003944773175543,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019612332246619012,
      "loss": 0.0755,
      "step": 3652
    },
    {
      "epoch": 0.900887573964497,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019611382463664037,
      "loss": 0.0661,
      "step": 3654
    },
    {
      "epoch": 0.9013806706114399,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001961043154170658,
      "loss": 0.0683,
      "step": 3656
    },
    {
      "epoch": 0.9018737672583826,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00019609479480859322,
      "loss": 0.0692,
      "step": 3658
    },
    {
      "epoch": 0.9023668639053254,
      "grad_norm": 0.390625,
      "learning_rate": 0.00019608526281235093,
      "loss": 0.0606,
      "step": 3660
    },
    {
      "epoch": 0.9028599605522682,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00019607571942946854,
      "loss": 0.0723,
      "step": 3662
    },
    {
      "epoch": 0.903353057199211,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00019606616466107699,
      "loss": 0.0706,
      "step": 3664
    },
    {
      "epoch": 0.9038461538461539,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019605659850830857,
      "loss": 0.0737,
      "step": 3666
    },
    {
      "epoch": 0.9043392504930966,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001960470209722969,
      "loss": 0.0824,
      "step": 3668
    },
    {
      "epoch": 0.9048323471400395,
      "grad_norm": 0.439453125,
      "learning_rate": 0.000196037432054177,
      "loss": 0.068,
      "step": 3670
    },
    {
      "epoch": 0.9053254437869822,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001960278317550852,
      "loss": 0.0703,
      "step": 3672
    },
    {
      "epoch": 0.9058185404339251,
      "grad_norm": 0.5,
      "learning_rate": 0.00019601822007615915,
      "loss": 0.0746,
      "step": 3674
    },
    {
      "epoch": 0.9063116370808678,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00019600859701853794,
      "loss": 0.0689,
      "step": 3676
    },
    {
      "epoch": 0.9068047337278107,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000195998962583362,
      "loss": 0.0723,
      "step": 3678
    },
    {
      "epoch": 0.9072978303747534,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001959893167717729,
      "loss": 0.074,
      "step": 3680
    },
    {
      "epoch": 0.9077909270216963,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001959796595849139,
      "loss": 0.0684,
      "step": 3682
    },
    {
      "epoch": 0.908284023668639,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019596999102392934,
      "loss": 0.0697,
      "step": 3684
    },
    {
      "epoch": 0.9087771203155819,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019596031108996503,
      "loss": 0.0702,
      "step": 3686
    },
    {
      "epoch": 0.9092702169625246,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019595061978416807,
      "loss": 0.0685,
      "step": 3688
    },
    {
      "epoch": 0.9097633136094675,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001959409171076869,
      "loss": 0.076,
      "step": 3690
    },
    {
      "epoch": 0.9102564102564102,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019593120306167147,
      "loss": 0.0687,
      "step": 3692
    },
    {
      "epoch": 0.9107495069033531,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00019592147764727283,
      "loss": 0.0746,
      "step": 3694
    },
    {
      "epoch": 0.9112426035502958,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001959117408656435,
      "loss": 0.0717,
      "step": 3696
    },
    {
      "epoch": 0.9117357001972387,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001959019927179374,
      "loss": 0.0624,
      "step": 3698
    },
    {
      "epoch": 0.9122287968441815,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00019589223320530975,
      "loss": 0.0751,
      "step": 3700
    },
    {
      "epoch": 0.9127218934911243,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000195882462328917,
      "loss": 0.0672,
      "step": 3702
    },
    {
      "epoch": 0.9132149901380671,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001958726800899172,
      "loss": 0.0747,
      "step": 3704
    },
    {
      "epoch": 0.9137080867850098,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019586288648946947,
      "loss": 0.0678,
      "step": 3706
    },
    {
      "epoch": 0.9142011834319527,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019585308152873448,
      "loss": 0.0722,
      "step": 3708
    },
    {
      "epoch": 0.9146942800788954,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00019584326520887414,
      "loss": 0.0743,
      "step": 3710
    },
    {
      "epoch": 0.9151873767258383,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00019583343753105177,
      "loss": 0.0636,
      "step": 3712
    },
    {
      "epoch": 0.915680473372781,
      "grad_norm": 0.453125,
      "learning_rate": 0.000195823598496432,
      "loss": 0.0694,
      "step": 3714
    },
    {
      "epoch": 0.9161735700197239,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00019581374810618077,
      "loss": 0.0713,
      "step": 3716
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.5,
      "learning_rate": 0.00019580388636146543,
      "loss": 0.0667,
      "step": 3718
    },
    {
      "epoch": 0.9171597633136095,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00019579401326345468,
      "loss": 0.0698,
      "step": 3720
    },
    {
      "epoch": 0.9176528599605522,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0001957841288133185,
      "loss": 0.0692,
      "step": 3722
    },
    {
      "epoch": 0.9181459566074951,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00019577423301222826,
      "loss": 0.0653,
      "step": 3724
    },
    {
      "epoch": 0.9186390532544378,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001957643258613567,
      "loss": 0.0655,
      "step": 3726
    },
    {
      "epoch": 0.9191321499013807,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019575440736187781,
      "loss": 0.074,
      "step": 3728
    },
    {
      "epoch": 0.9196252465483234,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00019574447751496705,
      "loss": 0.0739,
      "step": 3730
    },
    {
      "epoch": 0.9201183431952663,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019573453632180113,
      "loss": 0.073,
      "step": 3732
    },
    {
      "epoch": 0.9206114398422091,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00019572458378355813,
      "loss": 0.0701,
      "step": 3734
    },
    {
      "epoch": 0.9211045364891519,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019571461990141754,
      "loss": 0.0692,
      "step": 3736
    },
    {
      "epoch": 0.9215976331360947,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00019570464467656001,
      "loss": 0.072,
      "step": 3738
    },
    {
      "epoch": 0.9220907297830375,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001956946581101678,
      "loss": 0.0703,
      "step": 3740
    },
    {
      "epoch": 0.9225838264299803,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001956846602034243,
      "loss": 0.0741,
      "step": 3742
    },
    {
      "epoch": 0.9230769230769231,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001956746509575143,
      "loss": 0.0679,
      "step": 3744
    },
    {
      "epoch": 0.9235700197238659,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019566463037362403,
      "loss": 0.0723,
      "step": 3746
    },
    {
      "epoch": 0.9240631163708086,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019565459845294092,
      "loss": 0.0746,
      "step": 3748
    },
    {
      "epoch": 0.9245562130177515,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019564455519665384,
      "loss": 0.0815,
      "step": 3750
    },
    {
      "epoch": 0.9250493096646942,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001956345006059529,
      "loss": 0.0688,
      "step": 3752
    },
    {
      "epoch": 0.9255424063116371,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019562443468202975,
      "loss": 0.074,
      "step": 3754
    },
    {
      "epoch": 0.9260355029585798,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019561435742607717,
      "loss": 0.0737,
      "step": 3756
    },
    {
      "epoch": 0.9265285996055227,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001956042688392894,
      "loss": 0.0759,
      "step": 3758
    },
    {
      "epoch": 0.9270216962524654,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019559416892286197,
      "loss": 0.0731,
      "step": 3760
    },
    {
      "epoch": 0.9275147928994083,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001955840576779918,
      "loss": 0.0703,
      "step": 3762
    },
    {
      "epoch": 0.928007889546351,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001955739351058771,
      "loss": 0.0761,
      "step": 3764
    },
    {
      "epoch": 0.9285009861932939,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019556380120771752,
      "loss": 0.0748,
      "step": 3766
    },
    {
      "epoch": 0.9289940828402367,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001955536559847139,
      "loss": 0.0734,
      "step": 3768
    },
    {
      "epoch": 0.9294871794871795,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019554349943806857,
      "loss": 0.0679,
      "step": 3770
    },
    {
      "epoch": 0.9299802761341223,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001955333315689851,
      "loss": 0.067,
      "step": 3772
    },
    {
      "epoch": 0.9304733727810651,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019552315237866843,
      "loss": 0.0718,
      "step": 3774
    },
    {
      "epoch": 0.9309664694280079,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019551296186832488,
      "loss": 0.0705,
      "step": 3776
    },
    {
      "epoch": 0.9314595660749507,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019550276003916207,
      "loss": 0.0818,
      "step": 3778
    },
    {
      "epoch": 0.9319526627218935,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019549254689238898,
      "loss": 0.0736,
      "step": 3780
    },
    {
      "epoch": 0.9324457593688363,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019548232242921595,
      "loss": 0.0683,
      "step": 3782
    },
    {
      "epoch": 0.9329388560157791,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019547208665085457,
      "loss": 0.0709,
      "step": 3784
    },
    {
      "epoch": 0.9334319526627219,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019546183955851789,
      "loss": 0.0711,
      "step": 3786
    },
    {
      "epoch": 0.9339250493096647,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019545158115342025,
      "loss": 0.0738,
      "step": 3788
    },
    {
      "epoch": 0.9344181459566075,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00019544131143677727,
      "loss": 0.0729,
      "step": 3790
    },
    {
      "epoch": 0.9349112426035503,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019543103040980603,
      "loss": 0.0731,
      "step": 3792
    },
    {
      "epoch": 0.935404339250493,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019542073807372485,
      "loss": 0.0714,
      "step": 3794
    },
    {
      "epoch": 0.9358974358974359,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019541043442975346,
      "loss": 0.0654,
      "step": 3796
    },
    {
      "epoch": 0.9363905325443787,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001954001194791129,
      "loss": 0.0769,
      "step": 3798
    },
    {
      "epoch": 0.9368836291913215,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001953897932230255,
      "loss": 0.0668,
      "step": 3800
    },
    {
      "epoch": 0.9373767258382643,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019537945566271505,
      "loss": 0.0732,
      "step": 3802
    },
    {
      "epoch": 0.9378698224852071,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00019536910679940656,
      "loss": 0.0672,
      "step": 3804
    },
    {
      "epoch": 0.9383629191321499,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00019535874663432644,
      "loss": 0.0683,
      "step": 3806
    },
    {
      "epoch": 0.9388560157790927,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019534837516870245,
      "loss": 0.0746,
      "step": 3808
    },
    {
      "epoch": 0.9393491124260355,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00019533799240376362,
      "loss": 0.0656,
      "step": 3810
    },
    {
      "epoch": 0.9398422090729783,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019532759834074037,
      "loss": 0.0706,
      "step": 3812
    },
    {
      "epoch": 0.9403353057199211,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001953171929808645,
      "loss": 0.0735,
      "step": 3814
    },
    {
      "epoch": 0.9408284023668639,
      "grad_norm": 0.4375,
      "learning_rate": 0.00019530677632536908,
      "loss": 0.0714,
      "step": 3816
    },
    {
      "epoch": 0.9413214990138067,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019529634837548855,
      "loss": 0.0712,
      "step": 3818
    },
    {
      "epoch": 0.9418145956607495,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019528590913245867,
      "loss": 0.0686,
      "step": 3820
    },
    {
      "epoch": 0.9423076923076923,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019527545859751656,
      "loss": 0.0729,
      "step": 3822
    },
    {
      "epoch": 0.9428007889546351,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019526499677190062,
      "loss": 0.0714,
      "step": 3824
    },
    {
      "epoch": 0.9432938856015779,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001952545236568507,
      "loss": 0.0723,
      "step": 3826
    },
    {
      "epoch": 0.9437869822485208,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019524403925360787,
      "loss": 0.0732,
      "step": 3828
    },
    {
      "epoch": 0.9442800788954635,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019523354356341464,
      "loss": 0.0694,
      "step": 3830
    },
    {
      "epoch": 0.9447731755424064,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019522303658751477,
      "loss": 0.0747,
      "step": 3832
    },
    {
      "epoch": 0.9452662721893491,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001952125183271534,
      "loss": 0.0711,
      "step": 3834
    },
    {
      "epoch": 0.9457593688362919,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00019520198878357703,
      "loss": 0.0732,
      "step": 3836
    },
    {
      "epoch": 0.9462524654832347,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001951914479580334,
      "loss": 0.0664,
      "step": 3838
    },
    {
      "epoch": 0.9467455621301775,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00019518089585177174,
      "loss": 0.0709,
      "step": 3840
    },
    {
      "epoch": 0.9472386587771203,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001951703324660425,
      "loss": 0.0692,
      "step": 3842
    },
    {
      "epoch": 0.9477317554240631,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00019515975780209753,
      "loss": 0.068,
      "step": 3844
    },
    {
      "epoch": 0.9482248520710059,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001951491718611899,
      "loss": 0.0784,
      "step": 3846
    },
    {
      "epoch": 0.9487179487179487,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001951385746445742,
      "loss": 0.068,
      "step": 3848
    },
    {
      "epoch": 0.9492110453648915,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001951279661535062,
      "loss": 0.0677,
      "step": 3850
    },
    {
      "epoch": 0.9497041420118343,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019511734638924308,
      "loss": 0.0691,
      "step": 3852
    },
    {
      "epoch": 0.9501972386587771,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019510671535304333,
      "loss": 0.0691,
      "step": 3854
    },
    {
      "epoch": 0.9506903353057199,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019509607304616687,
      "loss": 0.0784,
      "step": 3856
    },
    {
      "epoch": 0.9511834319526628,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019508541946987475,
      "loss": 0.0766,
      "step": 3858
    },
    {
      "epoch": 0.9516765285996055,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019507475462542955,
      "loss": 0.0723,
      "step": 3860
    },
    {
      "epoch": 0.9521696252465484,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001950640785140951,
      "loss": 0.0721,
      "step": 3862
    },
    {
      "epoch": 0.9526627218934911,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019505339113713656,
      "loss": 0.0758,
      "step": 3864
    },
    {
      "epoch": 0.953155818540434,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019504269249582046,
      "loss": 0.0674,
      "step": 3866
    },
    {
      "epoch": 0.9536489151873767,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019503198259141467,
      "loss": 0.0742,
      "step": 3868
    },
    {
      "epoch": 0.9541420118343196,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019502126142518833,
      "loss": 0.0662,
      "step": 3870
    },
    {
      "epoch": 0.9546351084812623,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019501052899841201,
      "loss": 0.0694,
      "step": 3872
    },
    {
      "epoch": 0.9551282051282052,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001949997853123575,
      "loss": 0.0726,
      "step": 3874
    },
    {
      "epoch": 0.9556213017751479,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019498903036829806,
      "loss": 0.0704,
      "step": 3876
    },
    {
      "epoch": 0.9561143984220908,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001949782641675081,
      "loss": 0.0604,
      "step": 3878
    },
    {
      "epoch": 0.9566074950690335,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019496748671126362,
      "loss": 0.0699,
      "step": 3880
    },
    {
      "epoch": 0.9571005917159763,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001949566980008417,
      "loss": 0.0721,
      "step": 3882
    },
    {
      "epoch": 0.9575936883629191,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001949458980375209,
      "loss": 0.0709,
      "step": 3884
    },
    {
      "epoch": 0.9580867850098619,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019493508682258108,
      "loss": 0.0718,
      "step": 3886
    },
    {
      "epoch": 0.9585798816568047,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001949242643573034,
      "loss": 0.0802,
      "step": 3888
    },
    {
      "epoch": 0.9590729783037475,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019491343064297043,
      "loss": 0.0672,
      "step": 3890
    },
    {
      "epoch": 0.9595660749506904,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000194902585680866,
      "loss": 0.0725,
      "step": 3892
    },
    {
      "epoch": 0.9600591715976331,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019489172947227527,
      "loss": 0.0781,
      "step": 3894
    },
    {
      "epoch": 0.960552268244576,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001948808620184848,
      "loss": 0.071,
      "step": 3896
    },
    {
      "epoch": 0.9610453648915187,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00019486998332078246,
      "loss": 0.0726,
      "step": 3898
    },
    {
      "epoch": 0.9615384615384616,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001948590933804574,
      "loss": 0.0745,
      "step": 3900
    },
    {
      "epoch": 0.9620315581854043,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019484819219880013,
      "loss": 0.0673,
      "step": 3902
    },
    {
      "epoch": 0.9625246548323472,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019483727977710253,
      "loss": 0.0744,
      "step": 3904
    },
    {
      "epoch": 0.9630177514792899,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019482635611665775,
      "loss": 0.0733,
      "step": 3906
    },
    {
      "epoch": 0.9635108481262328,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019481542121876038,
      "loss": 0.0672,
      "step": 3908
    },
    {
      "epoch": 0.9640039447731755,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001948044750847062,
      "loss": 0.0751,
      "step": 3910
    },
    {
      "epoch": 0.9644970414201184,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019479351771579235,
      "loss": 0.0732,
      "step": 3912
    },
    {
      "epoch": 0.9649901380670611,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019478254911331743,
      "loss": 0.071,
      "step": 3914
    },
    {
      "epoch": 0.965483234714004,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019477156927858125,
      "loss": 0.0668,
      "step": 3916
    },
    {
      "epoch": 0.9659763313609467,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019476057821288495,
      "loss": 0.0728,
      "step": 3918
    },
    {
      "epoch": 0.9664694280078896,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019474957591753106,
      "loss": 0.0706,
      "step": 3920
    },
    {
      "epoch": 0.9669625246548323,
      "grad_norm": 0.5,
      "learning_rate": 0.00019473856239382345,
      "loss": 0.0694,
      "step": 3922
    },
    {
      "epoch": 0.9674556213017751,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001947275376430672,
      "loss": 0.0709,
      "step": 3924
    },
    {
      "epoch": 0.967948717948718,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019471650166656888,
      "loss": 0.0679,
      "step": 3926
    },
    {
      "epoch": 0.9684418145956607,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019470545446563625,
      "loss": 0.063,
      "step": 3928
    },
    {
      "epoch": 0.9689349112426036,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019469439604157852,
      "loss": 0.0715,
      "step": 3930
    },
    {
      "epoch": 0.9694280078895463,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019468332639570614,
      "loss": 0.0802,
      "step": 3932
    },
    {
      "epoch": 0.9699211045364892,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019467224552933095,
      "loss": 0.0689,
      "step": 3934
    },
    {
      "epoch": 0.9704142011834319,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019466115344376607,
      "loss": 0.0646,
      "step": 3936
    },
    {
      "epoch": 0.9709072978303748,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019465005014032602,
      "loss": 0.066,
      "step": 3938
    },
    {
      "epoch": 0.9714003944773175,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019463893562032654,
      "loss": 0.0749,
      "step": 3940
    },
    {
      "epoch": 0.9718934911242604,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001946278098850848,
      "loss": 0.0667,
      "step": 3942
    },
    {
      "epoch": 0.9723865877712031,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019461667293591925,
      "loss": 0.0764,
      "step": 3944
    },
    {
      "epoch": 0.972879684418146,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001946055247741497,
      "loss": 0.0618,
      "step": 3946
    },
    {
      "epoch": 0.9733727810650887,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019459436540109724,
      "loss": 0.0739,
      "step": 3948
    },
    {
      "epoch": 0.9738658777120316,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019458319481808434,
      "loss": 0.077,
      "step": 3950
    },
    {
      "epoch": 0.9743589743589743,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019457201302643476,
      "loss": 0.077,
      "step": 3952
    },
    {
      "epoch": 0.9748520710059172,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001945608200274736,
      "loss": 0.0685,
      "step": 3954
    },
    {
      "epoch": 0.97534516765286,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00019454961582252733,
      "loss": 0.0702,
      "step": 3956
    },
    {
      "epoch": 0.9758382642998028,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00019453840041292367,
      "loss": 0.0721,
      "step": 3958
    },
    {
      "epoch": 0.9763313609467456,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00019452717379999176,
      "loss": 0.0696,
      "step": 3960
    },
    {
      "epoch": 0.9768244575936884,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019451593598506195,
      "loss": 0.0645,
      "step": 3962
    },
    {
      "epoch": 0.9773175542406312,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019450468696946603,
      "loss": 0.0712,
      "step": 3964
    },
    {
      "epoch": 0.977810650887574,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019449342675453707,
      "loss": 0.0659,
      "step": 3966
    },
    {
      "epoch": 0.9783037475345168,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00019448215534160945,
      "loss": 0.067,
      "step": 3968
    },
    {
      "epoch": 0.9787968441814595,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001944708727320189,
      "loss": 0.0751,
      "step": 3970
    },
    {
      "epoch": 0.9792899408284024,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019445957892710245,
      "loss": 0.0716,
      "step": 3972
    },
    {
      "epoch": 0.9797830374753451,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019444827392819852,
      "loss": 0.0727,
      "step": 3974
    },
    {
      "epoch": 0.980276134122288,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001944369577366468,
      "loss": 0.0702,
      "step": 3976
    },
    {
      "epoch": 0.9807692307692307,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00019442563035378831,
      "loss": 0.0744,
      "step": 3978
    },
    {
      "epoch": 0.9812623274161736,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019441429178096542,
      "loss": 0.0678,
      "step": 3980
    },
    {
      "epoch": 0.9817554240631163,
      "grad_norm": 0.625,
      "learning_rate": 0.00019440294201952184,
      "loss": 0.067,
      "step": 3982
    },
    {
      "epoch": 0.9822485207100592,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001943915810708025,
      "loss": 0.0729,
      "step": 3984
    },
    {
      "epoch": 0.9827416173570019,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019438020893615383,
      "loss": 0.0654,
      "step": 3986
    },
    {
      "epoch": 0.9832347140039448,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019436882561692343,
      "loss": 0.0758,
      "step": 3988
    },
    {
      "epoch": 0.9837278106508875,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001943574311144603,
      "loss": 0.0684,
      "step": 3990
    },
    {
      "epoch": 0.9842209072978304,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019434602543011477,
      "loss": 0.0786,
      "step": 3992
    },
    {
      "epoch": 0.9847140039447732,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00019433460856523849,
      "loss": 0.0722,
      "step": 3994
    },
    {
      "epoch": 0.985207100591716,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00019432318052118436,
      "loss": 0.0714,
      "step": 3996
    },
    {
      "epoch": 0.9857001972386588,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019431174129930673,
      "loss": 0.0708,
      "step": 3998
    },
    {
      "epoch": 0.9861932938856016,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019430029090096115,
      "loss": 0.0628,
      "step": 4000
    },
    {
      "epoch": 0.9866863905325444,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019428882932750465,
      "loss": 0.0698,
      "step": 4002
    },
    {
      "epoch": 0.9871794871794872,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00019427735658029537,
      "loss": 0.0728,
      "step": 4004
    },
    {
      "epoch": 0.98767258382643,
      "grad_norm": 0.6640625,
      "learning_rate": 0.000194265872660693,
      "loss": 0.0675,
      "step": 4006
    },
    {
      "epoch": 0.9881656804733728,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001942543775700584,
      "loss": 0.0706,
      "step": 4008
    },
    {
      "epoch": 0.9886587771203156,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001942428713097538,
      "loss": 0.0733,
      "step": 4010
    },
    {
      "epoch": 0.9891518737672583,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001942313538811428,
      "loss": 0.0684,
      "step": 4012
    },
    {
      "epoch": 0.9896449704142012,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001942198252855902,
      "loss": 0.0731,
      "step": 4014
    },
    {
      "epoch": 0.9901380670611439,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019420828552446228,
      "loss": 0.068,
      "step": 4016
    },
    {
      "epoch": 0.9906311637080868,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001941967345991265,
      "loss": 0.0719,
      "step": 4018
    },
    {
      "epoch": 0.9911242603550295,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019418517251095177,
      "loss": 0.0718,
      "step": 4020
    },
    {
      "epoch": 0.9916173570019724,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001941735992613082,
      "loss": 0.072,
      "step": 4022
    },
    {
      "epoch": 0.9921104536489151,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00019416201485156735,
      "loss": 0.0654,
      "step": 4024
    },
    {
      "epoch": 0.992603550295858,
      "grad_norm": 0.546875,
      "learning_rate": 0.000194150419283102,
      "loss": 0.0665,
      "step": 4026
    },
    {
      "epoch": 0.9930966469428008,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001941388125572863,
      "loss": 0.0703,
      "step": 4028
    },
    {
      "epoch": 0.9935897435897436,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019412719467549571,
      "loss": 0.0736,
      "step": 4030
    },
    {
      "epoch": 0.9940828402366864,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000194115565639107,
      "loss": 0.0733,
      "step": 4032
    },
    {
      "epoch": 0.9945759368836292,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00019410392544949828,
      "loss": 0.0635,
      "step": 4034
    },
    {
      "epoch": 0.995069033530572,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000194092274108049,
      "loss": 0.0796,
      "step": 4036
    },
    {
      "epoch": 0.9955621301775148,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001940806116161399,
      "loss": 0.068,
      "step": 4038
    },
    {
      "epoch": 0.9960552268244576,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00019406893797515305,
      "loss": 0.0656,
      "step": 4040
    },
    {
      "epoch": 0.9965483234714004,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019405725318647182,
      "loss": 0.0709,
      "step": 4042
    },
    {
      "epoch": 0.9970414201183432,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019404555725148094,
      "loss": 0.0741,
      "step": 4044
    },
    {
      "epoch": 0.997534516765286,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019403385017156642,
      "loss": 0.0703,
      "step": 4046
    },
    {
      "epoch": 0.9980276134122288,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001940221319481157,
      "loss": 0.074,
      "step": 4048
    },
    {
      "epoch": 0.9985207100591716,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019401040258251734,
      "loss": 0.0785,
      "step": 4050
    },
    {
      "epoch": 0.9990138067061144,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001939986620761614,
      "loss": 0.0653,
      "step": 4052
    },
    {
      "epoch": 0.9995069033530573,
      "grad_norm": 0.5,
      "learning_rate": 0.0001939869104304392,
      "loss": 0.066,
      "step": 4054
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019397514764674334,
      "loss": 0.0715,
      "step": 4056
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.07079669833183289,
      "eval_runtime": 61.6782,
      "eval_samples_per_second": 258.26,
      "eval_steps_per_second": 2.027,
      "step": 4056
    },
    {
      "epoch": 1.0004930966469427,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001939633737264678,
      "loss": 0.0537,
      "step": 4058
    },
    {
      "epoch": 1.0009861932938855,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019395158867100784,
      "loss": 0.0474,
      "step": 4060
    },
    {
      "epoch": 1.0014792899408285,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00019393979248176005,
      "loss": 0.0485,
      "step": 4062
    },
    {
      "epoch": 1.0019723865877712,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00019392798516012239,
      "loss": 0.0542,
      "step": 4064
    },
    {
      "epoch": 1.002465483234714,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00019391616670749406,
      "loss": 0.0498,
      "step": 4066
    },
    {
      "epoch": 1.0029585798816567,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001939043371252756,
      "loss": 0.0454,
      "step": 4068
    },
    {
      "epoch": 1.0034516765285997,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001938924964148689,
      "loss": 0.058,
      "step": 4070
    },
    {
      "epoch": 1.0039447731755424,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00019388064457767713,
      "loss": 0.0587,
      "step": 4072
    },
    {
      "epoch": 1.0044378698224852,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019386878161510484,
      "loss": 0.0557,
      "step": 4074
    },
    {
      "epoch": 1.004930966469428,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019385690752855782,
      "loss": 0.0514,
      "step": 4076
    },
    {
      "epoch": 1.005424063116371,
      "grad_norm": 0.421875,
      "learning_rate": 0.00019384502231944325,
      "loss": 0.051,
      "step": 4078
    },
    {
      "epoch": 1.0059171597633136,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00019383312598916954,
      "loss": 0.0546,
      "step": 4080
    },
    {
      "epoch": 1.0064102564102564,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00019382121853914652,
      "loss": 0.0532,
      "step": 4082
    },
    {
      "epoch": 1.0069033530571991,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00019380929997078527,
      "loss": 0.0474,
      "step": 4084
    },
    {
      "epoch": 1.007396449704142,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001937973702854982,
      "loss": 0.0529,
      "step": 4086
    },
    {
      "epoch": 1.0078895463510849,
      "grad_norm": 0.3125,
      "learning_rate": 0.00019378542948469905,
      "loss": 0.0503,
      "step": 4088
    },
    {
      "epoch": 1.0083826429980276,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0001937734775698029,
      "loss": 0.0522,
      "step": 4090
    },
    {
      "epoch": 1.0088757396449703,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00019376151454222608,
      "loss": 0.0497,
      "step": 4092
    },
    {
      "epoch": 1.0093688362919133,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001937495404033863,
      "loss": 0.0528,
      "step": 4094
    },
    {
      "epoch": 1.009861932938856,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00019373755515470254,
      "loss": 0.0521,
      "step": 4096
    },
    {
      "epoch": 1.0103550295857988,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00019372555879759514,
      "loss": 0.0491,
      "step": 4098
    },
    {
      "epoch": 1.0108481262327416,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001937135513334857,
      "loss": 0.046,
      "step": 4100
    },
    {
      "epoch": 1.0113412228796843,
      "grad_norm": 0.4375,
      "learning_rate": 0.00019370153276379722,
      "loss": 0.0551,
      "step": 4102
    },
    {
      "epoch": 1.0118343195266273,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019368950308995395,
      "loss": 0.048,
      "step": 4104
    },
    {
      "epoch": 1.01232741617357,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00019367746231338146,
      "loss": 0.0468,
      "step": 4106
    },
    {
      "epoch": 1.0128205128205128,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019366541043550666,
      "loss": 0.0518,
      "step": 4108
    },
    {
      "epoch": 1.0133136094674555,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00019365334745775775,
      "loss": 0.0498,
      "step": 4110
    },
    {
      "epoch": 1.0138067061143985,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019364127338156428,
      "loss": 0.0498,
      "step": 4112
    },
    {
      "epoch": 1.0142998027613412,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001936291882083571,
      "loss": 0.047,
      "step": 4114
    },
    {
      "epoch": 1.014792899408284,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00019361709193956836,
      "loss": 0.0491,
      "step": 4116
    },
    {
      "epoch": 1.0152859960552267,
      "grad_norm": 0.625,
      "learning_rate": 0.0001936049845766315,
      "loss": 0.0542,
      "step": 4118
    },
    {
      "epoch": 1.0157790927021697,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00019359286612098139,
      "loss": 0.0482,
      "step": 4120
    },
    {
      "epoch": 1.0162721893491125,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019358073657405406,
      "loss": 0.0569,
      "step": 4122
    },
    {
      "epoch": 1.0167652859960552,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00019356859593728699,
      "loss": 0.0434,
      "step": 4124
    },
    {
      "epoch": 1.017258382642998,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019355644421211886,
      "loss": 0.0462,
      "step": 4126
    },
    {
      "epoch": 1.017751479289941,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001935442813999897,
      "loss": 0.0523,
      "step": 4128
    },
    {
      "epoch": 1.0182445759368837,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000193532107502341,
      "loss": 0.048,
      "step": 4130
    },
    {
      "epoch": 1.0187376725838264,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00019351992252061525,
      "loss": 0.0467,
      "step": 4132
    },
    {
      "epoch": 1.0192307692307692,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00019350772645625658,
      "loss": 0.0484,
      "step": 4134
    },
    {
      "epoch": 1.0197238658777121,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00019349551931071023,
      "loss": 0.0545,
      "step": 4136
    },
    {
      "epoch": 1.0202169625246549,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00019348330108542286,
      "loss": 0.0468,
      "step": 4138
    },
    {
      "epoch": 1.0207100591715976,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00019347107178184234,
      "loss": 0.0528,
      "step": 4140
    },
    {
      "epoch": 1.0212031558185404,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00019345883140141793,
      "loss": 0.0454,
      "step": 4142
    },
    {
      "epoch": 1.0216962524654831,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00019344657994560023,
      "loss": 0.0472,
      "step": 4144
    },
    {
      "epoch": 1.022189349112426,
      "grad_norm": 0.28515625,
      "learning_rate": 0.000193434317415841,
      "loss": 0.0483,
      "step": 4146
    },
    {
      "epoch": 1.0226824457593688,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00019342204381359355,
      "loss": 0.0449,
      "step": 4148
    },
    {
      "epoch": 1.0231755424063116,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001934097591403123,
      "loss": 0.0551,
      "step": 4150
    },
    {
      "epoch": 1.0236686390532543,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00019339746339745305,
      "loss": 0.0505,
      "step": 4152
    },
    {
      "epoch": 1.0241617357001973,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00019338515658647292,
      "loss": 0.0474,
      "step": 4154
    },
    {
      "epoch": 1.02465483234714,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001933728387088303,
      "loss": 0.0471,
      "step": 4156
    },
    {
      "epoch": 1.0251479289940828,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00019336050976598502,
      "loss": 0.0469,
      "step": 4158
    },
    {
      "epoch": 1.0256410256410255,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00019334816975939803,
      "loss": 0.0447,
      "step": 4160
    },
    {
      "epoch": 1.0261341222879685,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019333581869053178,
      "loss": 0.0494,
      "step": 4162
    },
    {
      "epoch": 1.0266272189349113,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00019332345656084988,
      "loss": 0.0477,
      "step": 4164
    },
    {
      "epoch": 1.027120315581854,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001933110833718173,
      "loss": 0.043,
      "step": 4166
    },
    {
      "epoch": 1.0276134122287968,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001932986991249004,
      "loss": 0.051,
      "step": 4168
    },
    {
      "epoch": 1.0281065088757397,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001932863038215667,
      "loss": 0.0443,
      "step": 4170
    },
    {
      "epoch": 1.0285996055226825,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019327389746328518,
      "loss": 0.0553,
      "step": 4172
    },
    {
      "epoch": 1.0290927021696252,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00019326148005152606,
      "loss": 0.0435,
      "step": 4174
    },
    {
      "epoch": 1.029585798816568,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019324905158776078,
      "loss": 0.0429,
      "step": 4176
    },
    {
      "epoch": 1.030078895463511,
      "grad_norm": 0.25390625,
      "learning_rate": 0.00019323661207346229,
      "loss": 0.0455,
      "step": 4178
    },
    {
      "epoch": 1.0305719921104537,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0001932241615101047,
      "loss": 0.049,
      "step": 4180
    },
    {
      "epoch": 1.0310650887573964,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001932116998991635,
      "loss": 0.0501,
      "step": 4182
    },
    {
      "epoch": 1.0315581854043392,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001931992272421154,
      "loss": 0.0428,
      "step": 4184
    },
    {
      "epoch": 1.032051282051282,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001931867435404385,
      "loss": 0.046,
      "step": 4186
    },
    {
      "epoch": 1.032544378698225,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00019317424879561224,
      "loss": 0.0476,
      "step": 4188
    },
    {
      "epoch": 1.0330374753451677,
      "grad_norm": 0.5,
      "learning_rate": 0.00019316174300911727,
      "loss": 0.0447,
      "step": 4190
    },
    {
      "epoch": 1.0335305719921104,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001931492261824356,
      "loss": 0.0467,
      "step": 4192
    },
    {
      "epoch": 1.0340236686390532,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00019313669831705054,
      "loss": 0.0454,
      "step": 4194
    },
    {
      "epoch": 1.0345167652859961,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001931241594144467,
      "loss": 0.0495,
      "step": 4196
    },
    {
      "epoch": 1.0350098619329389,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00019311160947611006,
      "loss": 0.0468,
      "step": 4198
    },
    {
      "epoch": 1.0355029585798816,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00019309904850352784,
      "loss": 0.0446,
      "step": 4200
    },
    {
      "epoch": 1.0359960552268244,
      "grad_norm": 0.28125,
      "learning_rate": 0.00019308647649818856,
      "loss": 0.0459,
      "step": 4202
    },
    {
      "epoch": 1.0364891518737673,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0001930738934615821,
      "loss": 0.0449,
      "step": 4204
    },
    {
      "epoch": 1.03698224852071,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0001930612993951996,
      "loss": 0.0498,
      "step": 4206
    },
    {
      "epoch": 1.0374753451676528,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00019304869430053351,
      "loss": 0.0436,
      "step": 4208
    },
    {
      "epoch": 1.0379684418145956,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00019303607817907766,
      "loss": 0.0392,
      "step": 4210
    },
    {
      "epoch": 1.0384615384615385,
      "grad_norm": 0.26953125,
      "learning_rate": 0.00019302345103232709,
      "loss": 0.0441,
      "step": 4212
    },
    {
      "epoch": 1.0389546351084813,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00019301081286177818,
      "loss": 0.0433,
      "step": 4214
    },
    {
      "epoch": 1.039447731755424,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00019299816366892864,
      "loss": 0.0406,
      "step": 4216
    },
    {
      "epoch": 1.0399408284023668,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00019298550345527748,
      "loss": 0.0501,
      "step": 4218
    },
    {
      "epoch": 1.0404339250493098,
      "grad_norm": 0.349609375,
      "learning_rate": 0.000192972832222325,
      "loss": 0.0427,
      "step": 4220
    },
    {
      "epoch": 1.0409270216962525,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001929601499715728,
      "loss": 0.0413,
      "step": 4222
    },
    {
      "epoch": 1.0414201183431953,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00019294745670452384,
      "loss": 0.0473,
      "step": 4224
    },
    {
      "epoch": 1.041913214990138,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00019293475242268223,
      "loss": 0.043,
      "step": 4226
    },
    {
      "epoch": 1.042406311637081,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00019292203712755366,
      "loss": 0.0456,
      "step": 4228
    },
    {
      "epoch": 1.0428994082840237,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00019290931082064484,
      "loss": 0.0435,
      "step": 4230
    },
    {
      "epoch": 1.0433925049309665,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00019289657350346395,
      "loss": 0.0489,
      "step": 4232
    },
    {
      "epoch": 1.0438856015779092,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019288382517752045,
      "loss": 0.0459,
      "step": 4234
    },
    {
      "epoch": 1.044378698224852,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00019287106584432503,
      "loss": 0.0447,
      "step": 4236
    },
    {
      "epoch": 1.044871794871795,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00019285829550538984,
      "loss": 0.0474,
      "step": 4238
    },
    {
      "epoch": 1.0453648915187377,
      "grad_norm": 0.328125,
      "learning_rate": 0.00019284551416222812,
      "loss": 0.0466,
      "step": 4240
    },
    {
      "epoch": 1.0458579881656804,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001928327218163546,
      "loss": 0.046,
      "step": 4242
    },
    {
      "epoch": 1.0463510848126232,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019281991846928527,
      "loss": 0.0458,
      "step": 4244
    },
    {
      "epoch": 1.0468441814595661,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00019280710412253733,
      "loss": 0.0468,
      "step": 4246
    },
    {
      "epoch": 1.047337278106509,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019279427877762938,
      "loss": 0.0429,
      "step": 4248
    },
    {
      "epoch": 1.0478303747534516,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001927814424360813,
      "loss": 0.0416,
      "step": 4250
    },
    {
      "epoch": 1.0483234714003944,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001927685950994143,
      "loss": 0.048,
      "step": 4252
    },
    {
      "epoch": 1.0488165680473374,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0001927557367691508,
      "loss": 0.0389,
      "step": 4254
    },
    {
      "epoch": 1.04930966469428,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00019274286744681463,
      "loss": 0.0434,
      "step": 4256
    },
    {
      "epoch": 1.0498027613412229,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00019272998713393082,
      "loss": 0.0471,
      "step": 4258
    },
    {
      "epoch": 1.0502958579881656,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00019271709583202582,
      "loss": 0.0438,
      "step": 4260
    },
    {
      "epoch": 1.0507889546351086,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001927041935426273,
      "loss": 0.0451,
      "step": 4262
    },
    {
      "epoch": 1.0512820512820513,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00019269128026726425,
      "loss": 0.0452,
      "step": 4264
    },
    {
      "epoch": 1.051775147928994,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019267835600746697,
      "loss": 0.0446,
      "step": 4266
    },
    {
      "epoch": 1.0522682445759368,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019266542076476704,
      "loss": 0.0444,
      "step": 4268
    },
    {
      "epoch": 1.0527613412228798,
      "grad_norm": 0.5,
      "learning_rate": 0.00019265247454069737,
      "loss": 0.0477,
      "step": 4270
    },
    {
      "epoch": 1.0532544378698225,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00019263951733679218,
      "loss": 0.0452,
      "step": 4272
    },
    {
      "epoch": 1.0537475345167653,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00019262654915458693,
      "loss": 0.0415,
      "step": 4274
    },
    {
      "epoch": 1.054240631163708,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00019261356999561845,
      "loss": 0.0447,
      "step": 4276
    },
    {
      "epoch": 1.054733727810651,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00019260057986142485,
      "loss": 0.0472,
      "step": 4278
    },
    {
      "epoch": 1.0552268244575937,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019258757875354553,
      "loss": 0.0403,
      "step": 4280
    },
    {
      "epoch": 1.0557199211045365,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00019257456667352117,
      "loss": 0.0434,
      "step": 4282
    },
    {
      "epoch": 1.0562130177514792,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001925615436228938,
      "loss": 0.0478,
      "step": 4284
    },
    {
      "epoch": 1.056706114398422,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001925485096032067,
      "loss": 0.0473,
      "step": 4286
    },
    {
      "epoch": 1.057199211045365,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0001925354646160045,
      "loss": 0.0448,
      "step": 4288
    },
    {
      "epoch": 1.0576923076923077,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00019252240866283308,
      "loss": 0.0442,
      "step": 4290
    },
    {
      "epoch": 1.0581854043392505,
      "grad_norm": 0.3125,
      "learning_rate": 0.00019250934174523966,
      "loss": 0.0408,
      "step": 4292
    },
    {
      "epoch": 1.0586785009861932,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00019249626386477273,
      "loss": 0.0417,
      "step": 4294
    },
    {
      "epoch": 1.0591715976331362,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0001924831750229821,
      "loss": 0.041,
      "step": 4296
    },
    {
      "epoch": 1.059664694280079,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00019247007522141886,
      "loss": 0.0433,
      "step": 4298
    },
    {
      "epoch": 1.0601577909270217,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00019245696446163545,
      "loss": 0.0458,
      "step": 4300
    },
    {
      "epoch": 1.0606508875739644,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00019244384274518553,
      "loss": 0.0382,
      "step": 4302
    },
    {
      "epoch": 1.0611439842209074,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00019243071007362408,
      "loss": 0.0476,
      "step": 4304
    },
    {
      "epoch": 1.0616370808678501,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00019241756644850744,
      "loss": 0.0434,
      "step": 4306
    },
    {
      "epoch": 1.0621301775147929,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00019240441187139316,
      "loss": 0.0475,
      "step": 4308
    },
    {
      "epoch": 1.0626232741617356,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001923912463438402,
      "loss": 0.0439,
      "step": 4310
    },
    {
      "epoch": 1.0631163708086786,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00019237806986740865,
      "loss": 0.0408,
      "step": 4312
    },
    {
      "epoch": 1.0636094674556213,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00019236488244366007,
      "loss": 0.0418,
      "step": 4314
    },
    {
      "epoch": 1.064102564102564,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001923516840741572,
      "loss": 0.0449,
      "step": 4316
    },
    {
      "epoch": 1.0645956607495068,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001923384747604642,
      "loss": 0.0396,
      "step": 4318
    },
    {
      "epoch": 1.0650887573964498,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019232525450414633,
      "loss": 0.046,
      "step": 4320
    },
    {
      "epoch": 1.0655818540433926,
      "grad_norm": 0.3125,
      "learning_rate": 0.00019231202330677034,
      "loss": 0.0442,
      "step": 4322
    },
    {
      "epoch": 1.0660749506903353,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001922987811699042,
      "loss": 0.0382,
      "step": 4324
    },
    {
      "epoch": 1.066568047337278,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00019228552809511714,
      "loss": 0.0422,
      "step": 4326
    },
    {
      "epoch": 1.0670611439842208,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00019227226408397976,
      "loss": 0.046,
      "step": 4328
    },
    {
      "epoch": 1.0675542406311638,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00019225898913806388,
      "loss": 0.0407,
      "step": 4330
    },
    {
      "epoch": 1.0680473372781065,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00019224570325894268,
      "loss": 0.0445,
      "step": 4332
    },
    {
      "epoch": 1.0685404339250493,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00019223240644819061,
      "loss": 0.0487,
      "step": 4334
    },
    {
      "epoch": 1.069033530571992,
      "grad_norm": 0.3125,
      "learning_rate": 0.00019221909870738345,
      "loss": 0.0414,
      "step": 4336
    },
    {
      "epoch": 1.069526627218935,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0001922057800380982,
      "loss": 0.0416,
      "step": 4338
    },
    {
      "epoch": 1.0700197238658777,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00019219245044191317,
      "loss": 0.0403,
      "step": 4340
    },
    {
      "epoch": 1.0705128205128205,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00019217910992040807,
      "loss": 0.0413,
      "step": 4342
    },
    {
      "epoch": 1.0710059171597632,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00019216575847516375,
      "loss": 0.0414,
      "step": 4344
    },
    {
      "epoch": 1.0714990138067062,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001921523961077625,
      "loss": 0.0454,
      "step": 4346
    },
    {
      "epoch": 1.071992110453649,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00019213902281978775,
      "loss": 0.043,
      "step": 4348
    },
    {
      "epoch": 1.0724852071005917,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00019212563861282437,
      "loss": 0.0432,
      "step": 4350
    },
    {
      "epoch": 1.0729783037475344,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00019211224348845846,
      "loss": 0.0472,
      "step": 4352
    },
    {
      "epoch": 1.0734714003944774,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00019209883744827742,
      "loss": 0.0427,
      "step": 4354
    },
    {
      "epoch": 1.0739644970414202,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0001920854204938699,
      "loss": 0.0431,
      "step": 4356
    },
    {
      "epoch": 1.074457593688363,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00019207199262682593,
      "loss": 0.0433,
      "step": 4358
    },
    {
      "epoch": 1.0749506903353057,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00019205855384873675,
      "loss": 0.044,
      "step": 4360
    },
    {
      "epoch": 1.0754437869822486,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00019204510416119498,
      "loss": 0.0374,
      "step": 4362
    },
    {
      "epoch": 1.0759368836291914,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019203164356579447,
      "loss": 0.0442,
      "step": 4364
    },
    {
      "epoch": 1.0764299802761341,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001920181720641303,
      "loss": 0.0448,
      "step": 4366
    },
    {
      "epoch": 1.0769230769230769,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00019200468965779904,
      "loss": 0.0419,
      "step": 4368
    },
    {
      "epoch": 1.0774161735700196,
      "grad_norm": 0.328125,
      "learning_rate": 0.00019199119634839831,
      "loss": 0.0461,
      "step": 4370
    },
    {
      "epoch": 1.0779092702169626,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00019197769213752722,
      "loss": 0.0433,
      "step": 4372
    },
    {
      "epoch": 1.0784023668639053,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00019196417702678606,
      "loss": 0.0449,
      "step": 4374
    },
    {
      "epoch": 1.078895463510848,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001919506510177765,
      "loss": 0.0424,
      "step": 4376
    },
    {
      "epoch": 1.0793885601577908,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00019193711411210138,
      "loss": 0.0434,
      "step": 4378
    },
    {
      "epoch": 1.0798816568047338,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00019192356631136492,
      "loss": 0.0441,
      "step": 4380
    },
    {
      "epoch": 1.0803747534516766,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00019191000761717265,
      "loss": 0.0426,
      "step": 4382
    },
    {
      "epoch": 1.0808678500986193,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00019189643803113127,
      "loss": 0.0387,
      "step": 4384
    },
    {
      "epoch": 1.081360946745562,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00019188285755484893,
      "loss": 0.0418,
      "step": 4386
    },
    {
      "epoch": 1.081854043392505,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00019186926618993498,
      "loss": 0.046,
      "step": 4388
    },
    {
      "epoch": 1.0823471400394478,
      "grad_norm": 0.3046875,
      "learning_rate": 0.000191855663938,
      "loss": 0.0436,
      "step": 4390
    },
    {
      "epoch": 1.0828402366863905,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00019184205080065604,
      "loss": 0.0424,
      "step": 4392
    },
    {
      "epoch": 1.0833333333333333,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00019182842677951625,
      "loss": 0.0408,
      "step": 4394
    },
    {
      "epoch": 1.0838264299802762,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00019181479187619518,
      "loss": 0.0475,
      "step": 4396
    },
    {
      "epoch": 1.084319526627219,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00019180114609230865,
      "loss": 0.043,
      "step": 4398
    },
    {
      "epoch": 1.0848126232741617,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00019178748942947377,
      "loss": 0.044,
      "step": 4400
    },
    {
      "epoch": 1.0853057199211045,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001917738218893089,
      "loss": 0.0458,
      "step": 4402
    },
    {
      "epoch": 1.0857988165680474,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00019176014347343373,
      "loss": 0.0433,
      "step": 4404
    },
    {
      "epoch": 1.0862919132149902,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00019174645418346925,
      "loss": 0.0448,
      "step": 4406
    },
    {
      "epoch": 1.086785009861933,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019173275402103773,
      "loss": 0.0421,
      "step": 4408
    },
    {
      "epoch": 1.0872781065088757,
      "grad_norm": 0.265625,
      "learning_rate": 0.00019171904298776265,
      "loss": 0.0387,
      "step": 4410
    },
    {
      "epoch": 1.0877712031558184,
      "grad_norm": 0.296875,
      "learning_rate": 0.00019170532108526888,
      "loss": 0.0441,
      "step": 4412
    },
    {
      "epoch": 1.0882642998027614,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00019169158831518256,
      "loss": 0.0385,
      "step": 4414
    },
    {
      "epoch": 1.0887573964497042,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00019167784467913107,
      "loss": 0.0457,
      "step": 4416
    },
    {
      "epoch": 1.089250493096647,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00019166409017874313,
      "loss": 0.0468,
      "step": 4418
    },
    {
      "epoch": 1.0897435897435896,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00019165032481564874,
      "loss": 0.042,
      "step": 4420
    },
    {
      "epoch": 1.0902366863905326,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0001916365485914791,
      "loss": 0.0403,
      "step": 4422
    },
    {
      "epoch": 1.0907297830374754,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00019162276150786687,
      "loss": 0.0438,
      "step": 4424
    },
    {
      "epoch": 1.0912228796844181,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019160896356644583,
      "loss": 0.0434,
      "step": 4426
    },
    {
      "epoch": 1.0917159763313609,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019159515476885113,
      "loss": 0.0385,
      "step": 4428
    },
    {
      "epoch": 1.0922090729783038,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00019158133511671917,
      "loss": 0.04,
      "step": 4430
    },
    {
      "epoch": 1.0927021696252466,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019156750461168768,
      "loss": 0.0415,
      "step": 4432
    },
    {
      "epoch": 1.0931952662721893,
      "grad_norm": 0.34375,
      "learning_rate": 0.00019155366325539563,
      "loss": 0.043,
      "step": 4434
    },
    {
      "epoch": 1.093688362919132,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019153981104948336,
      "loss": 0.043,
      "step": 4436
    },
    {
      "epoch": 1.094181459566075,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00019152594799559233,
      "loss": 0.0436,
      "step": 4438
    },
    {
      "epoch": 1.0946745562130178,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0001915120740953655,
      "loss": 0.0448,
      "step": 4440
    },
    {
      "epoch": 1.0951676528599605,
      "grad_norm": 0.5,
      "learning_rate": 0.00019149818935044693,
      "loss": 0.042,
      "step": 4442
    },
    {
      "epoch": 1.0956607495069033,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019148429376248207,
      "loss": 0.0416,
      "step": 4444
    },
    {
      "epoch": 1.0961538461538463,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0001914703873331176,
      "loss": 0.0445,
      "step": 4446
    },
    {
      "epoch": 1.096646942800789,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00019145647006400154,
      "loss": 0.0515,
      "step": 4448
    },
    {
      "epoch": 1.0971400394477318,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00019144254195678314,
      "loss": 0.0417,
      "step": 4450
    },
    {
      "epoch": 1.0976331360946745,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000191428603013113,
      "loss": 0.0422,
      "step": 4452
    },
    {
      "epoch": 1.0981262327416172,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00019141465323464286,
      "loss": 0.0492,
      "step": 4454
    },
    {
      "epoch": 1.0986193293885602,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00019140069262302598,
      "loss": 0.0487,
      "step": 4456
    },
    {
      "epoch": 1.099112426035503,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001913867211799167,
      "loss": 0.0397,
      "step": 4458
    },
    {
      "epoch": 1.0996055226824457,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0001913727389069707,
      "loss": 0.0376,
      "step": 4460
    },
    {
      "epoch": 1.1000986193293885,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00019135874580584499,
      "loss": 0.0481,
      "step": 4462
    },
    {
      "epoch": 1.1005917159763314,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019134474187819782,
      "loss": 0.0448,
      "step": 4464
    },
    {
      "epoch": 1.1010848126232742,
      "grad_norm": 0.328125,
      "learning_rate": 0.00019133072712568873,
      "loss": 0.039,
      "step": 4466
    },
    {
      "epoch": 1.101577909270217,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00019131670154997856,
      "loss": 0.039,
      "step": 4468
    },
    {
      "epoch": 1.1020710059171597,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00019130266515272938,
      "loss": 0.042,
      "step": 4470
    },
    {
      "epoch": 1.1025641025641026,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00019128861793560464,
      "loss": 0.0525,
      "step": 4472
    },
    {
      "epoch": 1.1030571992110454,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00019127455990026898,
      "loss": 0.0426,
      "step": 4474
    },
    {
      "epoch": 1.1035502958579881,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00019126049104838837,
      "loss": 0.04,
      "step": 4476
    },
    {
      "epoch": 1.1040433925049309,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00019124641138163002,
      "loss": 0.0548,
      "step": 4478
    },
    {
      "epoch": 1.1045364891518739,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00019123232090166248,
      "loss": 0.0555,
      "step": 4480
    },
    {
      "epoch": 1.1050295857988166,
      "grad_norm": 0.40625,
      "learning_rate": 0.00019121821961015553,
      "loss": 0.0422,
      "step": 4482
    },
    {
      "epoch": 1.1055226824457594,
      "grad_norm": 0.40625,
      "learning_rate": 0.00019120410750878027,
      "loss": 0.0372,
      "step": 4484
    },
    {
      "epoch": 1.106015779092702,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019118998459920902,
      "loss": 0.0516,
      "step": 4486
    },
    {
      "epoch": 1.106508875739645,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019117585088311548,
      "loss": 0.0574,
      "step": 4488
    },
    {
      "epoch": 1.1070019723865878,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00019116170636217453,
      "loss": 0.0427,
      "step": 4490
    },
    {
      "epoch": 1.1074950690335306,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00019114755103806242,
      "loss": 0.0416,
      "step": 4492
    },
    {
      "epoch": 1.1079881656804733,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00019113338491245658,
      "loss": 0.0554,
      "step": 4494
    },
    {
      "epoch": 1.108481262327416,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00019111920798703584,
      "loss": 0.0547,
      "step": 4496
    },
    {
      "epoch": 1.108974358974359,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001911050202634802,
      "loss": 0.0446,
      "step": 4498
    },
    {
      "epoch": 1.1094674556213018,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00019109082174347097,
      "loss": 0.0465,
      "step": 4500
    },
    {
      "epoch": 1.1099605522682445,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019107661242869078,
      "loss": 0.0568,
      "step": 4502
    },
    {
      "epoch": 1.1104536489151873,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019106239232082354,
      "loss": 0.0678,
      "step": 4504
    },
    {
      "epoch": 1.1109467455621302,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00019104816142155438,
      "loss": 0.0456,
      "step": 4506
    },
    {
      "epoch": 1.111439842209073,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001910339197325697,
      "loss": 0.0479,
      "step": 4508
    },
    {
      "epoch": 1.1119329388560157,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019101966725555732,
      "loss": 0.0581,
      "step": 4510
    },
    {
      "epoch": 1.1124260355029585,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019100540399220615,
      "loss": 0.0602,
      "step": 4512
    },
    {
      "epoch": 1.1129191321499015,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0001909911299442065,
      "loss": 0.0435,
      "step": 4514
    },
    {
      "epoch": 1.1134122287968442,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00019097684511324993,
      "loss": 0.0421,
      "step": 4516
    },
    {
      "epoch": 1.113905325443787,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00019096254950102927,
      "loss": 0.0583,
      "step": 4518
    },
    {
      "epoch": 1.1143984220907297,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001909482431092386,
      "loss": 0.0575,
      "step": 4520
    },
    {
      "epoch": 1.1148915187376727,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019093392593957334,
      "loss": 0.0454,
      "step": 4522
    },
    {
      "epoch": 1.1153846153846154,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00019091959799373014,
      "loss": 0.0446,
      "step": 4524
    },
    {
      "epoch": 1.1158777120315582,
      "grad_norm": 0.375,
      "learning_rate": 0.00019090525927340695,
      "loss": 0.0555,
      "step": 4526
    },
    {
      "epoch": 1.116370808678501,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00019089090978030302,
      "loss": 0.0619,
      "step": 4528
    },
    {
      "epoch": 1.1168639053254439,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019087654951611876,
      "loss": 0.04,
      "step": 4530
    },
    {
      "epoch": 1.1173570019723866,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00019086217848255602,
      "loss": 0.0546,
      "step": 4532
    },
    {
      "epoch": 1.1178500986193294,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00019084779668131782,
      "loss": 0.0586,
      "step": 4534
    },
    {
      "epoch": 1.1183431952662721,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00019083340411410848,
      "loss": 0.0548,
      "step": 4536
    },
    {
      "epoch": 1.1188362919132149,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001908190007826336,
      "loss": 0.0587,
      "step": 4538
    },
    {
      "epoch": 1.1193293885601578,
      "grad_norm": 0.3125,
      "learning_rate": 0.00019080458668860008,
      "loss": 0.0457,
      "step": 4540
    },
    {
      "epoch": 1.1198224852071006,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00019079016183371603,
      "loss": 0.059,
      "step": 4542
    },
    {
      "epoch": 1.1203155818540433,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001907757262196909,
      "loss": 0.0568,
      "step": 4544
    },
    {
      "epoch": 1.1208086785009863,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0001907612798482354,
      "loss": 0.0469,
      "step": 4546
    },
    {
      "epoch": 1.121301775147929,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00019074682272106146,
      "loss": 0.0503,
      "step": 4548
    },
    {
      "epoch": 1.1217948717948718,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001907323548398824,
      "loss": 0.0612,
      "step": 4550
    },
    {
      "epoch": 1.1222879684418146,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019071787620641265,
      "loss": 0.0612,
      "step": 4552
    },
    {
      "epoch": 1.1227810650887573,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00019070338682236813,
      "loss": 0.055,
      "step": 4554
    },
    {
      "epoch": 1.1232741617357003,
      "grad_norm": 0.375,
      "learning_rate": 0.00019068888668946583,
      "loss": 0.0516,
      "step": 4556
    },
    {
      "epoch": 1.123767258382643,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019067437580942408,
      "loss": 0.061,
      "step": 4558
    },
    {
      "epoch": 1.1242603550295858,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019065985418396258,
      "loss": 0.0594,
      "step": 4560
    },
    {
      "epoch": 1.1247534516765285,
      "grad_norm": 0.375,
      "learning_rate": 0.00019064532181480216,
      "loss": 0.0588,
      "step": 4562
    },
    {
      "epoch": 1.1252465483234715,
      "grad_norm": 0.5625,
      "learning_rate": 0.000190630778703665,
      "loss": 0.0541,
      "step": 4564
    },
    {
      "epoch": 1.1257396449704142,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00019061622485227458,
      "loss": 0.0606,
      "step": 4566
    },
    {
      "epoch": 1.126232741617357,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00019060166026235552,
      "loss": 0.0621,
      "step": 4568
    },
    {
      "epoch": 1.1267258382642997,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00019058708493563392,
      "loss": 0.0608,
      "step": 4570
    },
    {
      "epoch": 1.1272189349112427,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00019057249887383695,
      "loss": 0.0631,
      "step": 4572
    },
    {
      "epoch": 1.1277120315581854,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019055790207869318,
      "loss": 0.0638,
      "step": 4574
    },
    {
      "epoch": 1.1282051282051282,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0001905432945519324,
      "loss": 0.0711,
      "step": 4576
    },
    {
      "epoch": 1.128698224852071,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019052867629528567,
      "loss": 0.0579,
      "step": 4578
    },
    {
      "epoch": 1.1291913214990137,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001905140473104854,
      "loss": 0.0574,
      "step": 4580
    },
    {
      "epoch": 1.1296844181459567,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001904994075992651,
      "loss": 0.0592,
      "step": 4582
    },
    {
      "epoch": 1.1301775147928994,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00019048475716335975,
      "loss": 0.0643,
      "step": 4584
    },
    {
      "epoch": 1.1306706114398422,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00019047009600450546,
      "loss": 0.0562,
      "step": 4586
    },
    {
      "epoch": 1.1311637080867851,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019045542412443972,
      "loss": 0.051,
      "step": 4588
    },
    {
      "epoch": 1.1316568047337279,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019044074152490113,
      "loss": 0.0605,
      "step": 4590
    },
    {
      "epoch": 1.1321499013806706,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019042604820762975,
      "loss": 0.0672,
      "step": 4592
    },
    {
      "epoch": 1.1326429980276134,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019041134417436674,
      "loss": 0.059,
      "step": 4594
    },
    {
      "epoch": 1.1331360946745561,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019039662942685467,
      "loss": 0.0588,
      "step": 4596
    },
    {
      "epoch": 1.133629191321499,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001903819039668373,
      "loss": 0.0636,
      "step": 4598
    },
    {
      "epoch": 1.1341222879684418,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001903671677960597,
      "loss": 0.0673,
      "step": 4600
    },
    {
      "epoch": 1.1346153846153846,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019035242091626814,
      "loss": 0.0736,
      "step": 4602
    },
    {
      "epoch": 1.1351084812623273,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019033766332921028,
      "loss": 0.064,
      "step": 4604
    },
    {
      "epoch": 1.1356015779092703,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001903228950366349,
      "loss": 0.066,
      "step": 4606
    },
    {
      "epoch": 1.136094674556213,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019030811604029218,
      "loss": 0.065,
      "step": 4608
    },
    {
      "epoch": 1.1365877712031558,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001902933263419335,
      "loss": 0.0618,
      "step": 4610
    },
    {
      "epoch": 1.1370808678500985,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00019027852594331147,
      "loss": 0.06,
      "step": 4612
    },
    {
      "epoch": 1.1375739644970415,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001902637148461801,
      "loss": 0.0709,
      "step": 4614
    },
    {
      "epoch": 1.1380670611439843,
      "grad_norm": 0.421875,
      "learning_rate": 0.00019024889305229456,
      "loss": 0.0653,
      "step": 4616
    },
    {
      "epoch": 1.138560157790927,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019023406056341128,
      "loss": 0.0576,
      "step": 4618
    },
    {
      "epoch": 1.1390532544378698,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00019021921738128806,
      "loss": 0.0599,
      "step": 4620
    },
    {
      "epoch": 1.1395463510848125,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019020436350768383,
      "loss": 0.0652,
      "step": 4622
    },
    {
      "epoch": 1.1400394477317555,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001901894989443589,
      "loss": 0.0719,
      "step": 4624
    },
    {
      "epoch": 1.1405325443786982,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00019017462369307476,
      "loss": 0.0631,
      "step": 4626
    },
    {
      "epoch": 1.141025641025641,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001901597377555943,
      "loss": 0.0677,
      "step": 4628
    },
    {
      "epoch": 1.141518737672584,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019014484113368148,
      "loss": 0.0802,
      "step": 4630
    },
    {
      "epoch": 1.1420118343195267,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001901299338291017,
      "loss": 0.0675,
      "step": 4632
    },
    {
      "epoch": 1.1425049309664694,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019011501584362153,
      "loss": 0.067,
      "step": 4634
    },
    {
      "epoch": 1.1429980276134122,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019010008717900884,
      "loss": 0.0705,
      "step": 4636
    },
    {
      "epoch": 1.143491124260355,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019008514783703278,
      "loss": 0.0717,
      "step": 4638
    },
    {
      "epoch": 1.143984220907298,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019007019781946375,
      "loss": 0.0727,
      "step": 4640
    },
    {
      "epoch": 1.1444773175542406,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00019005523712807337,
      "loss": 0.0636,
      "step": 4642
    },
    {
      "epoch": 1.1449704142011834,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019004026576463457,
      "loss": 0.0594,
      "step": 4644
    },
    {
      "epoch": 1.1454635108481261,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001900252837309216,
      "loss": 0.0692,
      "step": 4646
    },
    {
      "epoch": 1.1459566074950691,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019001029102870982,
      "loss": 0.0664,
      "step": 4648
    },
    {
      "epoch": 1.1464497041420119,
      "grad_norm": 0.5625,
      "learning_rate": 0.000189995287659776,
      "loss": 0.0637,
      "step": 4650
    },
    {
      "epoch": 1.1469428007889546,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00018998027362589816,
      "loss": 0.0608,
      "step": 4652
    },
    {
      "epoch": 1.1474358974358974,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001899652489288555,
      "loss": 0.0697,
      "step": 4654
    },
    {
      "epoch": 1.1479289940828403,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00018995021357042852,
      "loss": 0.0698,
      "step": 4656
    },
    {
      "epoch": 1.148422090729783,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00018993516755239903,
      "loss": 0.072,
      "step": 4658
    },
    {
      "epoch": 1.1489151873767258,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018992011087655007,
      "loss": 0.0687,
      "step": 4660
    },
    {
      "epoch": 1.1494082840236686,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001899050435446659,
      "loss": 0.071,
      "step": 4662
    },
    {
      "epoch": 1.1499013806706113,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018988996555853208,
      "loss": 0.0766,
      "step": 4664
    },
    {
      "epoch": 1.1503944773175543,
      "grad_norm": 0.640625,
      "learning_rate": 0.00018987487691993552,
      "loss": 0.067,
      "step": 4666
    },
    {
      "epoch": 1.150887573964497,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0001898597776306642,
      "loss": 0.0649,
      "step": 4668
    },
    {
      "epoch": 1.1513806706114398,
      "grad_norm": 0.625,
      "learning_rate": 0.00018984466769250753,
      "loss": 0.0708,
      "step": 4670
    },
    {
      "epoch": 1.1518737672583828,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00018982954710725612,
      "loss": 0.0659,
      "step": 4672
    },
    {
      "epoch": 1.1523668639053255,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018981441587670182,
      "loss": 0.069,
      "step": 4674
    },
    {
      "epoch": 1.1528599605522682,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001897992740026378,
      "loss": 0.0639,
      "step": 4676
    },
    {
      "epoch": 1.153353057199211,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001897841214868584,
      "loss": 0.0703,
      "step": 4678
    },
    {
      "epoch": 1.1538461538461537,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00018976895833115935,
      "loss": 0.0734,
      "step": 4680
    },
    {
      "epoch": 1.1543392504930967,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018975378453733752,
      "loss": 0.0643,
      "step": 4682
    },
    {
      "epoch": 1.1548323471400395,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018973860010719108,
      "loss": 0.0673,
      "step": 4684
    },
    {
      "epoch": 1.1553254437869822,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001897234050425195,
      "loss": 0.0673,
      "step": 4686
    },
    {
      "epoch": 1.155818540433925,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001897081993451235,
      "loss": 0.0781,
      "step": 4688
    },
    {
      "epoch": 1.156311637080868,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00018969298301680499,
      "loss": 0.0617,
      "step": 4690
    },
    {
      "epoch": 1.1568047337278107,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001896777560593672,
      "loss": 0.0709,
      "step": 4692
    },
    {
      "epoch": 1.1572978303747534,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018966251847461462,
      "loss": 0.0645,
      "step": 4694
    },
    {
      "epoch": 1.1577909270216962,
      "grad_norm": 0.421875,
      "learning_rate": 0.00018964727026435303,
      "loss": 0.0709,
      "step": 4696
    },
    {
      "epoch": 1.1582840236686391,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00018963201143038937,
      "loss": 0.0683,
      "step": 4698
    },
    {
      "epoch": 1.1587771203155819,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00018961674197453187,
      "loss": 0.0708,
      "step": 4700
    },
    {
      "epoch": 1.1592702169625246,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018960146189859014,
      "loss": 0.0782,
      "step": 4702
    },
    {
      "epoch": 1.1597633136094674,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001895861712043749,
      "loss": 0.0732,
      "step": 4704
    },
    {
      "epoch": 1.1602564102564104,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018957086989369816,
      "loss": 0.0656,
      "step": 4706
    },
    {
      "epoch": 1.160749506903353,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00018955555796837324,
      "loss": 0.0712,
      "step": 4708
    },
    {
      "epoch": 1.1612426035502958,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00018954023543021472,
      "loss": 0.0722,
      "step": 4710
    },
    {
      "epoch": 1.1617357001972386,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00018952490228103837,
      "loss": 0.0817,
      "step": 4712
    },
    {
      "epoch": 1.1622287968441816,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001895095585226612,
      "loss": 0.0674,
      "step": 4714
    },
    {
      "epoch": 1.1627218934911243,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00018949420415690166,
      "loss": 0.0661,
      "step": 4716
    },
    {
      "epoch": 1.163214990138067,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001894788391855792,
      "loss": 0.0707,
      "step": 4718
    },
    {
      "epoch": 1.1637080867850098,
      "grad_norm": 0.390625,
      "learning_rate": 0.00018946346361051474,
      "loss": 0.0748,
      "step": 4720
    },
    {
      "epoch": 1.1642011834319526,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00018944807743353034,
      "loss": 0.0676,
      "step": 4722
    },
    {
      "epoch": 1.1646942800788955,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018943268065644935,
      "loss": 0.0647,
      "step": 4724
    },
    {
      "epoch": 1.1651873767258383,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00018941727328109637,
      "loss": 0.0695,
      "step": 4726
    },
    {
      "epoch": 1.165680473372781,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018940185530929725,
      "loss": 0.0728,
      "step": 4728
    },
    {
      "epoch": 1.1661735700197238,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018938642674287914,
      "loss": 0.0646,
      "step": 4730
    },
    {
      "epoch": 1.1666666666666667,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001893709875836704,
      "loss": 0.0707,
      "step": 4732
    },
    {
      "epoch": 1.1671597633136095,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00018935553783350063,
      "loss": 0.0729,
      "step": 4734
    },
    {
      "epoch": 1.1676528599605522,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018934007749420073,
      "loss": 0.0726,
      "step": 4736
    },
    {
      "epoch": 1.168145956607495,
      "grad_norm": 0.5,
      "learning_rate": 0.00018932460656760283,
      "loss": 0.0621,
      "step": 4738
    },
    {
      "epoch": 1.168639053254438,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00018930912505554033,
      "loss": 0.0696,
      "step": 4740
    },
    {
      "epoch": 1.1691321499013807,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001892936329598479,
      "loss": 0.0746,
      "step": 4742
    },
    {
      "epoch": 1.1696252465483234,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018927813028236137,
      "loss": 0.0762,
      "step": 4744
    },
    {
      "epoch": 1.1701183431952662,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018926261702491797,
      "loss": 0.0712,
      "step": 4746
    },
    {
      "epoch": 1.1706114398422092,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00018924709318935605,
      "loss": 0.0744,
      "step": 4748
    },
    {
      "epoch": 1.171104536489152,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001892315587775153,
      "loss": 0.0718,
      "step": 4750
    },
    {
      "epoch": 1.1715976331360947,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018921601379123665,
      "loss": 0.0778,
      "step": 4752
    },
    {
      "epoch": 1.1720907297830374,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018920045823236223,
      "loss": 0.0751,
      "step": 4754
    },
    {
      "epoch": 1.1725838264299804,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001891848921027355,
      "loss": 0.0646,
      "step": 4756
    },
    {
      "epoch": 1.1730769230769231,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018916931540420107,
      "loss": 0.0806,
      "step": 4758
    },
    {
      "epoch": 1.1735700197238659,
      "grad_norm": 0.71875,
      "learning_rate": 0.00018915372813860492,
      "loss": 0.0801,
      "step": 4760
    },
    {
      "epoch": 1.1740631163708086,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018913813030779425,
      "loss": 0.0718,
      "step": 4762
    },
    {
      "epoch": 1.1745562130177514,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00018912252191361742,
      "loss": 0.064,
      "step": 4764
    },
    {
      "epoch": 1.1750493096646943,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018910690295792412,
      "loss": 0.0831,
      "step": 4766
    },
    {
      "epoch": 1.175542406311637,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018909127344256537,
      "loss": 0.0735,
      "step": 4768
    },
    {
      "epoch": 1.1760355029585798,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00018907563336939326,
      "loss": 0.0768,
      "step": 4770
    },
    {
      "epoch": 1.1765285996055228,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018905998274026126,
      "loss": 0.0739,
      "step": 4772
    },
    {
      "epoch": 1.1770216962524656,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00018904432155702406,
      "loss": 0.0755,
      "step": 4774
    },
    {
      "epoch": 1.1775147928994083,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018902864982153764,
      "loss": 0.0777,
      "step": 4776
    },
    {
      "epoch": 1.178007889546351,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001890129675356591,
      "loss": 0.0651,
      "step": 4778
    },
    {
      "epoch": 1.1785009861932938,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00018899727470124693,
      "loss": 0.0659,
      "step": 4780
    },
    {
      "epoch": 1.1789940828402368,
      "grad_norm": 0.484375,
      "learning_rate": 0.00018898157132016082,
      "loss": 0.0674,
      "step": 4782
    },
    {
      "epoch": 1.1794871794871795,
      "grad_norm": 0.59375,
      "learning_rate": 0.00018896585739426169,
      "loss": 0.0723,
      "step": 4784
    },
    {
      "epoch": 1.1799802761341223,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00018895013292541174,
      "loss": 0.0694,
      "step": 4786
    },
    {
      "epoch": 1.180473372781065,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001889343979154744,
      "loss": 0.0726,
      "step": 4788
    },
    {
      "epoch": 1.180966469428008,
      "grad_norm": 0.40625,
      "learning_rate": 0.00018891865236631437,
      "loss": 0.0713,
      "step": 4790
    },
    {
      "epoch": 1.1814595660749507,
      "grad_norm": 0.796875,
      "learning_rate": 0.00018890289627979755,
      "loss": 0.0701,
      "step": 4792
    },
    {
      "epoch": 1.1819526627218935,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018888712965779117,
      "loss": 0.0671,
      "step": 4794
    },
    {
      "epoch": 1.1824457593688362,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018887135250216365,
      "loss": 0.0671,
      "step": 4796
    },
    {
      "epoch": 1.1829388560157792,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018885556481478466,
      "loss": 0.0755,
      "step": 4798
    },
    {
      "epoch": 1.183431952662722,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018883976659752512,
      "loss": 0.0732,
      "step": 4800
    },
    {
      "epoch": 1.1839250493096647,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018882395785225725,
      "loss": 0.0666,
      "step": 4802
    },
    {
      "epoch": 1.1844181459566074,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001888081385808544,
      "loss": 0.0721,
      "step": 4804
    },
    {
      "epoch": 1.1849112426035502,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001887923087851913,
      "loss": 0.079,
      "step": 4806
    },
    {
      "epoch": 1.1854043392504932,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00018877646846714388,
      "loss": 0.0718,
      "step": 4808
    },
    {
      "epoch": 1.185897435897436,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001887606176285893,
      "loss": 0.078,
      "step": 4810
    },
    {
      "epoch": 1.1863905325443787,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001887447562714059,
      "loss": 0.0703,
      "step": 4812
    },
    {
      "epoch": 1.1868836291913216,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018872888439747345,
      "loss": 0.0738,
      "step": 4814
    },
    {
      "epoch": 1.1873767258382644,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018871300200867274,
      "loss": 0.0784,
      "step": 4816
    },
    {
      "epoch": 1.1878698224852071,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018869710910688603,
      "loss": 0.0754,
      "step": 4818
    },
    {
      "epoch": 1.1883629191321499,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00018868120569399664,
      "loss": 0.0677,
      "step": 4820
    },
    {
      "epoch": 1.1888560157790926,
      "grad_norm": 0.640625,
      "learning_rate": 0.00018866529177188928,
      "loss": 0.0749,
      "step": 4822
    },
    {
      "epoch": 1.1893491124260356,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018864936734244977,
      "loss": 0.0796,
      "step": 4824
    },
    {
      "epoch": 1.1898422090729783,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00018863343240756527,
      "loss": 0.0821,
      "step": 4826
    },
    {
      "epoch": 1.190335305719921,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018861748696912418,
      "loss": 0.0697,
      "step": 4828
    },
    {
      "epoch": 1.1908284023668638,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001886015310290161,
      "loss": 0.0794,
      "step": 4830
    },
    {
      "epoch": 1.1913214990138068,
      "grad_norm": 1.4375,
      "learning_rate": 0.00018858556458913191,
      "loss": 0.0729,
      "step": 4832
    },
    {
      "epoch": 1.1918145956607495,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018856958765136374,
      "loss": 0.0744,
      "step": 4834
    },
    {
      "epoch": 1.1923076923076923,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001885536002176049,
      "loss": 0.0759,
      "step": 4836
    },
    {
      "epoch": 1.192800788954635,
      "grad_norm": 0.796875,
      "learning_rate": 0.00018853760228975002,
      "loss": 0.0744,
      "step": 4838
    },
    {
      "epoch": 1.193293885601578,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00018852159386969494,
      "loss": 0.0729,
      "step": 4840
    },
    {
      "epoch": 1.1937869822485208,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018850557495933673,
      "loss": 0.0781,
      "step": 4842
    },
    {
      "epoch": 1.1942800788954635,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018848954556057373,
      "loss": 0.0743,
      "step": 4844
    },
    {
      "epoch": 1.1947731755424063,
      "grad_norm": 0.375,
      "learning_rate": 0.00018847350567530554,
      "loss": 0.0758,
      "step": 4846
    },
    {
      "epoch": 1.195266272189349,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00018845745530543295,
      "loss": 0.0777,
      "step": 4848
    },
    {
      "epoch": 1.195759368836292,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000188441394452858,
      "loss": 0.0739,
      "step": 4850
    },
    {
      "epoch": 1.1962524654832347,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00018842532311948404,
      "loss": 0.0791,
      "step": 4852
    },
    {
      "epoch": 1.1967455621301775,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001884092413072156,
      "loss": 0.0791,
      "step": 4854
    },
    {
      "epoch": 1.1972386587771204,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00018839314901795842,
      "loss": 0.0815,
      "step": 4856
    },
    {
      "epoch": 1.1977317554240632,
      "grad_norm": 0.625,
      "learning_rate": 0.00018837704625361958,
      "loss": 0.0682,
      "step": 4858
    },
    {
      "epoch": 1.198224852071006,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00018836093301610732,
      "loss": 0.0622,
      "step": 4860
    },
    {
      "epoch": 1.1987179487179487,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018834480930733115,
      "loss": 0.0731,
      "step": 4862
    },
    {
      "epoch": 1.1992110453648914,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018832867512920181,
      "loss": 0.0712,
      "step": 4864
    },
    {
      "epoch": 1.1997041420118344,
      "grad_norm": 0.703125,
      "learning_rate": 0.00018831253048363133,
      "loss": 0.0709,
      "step": 4866
    },
    {
      "epoch": 1.2001972386587771,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018829637537253288,
      "loss": 0.0741,
      "step": 4868
    },
    {
      "epoch": 1.20069033530572,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000188280209797821,
      "loss": 0.0795,
      "step": 4870
    },
    {
      "epoch": 1.2011834319526626,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00018826403376141136,
      "loss": 0.0676,
      "step": 4872
    },
    {
      "epoch": 1.2016765285996056,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001882478472652209,
      "loss": 0.0682,
      "step": 4874
    },
    {
      "epoch": 1.2021696252465484,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001882316503111678,
      "loss": 0.0781,
      "step": 4876
    },
    {
      "epoch": 1.202662721893491,
      "grad_norm": 0.453125,
      "learning_rate": 0.00018821544290117153,
      "loss": 0.08,
      "step": 4878
    },
    {
      "epoch": 1.2031558185404339,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001881992250371528,
      "loss": 0.0741,
      "step": 4880
    },
    {
      "epoch": 1.2036489151873768,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001881829967210334,
      "loss": 0.0683,
      "step": 4882
    },
    {
      "epoch": 1.2041420118343196,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00018816675795473655,
      "loss": 0.0651,
      "step": 4884
    },
    {
      "epoch": 1.2046351084812623,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018815050874018666,
      "loss": 0.0752,
      "step": 4886
    },
    {
      "epoch": 1.205128205128205,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018813424907930925,
      "loss": 0.0733,
      "step": 4888
    },
    {
      "epoch": 1.2056213017751478,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001881179789740313,
      "loss": 0.0738,
      "step": 4890
    },
    {
      "epoch": 1.2061143984220908,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018810169842628085,
      "loss": 0.0766,
      "step": 4892
    },
    {
      "epoch": 1.2066074950690335,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001880854074379872,
      "loss": 0.0742,
      "step": 4894
    },
    {
      "epoch": 1.2071005917159763,
      "grad_norm": 0.484375,
      "learning_rate": 0.00018806910601108098,
      "loss": 0.077,
      "step": 4896
    },
    {
      "epoch": 1.2075936883629192,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018805279414749402,
      "loss": 0.0733,
      "step": 4898
    },
    {
      "epoch": 1.208086785009862,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001880364718491593,
      "loss": 0.0707,
      "step": 4900
    },
    {
      "epoch": 1.2085798816568047,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00018802013911801112,
      "loss": 0.0762,
      "step": 4902
    },
    {
      "epoch": 1.2090729783037475,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018800379595598502,
      "loss": 0.0831,
      "step": 4904
    },
    {
      "epoch": 1.2095660749506902,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00018798744236501777,
      "loss": 0.0805,
      "step": 4906
    },
    {
      "epoch": 1.2100591715976332,
      "grad_norm": 0.484375,
      "learning_rate": 0.00018797107834704732,
      "loss": 0.0786,
      "step": 4908
    },
    {
      "epoch": 1.210552268244576,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001879547039040129,
      "loss": 0.0845,
      "step": 4910
    },
    {
      "epoch": 1.2110453648915187,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018793831903785504,
      "loss": 0.0824,
      "step": 4912
    },
    {
      "epoch": 1.2115384615384615,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00018792192375051536,
      "loss": 0.0706,
      "step": 4914
    },
    {
      "epoch": 1.2120315581854044,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018790551804393683,
      "loss": 0.0769,
      "step": 4916
    },
    {
      "epoch": 1.2125246548323472,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018788910192006362,
      "loss": 0.0758,
      "step": 4918
    },
    {
      "epoch": 1.21301775147929,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018787267538084108,
      "loss": 0.0851,
      "step": 4920
    },
    {
      "epoch": 1.2135108481262327,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018785623842821594,
      "loss": 0.0782,
      "step": 4922
    },
    {
      "epoch": 1.2140039447731756,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000187839791064136,
      "loss": 0.0769,
      "step": 4924
    },
    {
      "epoch": 1.2144970414201184,
      "grad_norm": 0.625,
      "learning_rate": 0.00018782333329055037,
      "loss": 0.0713,
      "step": 4926
    },
    {
      "epoch": 1.2149901380670611,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00018780686510940943,
      "loss": 0.0785,
      "step": 4928
    },
    {
      "epoch": 1.2154832347140039,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018779038652266472,
      "loss": 0.0735,
      "step": 4930
    },
    {
      "epoch": 1.2159763313609466,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000187773897532269,
      "loss": 0.0793,
      "step": 4932
    },
    {
      "epoch": 1.2164694280078896,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001877573981401764,
      "loss": 0.0755,
      "step": 4934
    },
    {
      "epoch": 1.2169625246548323,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018774088834834217,
      "loss": 0.0744,
      "step": 4936
    },
    {
      "epoch": 1.217455621301775,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00018772436815872275,
      "loss": 0.0745,
      "step": 4938
    },
    {
      "epoch": 1.217948717948718,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018770783757327594,
      "loss": 0.0779,
      "step": 4940
    },
    {
      "epoch": 1.2184418145956608,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018769129659396063,
      "loss": 0.0768,
      "step": 4942
    },
    {
      "epoch": 1.2189349112426036,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018767474522273712,
      "loss": 0.0804,
      "step": 4944
    },
    {
      "epoch": 1.2194280078895463,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00018765818346156676,
      "loss": 0.0791,
      "step": 4946
    },
    {
      "epoch": 1.219921104536489,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018764161131241223,
      "loss": 0.0749,
      "step": 4948
    },
    {
      "epoch": 1.220414201183432,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00018762502877723745,
      "loss": 0.0741,
      "step": 4950
    },
    {
      "epoch": 1.2209072978303748,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001876084358580075,
      "loss": 0.0734,
      "step": 4952
    },
    {
      "epoch": 1.2214003944773175,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001875918325566888,
      "loss": 0.0772,
      "step": 4954
    },
    {
      "epoch": 1.2218934911242603,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00018757521887524886,
      "loss": 0.0728,
      "step": 4956
    },
    {
      "epoch": 1.2223865877712032,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00018755859481565655,
      "loss": 0.0737,
      "step": 4958
    },
    {
      "epoch": 1.222879684418146,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018754196037988185,
      "loss": 0.0757,
      "step": 4960
    },
    {
      "epoch": 1.2233727810650887,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018752531556989613,
      "loss": 0.0774,
      "step": 4962
    },
    {
      "epoch": 1.2238658777120315,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018750866038767181,
      "loss": 0.0721,
      "step": 4964
    },
    {
      "epoch": 1.2243589743589745,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00018749199483518263,
      "loss": 0.0805,
      "step": 4966
    },
    {
      "epoch": 1.2248520710059172,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018747531891440358,
      "loss": 0.0653,
      "step": 4968
    },
    {
      "epoch": 1.22534516765286,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018745863262731085,
      "loss": 0.0799,
      "step": 4970
    },
    {
      "epoch": 1.2258382642998027,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00018744193597588184,
      "loss": 0.0782,
      "step": 4972
    },
    {
      "epoch": 1.2263313609467454,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001874252289620952,
      "loss": 0.0773,
      "step": 4974
    },
    {
      "epoch": 1.2268244575936884,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001874085115879308,
      "loss": 0.0758,
      "step": 4976
    },
    {
      "epoch": 1.2273175542406312,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018739178385536977,
      "loss": 0.0785,
      "step": 4978
    },
    {
      "epoch": 1.227810650887574,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00018737504576639443,
      "loss": 0.0782,
      "step": 4980
    },
    {
      "epoch": 1.2283037475345169,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001873582973229883,
      "loss": 0.0777,
      "step": 4982
    },
    {
      "epoch": 1.2287968441814596,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018734153852713618,
      "loss": 0.0753,
      "step": 4984
    },
    {
      "epoch": 1.2292899408284024,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00018732476938082413,
      "loss": 0.0792,
      "step": 4986
    },
    {
      "epoch": 1.2297830374753451,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018730798988603932,
      "loss": 0.0767,
      "step": 4988
    },
    {
      "epoch": 1.2302761341222879,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018729120004477027,
      "loss": 0.0833,
      "step": 4990
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001872743998590066,
      "loss": 0.0772,
      "step": 4992
    },
    {
      "epoch": 1.2312623274161736,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001872575893307393,
      "loss": 0.0753,
      "step": 4994
    },
    {
      "epoch": 1.2317554240631163,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001872407684619605,
      "loss": 0.0767,
      "step": 4996
    },
    {
      "epoch": 1.232248520710059,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001872239372546635,
      "loss": 0.0726,
      "step": 4998
    },
    {
      "epoch": 1.232741617357002,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018720709571084298,
      "loss": 0.0792,
      "step": 5000
    },
    {
      "epoch": 1.2332347140039448,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018719024383249467,
      "loss": 0.0777,
      "step": 5002
    },
    {
      "epoch": 1.2337278106508875,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018717338162161572,
      "loss": 0.0721,
      "step": 5004
    },
    {
      "epoch": 1.2342209072978303,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018715650908020427,
      "loss": 0.0749,
      "step": 5006
    },
    {
      "epoch": 1.2347140039447733,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001871396262102599,
      "loss": 0.0717,
      "step": 5008
    },
    {
      "epoch": 1.235207100591716,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001871227330137833,
      "loss": 0.0712,
      "step": 5010
    },
    {
      "epoch": 1.2357001972386588,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001871058294927764,
      "loss": 0.0766,
      "step": 5012
    },
    {
      "epoch": 1.2361932938856015,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001870889156492424,
      "loss": 0.0718,
      "step": 5014
    },
    {
      "epoch": 1.2366863905325443,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00018707199148518565,
      "loss": 0.0809,
      "step": 5016
    },
    {
      "epoch": 1.2371794871794872,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00018705505700261171,
      "loss": 0.0776,
      "step": 5018
    },
    {
      "epoch": 1.23767258382643,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00018703811220352752,
      "loss": 0.076,
      "step": 5020
    },
    {
      "epoch": 1.2381656804733727,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018702115708994104,
      "loss": 0.0755,
      "step": 5022
    },
    {
      "epoch": 1.2386587771203157,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001870041916638616,
      "loss": 0.0811,
      "step": 5024
    },
    {
      "epoch": 1.2391518737672584,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018698721592729967,
      "loss": 0.0737,
      "step": 5026
    },
    {
      "epoch": 1.2396449704142012,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018697022988226698,
      "loss": 0.0757,
      "step": 5028
    },
    {
      "epoch": 1.240138067061144,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00018695323353077646,
      "loss": 0.0786,
      "step": 5030
    },
    {
      "epoch": 1.2406311637080867,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018693622687484228,
      "loss": 0.0752,
      "step": 5032
    },
    {
      "epoch": 1.2411242603550297,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00018691920991647986,
      "loss": 0.0746,
      "step": 5034
    },
    {
      "epoch": 1.2416173570019724,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001869021826577057,
      "loss": 0.076,
      "step": 5036
    },
    {
      "epoch": 1.2421104536489151,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018688514510053778,
      "loss": 0.0756,
      "step": 5038
    },
    {
      "epoch": 1.242603550295858,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00018686809724699502,
      "loss": 0.0794,
      "step": 5040
    },
    {
      "epoch": 1.2430966469428009,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018685103909909772,
      "loss": 0.0705,
      "step": 5042
    },
    {
      "epoch": 1.2435897435897436,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00018683397065886737,
      "loss": 0.075,
      "step": 5044
    },
    {
      "epoch": 1.2440828402366864,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0001868168919283267,
      "loss": 0.0768,
      "step": 5046
    },
    {
      "epoch": 1.244575936883629,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001867998029094996,
      "loss": 0.0735,
      "step": 5048
    },
    {
      "epoch": 1.245069033530572,
      "grad_norm": 0.6875,
      "learning_rate": 0.00018678270360441123,
      "loss": 0.0768,
      "step": 5050
    },
    {
      "epoch": 1.2455621301775148,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00018676559401508796,
      "loss": 0.0698,
      "step": 5052
    },
    {
      "epoch": 1.2460552268244576,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018674847414355735,
      "loss": 0.0785,
      "step": 5054
    },
    {
      "epoch": 1.2465483234714003,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018673134399184823,
      "loss": 0.0705,
      "step": 5056
    },
    {
      "epoch": 1.2470414201183433,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018671420356199058,
      "loss": 0.0744,
      "step": 5058
    },
    {
      "epoch": 1.247534516765286,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00018669705285601564,
      "loss": 0.0835,
      "step": 5060
    },
    {
      "epoch": 1.2480276134122288,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001866798918759559,
      "loss": 0.0723,
      "step": 5062
    },
    {
      "epoch": 1.2485207100591715,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00018666272062384503,
      "loss": 0.0826,
      "step": 5064
    },
    {
      "epoch": 1.2490138067061145,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001866455391017179,
      "loss": 0.0772,
      "step": 5066
    },
    {
      "epoch": 1.2495069033530573,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001866283473116106,
      "loss": 0.0829,
      "step": 5068
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001866111452555605,
      "loss": 0.0761,
      "step": 5070
    },
    {
      "epoch": 1.2504930966469427,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001865939329356061,
      "loss": 0.0741,
      "step": 5072
    },
    {
      "epoch": 1.2509861932938855,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018657671035378717,
      "loss": 0.0723,
      "step": 5074
    },
    {
      "epoch": 1.2514792899408285,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018655947751214467,
      "loss": 0.0723,
      "step": 5076
    },
    {
      "epoch": 1.2519723865877712,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001865422344127208,
      "loss": 0.0812,
      "step": 5078
    },
    {
      "epoch": 1.252465483234714,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00018652498105755898,
      "loss": 0.0743,
      "step": 5080
    },
    {
      "epoch": 1.252958579881657,
      "grad_norm": 0.71875,
      "learning_rate": 0.00018650771744870378,
      "loss": 0.0751,
      "step": 5082
    },
    {
      "epoch": 1.2534516765285997,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001864904435882011,
      "loss": 0.0686,
      "step": 5084
    },
    {
      "epoch": 1.2539447731755424,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00018647315947809792,
      "loss": 0.0759,
      "step": 5086
    },
    {
      "epoch": 1.2544378698224852,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00018645586512044257,
      "loss": 0.0751,
      "step": 5088
    },
    {
      "epoch": 1.254930966469428,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018643856051728448,
      "loss": 0.0759,
      "step": 5090
    },
    {
      "epoch": 1.255424063116371,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00018642124567067435,
      "loss": 0.0829,
      "step": 5092
    },
    {
      "epoch": 1.2559171597633136,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001864039205826641,
      "loss": 0.0715,
      "step": 5094
    },
    {
      "epoch": 1.2564102564102564,
      "grad_norm": 0.40625,
      "learning_rate": 0.00018638658525530686,
      "loss": 0.0826,
      "step": 5096
    },
    {
      "epoch": 1.2569033530571991,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018636923969065695,
      "loss": 0.0797,
      "step": 5098
    },
    {
      "epoch": 1.2573964497041419,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00018635188389076987,
      "loss": 0.0788,
      "step": 5100
    },
    {
      "epoch": 1.2578895463510849,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00018633451785770246,
      "loss": 0.0834,
      "step": 5102
    },
    {
      "epoch": 1.2583826429980276,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018631714159351263,
      "loss": 0.0797,
      "step": 5104
    },
    {
      "epoch": 1.2588757396449703,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018629975510025962,
      "loss": 0.0724,
      "step": 5106
    },
    {
      "epoch": 1.2593688362919133,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00018628235838000375,
      "loss": 0.0815,
      "step": 5108
    },
    {
      "epoch": 1.259861932938856,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00018626495143480668,
      "loss": 0.0742,
      "step": 5110
    },
    {
      "epoch": 1.2603550295857988,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018624753426673126,
      "loss": 0.0801,
      "step": 5112
    },
    {
      "epoch": 1.2608481262327416,
      "grad_norm": 0.484375,
      "learning_rate": 0.00018623010687784147,
      "loss": 0.0799,
      "step": 5114
    },
    {
      "epoch": 1.2613412228796843,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00018621266927020254,
      "loss": 0.0698,
      "step": 5116
    },
    {
      "epoch": 1.2618343195266273,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00018619522144588098,
      "loss": 0.0765,
      "step": 5118
    },
    {
      "epoch": 1.26232741617357,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018617776340694442,
      "loss": 0.0759,
      "step": 5120
    },
    {
      "epoch": 1.2628205128205128,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00018616029515546175,
      "loss": 0.0718,
      "step": 5122
    },
    {
      "epoch": 1.2633136094674557,
      "grad_norm": 0.43359375,
      "learning_rate": 0.000186142816693503,
      "loss": 0.0731,
      "step": 5124
    },
    {
      "epoch": 1.2638067061143985,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018612532802313956,
      "loss": 0.0785,
      "step": 5126
    },
    {
      "epoch": 1.2642998027613412,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00018610782914644389,
      "loss": 0.078,
      "step": 5128
    },
    {
      "epoch": 1.264792899408284,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00018609032006548968,
      "loss": 0.0809,
      "step": 5130
    },
    {
      "epoch": 1.2652859960552267,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018607280078235188,
      "loss": 0.0751,
      "step": 5132
    },
    {
      "epoch": 1.2657790927021697,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018605527129910663,
      "loss": 0.0837,
      "step": 5134
    },
    {
      "epoch": 1.2662721893491125,
      "grad_norm": 0.484375,
      "learning_rate": 0.00018603773161783124,
      "loss": 0.0762,
      "step": 5136
    },
    {
      "epoch": 1.2667652859960552,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001860201817406043,
      "loss": 0.0741,
      "step": 5138
    },
    {
      "epoch": 1.267258382642998,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00018600262166950553,
      "loss": 0.08,
      "step": 5140
    },
    {
      "epoch": 1.2677514792899407,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018598505140661592,
      "loss": 0.0774,
      "step": 5142
    },
    {
      "epoch": 1.2682445759368837,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018596747095401766,
      "loss": 0.0871,
      "step": 5144
    },
    {
      "epoch": 1.2687376725838264,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001859498803137941,
      "loss": 0.0755,
      "step": 5146
    },
    {
      "epoch": 1.2692307692307692,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018593227948802981,
      "loss": 0.0743,
      "step": 5148
    },
    {
      "epoch": 1.2697238658777121,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018591466847881066,
      "loss": 0.0781,
      "step": 5150
    },
    {
      "epoch": 1.2702169625246549,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001858970472882236,
      "loss": 0.0703,
      "step": 5152
    },
    {
      "epoch": 1.2707100591715976,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00018587941591835682,
      "loss": 0.0804,
      "step": 5154
    },
    {
      "epoch": 1.2712031558185404,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018586177437129978,
      "loss": 0.0762,
      "step": 5156
    },
    {
      "epoch": 1.2716962524654831,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001858441226491431,
      "loss": 0.0692,
      "step": 5158
    },
    {
      "epoch": 1.272189349112426,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018582646075397858,
      "loss": 0.0822,
      "step": 5160
    },
    {
      "epoch": 1.2726824457593688,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00018580878868789928,
      "loss": 0.0767,
      "step": 5162
    },
    {
      "epoch": 1.2731755424063116,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001857911064529994,
      "loss": 0.0819,
      "step": 5164
    },
    {
      "epoch": 1.2736686390532546,
      "grad_norm": 0.625,
      "learning_rate": 0.00018577341405137443,
      "loss": 0.079,
      "step": 5166
    },
    {
      "epoch": 1.2741617357001973,
      "grad_norm": 0.470703125,
      "learning_rate": 0.000185755711485121,
      "loss": 0.0785,
      "step": 5168
    },
    {
      "epoch": 1.27465483234714,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018573799875633694,
      "loss": 0.0685,
      "step": 5170
    },
    {
      "epoch": 1.2751479289940828,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00018572027586712136,
      "loss": 0.0738,
      "step": 5172
    },
    {
      "epoch": 1.2756410256410255,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018570254281957444,
      "loss": 0.0813,
      "step": 5174
    },
    {
      "epoch": 1.2761341222879685,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018568479961579774,
      "loss": 0.074,
      "step": 5176
    },
    {
      "epoch": 1.2766272189349113,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018566704625789387,
      "loss": 0.0769,
      "step": 5178
    },
    {
      "epoch": 1.277120315581854,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001856492827479667,
      "loss": 0.076,
      "step": 5180
    },
    {
      "epoch": 1.2776134122287968,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018563150908812133,
      "loss": 0.0768,
      "step": 5182
    },
    {
      "epoch": 1.2781065088757395,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00018561372528046402,
      "loss": 0.0782,
      "step": 5184
    },
    {
      "epoch": 1.2785996055226825,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018559593132710229,
      "loss": 0.0665,
      "step": 5186
    },
    {
      "epoch": 1.2790927021696252,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018557812723014476,
      "loss": 0.0697,
      "step": 5188
    },
    {
      "epoch": 1.279585798816568,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018556031299170136,
      "loss": 0.0719,
      "step": 5190
    },
    {
      "epoch": 1.280078895463511,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018554248861388315,
      "loss": 0.0729,
      "step": 5192
    },
    {
      "epoch": 1.2805719921104537,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00018552465409880245,
      "loss": 0.0758,
      "step": 5194
    },
    {
      "epoch": 1.2810650887573964,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001855068094485727,
      "loss": 0.0772,
      "step": 5196
    },
    {
      "epoch": 1.2815581854043392,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018548895466530865,
      "loss": 0.0746,
      "step": 5198
    },
    {
      "epoch": 1.282051282051282,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018547108975112617,
      "loss": 0.0763,
      "step": 5200
    },
    {
      "epoch": 1.282544378698225,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00018545321470814233,
      "loss": 0.0841,
      "step": 5202
    },
    {
      "epoch": 1.2830374753451677,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00018543532953847544,
      "loss": 0.0771,
      "step": 5204
    },
    {
      "epoch": 1.2835305719921104,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00018541743424424498,
      "loss": 0.0809,
      "step": 5206
    },
    {
      "epoch": 1.2840236686390534,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018539952882757165,
      "loss": 0.0759,
      "step": 5208
    },
    {
      "epoch": 1.2845167652859961,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018538161329057737,
      "loss": 0.0793,
      "step": 5210
    },
    {
      "epoch": 1.2850098619329389,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00018536368763538518,
      "loss": 0.0737,
      "step": 5212
    },
    {
      "epoch": 1.2855029585798816,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001853457518641194,
      "loss": 0.0775,
      "step": 5214
    },
    {
      "epoch": 1.2859960552268244,
      "grad_norm": 0.453125,
      "learning_rate": 0.00018532780597890552,
      "loss": 0.0695,
      "step": 5216
    },
    {
      "epoch": 1.2864891518737673,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018530984998187022,
      "loss": 0.0784,
      "step": 5218
    },
    {
      "epoch": 1.28698224852071,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001852918838751414,
      "loss": 0.079,
      "step": 5220
    },
    {
      "epoch": 1.2874753451676528,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001852739076608481,
      "loss": 0.0761,
      "step": 5222
    },
    {
      "epoch": 1.2879684418145958,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018525592134112064,
      "loss": 0.0815,
      "step": 5224
    },
    {
      "epoch": 1.2884615384615383,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00018523792491809054,
      "loss": 0.0742,
      "step": 5226
    },
    {
      "epoch": 1.2889546351084813,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001852199183938904,
      "loss": 0.0739,
      "step": 5228
    },
    {
      "epoch": 1.289447731755424,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001852019017706541,
      "loss": 0.0721,
      "step": 5230
    },
    {
      "epoch": 1.2899408284023668,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018518387505051675,
      "loss": 0.0758,
      "step": 5232
    },
    {
      "epoch": 1.2904339250493098,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001851658382356146,
      "loss": 0.0719,
      "step": 5234
    },
    {
      "epoch": 1.2909270216962525,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018514779132808513,
      "loss": 0.078,
      "step": 5236
    },
    {
      "epoch": 1.2914201183431953,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018512973433006696,
      "loss": 0.0755,
      "step": 5238
    },
    {
      "epoch": 1.291913214990138,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018511166724369997,
      "loss": 0.0739,
      "step": 5240
    },
    {
      "epoch": 1.2924063116370808,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018509359007112523,
      "loss": 0.0744,
      "step": 5242
    },
    {
      "epoch": 1.2928994082840237,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018507550281448497,
      "loss": 0.0735,
      "step": 5244
    },
    {
      "epoch": 1.2933925049309665,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001850574054759226,
      "loss": 0.0766,
      "step": 5246
    },
    {
      "epoch": 1.2938856015779092,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018503929805758278,
      "loss": 0.0704,
      "step": 5248
    },
    {
      "epoch": 1.2943786982248522,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018502118056161135,
      "loss": 0.0777,
      "step": 5250
    },
    {
      "epoch": 1.294871794871795,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018500305299015533,
      "loss": 0.0762,
      "step": 5252
    },
    {
      "epoch": 1.2953648915187377,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018498491534536292,
      "loss": 0.0775,
      "step": 5254
    },
    {
      "epoch": 1.2958579881656804,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001849667676293836,
      "loss": 0.0746,
      "step": 5256
    },
    {
      "epoch": 1.2963510848126232,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00018494860984436784,
      "loss": 0.0767,
      "step": 5258
    },
    {
      "epoch": 1.2968441814595661,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00018493044199246756,
      "loss": 0.0753,
      "step": 5260
    },
    {
      "epoch": 1.297337278106509,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018491226407583574,
      "loss": 0.0686,
      "step": 5262
    },
    {
      "epoch": 1.2978303747534516,
      "grad_norm": 0.5,
      "learning_rate": 0.0001848940760966265,
      "loss": 0.0766,
      "step": 5264
    },
    {
      "epoch": 1.2983234714003946,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018487587805699526,
      "loss": 0.0732,
      "step": 5266
    },
    {
      "epoch": 1.2988165680473374,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001848576699590986,
      "loss": 0.0709,
      "step": 5268
    },
    {
      "epoch": 1.29930966469428,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00018483945180509424,
      "loss": 0.0757,
      "step": 5270
    },
    {
      "epoch": 1.2998027613412229,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001848212235971412,
      "loss": 0.0747,
      "step": 5272
    },
    {
      "epoch": 1.3002958579881656,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00018480298533739956,
      "loss": 0.0697,
      "step": 5274
    },
    {
      "epoch": 1.3007889546351086,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018478473702803069,
      "loss": 0.0754,
      "step": 5276
    },
    {
      "epoch": 1.3012820512820513,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018476647867119708,
      "loss": 0.078,
      "step": 5278
    },
    {
      "epoch": 1.301775147928994,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018474821026906246,
      "loss": 0.0713,
      "step": 5280
    },
    {
      "epoch": 1.3022682445759368,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00018472993182379178,
      "loss": 0.0703,
      "step": 5282
    },
    {
      "epoch": 1.3027613412228796,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001847116433375511,
      "loss": 0.0791,
      "step": 5284
    },
    {
      "epoch": 1.3032544378698225,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018469334481250772,
      "loss": 0.0786,
      "step": 5286
    },
    {
      "epoch": 1.3037475345167653,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018467503625083009,
      "loss": 0.0861,
      "step": 5288
    },
    {
      "epoch": 1.304240631163708,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00018465671765468793,
      "loss": 0.08,
      "step": 5290
    },
    {
      "epoch": 1.304733727810651,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018463838902625207,
      "loss": 0.0743,
      "step": 5292
    },
    {
      "epoch": 1.3052268244575937,
      "grad_norm": 1.265625,
      "learning_rate": 0.00018462005036769453,
      "loss": 0.0772,
      "step": 5294
    },
    {
      "epoch": 1.3057199211045365,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018460170168118855,
      "loss": 0.0776,
      "step": 5296
    },
    {
      "epoch": 1.3062130177514792,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018458334296890858,
      "loss": 0.0759,
      "step": 5298
    },
    {
      "epoch": 1.306706114398422,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001845649742330302,
      "loss": 0.0703,
      "step": 5300
    },
    {
      "epoch": 1.307199211045365,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018454659547573025,
      "loss": 0.0741,
      "step": 5302
    },
    {
      "epoch": 1.3076923076923077,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001845282066991867,
      "loss": 0.0727,
      "step": 5304
    },
    {
      "epoch": 1.3081854043392505,
      "grad_norm": 0.65625,
      "learning_rate": 0.00018450980790557866,
      "loss": 0.078,
      "step": 5306
    },
    {
      "epoch": 1.3086785009861934,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001844913990970866,
      "loss": 0.0699,
      "step": 5308
    },
    {
      "epoch": 1.3091715976331362,
      "grad_norm": 0.59375,
      "learning_rate": 0.00018447298027589196,
      "loss": 0.0729,
      "step": 5310
    },
    {
      "epoch": 1.309664694280079,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00018445455144417753,
      "loss": 0.0803,
      "step": 5312
    },
    {
      "epoch": 1.3101577909270217,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018443611260412723,
      "loss": 0.0803,
      "step": 5314
    },
    {
      "epoch": 1.3106508875739644,
      "grad_norm": 0.5,
      "learning_rate": 0.00018441766375792615,
      "loss": 0.0747,
      "step": 5316
    },
    {
      "epoch": 1.3111439842209074,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018439920490776062,
      "loss": 0.0801,
      "step": 5318
    },
    {
      "epoch": 1.3116370808678501,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00018438073605581801,
      "loss": 0.0792,
      "step": 5320
    },
    {
      "epoch": 1.3121301775147929,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018436225720428712,
      "loss": 0.0841,
      "step": 5322
    },
    {
      "epoch": 1.3126232741617356,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018434376835535773,
      "loss": 0.0726,
      "step": 5324
    },
    {
      "epoch": 1.3131163708086784,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00018432526951122084,
      "loss": 0.0756,
      "step": 5326
    },
    {
      "epoch": 1.3136094674556213,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001843067606740687,
      "loss": 0.0693,
      "step": 5328
    },
    {
      "epoch": 1.314102564102564,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018428824184609475,
      "loss": 0.0739,
      "step": 5330
    },
    {
      "epoch": 1.3145956607495068,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00018426971302949353,
      "loss": 0.0789,
      "step": 5332
    },
    {
      "epoch": 1.3150887573964498,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00018425117422646075,
      "loss": 0.076,
      "step": 5334
    },
    {
      "epoch": 1.3155818540433926,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001842326254391935,
      "loss": 0.0713,
      "step": 5336
    },
    {
      "epoch": 1.3160749506903353,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001842140666698898,
      "loss": 0.0705,
      "step": 5338
    },
    {
      "epoch": 1.316568047337278,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00018419549792074905,
      "loss": 0.0738,
      "step": 5340
    },
    {
      "epoch": 1.3170611439842208,
      "grad_norm": 0.5,
      "learning_rate": 0.00018417691919397168,
      "loss": 0.0713,
      "step": 5342
    },
    {
      "epoch": 1.3175542406311638,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018415833049175941,
      "loss": 0.0705,
      "step": 5344
    },
    {
      "epoch": 1.3180473372781065,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001841397318163151,
      "loss": 0.0758,
      "step": 5346
    },
    {
      "epoch": 1.3185404339250493,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00018412112316984277,
      "loss": 0.0704,
      "step": 5348
    },
    {
      "epoch": 1.3190335305719922,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001841025045545477,
      "loss": 0.0741,
      "step": 5350
    },
    {
      "epoch": 1.319526627218935,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00018408387597263626,
      "loss": 0.0732,
      "step": 5352
    },
    {
      "epoch": 1.3200197238658777,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018406523742631604,
      "loss": 0.0802,
      "step": 5354
    },
    {
      "epoch": 1.3205128205128205,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00018404658891779584,
      "loss": 0.0752,
      "step": 5356
    },
    {
      "epoch": 1.3210059171597632,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001840279304492856,
      "loss": 0.0721,
      "step": 5358
    },
    {
      "epoch": 1.3214990138067062,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00018400926202299643,
      "loss": 0.0728,
      "step": 5360
    },
    {
      "epoch": 1.321992110453649,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018399058364114068,
      "loss": 0.0729,
      "step": 5362
    },
    {
      "epoch": 1.3224852071005917,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018397189530593183,
      "loss": 0.0735,
      "step": 5364
    },
    {
      "epoch": 1.3229783037475344,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00018395319701958453,
      "loss": 0.0757,
      "step": 5366
    },
    {
      "epoch": 1.3234714003944772,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018393448878431466,
      "loss": 0.0746,
      "step": 5368
    },
    {
      "epoch": 1.3239644970414202,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00018391577060233925,
      "loss": 0.0695,
      "step": 5370
    },
    {
      "epoch": 1.324457593688363,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001838970424758765,
      "loss": 0.0649,
      "step": 5372
    },
    {
      "epoch": 1.3249506903353057,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00018387830440714576,
      "loss": 0.0751,
      "step": 5374
    },
    {
      "epoch": 1.3254437869822486,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00018385955639836765,
      "loss": 0.0709,
      "step": 5376
    },
    {
      "epoch": 1.3259368836291914,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001838407984517639,
      "loss": 0.0714,
      "step": 5378
    },
    {
      "epoch": 1.3264299802761341,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001838220305695574,
      "loss": 0.078,
      "step": 5380
    },
    {
      "epoch": 1.3269230769230769,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001838032527539723,
      "loss": 0.0707,
      "step": 5382
    },
    {
      "epoch": 1.3274161735700196,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00018378446500723388,
      "loss": 0.0686,
      "step": 5384
    },
    {
      "epoch": 1.3279092702169626,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018376566733156852,
      "loss": 0.0718,
      "step": 5386
    },
    {
      "epoch": 1.3284023668639053,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018374685972920388,
      "loss": 0.0754,
      "step": 5388
    },
    {
      "epoch": 1.328895463510848,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018372804220236882,
      "loss": 0.0728,
      "step": 5390
    },
    {
      "epoch": 1.329388560157791,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00018370921475329327,
      "loss": 0.0699,
      "step": 5392
    },
    {
      "epoch": 1.3298816568047338,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00018369037738420844,
      "loss": 0.0757,
      "step": 5394
    },
    {
      "epoch": 1.3303747534516766,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00018367153009734655,
      "loss": 0.069,
      "step": 5396
    },
    {
      "epoch": 1.3308678500986193,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001836526728949412,
      "loss": 0.074,
      "step": 5398
    },
    {
      "epoch": 1.331360946745562,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018363380577922708,
      "loss": 0.0684,
      "step": 5400
    },
    {
      "epoch": 1.331854043392505,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00018361492875244002,
      "loss": 0.0769,
      "step": 5402
    },
    {
      "epoch": 1.3323471400394478,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018359604181681705,
      "loss": 0.0741,
      "step": 5404
    },
    {
      "epoch": 1.3328402366863905,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00018357714497459638,
      "loss": 0.072,
      "step": 5406
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001835582382280174,
      "loss": 0.0743,
      "step": 5408
    },
    {
      "epoch": 1.333826429980276,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00018353932157932063,
      "loss": 0.0691,
      "step": 5410
    },
    {
      "epoch": 1.334319526627219,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018352039503074786,
      "loss": 0.0637,
      "step": 5412
    },
    {
      "epoch": 1.3348126232741617,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018350145858454195,
      "loss": 0.0704,
      "step": 5414
    },
    {
      "epoch": 1.3353057199211045,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000183482512242947,
      "loss": 0.0721,
      "step": 5416
    },
    {
      "epoch": 1.3357988165680474,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018346355600820823,
      "loss": 0.0657,
      "step": 5418
    },
    {
      "epoch": 1.3362919132149902,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001834445898825721,
      "loss": 0.0725,
      "step": 5420
    },
    {
      "epoch": 1.336785009861933,
      "grad_norm": 0.390625,
      "learning_rate": 0.00018342561386828615,
      "loss": 0.0698,
      "step": 5422
    },
    {
      "epoch": 1.3372781065088757,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00018340662796759914,
      "loss": 0.0755,
      "step": 5424
    },
    {
      "epoch": 1.3377712031558184,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018338763218276108,
      "loss": 0.0737,
      "step": 5426
    },
    {
      "epoch": 1.3382642998027614,
      "grad_norm": 0.5625,
      "learning_rate": 0.000183368626516023,
      "loss": 0.0759,
      "step": 5428
    },
    {
      "epoch": 1.3387573964497042,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018334961096963724,
      "loss": 0.0697,
      "step": 5430
    },
    {
      "epoch": 1.339250493096647,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018333058554585718,
      "loss": 0.0687,
      "step": 5432
    },
    {
      "epoch": 1.3397435897435899,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001833115502469375,
      "loss": 0.0738,
      "step": 5434
    },
    {
      "epoch": 1.3402366863905326,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00018329250507513397,
      "loss": 0.0695,
      "step": 5436
    },
    {
      "epoch": 1.3407297830374754,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00018327345003270348,
      "loss": 0.0688,
      "step": 5438
    },
    {
      "epoch": 1.3412228796844181,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018325438512190427,
      "loss": 0.0767,
      "step": 5440
    },
    {
      "epoch": 1.3417159763313609,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00018323531034499562,
      "loss": 0.0689,
      "step": 5442
    },
    {
      "epoch": 1.3422090729783038,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018321622570423792,
      "loss": 0.0727,
      "step": 5444
    },
    {
      "epoch": 1.3427021696252466,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018319713120189287,
      "loss": 0.0686,
      "step": 5446
    },
    {
      "epoch": 1.3431952662721893,
      "grad_norm": 0.453125,
      "learning_rate": 0.00018317802684022325,
      "loss": 0.0711,
      "step": 5448
    },
    {
      "epoch": 1.343688362919132,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018315891262149305,
      "loss": 0.0803,
      "step": 5450
    },
    {
      "epoch": 1.3441814595660748,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001831397885479674,
      "loss": 0.0756,
      "step": 5452
    },
    {
      "epoch": 1.3446745562130178,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018312065462191263,
      "loss": 0.0773,
      "step": 5454
    },
    {
      "epoch": 1.3451676528599605,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00018310151084559618,
      "loss": 0.0744,
      "step": 5456
    },
    {
      "epoch": 1.3456607495069033,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018308235722128672,
      "loss": 0.0752,
      "step": 5458
    },
    {
      "epoch": 1.3461538461538463,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018306319375125407,
      "loss": 0.0751,
      "step": 5460
    },
    {
      "epoch": 1.346646942800789,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001830440204377692,
      "loss": 0.0757,
      "step": 5462
    },
    {
      "epoch": 1.3471400394477318,
      "grad_norm": 0.40625,
      "learning_rate": 0.00018302483728310423,
      "loss": 0.0709,
      "step": 5464
    },
    {
      "epoch": 1.3476331360946745,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018300564428953248,
      "loss": 0.0758,
      "step": 5466
    },
    {
      "epoch": 1.3481262327416172,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00018298644145932848,
      "loss": 0.0631,
      "step": 5468
    },
    {
      "epoch": 1.3486193293885602,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00018296722879476777,
      "loss": 0.07,
      "step": 5470
    },
    {
      "epoch": 1.349112426035503,
      "grad_norm": 0.6875,
      "learning_rate": 0.00018294800629812727,
      "loss": 0.0678,
      "step": 5472
    },
    {
      "epoch": 1.3496055226824457,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018292877397168487,
      "loss": 0.0662,
      "step": 5474
    },
    {
      "epoch": 1.3500986193293887,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018290953181771978,
      "loss": 0.0664,
      "step": 5476
    },
    {
      "epoch": 1.3505917159763314,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00018289027983851223,
      "loss": 0.0712,
      "step": 5478
    },
    {
      "epoch": 1.3510848126232742,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018287101803634372,
      "loss": 0.0728,
      "step": 5480
    },
    {
      "epoch": 1.351577909270217,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001828517464134969,
      "loss": 0.0728,
      "step": 5482
    },
    {
      "epoch": 1.3520710059171597,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00018283246497225554,
      "loss": 0.0712,
      "step": 5484
    },
    {
      "epoch": 1.3525641025641026,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001828131737149046,
      "loss": 0.0685,
      "step": 5486
    },
    {
      "epoch": 1.3530571992110454,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018279387264373017,
      "loss": 0.073,
      "step": 5488
    },
    {
      "epoch": 1.3535502958579881,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001827745617610196,
      "loss": 0.0702,
      "step": 5490
    },
    {
      "epoch": 1.3540433925049309,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001827552410690613,
      "loss": 0.0662,
      "step": 5492
    },
    {
      "epoch": 1.3545364891518736,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001827359105701449,
      "loss": 0.0679,
      "step": 5494
    },
    {
      "epoch": 1.3550295857988166,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018271657026656114,
      "loss": 0.069,
      "step": 5496
    },
    {
      "epoch": 1.3555226824457594,
      "grad_norm": 0.421875,
      "learning_rate": 0.000182697220160602,
      "loss": 0.0661,
      "step": 5498
    },
    {
      "epoch": 1.356015779092702,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018267786025456052,
      "loss": 0.0705,
      "step": 5500
    },
    {
      "epoch": 1.356508875739645,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00018265849055073102,
      "loss": 0.065,
      "step": 5502
    },
    {
      "epoch": 1.3570019723865878,
      "grad_norm": 0.375,
      "learning_rate": 0.00018263911105140886,
      "loss": 0.0698,
      "step": 5504
    },
    {
      "epoch": 1.3574950690335306,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018261972175889063,
      "loss": 0.0785,
      "step": 5506
    },
    {
      "epoch": 1.3579881656804733,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00018260032267547412,
      "loss": 0.0644,
      "step": 5508
    },
    {
      "epoch": 1.358481262327416,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00018258091380345817,
      "loss": 0.0705,
      "step": 5510
    },
    {
      "epoch": 1.358974358974359,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018256149514514285,
      "loss": 0.069,
      "step": 5512
    },
    {
      "epoch": 1.3594674556213018,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00018254206670282938,
      "loss": 0.0612,
      "step": 5514
    },
    {
      "epoch": 1.3599605522682445,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001825226284788202,
      "loss": 0.0632,
      "step": 5516
    },
    {
      "epoch": 1.3604536489151875,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00018250318047541877,
      "loss": 0.0736,
      "step": 5518
    },
    {
      "epoch": 1.3609467455621302,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018248372269492983,
      "loss": 0.0748,
      "step": 5520
    },
    {
      "epoch": 1.361439842209073,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001824642551396592,
      "loss": 0.0714,
      "step": 5522
    },
    {
      "epoch": 1.3619329388560157,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018244477781191392,
      "loss": 0.0589,
      "step": 5524
    },
    {
      "epoch": 1.3624260355029585,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018242529071400214,
      "loss": 0.0706,
      "step": 5526
    },
    {
      "epoch": 1.3629191321499015,
      "grad_norm": 0.625,
      "learning_rate": 0.00018240579384823323,
      "loss": 0.0674,
      "step": 5528
    },
    {
      "epoch": 1.3634122287968442,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018238628721691766,
      "loss": 0.0702,
      "step": 5530
    },
    {
      "epoch": 1.363905325443787,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018236677082236703,
      "loss": 0.0697,
      "step": 5532
    },
    {
      "epoch": 1.3643984220907297,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018234724466689421,
      "loss": 0.0707,
      "step": 5534
    },
    {
      "epoch": 1.3648915187376724,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00018232770875281314,
      "loss": 0.0653,
      "step": 5536
    },
    {
      "epoch": 1.3653846153846154,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001823081630824389,
      "loss": 0.0733,
      "step": 5538
    },
    {
      "epoch": 1.3658777120315582,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018228860765808782,
      "loss": 0.0691,
      "step": 5540
    },
    {
      "epoch": 1.366370808678501,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018226904248207725,
      "loss": 0.0729,
      "step": 5542
    },
    {
      "epoch": 1.3668639053254439,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018224946755672584,
      "loss": 0.0677,
      "step": 5544
    },
    {
      "epoch": 1.3673570019723866,
      "grad_norm": 0.625,
      "learning_rate": 0.00018222988288435328,
      "loss": 0.0659,
      "step": 5546
    },
    {
      "epoch": 1.3678500986193294,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001822102884672805,
      "loss": 0.0615,
      "step": 5548
    },
    {
      "epoch": 1.3683431952662721,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018219068430782952,
      "loss": 0.0643,
      "step": 5550
    },
    {
      "epoch": 1.3688362919132149,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00018217107040832358,
      "loss": 0.0632,
      "step": 5552
    },
    {
      "epoch": 1.3693293885601578,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00018215144677108698,
      "loss": 0.0632,
      "step": 5554
    },
    {
      "epoch": 1.3698224852071006,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001821318133984453,
      "loss": 0.0685,
      "step": 5556
    },
    {
      "epoch": 1.3703155818540433,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018211217029272513,
      "loss": 0.0657,
      "step": 5558
    },
    {
      "epoch": 1.3708086785009863,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018209251745625435,
      "loss": 0.0677,
      "step": 5560
    },
    {
      "epoch": 1.371301775147929,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001820728548913619,
      "loss": 0.0717,
      "step": 5562
    },
    {
      "epoch": 1.3717948717948718,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001820531826003779,
      "loss": 0.0679,
      "step": 5564
    },
    {
      "epoch": 1.3722879684418146,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018203350058563364,
      "loss": 0.0671,
      "step": 5566
    },
    {
      "epoch": 1.3727810650887573,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018201380884946158,
      "loss": 0.0716,
      "step": 5568
    },
    {
      "epoch": 1.3732741617357003,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00018199410739419523,
      "loss": 0.0738,
      "step": 5570
    },
    {
      "epoch": 1.373767258382643,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00018197439622216934,
      "loss": 0.0696,
      "step": 5572
    },
    {
      "epoch": 1.3742603550295858,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018195467533571984,
      "loss": 0.0704,
      "step": 5574
    },
    {
      "epoch": 1.3747534516765287,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018193494473718374,
      "loss": 0.0624,
      "step": 5576
    },
    {
      "epoch": 1.3752465483234713,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001819152044288992,
      "loss": 0.0686,
      "step": 5578
    },
    {
      "epoch": 1.3757396449704142,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018189545441320556,
      "loss": 0.0683,
      "step": 5580
    },
    {
      "epoch": 1.376232741617357,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018187569469244337,
      "loss": 0.0714,
      "step": 5582
    },
    {
      "epoch": 1.3767258382642997,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00018185592526895423,
      "loss": 0.065,
      "step": 5584
    },
    {
      "epoch": 1.3772189349112427,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018183614614508087,
      "loss": 0.0646,
      "step": 5586
    },
    {
      "epoch": 1.3777120315581854,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018181635732316732,
      "loss": 0.0604,
      "step": 5588
    },
    {
      "epoch": 1.3782051282051282,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018179655880555862,
      "loss": 0.0654,
      "step": 5590
    },
    {
      "epoch": 1.378698224852071,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000181776750594601,
      "loss": 0.0679,
      "step": 5592
    },
    {
      "epoch": 1.3791913214990137,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018175693269264183,
      "loss": 0.0716,
      "step": 5594
    },
    {
      "epoch": 1.3796844181459567,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018173710510202968,
      "loss": 0.0644,
      "step": 5596
    },
    {
      "epoch": 1.3801775147928994,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001817172678251142,
      "loss": 0.0699,
      "step": 5598
    },
    {
      "epoch": 1.3806706114398422,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001816974208642463,
      "loss": 0.0689,
      "step": 5600
    },
    {
      "epoch": 1.3811637080867851,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001816775642217778,
      "loss": 0.0663,
      "step": 5602
    },
    {
      "epoch": 1.3816568047337279,
      "grad_norm": 0.375,
      "learning_rate": 0.00018165769790006199,
      "loss": 0.0656,
      "step": 5604
    },
    {
      "epoch": 1.3821499013806706,
      "grad_norm": 0.37890625,
      "learning_rate": 0.000181637821901453,
      "loss": 0.0637,
      "step": 5606
    },
    {
      "epoch": 1.3826429980276134,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00018161793622830633,
      "loss": 0.0674,
      "step": 5608
    },
    {
      "epoch": 1.3831360946745561,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001815980408829785,
      "loss": 0.0628,
      "step": 5610
    },
    {
      "epoch": 1.383629191321499,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00018157813586782727,
      "loss": 0.0646,
      "step": 5612
    },
    {
      "epoch": 1.3841222879684418,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001815582211852114,
      "loss": 0.0701,
      "step": 5614
    },
    {
      "epoch": 1.3846153846153846,
      "grad_norm": 0.625,
      "learning_rate": 0.00018153829683749105,
      "loss": 0.0723,
      "step": 5616
    },
    {
      "epoch": 1.3851084812623276,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00018151836282702722,
      "loss": 0.0678,
      "step": 5618
    },
    {
      "epoch": 1.3856015779092703,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001814984191561823,
      "loss": 0.0648,
      "step": 5620
    },
    {
      "epoch": 1.386094674556213,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018147846582731962,
      "loss": 0.0602,
      "step": 5622
    },
    {
      "epoch": 1.3865877712031558,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00018145850284280384,
      "loss": 0.0597,
      "step": 5624
    },
    {
      "epoch": 1.3870808678500985,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001814385302050007,
      "loss": 0.0695,
      "step": 5626
    },
    {
      "epoch": 1.3875739644970415,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00018141854791627702,
      "loss": 0.0636,
      "step": 5628
    },
    {
      "epoch": 1.3880670611439843,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001813985559790008,
      "loss": 0.0698,
      "step": 5630
    },
    {
      "epoch": 1.388560157790927,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018137855439554124,
      "loss": 0.0736,
      "step": 5632
    },
    {
      "epoch": 1.3890532544378698,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018135854316826864,
      "loss": 0.0692,
      "step": 5634
    },
    {
      "epoch": 1.3895463510848125,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018133852229955443,
      "loss": 0.0627,
      "step": 5636
    },
    {
      "epoch": 1.3900394477317555,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018131849179177115,
      "loss": 0.0638,
      "step": 5638
    },
    {
      "epoch": 1.3905325443786982,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00018129845164729263,
      "loss": 0.0718,
      "step": 5640
    },
    {
      "epoch": 1.391025641025641,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018127840186849364,
      "loss": 0.0647,
      "step": 5642
    },
    {
      "epoch": 1.391518737672584,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001812583424577502,
      "loss": 0.0661,
      "step": 5644
    },
    {
      "epoch": 1.3920118343195267,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018123827341743956,
      "loss": 0.0626,
      "step": 5646
    },
    {
      "epoch": 1.3925049309664694,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001812181947499399,
      "loss": 0.0633,
      "step": 5648
    },
    {
      "epoch": 1.3929980276134122,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001811981064576307,
      "loss": 0.064,
      "step": 5650
    },
    {
      "epoch": 1.393491124260355,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00018117800854289254,
      "loss": 0.0688,
      "step": 5652
    },
    {
      "epoch": 1.393984220907298,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001811579010081071,
      "loss": 0.0713,
      "step": 5654
    },
    {
      "epoch": 1.3944773175542406,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00018113778385565733,
      "loss": 0.0676,
      "step": 5656
    },
    {
      "epoch": 1.3949704142011834,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001811176570879271,
      "loss": 0.0616,
      "step": 5658
    },
    {
      "epoch": 1.3954635108481264,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018109752070730162,
      "loss": 0.0602,
      "step": 5660
    },
    {
      "epoch": 1.3959566074950691,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00018107737471616714,
      "loss": 0.0678,
      "step": 5662
    },
    {
      "epoch": 1.3964497041420119,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001810572191169111,
      "loss": 0.0658,
      "step": 5664
    },
    {
      "epoch": 1.3969428007889546,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00018103705391192199,
      "loss": 0.0668,
      "step": 5666
    },
    {
      "epoch": 1.3974358974358974,
      "grad_norm": 0.40625,
      "learning_rate": 0.00018101687910358956,
      "loss": 0.0664,
      "step": 5668
    },
    {
      "epoch": 1.3979289940828403,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001809966946943046,
      "loss": 0.0553,
      "step": 5670
    },
    {
      "epoch": 1.398422090729783,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00018097650068645913,
      "loss": 0.075,
      "step": 5672
    },
    {
      "epoch": 1.3989151873767258,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001809562970824462,
      "loss": 0.0655,
      "step": 5674
    },
    {
      "epoch": 1.3994082840236686,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018093608388466005,
      "loss": 0.0649,
      "step": 5676
    },
    {
      "epoch": 1.3999013806706113,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00018091586109549604,
      "loss": 0.0619,
      "step": 5678
    },
    {
      "epoch": 1.4003944773175543,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018089562871735076,
      "loss": 0.0614,
      "step": 5680
    },
    {
      "epoch": 1.400887573964497,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001808753867526218,
      "loss": 0.0666,
      "step": 5682
    },
    {
      "epoch": 1.4013806706114398,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00018085513520370796,
      "loss": 0.0695,
      "step": 5684
    },
    {
      "epoch": 1.4018737672583828,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018083487407300917,
      "loss": 0.071,
      "step": 5686
    },
    {
      "epoch": 1.4023668639053255,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018081460336292647,
      "loss": 0.0709,
      "step": 5688
    },
    {
      "epoch": 1.4028599605522682,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018079432307586206,
      "loss": 0.0633,
      "step": 5690
    },
    {
      "epoch": 1.403353057199211,
      "grad_norm": 0.703125,
      "learning_rate": 0.00018077403321421927,
      "loss": 0.0632,
      "step": 5692
    },
    {
      "epoch": 1.4038461538461537,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00018075373378040256,
      "loss": 0.0645,
      "step": 5694
    },
    {
      "epoch": 1.4043392504930967,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001807334247768175,
      "loss": 0.0622,
      "step": 5696
    },
    {
      "epoch": 1.4048323471400395,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001807131062058709,
      "loss": 0.0618,
      "step": 5698
    },
    {
      "epoch": 1.4053254437869822,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00018069277806997052,
      "loss": 0.0616,
      "step": 5700
    },
    {
      "epoch": 1.4058185404339252,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00018067244037152545,
      "loss": 0.0623,
      "step": 5702
    },
    {
      "epoch": 1.406311637080868,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018065209311294574,
      "loss": 0.0679,
      "step": 5704
    },
    {
      "epoch": 1.4068047337278107,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00018063173629664271,
      "loss": 0.0667,
      "step": 5706
    },
    {
      "epoch": 1.4072978303747534,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018061136992502874,
      "loss": 0.065,
      "step": 5708
    },
    {
      "epoch": 1.4077909270216962,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018059099400051734,
      "loss": 0.0638,
      "step": 5710
    },
    {
      "epoch": 1.4082840236686391,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001805706085255232,
      "loss": 0.0592,
      "step": 5712
    },
    {
      "epoch": 1.4087771203155819,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00018055021350246213,
      "loss": 0.0625,
      "step": 5714
    },
    {
      "epoch": 1.4092702169625246,
      "grad_norm": 0.38671875,
      "learning_rate": 0.000180529808933751,
      "loss": 0.0667,
      "step": 5716
    },
    {
      "epoch": 1.4097633136094674,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001805093948218079,
      "loss": 0.0613,
      "step": 5718
    },
    {
      "epoch": 1.4102564102564101,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000180488971169052,
      "loss": 0.0658,
      "step": 5720
    },
    {
      "epoch": 1.410749506903353,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001804685379779036,
      "loss": 0.0631,
      "step": 5722
    },
    {
      "epoch": 1.4112426035502958,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00018044809525078422,
      "loss": 0.0556,
      "step": 5724
    },
    {
      "epoch": 1.4117357001972386,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00018042764299011635,
      "loss": 0.0601,
      "step": 5726
    },
    {
      "epoch": 1.4122287968441816,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00018040718119832376,
      "loss": 0.0627,
      "step": 5728
    },
    {
      "epoch": 1.4127218934911243,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00018038670987783123,
      "loss": 0.0682,
      "step": 5730
    },
    {
      "epoch": 1.413214990138067,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018036622903106476,
      "loss": 0.069,
      "step": 5732
    },
    {
      "epoch": 1.4137080867850098,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00018034573866045146,
      "loss": 0.0643,
      "step": 5734
    },
    {
      "epoch": 1.4142011834319526,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00018032523876841957,
      "loss": 0.0629,
      "step": 5736
    },
    {
      "epoch": 1.4146942800788955,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00018030472935739834,
      "loss": 0.0715,
      "step": 5738
    },
    {
      "epoch": 1.4151873767258383,
      "grad_norm": 0.40625,
      "learning_rate": 0.00018028421042981836,
      "loss": 0.0681,
      "step": 5740
    },
    {
      "epoch": 1.415680473372781,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018026368198811117,
      "loss": 0.0629,
      "step": 5742
    },
    {
      "epoch": 1.416173570019724,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00018024314403470955,
      "loss": 0.0663,
      "step": 5744
    },
    {
      "epoch": 1.4166666666666667,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001802225965720473,
      "loss": 0.0644,
      "step": 5746
    },
    {
      "epoch": 1.4171597633136095,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018020203960255953,
      "loss": 0.0663,
      "step": 5748
    },
    {
      "epoch": 1.4176528599605522,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00018018147312868222,
      "loss": 0.0601,
      "step": 5750
    },
    {
      "epoch": 1.418145956607495,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018016089715285267,
      "loss": 0.0669,
      "step": 5752
    },
    {
      "epoch": 1.418639053254438,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018014031167750923,
      "loss": 0.0582,
      "step": 5754
    },
    {
      "epoch": 1.4191321499013807,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00018011971670509148,
      "loss": 0.0569,
      "step": 5756
    },
    {
      "epoch": 1.4196252465483234,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001800991122380399,
      "loss": 0.0603,
      "step": 5758
    },
    {
      "epoch": 1.4201183431952662,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00018007849827879634,
      "loss": 0.0546,
      "step": 5760
    },
    {
      "epoch": 1.420611439842209,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00018005787482980358,
      "loss": 0.0651,
      "step": 5762
    },
    {
      "epoch": 1.421104536489152,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018003724189350574,
      "loss": 0.064,
      "step": 5764
    },
    {
      "epoch": 1.4215976331360947,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001800165994723478,
      "loss": 0.0652,
      "step": 5766
    },
    {
      "epoch": 1.4220907297830374,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00017999594756877612,
      "loss": 0.0653,
      "step": 5768
    },
    {
      "epoch": 1.4225838264299804,
      "grad_norm": 0.5390625,
      "learning_rate": 0.000179975286185238,
      "loss": 0.0618,
      "step": 5770
    },
    {
      "epoch": 1.4230769230769231,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00017995461532418193,
      "loss": 0.0685,
      "step": 5772
    },
    {
      "epoch": 1.4235700197238659,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00017993393498805753,
      "loss": 0.0649,
      "step": 5774
    },
    {
      "epoch": 1.4240631163708086,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00017991324517931557,
      "loss": 0.0654,
      "step": 5776
    },
    {
      "epoch": 1.4245562130177514,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00017989254590040785,
      "loss": 0.0648,
      "step": 5778
    },
    {
      "epoch": 1.4250493096646943,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00017987183715378736,
      "loss": 0.0587,
      "step": 5780
    },
    {
      "epoch": 1.425542406311637,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00017985111894190827,
      "loss": 0.068,
      "step": 5782
    },
    {
      "epoch": 1.4260355029585798,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017983039126722572,
      "loss": 0.0671,
      "step": 5784
    },
    {
      "epoch": 1.4265285996055228,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001798096541321961,
      "loss": 0.0646,
      "step": 5786
    },
    {
      "epoch": 1.4270216962524656,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00017978890753927686,
      "loss": 0.0655,
      "step": 5788
    },
    {
      "epoch": 1.4275147928994083,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001797681514909266,
      "loss": 0.0613,
      "step": 5790
    },
    {
      "epoch": 1.428007889546351,
      "grad_norm": 0.462890625,
      "learning_rate": 0.000179747385989605,
      "loss": 0.0627,
      "step": 5792
    },
    {
      "epoch": 1.4285009861932938,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00017972661103777293,
      "loss": 0.0621,
      "step": 5794
    },
    {
      "epoch": 1.4289940828402368,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001797058266378923,
      "loss": 0.0625,
      "step": 5796
    },
    {
      "epoch": 1.4294871794871795,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001796850327924262,
      "loss": 0.0633,
      "step": 5798
    },
    {
      "epoch": 1.4299802761341223,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00017966422950383882,
      "loss": 0.0601,
      "step": 5800
    },
    {
      "epoch": 1.430473372781065,
      "grad_norm": 0.40625,
      "learning_rate": 0.00017964341677459541,
      "loss": 0.0582,
      "step": 5802
    },
    {
      "epoch": 1.4309664694280078,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00017962259460716252,
      "loss": 0.0623,
      "step": 5804
    },
    {
      "epoch": 1.4314595660749507,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00017960176300400757,
      "loss": 0.0616,
      "step": 5806
    },
    {
      "epoch": 1.4319526627218935,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00017958092196759923,
      "loss": 0.0559,
      "step": 5808
    },
    {
      "epoch": 1.4324457593688362,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00017956007150040737,
      "loss": 0.0653,
      "step": 5810
    },
    {
      "epoch": 1.4329388560157792,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00017953921160490278,
      "loss": 0.0601,
      "step": 5812
    },
    {
      "epoch": 1.433431952662722,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00017951834228355757,
      "loss": 0.0643,
      "step": 5814
    },
    {
      "epoch": 1.4339250493096647,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00017949746353884482,
      "loss": 0.0616,
      "step": 5816
    },
    {
      "epoch": 1.4344181459566074,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00017947657537323877,
      "loss": 0.0679,
      "step": 5818
    },
    {
      "epoch": 1.4349112426035502,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001794556777892148,
      "loss": 0.0669,
      "step": 5820
    },
    {
      "epoch": 1.4354043392504932,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001794347707892494,
      "loss": 0.0614,
      "step": 5822
    },
    {
      "epoch": 1.435897435897436,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00017941385437582016,
      "loss": 0.0654,
      "step": 5824
    },
    {
      "epoch": 1.4363905325443787,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00017939292855140575,
      "loss": 0.0671,
      "step": 5826
    },
    {
      "epoch": 1.4368836291913216,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00017937199331848603,
      "loss": 0.0648,
      "step": 5828
    },
    {
      "epoch": 1.4373767258382644,
      "grad_norm": 0.310546875,
      "learning_rate": 0.000179351048679542,
      "loss": 0.0572,
      "step": 5830
    },
    {
      "epoch": 1.4378698224852071,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001793300946370556,
      "loss": 0.0553,
      "step": 5832
    },
    {
      "epoch": 1.4383629191321499,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0001793091311935101,
      "loss": 0.061,
      "step": 5834
    },
    {
      "epoch": 1.4388560157790926,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00017928815835138978,
      "loss": 0.0618,
      "step": 5836
    },
    {
      "epoch": 1.4393491124260356,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00017926717611317994,
      "loss": 0.0608,
      "step": 5838
    },
    {
      "epoch": 1.4398422090729783,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00017924618448136718,
      "loss": 0.0614,
      "step": 5840
    },
    {
      "epoch": 1.440335305719921,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00017922518345843912,
      "loss": 0.0631,
      "step": 5842
    },
    {
      "epoch": 1.4408284023668638,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001792041730468845,
      "loss": 0.0625,
      "step": 5844
    },
    {
      "epoch": 1.4413214990138066,
      "grad_norm": 0.515625,
      "learning_rate": 0.00017918315324919312,
      "loss": 0.0596,
      "step": 5846
    },
    {
      "epoch": 1.4418145956607495,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000179162124067856,
      "loss": 0.061,
      "step": 5848
    },
    {
      "epoch": 1.4423076923076923,
      "grad_norm": 0.875,
      "learning_rate": 0.00017914108550536517,
      "loss": 0.0616,
      "step": 5850
    },
    {
      "epoch": 1.442800788954635,
      "grad_norm": 0.40625,
      "learning_rate": 0.00017912003756421384,
      "loss": 0.0561,
      "step": 5852
    },
    {
      "epoch": 1.443293885601578,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00017909898024689633,
      "loss": 0.0611,
      "step": 5854
    },
    {
      "epoch": 1.4437869822485208,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000179077913555908,
      "loss": 0.0556,
      "step": 5856
    },
    {
      "epoch": 1.4442800788954635,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00017905683749374544,
      "loss": 0.0625,
      "step": 5858
    },
    {
      "epoch": 1.4447731755424063,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00017903575206290625,
      "loss": 0.0587,
      "step": 5860
    },
    {
      "epoch": 1.445266272189349,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001790146572658891,
      "loss": 0.0619,
      "step": 5862
    },
    {
      "epoch": 1.445759368836292,
      "grad_norm": 0.53125,
      "learning_rate": 0.00017899355310519393,
      "loss": 0.0613,
      "step": 5864
    },
    {
      "epoch": 1.4462524654832347,
      "grad_norm": 0.46875,
      "learning_rate": 0.00017897243958332168,
      "loss": 0.0602,
      "step": 5866
    },
    {
      "epoch": 1.4467455621301775,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001789513167027744,
      "loss": 0.0558,
      "step": 5868
    },
    {
      "epoch": 1.4472386587771204,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00017893018446605533,
      "loss": 0.0684,
      "step": 5870
    },
    {
      "epoch": 1.4477317554240632,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00017890904287566866,
      "loss": 0.0575,
      "step": 5872
    },
    {
      "epoch": 1.448224852071006,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00017888789193411982,
      "loss": 0.058,
      "step": 5874
    },
    {
      "epoch": 1.4487179487179487,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001788667316439154,
      "loss": 0.057,
      "step": 5876
    },
    {
      "epoch": 1.4492110453648914,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001788455620075629,
      "loss": 0.0629,
      "step": 5878
    },
    {
      "epoch": 1.4497041420118344,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001788243830275711,
      "loss": 0.0605,
      "step": 5880
    },
    {
      "epoch": 1.4501972386587771,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001788031947064498,
      "loss": 0.0614,
      "step": 5882
    },
    {
      "epoch": 1.45069033530572,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00017878199704670996,
      "loss": 0.0672,
      "step": 5884
    },
    {
      "epoch": 1.4511834319526626,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001787607900508636,
      "loss": 0.0541,
      "step": 5886
    },
    {
      "epoch": 1.4516765285996054,
      "grad_norm": 0.515625,
      "learning_rate": 0.00017873957372142388,
      "loss": 0.0628,
      "step": 5888
    },
    {
      "epoch": 1.4521696252465484,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00017871834806090501,
      "loss": 0.0672,
      "step": 5890
    },
    {
      "epoch": 1.452662721893491,
      "grad_norm": 0.546875,
      "learning_rate": 0.00017869711307182245,
      "loss": 0.0644,
      "step": 5892
    },
    {
      "epoch": 1.4531558185404339,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00017867586875669255,
      "loss": 0.0559,
      "step": 5894
    },
    {
      "epoch": 1.4536489151873768,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00017865461511803297,
      "loss": 0.0601,
      "step": 5896
    },
    {
      "epoch": 1.4541420118343196,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00017863335215836235,
      "loss": 0.0628,
      "step": 5898
    },
    {
      "epoch": 1.4546351084812623,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00017861207988020044,
      "loss": 0.0648,
      "step": 5900
    },
    {
      "epoch": 1.455128205128205,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00017859079828606817,
      "loss": 0.0598,
      "step": 5902
    },
    {
      "epoch": 1.4556213017751478,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00017856950737848748,
      "loss": 0.0605,
      "step": 5904
    },
    {
      "epoch": 1.4561143984220908,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001785482071599815,
      "loss": 0.057,
      "step": 5906
    },
    {
      "epoch": 1.4566074950690335,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00017852689763307443,
      "loss": 0.0613,
      "step": 5908
    },
    {
      "epoch": 1.4571005917159763,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00017850557880029158,
      "loss": 0.0629,
      "step": 5910
    },
    {
      "epoch": 1.4575936883629192,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00017848425066415926,
      "loss": 0.0635,
      "step": 5912
    },
    {
      "epoch": 1.458086785009862,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00017846291322720507,
      "loss": 0.0643,
      "step": 5914
    },
    {
      "epoch": 1.4585798816568047,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00017844156649195759,
      "loss": 0.0584,
      "step": 5916
    },
    {
      "epoch": 1.4590729783037475,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00017842021046094648,
      "loss": 0.0557,
      "step": 5918
    },
    {
      "epoch": 1.4595660749506902,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00017839884513670264,
      "loss": 0.0615,
      "step": 5920
    },
    {
      "epoch": 1.4600591715976332,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001783774705217579,
      "loss": 0.0593,
      "step": 5922
    },
    {
      "epoch": 1.460552268244576,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001783560866186453,
      "loss": 0.0604,
      "step": 5924
    },
    {
      "epoch": 1.4610453648915187,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00017833469342989894,
      "loss": 0.057,
      "step": 5926
    },
    {
      "epoch": 1.4615384615384617,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00017831329095805408,
      "loss": 0.0609,
      "step": 5928
    },
    {
      "epoch": 1.4620315581854042,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00017829187920564697,
      "loss": 0.0628,
      "step": 5930
    },
    {
      "epoch": 1.4625246548323472,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017827045817521504,
      "loss": 0.063,
      "step": 5932
    },
    {
      "epoch": 1.46301775147929,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00017824902786929683,
      "loss": 0.0599,
      "step": 5934
    },
    {
      "epoch": 1.4635108481262327,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00017822758829043193,
      "loss": 0.0715,
      "step": 5936
    },
    {
      "epoch": 1.4640039447731756,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00017820613944116102,
      "loss": 0.06,
      "step": 5938
    },
    {
      "epoch": 1.4644970414201184,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00017818468132402597,
      "loss": 0.067,
      "step": 5940
    },
    {
      "epoch": 1.4649901380670611,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00017816321394156966,
      "loss": 0.0627,
      "step": 5942
    },
    {
      "epoch": 1.4654832347140039,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00017814173729633606,
      "loss": 0.0631,
      "step": 5944
    },
    {
      "epoch": 1.4659763313609466,
      "grad_norm": 0.5625,
      "learning_rate": 0.00017812025139087031,
      "loss": 0.0616,
      "step": 5946
    },
    {
      "epoch": 1.4664694280078896,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00017809875622771864,
      "loss": 0.064,
      "step": 5948
    },
    {
      "epoch": 1.4669625246548323,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001780772518094283,
      "loss": 0.0592,
      "step": 5950
    },
    {
      "epoch": 1.467455621301775,
      "grad_norm": 0.375,
      "learning_rate": 0.00017805573813854767,
      "loss": 0.063,
      "step": 5952
    },
    {
      "epoch": 1.467948717948718,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00017803421521762626,
      "loss": 0.0658,
      "step": 5954
    },
    {
      "epoch": 1.4684418145956608,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00017801268304921467,
      "loss": 0.0677,
      "step": 5956
    },
    {
      "epoch": 1.4689349112426036,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017799114163586456,
      "loss": 0.0609,
      "step": 5958
    },
    {
      "epoch": 1.4694280078895463,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017796959098012873,
      "loss": 0.0571,
      "step": 5960
    },
    {
      "epoch": 1.469921104536489,
      "grad_norm": 0.296875,
      "learning_rate": 0.00017794803108456104,
      "loss": 0.0619,
      "step": 5962
    },
    {
      "epoch": 1.470414201183432,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00017792646195171648,
      "loss": 0.061,
      "step": 5964
    },
    {
      "epoch": 1.4709072978303748,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017790488358415107,
      "loss": 0.0594,
      "step": 5966
    },
    {
      "epoch": 1.4714003944773175,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00017788329598442197,
      "loss": 0.061,
      "step": 5968
    },
    {
      "epoch": 1.4718934911242605,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001778616991550875,
      "loss": 0.0587,
      "step": 5970
    },
    {
      "epoch": 1.4723865877712032,
      "grad_norm": 0.375,
      "learning_rate": 0.00017784009309870693,
      "loss": 0.0613,
      "step": 5972
    },
    {
      "epoch": 1.472879684418146,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001778184778178407,
      "loss": 0.0631,
      "step": 5974
    },
    {
      "epoch": 1.4733727810650887,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00017779685331505043,
      "loss": 0.0566,
      "step": 5976
    },
    {
      "epoch": 1.4738658777120315,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00017777521959289866,
      "loss": 0.0577,
      "step": 5978
    },
    {
      "epoch": 1.4743589743589745,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001777535766539491,
      "loss": 0.0574,
      "step": 5980
    },
    {
      "epoch": 1.4748520710059172,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00017773192450076656,
      "loss": 0.0607,
      "step": 5982
    },
    {
      "epoch": 1.47534516765286,
      "grad_norm": 0.341796875,
      "learning_rate": 0.000177710263135917,
      "loss": 0.0622,
      "step": 5984
    },
    {
      "epoch": 1.4758382642998027,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001776885925619674,
      "loss": 0.058,
      "step": 5986
    },
    {
      "epoch": 1.4763313609467454,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00017766691278148583,
      "loss": 0.0598,
      "step": 5988
    },
    {
      "epoch": 1.4768244575936884,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00017764522379704143,
      "loss": 0.0608,
      "step": 5990
    },
    {
      "epoch": 1.4773175542406312,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017762352561120448,
      "loss": 0.0611,
      "step": 5992
    },
    {
      "epoch": 1.477810650887574,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00017760181822654638,
      "loss": 0.0565,
      "step": 5994
    },
    {
      "epoch": 1.4783037475345169,
      "grad_norm": 0.5,
      "learning_rate": 0.00017758010164563954,
      "loss": 0.064,
      "step": 5996
    },
    {
      "epoch": 1.4787968441814596,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001775583758710575,
      "loss": 0.0593,
      "step": 5998
    },
    {
      "epoch": 1.4792899408284024,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001775366409053749,
      "loss": 0.0567,
      "step": 6000
    },
    {
      "epoch": 1.4797830374753451,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00017751489675116742,
      "loss": 0.0581,
      "step": 6002
    },
    {
      "epoch": 1.4802761341222879,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001774931434110119,
      "loss": 0.0616,
      "step": 6004
    },
    {
      "epoch": 1.4807692307692308,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00017747138088748622,
      "loss": 0.0608,
      "step": 6006
    },
    {
      "epoch": 1.4812623274161736,
      "grad_norm": 0.578125,
      "learning_rate": 0.00017744960918316936,
      "loss": 0.0625,
      "step": 6008
    },
    {
      "epoch": 1.4817554240631163,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001774278283006414,
      "loss": 0.0593,
      "step": 6010
    },
    {
      "epoch": 1.4822485207100593,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00017740603824248347,
      "loss": 0.0575,
      "step": 6012
    },
    {
      "epoch": 1.482741617357002,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00017738423901127783,
      "loss": 0.0539,
      "step": 6014
    },
    {
      "epoch": 1.4832347140039448,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001773624306096078,
      "loss": 0.0624,
      "step": 6016
    },
    {
      "epoch": 1.4837278106508875,
      "grad_norm": 0.375,
      "learning_rate": 0.00017734061304005785,
      "loss": 0.0592,
      "step": 6018
    },
    {
      "epoch": 1.4842209072978303,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001773187863052134,
      "loss": 0.0592,
      "step": 6020
    },
    {
      "epoch": 1.4847140039447733,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001772969504076611,
      "loss": 0.0588,
      "step": 6022
    },
    {
      "epoch": 1.485207100591716,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001772751053499886,
      "loss": 0.0611,
      "step": 6024
    },
    {
      "epoch": 1.4857001972386588,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00017725325113478464,
      "loss": 0.0583,
      "step": 6026
    },
    {
      "epoch": 1.4861932938856015,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00017723138776463913,
      "loss": 0.0534,
      "step": 6028
    },
    {
      "epoch": 1.4866863905325443,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00017720951524214293,
      "loss": 0.0584,
      "step": 6030
    },
    {
      "epoch": 1.4871794871794872,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001771876335698881,
      "loss": 0.0616,
      "step": 6032
    },
    {
      "epoch": 1.48767258382643,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00017716574275046773,
      "loss": 0.0612,
      "step": 6034
    },
    {
      "epoch": 1.4881656804733727,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00017714384278647605,
      "loss": 0.0568,
      "step": 6036
    },
    {
      "epoch": 1.4886587771203157,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00017712193368050823,
      "loss": 0.0603,
      "step": 6038
    },
    {
      "epoch": 1.4891518737672584,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001771000154351607,
      "loss": 0.0606,
      "step": 6040
    },
    {
      "epoch": 1.4896449704142012,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00017707808805303085,
      "loss": 0.0642,
      "step": 6042
    },
    {
      "epoch": 1.490138067061144,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00017705615153671724,
      "loss": 0.0546,
      "step": 6044
    },
    {
      "epoch": 1.4906311637080867,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017703420588881946,
      "loss": 0.0587,
      "step": 6046
    },
    {
      "epoch": 1.4911242603550297,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00017701225111193815,
      "loss": 0.0536,
      "step": 6048
    },
    {
      "epoch": 1.4916173570019724,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001769902872086751,
      "loss": 0.0582,
      "step": 6050
    },
    {
      "epoch": 1.4921104536489151,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001769683141816332,
      "loss": 0.0592,
      "step": 6052
    },
    {
      "epoch": 1.4926035502958581,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001769463320334163,
      "loss": 0.06,
      "step": 6054
    },
    {
      "epoch": 1.4930966469428009,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001769243407666294,
      "loss": 0.0585,
      "step": 6056
    },
    {
      "epoch": 1.4935897435897436,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00017690234038387868,
      "loss": 0.0618,
      "step": 6058
    },
    {
      "epoch": 1.4940828402366864,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00017688033088777126,
      "loss": 0.0503,
      "step": 6060
    },
    {
      "epoch": 1.494575936883629,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00017685831228091535,
      "loss": 0.0594,
      "step": 6062
    },
    {
      "epoch": 1.495069033530572,
      "grad_norm": 0.359375,
      "learning_rate": 0.00017683628456592035,
      "loss": 0.0589,
      "step": 6064
    },
    {
      "epoch": 1.4955621301775148,
      "grad_norm": 0.53125,
      "learning_rate": 0.00017681424774539662,
      "loss": 0.067,
      "step": 6066
    },
    {
      "epoch": 1.4960552268244576,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017679220182195563,
      "loss": 0.0567,
      "step": 6068
    },
    {
      "epoch": 1.4965483234714003,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00017677014679821,
      "loss": 0.0654,
      "step": 6070
    },
    {
      "epoch": 1.497041420118343,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00017674808267677334,
      "loss": 0.0604,
      "step": 6072
    },
    {
      "epoch": 1.497534516765286,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00017672600946026036,
      "loss": 0.0568,
      "step": 6074
    },
    {
      "epoch": 1.4980276134122288,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001767039271512869,
      "loss": 0.0596,
      "step": 6076
    },
    {
      "epoch": 1.4985207100591715,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001766818357524698,
      "loss": 0.0604,
      "step": 6078
    },
    {
      "epoch": 1.4990138067061145,
      "grad_norm": 0.53125,
      "learning_rate": 0.00017665973526642703,
      "loss": 0.0594,
      "step": 6080
    },
    {
      "epoch": 1.4995069033530573,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00017663762569577763,
      "loss": 0.0705,
      "step": 6082
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00017661550704314166,
      "loss": 0.0694,
      "step": 6084
    },
    {
      "epoch": 1.5004930966469427,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00017659337931114037,
      "loss": 0.0585,
      "step": 6086
    },
    {
      "epoch": 1.5009861932938855,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00017657124250239597,
      "loss": 0.0563,
      "step": 6088
    },
    {
      "epoch": 1.5014792899408285,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00017654909661953184,
      "loss": 0.0601,
      "step": 6090
    },
    {
      "epoch": 1.5019723865877712,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00017652694166517234,
      "loss": 0.0623,
      "step": 6092
    },
    {
      "epoch": 1.502465483234714,
      "grad_norm": 0.4375,
      "learning_rate": 0.00017650477764194296,
      "loss": 0.0601,
      "step": 6094
    },
    {
      "epoch": 1.502958579881657,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00017648260455247034,
      "loss": 0.0584,
      "step": 6096
    },
    {
      "epoch": 1.5034516765285995,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017646042239938202,
      "loss": 0.0652,
      "step": 6098
    },
    {
      "epoch": 1.5039447731755424,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00017643823118530675,
      "loss": 0.0574,
      "step": 6100
    },
    {
      "epoch": 1.5044378698224852,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001764160309128743,
      "loss": 0.0588,
      "step": 6102
    },
    {
      "epoch": 1.504930966469428,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00017639382158471555,
      "loss": 0.0574,
      "step": 6104
    },
    {
      "epoch": 1.505424063116371,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001763716032034624,
      "loss": 0.057,
      "step": 6106
    },
    {
      "epoch": 1.5059171597633136,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001763493757717479,
      "loss": 0.0625,
      "step": 6108
    },
    {
      "epoch": 1.5064102564102564,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017632713929220605,
      "loss": 0.0601,
      "step": 6110
    },
    {
      "epoch": 1.5069033530571994,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00017630489376747204,
      "loss": 0.0626,
      "step": 6112
    },
    {
      "epoch": 1.5073964497041419,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00017628263920018214,
      "loss": 0.0622,
      "step": 6114
    },
    {
      "epoch": 1.5078895463510849,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00017626037559297352,
      "loss": 0.0559,
      "step": 6116
    },
    {
      "epoch": 1.5083826429980276,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00017623810294848465,
      "loss": 0.0575,
      "step": 6118
    },
    {
      "epoch": 1.5088757396449703,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001762158212693549,
      "loss": 0.0591,
      "step": 6120
    },
    {
      "epoch": 1.5093688362919133,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001761935305582248,
      "loss": 0.058,
      "step": 6122
    },
    {
      "epoch": 1.509861932938856,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017617123081773591,
      "loss": 0.0521,
      "step": 6124
    },
    {
      "epoch": 1.5103550295857988,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00017614892205053092,
      "loss": 0.0593,
      "step": 6126
    },
    {
      "epoch": 1.5108481262327418,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00017612660425925347,
      "loss": 0.0624,
      "step": 6128
    },
    {
      "epoch": 1.5113412228796843,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00017610427744654838,
      "loss": 0.06,
      "step": 6130
    },
    {
      "epoch": 1.5118343195266273,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001760819416150615,
      "loss": 0.0588,
      "step": 6132
    },
    {
      "epoch": 1.51232741617357,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00017605959676743977,
      "loss": 0.0587,
      "step": 6134
    },
    {
      "epoch": 1.5128205128205128,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00017603724290633115,
      "loss": 0.0598,
      "step": 6136
    },
    {
      "epoch": 1.5133136094674557,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017601488003438472,
      "loss": 0.0644,
      "step": 6138
    },
    {
      "epoch": 1.5138067061143983,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00017599250815425056,
      "loss": 0.0625,
      "step": 6140
    },
    {
      "epoch": 1.5142998027613412,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00017597012726857994,
      "loss": 0.0552,
      "step": 6142
    },
    {
      "epoch": 1.514792899408284,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00017594773738002504,
      "loss": 0.0592,
      "step": 6144
    },
    {
      "epoch": 1.5152859960552267,
      "grad_norm": 0.65625,
      "learning_rate": 0.00017592533849123925,
      "loss": 0.0589,
      "step": 6146
    },
    {
      "epoch": 1.5157790927021697,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001759029306048769,
      "loss": 0.0571,
      "step": 6148
    },
    {
      "epoch": 1.5162721893491125,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00017588051372359353,
      "loss": 0.0596,
      "step": 6150
    },
    {
      "epoch": 1.5167652859960552,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00017585808785004558,
      "loss": 0.063,
      "step": 6152
    },
    {
      "epoch": 1.5172583826429982,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00017583565298689071,
      "loss": 0.0586,
      "step": 6154
    },
    {
      "epoch": 1.5177514792899407,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00017581320913678754,
      "loss": 0.0583,
      "step": 6156
    },
    {
      "epoch": 1.5182445759368837,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00017579075630239581,
      "loss": 0.0612,
      "step": 6158
    },
    {
      "epoch": 1.5187376725838264,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00017576829448637629,
      "loss": 0.055,
      "step": 6160
    },
    {
      "epoch": 1.5192307692307692,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017574582369139083,
      "loss": 0.0605,
      "step": 6162
    },
    {
      "epoch": 1.5197238658777121,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00017572334392010236,
      "loss": 0.0614,
      "step": 6164
    },
    {
      "epoch": 1.5202169625246549,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00017570085517517483,
      "loss": 0.0598,
      "step": 6166
    },
    {
      "epoch": 1.5207100591715976,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00017567835745927333,
      "loss": 0.0601,
      "step": 6168
    },
    {
      "epoch": 1.5212031558185406,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00017565585077506394,
      "loss": 0.0542,
      "step": 6170
    },
    {
      "epoch": 1.5216962524654831,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00017563333512521378,
      "loss": 0.0562,
      "step": 6172
    },
    {
      "epoch": 1.522189349112426,
      "grad_norm": 0.328125,
      "learning_rate": 0.00017561081051239118,
      "loss": 0.0566,
      "step": 6174
    },
    {
      "epoch": 1.5226824457593688,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00017558827693926534,
      "loss": 0.0602,
      "step": 6176
    },
    {
      "epoch": 1.5231755424063116,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00017556573440850667,
      "loss": 0.0561,
      "step": 6178
    },
    {
      "epoch": 1.5236686390532546,
      "grad_norm": 0.546875,
      "learning_rate": 0.00017554318292278657,
      "loss": 0.0599,
      "step": 6180
    },
    {
      "epoch": 1.524161735700197,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001755206224847775,
      "loss": 0.0632,
      "step": 6182
    },
    {
      "epoch": 1.52465483234714,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017549805309715304,
      "loss": 0.0615,
      "step": 6184
    },
    {
      "epoch": 1.5251479289940828,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017547547476258775,
      "loss": 0.0577,
      "step": 6186
    },
    {
      "epoch": 1.5256410256410255,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001754528874837573,
      "loss": 0.0604,
      "step": 6188
    },
    {
      "epoch": 1.5261341222879685,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00017543029126333836,
      "loss": 0.0589,
      "step": 6190
    },
    {
      "epoch": 1.5266272189349113,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00017540768610400882,
      "loss": 0.0605,
      "step": 6192
    },
    {
      "epoch": 1.527120315581854,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00017538507200844742,
      "loss": 0.0621,
      "step": 6194
    },
    {
      "epoch": 1.527613412228797,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001753624489793341,
      "loss": 0.0541,
      "step": 6196
    },
    {
      "epoch": 1.5281065088757395,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001753398170193498,
      "loss": 0.0557,
      "step": 6198
    },
    {
      "epoch": 1.5285996055226825,
      "grad_norm": 0.359375,
      "learning_rate": 0.00017531717613117654,
      "loss": 0.0589,
      "step": 6200
    },
    {
      "epoch": 1.5290927021696252,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00017529452631749742,
      "loss": 0.0592,
      "step": 6202
    },
    {
      "epoch": 1.529585798816568,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00017527186758099653,
      "loss": 0.053,
      "step": 6204
    },
    {
      "epoch": 1.530078895463511,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000175249199924359,
      "loss": 0.0633,
      "step": 6206
    },
    {
      "epoch": 1.5305719921104537,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00017522652335027122,
      "loss": 0.0621,
      "step": 6208
    },
    {
      "epoch": 1.5310650887573964,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00017520383786142038,
      "loss": 0.0589,
      "step": 6210
    },
    {
      "epoch": 1.5315581854043394,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001751811434604949,
      "loss": 0.0661,
      "step": 6212
    },
    {
      "epoch": 1.532051282051282,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00017515844015018414,
      "loss": 0.0581,
      "step": 6214
    },
    {
      "epoch": 1.532544378698225,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001751357279331786,
      "loss": 0.0573,
      "step": 6216
    },
    {
      "epoch": 1.5330374753451677,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00017511300681216977,
      "loss": 0.0564,
      "step": 6218
    },
    {
      "epoch": 1.5335305719921104,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00017509027678985027,
      "loss": 0.0564,
      "step": 6220
    },
    {
      "epoch": 1.5340236686390534,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001750675378689137,
      "loss": 0.0562,
      "step": 6222
    },
    {
      "epoch": 1.534516765285996,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001750447900520548,
      "loss": 0.054,
      "step": 6224
    },
    {
      "epoch": 1.5350098619329389,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001750220333419693,
      "loss": 0.0524,
      "step": 6226
    },
    {
      "epoch": 1.5355029585798816,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017499926774135395,
      "loss": 0.0618,
      "step": 6228
    },
    {
      "epoch": 1.5359960552268244,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00017497649325290662,
      "loss": 0.058,
      "step": 6230
    },
    {
      "epoch": 1.5364891518737673,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00017495370987932624,
      "loss": 0.0574,
      "step": 6232
    },
    {
      "epoch": 1.53698224852071,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00017493091762331276,
      "loss": 0.0544,
      "step": 6234
    },
    {
      "epoch": 1.5374753451676528,
      "grad_norm": 0.421875,
      "learning_rate": 0.00017490811648756716,
      "loss": 0.0554,
      "step": 6236
    },
    {
      "epoch": 1.5379684418145958,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00017488530647479154,
      "loss": 0.0562,
      "step": 6238
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00017486248758768897,
      "loss": 0.0562,
      "step": 6240
    },
    {
      "epoch": 1.5389546351084813,
      "grad_norm": 0.65625,
      "learning_rate": 0.00017483965982896367,
      "loss": 0.0604,
      "step": 6242
    },
    {
      "epoch": 1.539447731755424,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00017481682320132083,
      "loss": 0.0629,
      "step": 6244
    },
    {
      "epoch": 1.5399408284023668,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001747939777074667,
      "loss": 0.0618,
      "step": 6246
    },
    {
      "epoch": 1.5404339250493098,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00017477112335010864,
      "loss": 0.0574,
      "step": 6248
    },
    {
      "epoch": 1.5409270216962525,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00017474826013195495,
      "loss": 0.0617,
      "step": 6250
    },
    {
      "epoch": 1.5414201183431953,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00017472538805571514,
      "loss": 0.057,
      "step": 6252
    },
    {
      "epoch": 1.5419132149901382,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001747025071240996,
      "loss": 0.0568,
      "step": 6254
    },
    {
      "epoch": 1.5424063116370808,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001746796173398199,
      "loss": 0.0583,
      "step": 6256
    },
    {
      "epoch": 1.5428994082840237,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017465671870558856,
      "loss": 0.0609,
      "step": 6258
    },
    {
      "epoch": 1.5433925049309665,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00017463381122411926,
      "loss": 0.057,
      "step": 6260
    },
    {
      "epoch": 1.5438856015779092,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00017461089489812661,
      "loss": 0.0545,
      "step": 6262
    },
    {
      "epoch": 1.5443786982248522,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00017458796973032637,
      "loss": 0.0546,
      "step": 6264
    },
    {
      "epoch": 1.5448717948717947,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00017456503572343523,
      "loss": 0.054,
      "step": 6266
    },
    {
      "epoch": 1.5453648915187377,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017454209288017105,
      "loss": 0.0599,
      "step": 6268
    },
    {
      "epoch": 1.5458579881656804,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001745191412032527,
      "loss": 0.0514,
      "step": 6270
    },
    {
      "epoch": 1.5463510848126232,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00017449618069540003,
      "loss": 0.0613,
      "step": 6272
    },
    {
      "epoch": 1.5468441814595661,
      "grad_norm": 0.40625,
      "learning_rate": 0.000174473211359334,
      "loss": 0.06,
      "step": 6274
    },
    {
      "epoch": 1.547337278106509,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00017445023319777664,
      "loss": 0.0633,
      "step": 6276
    },
    {
      "epoch": 1.5478303747534516,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00017442724621345097,
      "loss": 0.0552,
      "step": 6278
    },
    {
      "epoch": 1.5483234714003946,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00017440425040908108,
      "loss": 0.0614,
      "step": 6280
    },
    {
      "epoch": 1.5488165680473371,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00017438124578739208,
      "loss": 0.0592,
      "step": 6282
    },
    {
      "epoch": 1.54930966469428,
      "grad_norm": 0.328125,
      "learning_rate": 0.00017435823235111017,
      "loss": 0.0554,
      "step": 6284
    },
    {
      "epoch": 1.5498027613412229,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00017433521010296255,
      "loss": 0.0594,
      "step": 6286
    },
    {
      "epoch": 1.5502958579881656,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00017431217904567752,
      "loss": 0.0633,
      "step": 6288
    },
    {
      "epoch": 1.5507889546351086,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00017428913918198435,
      "loss": 0.0569,
      "step": 6290
    },
    {
      "epoch": 1.5512820512820513,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00017426609051461345,
      "loss": 0.0525,
      "step": 6292
    },
    {
      "epoch": 1.551775147928994,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00017424303304629612,
      "loss": 0.0578,
      "step": 6294
    },
    {
      "epoch": 1.552268244575937,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00017421996677976487,
      "loss": 0.0594,
      "step": 6296
    },
    {
      "epoch": 1.5527613412228796,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00017419689171775318,
      "loss": 0.0574,
      "step": 6298
    },
    {
      "epoch": 1.5532544378698225,
      "grad_norm": 0.375,
      "learning_rate": 0.00017417380786299556,
      "loss": 0.062,
      "step": 6300
    },
    {
      "epoch": 1.5537475345167653,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00017415071521822756,
      "loss": 0.0554,
      "step": 6302
    },
    {
      "epoch": 1.554240631163708,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001741276137861858,
      "loss": 0.0558,
      "step": 6304
    },
    {
      "epoch": 1.554733727810651,
      "grad_norm": 0.453125,
      "learning_rate": 0.00017410450356960795,
      "loss": 0.0552,
      "step": 6306
    },
    {
      "epoch": 1.5552268244575935,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017408138457123266,
      "loss": 0.0594,
      "step": 6308
    },
    {
      "epoch": 1.5557199211045365,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001740582567937997,
      "loss": 0.0563,
      "step": 6310
    },
    {
      "epoch": 1.5562130177514792,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00017403512024004983,
      "loss": 0.0568,
      "step": 6312
    },
    {
      "epoch": 1.556706114398422,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00017401197491272483,
      "loss": 0.0607,
      "step": 6314
    },
    {
      "epoch": 1.557199211045365,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00017398882081456755,
      "loss": 0.0571,
      "step": 6316
    },
    {
      "epoch": 1.5576923076923077,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00017396565794832193,
      "loss": 0.0574,
      "step": 6318
    },
    {
      "epoch": 1.5581854043392505,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00017394248631673288,
      "loss": 0.0591,
      "step": 6320
    },
    {
      "epoch": 1.5586785009861934,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017391930592254636,
      "loss": 0.0574,
      "step": 6322
    },
    {
      "epoch": 1.559171597633136,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00017389611676850938,
      "loss": 0.0619,
      "step": 6324
    },
    {
      "epoch": 1.559664694280079,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00017387291885736996,
      "loss": 0.0542,
      "step": 6326
    },
    {
      "epoch": 1.5601577909270217,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001738497121918772,
      "loss": 0.0557,
      "step": 6328
    },
    {
      "epoch": 1.5606508875739644,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017382649677478122,
      "loss": 0.0618,
      "step": 6330
    },
    {
      "epoch": 1.5611439842209074,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00017380327260883317,
      "loss": 0.0608,
      "step": 6332
    },
    {
      "epoch": 1.5616370808678501,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00017378003969678528,
      "loss": 0.0559,
      "step": 6334
    },
    {
      "epoch": 1.5621301775147929,
      "grad_norm": 0.421875,
      "learning_rate": 0.00017375679804139076,
      "loss": 0.0549,
      "step": 6336
    },
    {
      "epoch": 1.5626232741617359,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00017373354764540385,
      "loss": 0.0603,
      "step": 6338
    },
    {
      "epoch": 1.5631163708086784,
      "grad_norm": 0.515625,
      "learning_rate": 0.00017371028851157985,
      "loss": 0.0611,
      "step": 6340
    },
    {
      "epoch": 1.5636094674556213,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00017368702064267516,
      "loss": 0.0478,
      "step": 6342
    },
    {
      "epoch": 1.564102564102564,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00017366374404144706,
      "loss": 0.0556,
      "step": 6344
    },
    {
      "epoch": 1.5645956607495068,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000173640458710654,
      "loss": 0.056,
      "step": 6346
    },
    {
      "epoch": 1.5650887573964498,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00017361716465305547,
      "loss": 0.0577,
      "step": 6348
    },
    {
      "epoch": 1.5655818540433923,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001735938618714119,
      "loss": 0.0542,
      "step": 6350
    },
    {
      "epoch": 1.5660749506903353,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00017357055036848476,
      "loss": 0.0551,
      "step": 6352
    },
    {
      "epoch": 1.566568047337278,
      "grad_norm": 0.3125,
      "learning_rate": 0.00017354723014703665,
      "loss": 0.055,
      "step": 6354
    },
    {
      "epoch": 1.5670611439842208,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00017352390120983118,
      "loss": 0.0572,
      "step": 6356
    },
    {
      "epoch": 1.5675542406311638,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00017350056355963287,
      "loss": 0.0583,
      "step": 6358
    },
    {
      "epoch": 1.5680473372781065,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001734772171992074,
      "loss": 0.0573,
      "step": 6360
    },
    {
      "epoch": 1.5685404339250493,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00017345386213132147,
      "loss": 0.057,
      "step": 6362
    },
    {
      "epoch": 1.5690335305719922,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00017343049835874275,
      "loss": 0.0547,
      "step": 6364
    },
    {
      "epoch": 1.5695266272189348,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00017340712588424,
      "loss": 0.0564,
      "step": 6366
    },
    {
      "epoch": 1.5700197238658777,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00017338374471058296,
      "loss": 0.0612,
      "step": 6368
    },
    {
      "epoch": 1.5705128205128205,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00017336035484054248,
      "loss": 0.0606,
      "step": 6370
    },
    {
      "epoch": 1.5710059171597632,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00017333695627689034,
      "loss": 0.0643,
      "step": 6372
    },
    {
      "epoch": 1.5714990138067062,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00017331354902239943,
      "loss": 0.0645,
      "step": 6374
    },
    {
      "epoch": 1.571992110453649,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001732901330798436,
      "loss": 0.0552,
      "step": 6376
    },
    {
      "epoch": 1.5724852071005917,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00017326670845199785,
      "loss": 0.0577,
      "step": 6378
    },
    {
      "epoch": 1.5729783037475347,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00017324327514163802,
      "loss": 0.0598,
      "step": 6380
    },
    {
      "epoch": 1.5734714003944772,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001732198331515412,
      "loss": 0.0605,
      "step": 6382
    },
    {
      "epoch": 1.5739644970414202,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00017319638248448533,
      "loss": 0.0597,
      "step": 6384
    },
    {
      "epoch": 1.574457593688363,
      "grad_norm": 0.328125,
      "learning_rate": 0.00017317292314324945,
      "loss": 0.0604,
      "step": 6386
    },
    {
      "epoch": 1.5749506903353057,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00017314945513061364,
      "loss": 0.0596,
      "step": 6388
    },
    {
      "epoch": 1.5754437869822486,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000173125978449359,
      "loss": 0.0525,
      "step": 6390
    },
    {
      "epoch": 1.5759368836291914,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00017310249310226758,
      "loss": 0.0592,
      "step": 6392
    },
    {
      "epoch": 1.5764299802761341,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001730789990921226,
      "loss": 0.0586,
      "step": 6394
    },
    {
      "epoch": 1.5769230769230769,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00017305549642170819,
      "loss": 0.0615,
      "step": 6396
    },
    {
      "epoch": 1.5774161735700196,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001730319850938096,
      "loss": 0.0535,
      "step": 6398
    },
    {
      "epoch": 1.5779092702169626,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00017300846511121302,
      "loss": 0.0619,
      "step": 6400
    },
    {
      "epoch": 1.5784023668639053,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001729849364767057,
      "loss": 0.0636,
      "step": 6402
    },
    {
      "epoch": 1.578895463510848,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00017296139919307585,
      "loss": 0.0618,
      "step": 6404
    },
    {
      "epoch": 1.579388560157791,
      "grad_norm": 0.4375,
      "learning_rate": 0.00017293785326311288,
      "loss": 0.0607,
      "step": 6406
    },
    {
      "epoch": 1.5798816568047336,
      "grad_norm": 0.390625,
      "learning_rate": 0.00017291429868960707,
      "loss": 0.0557,
      "step": 6408
    },
    {
      "epoch": 1.5803747534516766,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00017289073547534976,
      "loss": 0.0608,
      "step": 6410
    },
    {
      "epoch": 1.5808678500986193,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00017286716362313332,
      "loss": 0.0567,
      "step": 6412
    },
    {
      "epoch": 1.581360946745562,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00017284358313575116,
      "loss": 0.0599,
      "step": 6414
    },
    {
      "epoch": 1.581854043392505,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00017281999401599766,
      "loss": 0.0548,
      "step": 6416
    },
    {
      "epoch": 1.5823471400394478,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001727963962666683,
      "loss": 0.0591,
      "step": 6418
    },
    {
      "epoch": 1.5828402366863905,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00017277278989055958,
      "loss": 0.0554,
      "step": 6420
    },
    {
      "epoch": 1.5833333333333335,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00017274917489046892,
      "loss": 0.0538,
      "step": 6422
    },
    {
      "epoch": 1.583826429980276,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00017272555126919488,
      "loss": 0.0548,
      "step": 6424
    },
    {
      "epoch": 1.584319526627219,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00017270191902953695,
      "loss": 0.061,
      "step": 6426
    },
    {
      "epoch": 1.5848126232741617,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00017267827817429567,
      "loss": 0.062,
      "step": 6428
    },
    {
      "epoch": 1.5853057199211045,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001726546287062727,
      "loss": 0.0517,
      "step": 6430
    },
    {
      "epoch": 1.5857988165680474,
      "grad_norm": 0.34375,
      "learning_rate": 0.00017263097062827053,
      "loss": 0.0584,
      "step": 6432
    },
    {
      "epoch": 1.5862919132149902,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00017260730394309283,
      "loss": 0.065,
      "step": 6434
    },
    {
      "epoch": 1.586785009861933,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00017258362865354426,
      "loss": 0.0635,
      "step": 6436
    },
    {
      "epoch": 1.5872781065088757,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001725599447624304,
      "loss": 0.054,
      "step": 6438
    },
    {
      "epoch": 1.5877712031558184,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00017253625227255798,
      "loss": 0.0576,
      "step": 6440
    },
    {
      "epoch": 1.5882642998027614,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00017251255118673467,
      "loss": 0.0611,
      "step": 6442
    },
    {
      "epoch": 1.5887573964497042,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001724888415077692,
      "loss": 0.0537,
      "step": 6444
    },
    {
      "epoch": 1.589250493096647,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00017246512323847125,
      "loss": 0.0619,
      "step": 6446
    },
    {
      "epoch": 1.5897435897435899,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00017244139638165163,
      "loss": 0.0591,
      "step": 6448
    },
    {
      "epoch": 1.5902366863905324,
      "grad_norm": 0.546875,
      "learning_rate": 0.00017241766094012206,
      "loss": 0.0605,
      "step": 6450
    },
    {
      "epoch": 1.5907297830374754,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00017239391691669536,
      "loss": 0.057,
      "step": 6452
    },
    {
      "epoch": 1.5912228796844181,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001723701643141853,
      "loss": 0.0507,
      "step": 6454
    },
    {
      "epoch": 1.5917159763313609,
      "grad_norm": 0.4375,
      "learning_rate": 0.00017234640313540674,
      "loss": 0.0583,
      "step": 6456
    },
    {
      "epoch": 1.5922090729783038,
      "grad_norm": 0.515625,
      "learning_rate": 0.00017232263338317542,
      "loss": 0.0614,
      "step": 6458
    },
    {
      "epoch": 1.5927021696252466,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00017229885506030827,
      "loss": 0.057,
      "step": 6460
    },
    {
      "epoch": 1.5931952662721893,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00017227506816962316,
      "loss": 0.0614,
      "step": 6462
    },
    {
      "epoch": 1.5936883629191323,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001722512727139389,
      "loss": 0.0623,
      "step": 6464
    },
    {
      "epoch": 1.5941814595660748,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00017222746869607542,
      "loss": 0.0579,
      "step": 6466
    },
    {
      "epoch": 1.5946745562130178,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00017220365611885367,
      "loss": 0.055,
      "step": 6468
    },
    {
      "epoch": 1.5951676528599605,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001721798349850955,
      "loss": 0.0608,
      "step": 6470
    },
    {
      "epoch": 1.5956607495069033,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001721560052976239,
      "loss": 0.0612,
      "step": 6472
    },
    {
      "epoch": 1.5961538461538463,
      "grad_norm": 0.359375,
      "learning_rate": 0.00017213216705926282,
      "loss": 0.056,
      "step": 6474
    },
    {
      "epoch": 1.596646942800789,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001721083202728372,
      "loss": 0.0617,
      "step": 6476
    },
    {
      "epoch": 1.5971400394477318,
      "grad_norm": 0.328125,
      "learning_rate": 0.000172084464941173,
      "loss": 0.0576,
      "step": 6478
    },
    {
      "epoch": 1.5976331360946747,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00017206060106709727,
      "loss": 0.0576,
      "step": 6480
    },
    {
      "epoch": 1.5981262327416172,
      "grad_norm": 0.34375,
      "learning_rate": 0.00017203672865343794,
      "loss": 0.0523,
      "step": 6482
    },
    {
      "epoch": 1.5986193293885602,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001720128477030241,
      "loss": 0.0558,
      "step": 6484
    },
    {
      "epoch": 1.599112426035503,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00017198895821868572,
      "loss": 0.0568,
      "step": 6486
    },
    {
      "epoch": 1.5996055226824457,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00017196506020325387,
      "loss": 0.0549,
      "step": 6488
    },
    {
      "epoch": 1.6000986193293887,
      "grad_norm": 0.5625,
      "learning_rate": 0.00017194115365956058,
      "loss": 0.0605,
      "step": 6490
    },
    {
      "epoch": 1.6005917159763312,
      "grad_norm": 0.359375,
      "learning_rate": 0.00017191723859043888,
      "loss": 0.0584,
      "step": 6492
    },
    {
      "epoch": 1.6010848126232742,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001718933149987229,
      "loss": 0.0587,
      "step": 6494
    },
    {
      "epoch": 1.601577909270217,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001718693828872477,
      "loss": 0.0585,
      "step": 6496
    },
    {
      "epoch": 1.6020710059171597,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00017184544225884937,
      "loss": 0.0595,
      "step": 6498
    },
    {
      "epoch": 1.6025641025641026,
      "grad_norm": 0.373046875,
      "learning_rate": 0.000171821493116365,
      "loss": 0.0594,
      "step": 6500
    },
    {
      "epoch": 1.6030571992110454,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00017179753546263266,
      "loss": 0.062,
      "step": 6502
    },
    {
      "epoch": 1.6035502958579881,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00017177356930049152,
      "loss": 0.058,
      "step": 6504
    },
    {
      "epoch": 1.604043392504931,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00017174959463278172,
      "loss": 0.0588,
      "step": 6506
    },
    {
      "epoch": 1.6045364891518736,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001717256114623443,
      "loss": 0.0556,
      "step": 6508
    },
    {
      "epoch": 1.6050295857988166,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00017170161979202149,
      "loss": 0.0567,
      "step": 6510
    },
    {
      "epoch": 1.6055226824457594,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00017167761962465637,
      "loss": 0.0584,
      "step": 6512
    },
    {
      "epoch": 1.606015779092702,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00017165361096309318,
      "loss": 0.0622,
      "step": 6514
    },
    {
      "epoch": 1.606508875739645,
      "grad_norm": 0.380859375,
      "learning_rate": 0.000171629593810177,
      "loss": 0.0563,
      "step": 6516
    },
    {
      "epoch": 1.6070019723865878,
      "grad_norm": 0.41015625,
      "learning_rate": 0.000171605568168754,
      "loss": 0.0567,
      "step": 6518
    },
    {
      "epoch": 1.6074950690335306,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00017158153404167142,
      "loss": 0.0576,
      "step": 6520
    },
    {
      "epoch": 1.6079881656804735,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001715574914317774,
      "loss": 0.0581,
      "step": 6522
    },
    {
      "epoch": 1.608481262327416,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001715334403419211,
      "loss": 0.0567,
      "step": 6524
    },
    {
      "epoch": 1.608974358974359,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00017150938077495272,
      "loss": 0.054,
      "step": 6526
    },
    {
      "epoch": 1.6094674556213018,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00017148531273372347,
      "loss": 0.0589,
      "step": 6528
    },
    {
      "epoch": 1.6099605522682445,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00017146123622108553,
      "loss": 0.0595,
      "step": 6530
    },
    {
      "epoch": 1.6104536489151875,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00017143715123989212,
      "loss": 0.0609,
      "step": 6532
    },
    {
      "epoch": 1.61094674556213,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00017141305779299745,
      "loss": 0.0504,
      "step": 6534
    },
    {
      "epoch": 1.611439842209073,
      "grad_norm": 0.453125,
      "learning_rate": 0.00017138895588325668,
      "loss": 0.0569,
      "step": 6536
    },
    {
      "epoch": 1.6119329388560157,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00017136484551352608,
      "loss": 0.0582,
      "step": 6538
    },
    {
      "epoch": 1.6124260355029585,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001713407266866628,
      "loss": 0.059,
      "step": 6540
    },
    {
      "epoch": 1.6129191321499015,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001713165994055251,
      "loss": 0.0561,
      "step": 6542
    },
    {
      "epoch": 1.6134122287968442,
      "grad_norm": 0.328125,
      "learning_rate": 0.00017129246367297222,
      "loss": 0.0555,
      "step": 6544
    },
    {
      "epoch": 1.613905325443787,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00017126831949186433,
      "loss": 0.0537,
      "step": 6546
    },
    {
      "epoch": 1.61439842209073,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00017124416686506268,
      "loss": 0.0614,
      "step": 6548
    },
    {
      "epoch": 1.6148915187376724,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00017122000579542946,
      "loss": 0.0571,
      "step": 6550
    },
    {
      "epoch": 1.6153846153846154,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001711958362858279,
      "loss": 0.058,
      "step": 6552
    },
    {
      "epoch": 1.6158777120315582,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001711716583391223,
      "loss": 0.0537,
      "step": 6554
    },
    {
      "epoch": 1.616370808678501,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00017114747195817777,
      "loss": 0.0609,
      "step": 6556
    },
    {
      "epoch": 1.6168639053254439,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001711232771458606,
      "loss": 0.0544,
      "step": 6558
    },
    {
      "epoch": 1.6173570019723866,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00017109907390503796,
      "loss": 0.055,
      "step": 6560
    },
    {
      "epoch": 1.6178500986193294,
      "grad_norm": 0.515625,
      "learning_rate": 0.00017107486223857807,
      "loss": 0.0579,
      "step": 6562
    },
    {
      "epoch": 1.6183431952662723,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001710506421493502,
      "loss": 0.0603,
      "step": 6564
    },
    {
      "epoch": 1.6188362919132149,
      "grad_norm": 0.390625,
      "learning_rate": 0.00017102641364022457,
      "loss": 0.0561,
      "step": 6566
    },
    {
      "epoch": 1.6193293885601578,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00017100217671407232,
      "loss": 0.0515,
      "step": 6568
    },
    {
      "epoch": 1.6198224852071006,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001709779313737657,
      "loss": 0.0546,
      "step": 6570
    },
    {
      "epoch": 1.6203155818540433,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00017095367762217796,
      "loss": 0.0559,
      "step": 6572
    },
    {
      "epoch": 1.6208086785009863,
      "grad_norm": 0.5,
      "learning_rate": 0.00017092941546218324,
      "loss": 0.0616,
      "step": 6574
    },
    {
      "epoch": 1.6213017751479288,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001709051448966568,
      "loss": 0.0543,
      "step": 6576
    },
    {
      "epoch": 1.6217948717948718,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00017088086592847478,
      "loss": 0.0647,
      "step": 6578
    },
    {
      "epoch": 1.6222879684418146,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00017085657856051438,
      "loss": 0.0577,
      "step": 6580
    },
    {
      "epoch": 1.6227810650887573,
      "grad_norm": 0.46875,
      "learning_rate": 0.00017083228279565383,
      "loss": 0.0564,
      "step": 6582
    },
    {
      "epoch": 1.6232741617357003,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00017080797863677232,
      "loss": 0.0503,
      "step": 6584
    },
    {
      "epoch": 1.623767258382643,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00017078366608674997,
      "loss": 0.0555,
      "step": 6586
    },
    {
      "epoch": 1.6242603550295858,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00017075934514846798,
      "loss": 0.0587,
      "step": 6588
    },
    {
      "epoch": 1.6247534516765287,
      "grad_norm": 0.515625,
      "learning_rate": 0.00017073501582480854,
      "loss": 0.0533,
      "step": 6590
    },
    {
      "epoch": 1.6252465483234713,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00017071067811865476,
      "loss": 0.0523,
      "step": 6592
    },
    {
      "epoch": 1.6257396449704142,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017068633203289085,
      "loss": 0.0589,
      "step": 6594
    },
    {
      "epoch": 1.626232741617357,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00017066197757040192,
      "loss": 0.0569,
      "step": 6596
    },
    {
      "epoch": 1.6267258382642997,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00017063761473407412,
      "loss": 0.0568,
      "step": 6598
    },
    {
      "epoch": 1.6272189349112427,
      "grad_norm": 0.390625,
      "learning_rate": 0.0001706132435267946,
      "loss": 0.0567,
      "step": 6600
    },
    {
      "epoch": 1.6277120315581854,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00017058886395145145,
      "loss": 0.0544,
      "step": 6602
    },
    {
      "epoch": 1.6282051282051282,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00017056447601093382,
      "loss": 0.0579,
      "step": 6604
    },
    {
      "epoch": 1.6286982248520712,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00017054007970813178,
      "loss": 0.0624,
      "step": 6606
    },
    {
      "epoch": 1.6291913214990137,
      "grad_norm": 0.421875,
      "learning_rate": 0.00017051567504593648,
      "loss": 0.0625,
      "step": 6608
    },
    {
      "epoch": 1.6296844181459567,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00017049126202723998,
      "loss": 0.06,
      "step": 6610
    },
    {
      "epoch": 1.6301775147928994,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00017046684065493532,
      "loss": 0.0566,
      "step": 6612
    },
    {
      "epoch": 1.6306706114398422,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00017044241093191662,
      "loss": 0.0532,
      "step": 6614
    },
    {
      "epoch": 1.6311637080867851,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00017041797286107893,
      "loss": 0.055,
      "step": 6616
    },
    {
      "epoch": 1.6316568047337277,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001703935264453183,
      "loss": 0.061,
      "step": 6618
    },
    {
      "epoch": 1.6321499013806706,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00017036907168753175,
      "loss": 0.0569,
      "step": 6620
    },
    {
      "epoch": 1.6326429980276134,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001703446085906173,
      "loss": 0.0579,
      "step": 6622
    },
    {
      "epoch": 1.6331360946745561,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00017032013715747403,
      "loss": 0.0572,
      "step": 6624
    },
    {
      "epoch": 1.633629191321499,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00017029565739100185,
      "loss": 0.0512,
      "step": 6626
    },
    {
      "epoch": 1.6341222879684418,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0001702711692941018,
      "loss": 0.0567,
      "step": 6628
    },
    {
      "epoch": 1.6346153846153846,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001702466728696758,
      "loss": 0.0589,
      "step": 6630
    },
    {
      "epoch": 1.6351084812623276,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00017022216812062691,
      "loss": 0.0563,
      "step": 6632
    },
    {
      "epoch": 1.63560157790927,
      "grad_norm": 0.27734375,
      "learning_rate": 0.000170197655049859,
      "loss": 0.0564,
      "step": 6634
    },
    {
      "epoch": 1.636094674556213,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00017017313366027705,
      "loss": 0.0581,
      "step": 6636
    },
    {
      "epoch": 1.6365877712031558,
      "grad_norm": 0.359375,
      "learning_rate": 0.00017014860395478696,
      "loss": 0.0527,
      "step": 6638
    },
    {
      "epoch": 1.6370808678500985,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00017012406593629563,
      "loss": 0.057,
      "step": 6640
    },
    {
      "epoch": 1.6375739644970415,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000170099519607711,
      "loss": 0.0608,
      "step": 6642
    },
    {
      "epoch": 1.6380670611439843,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00017007496497194188,
      "loss": 0.0567,
      "step": 6644
    },
    {
      "epoch": 1.638560157790927,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00017005040203189818,
      "loss": 0.0594,
      "step": 6646
    },
    {
      "epoch": 1.63905325443787,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00017002583079049068,
      "loss": 0.0597,
      "step": 6648
    },
    {
      "epoch": 1.6395463510848125,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001700012512506313,
      "loss": 0.0561,
      "step": 6650
    },
    {
      "epoch": 1.6400394477317555,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001699766634152328,
      "loss": 0.0518,
      "step": 6652
    },
    {
      "epoch": 1.6405325443786982,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000169952067287209,
      "loss": 0.0565,
      "step": 6654
    },
    {
      "epoch": 1.641025641025641,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00016992746286947466,
      "loss": 0.0504,
      "step": 6656
    },
    {
      "epoch": 1.641518737672584,
      "grad_norm": 0.390625,
      "learning_rate": 0.0001699028501649455,
      "loss": 0.057,
      "step": 6658
    },
    {
      "epoch": 1.6420118343195265,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00016987822917653839,
      "loss": 0.0569,
      "step": 6660
    },
    {
      "epoch": 1.6425049309664694,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00016985359990717092,
      "loss": 0.0557,
      "step": 6662
    },
    {
      "epoch": 1.6429980276134122,
      "grad_norm": 0.4375,
      "learning_rate": 0.00016982896235976185,
      "loss": 0.057,
      "step": 6664
    },
    {
      "epoch": 1.643491124260355,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00016980431653723085,
      "loss": 0.0537,
      "step": 6666
    },
    {
      "epoch": 1.643984220907298,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00016977966244249864,
      "loss": 0.0558,
      "step": 6668
    },
    {
      "epoch": 1.6444773175542406,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001697550000784868,
      "loss": 0.057,
      "step": 6670
    },
    {
      "epoch": 1.6449704142011834,
      "grad_norm": 0.390625,
      "learning_rate": 0.00016973032944811803,
      "loss": 0.0546,
      "step": 6672
    },
    {
      "epoch": 1.6454635108481264,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00016970565055431588,
      "loss": 0.0506,
      "step": 6674
    },
    {
      "epoch": 1.645956607495069,
      "grad_norm": 0.390625,
      "learning_rate": 0.00016968096340000492,
      "loss": 0.0521,
      "step": 6676
    },
    {
      "epoch": 1.6464497041420119,
      "grad_norm": 0.359375,
      "learning_rate": 0.00016965626798811075,
      "loss": 0.0547,
      "step": 6678
    },
    {
      "epoch": 1.6469428007889546,
      "grad_norm": 0.453125,
      "learning_rate": 0.00016963156432155994,
      "loss": 0.06,
      "step": 6680
    },
    {
      "epoch": 1.6474358974358974,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00016960685240327997,
      "loss": 0.056,
      "step": 6682
    },
    {
      "epoch": 1.6479289940828403,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001695821322361994,
      "loss": 0.0551,
      "step": 6684
    },
    {
      "epoch": 1.648422090729783,
      "grad_norm": 0.515625,
      "learning_rate": 0.00016955740382324763,
      "loss": 0.0568,
      "step": 6686
    },
    {
      "epoch": 1.6489151873767258,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00016953266716735515,
      "loss": 0.0504,
      "step": 6688
    },
    {
      "epoch": 1.6494082840236688,
      "grad_norm": 0.5,
      "learning_rate": 0.00016950792227145338,
      "loss": 0.0577,
      "step": 6690
    },
    {
      "epoch": 1.6499013806706113,
      "grad_norm": 0.546875,
      "learning_rate": 0.00016948316913847474,
      "loss": 0.0598,
      "step": 6692
    },
    {
      "epoch": 1.6503944773175543,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00016945840777135262,
      "loss": 0.0525,
      "step": 6694
    },
    {
      "epoch": 1.650887573964497,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00016943363817302135,
      "loss": 0.0557,
      "step": 6696
    },
    {
      "epoch": 1.6513806706114398,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00016940886034641633,
      "loss": 0.059,
      "step": 6698
    },
    {
      "epoch": 1.6518737672583828,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001693840742944738,
      "loss": 0.0583,
      "step": 6700
    },
    {
      "epoch": 1.6523668639053253,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00016935928002013108,
      "loss": 0.0592,
      "step": 6702
    },
    {
      "epoch": 1.6528599605522682,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00016933447752632644,
      "loss": 0.0632,
      "step": 6704
    },
    {
      "epoch": 1.653353057199211,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00016930966681599907,
      "loss": 0.0572,
      "step": 6706
    },
    {
      "epoch": 1.6538461538461537,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00016928484789208924,
      "loss": 0.0568,
      "step": 6708
    },
    {
      "epoch": 1.6543392504930967,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00016926002075753807,
      "loss": 0.0542,
      "step": 6710
    },
    {
      "epoch": 1.6548323471400395,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00016923518541528775,
      "loss": 0.0579,
      "step": 6712
    },
    {
      "epoch": 1.6553254437869822,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001692103418682814,
      "loss": 0.0575,
      "step": 6714
    },
    {
      "epoch": 1.6558185404339252,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001691854901194631,
      "loss": 0.0572,
      "step": 6716
    },
    {
      "epoch": 1.6563116370808677,
      "grad_norm": 0.3671875,
      "learning_rate": 0.000169160630171778,
      "loss": 0.0566,
      "step": 6718
    },
    {
      "epoch": 1.6568047337278107,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00016913576202817204,
      "loss": 0.0606,
      "step": 6720
    },
    {
      "epoch": 1.6572978303747534,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00016911088569159226,
      "loss": 0.0552,
      "step": 6722
    },
    {
      "epoch": 1.6577909270216962,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001690860011649867,
      "loss": 0.0548,
      "step": 6724
    },
    {
      "epoch": 1.6582840236686391,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001690611084513042,
      "loss": 0.0539,
      "step": 6726
    },
    {
      "epoch": 1.6587771203155819,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001690362075534948,
      "loss": 0.0514,
      "step": 6728
    },
    {
      "epoch": 1.6592702169625246,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00016901129847450938,
      "loss": 0.0579,
      "step": 6730
    },
    {
      "epoch": 1.6597633136094676,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00016898638121729976,
      "loss": 0.0569,
      "step": 6732
    },
    {
      "epoch": 1.6602564102564101,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00016896145578481882,
      "loss": 0.0536,
      "step": 6734
    },
    {
      "epoch": 1.660749506903353,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00016893652218002032,
      "loss": 0.0592,
      "step": 6736
    },
    {
      "epoch": 1.6612426035502958,
      "grad_norm": 0.640625,
      "learning_rate": 0.00016891158040585903,
      "loss": 0.0562,
      "step": 6738
    },
    {
      "epoch": 1.6617357001972386,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00016888663046529074,
      "loss": 0.0514,
      "step": 6740
    },
    {
      "epoch": 1.6622287968441816,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001688616723612721,
      "loss": 0.0559,
      "step": 6742
    },
    {
      "epoch": 1.6627218934911243,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00016883670609676082,
      "loss": 0.0544,
      "step": 6744
    },
    {
      "epoch": 1.663214990138067,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00016881173167471555,
      "loss": 0.0584,
      "step": 6746
    },
    {
      "epoch": 1.6637080867850098,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00016878674909809587,
      "loss": 0.0618,
      "step": 6748
    },
    {
      "epoch": 1.6642011834319526,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00016876175836986237,
      "loss": 0.0605,
      "step": 6750
    },
    {
      "epoch": 1.6646942800788955,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016873675949297657,
      "loss": 0.0577,
      "step": 6752
    },
    {
      "epoch": 1.6651873767258383,
      "grad_norm": 0.515625,
      "learning_rate": 0.000168711752470401,
      "loss": 0.0615,
      "step": 6754
    },
    {
      "epoch": 1.665680473372781,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00016868673730509914,
      "loss": 0.0589,
      "step": 6756
    },
    {
      "epoch": 1.666173570019724,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001686617140000354,
      "loss": 0.0697,
      "step": 6758
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.34375,
      "learning_rate": 0.00016863668255817518,
      "loss": 0.0532,
      "step": 6760
    },
    {
      "epoch": 1.6671597633136095,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001686116429824849,
      "loss": 0.0581,
      "step": 6762
    },
    {
      "epoch": 1.6676528599605522,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001685865952759318,
      "loss": 0.0521,
      "step": 6764
    },
    {
      "epoch": 1.668145956607495,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00016856153944148425,
      "loss": 0.0596,
      "step": 6766
    },
    {
      "epoch": 1.668639053254438,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00016853647548211148,
      "loss": 0.0508,
      "step": 6768
    },
    {
      "epoch": 1.6691321499013807,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016851140340078368,
      "loss": 0.0554,
      "step": 6770
    },
    {
      "epoch": 1.6696252465483234,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00016848632320047208,
      "loss": 0.054,
      "step": 6772
    },
    {
      "epoch": 1.6701183431952664,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00016846123488414876,
      "loss": 0.055,
      "step": 6774
    },
    {
      "epoch": 1.670611439842209,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00016843613845478694,
      "loss": 0.052,
      "step": 6776
    },
    {
      "epoch": 1.671104536489152,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00016841103391536058,
      "loss": 0.0652,
      "step": 6778
    },
    {
      "epoch": 1.6715976331360947,
      "grad_norm": 0.375,
      "learning_rate": 0.00016838592126884475,
      "loss": 0.0624,
      "step": 6780
    },
    {
      "epoch": 1.6720907297830374,
      "grad_norm": 0.40625,
      "learning_rate": 0.00016836080051821543,
      "loss": 0.0547,
      "step": 6782
    },
    {
      "epoch": 1.6725838264299804,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00016833567166644957,
      "loss": 0.0597,
      "step": 6784
    },
    {
      "epoch": 1.6730769230769231,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00016831053471652507,
      "loss": 0.0563,
      "step": 6786
    },
    {
      "epoch": 1.6735700197238659,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00016828538967142081,
      "loss": 0.0538,
      "step": 6788
    },
    {
      "epoch": 1.6740631163708086,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001682602365341166,
      "loss": 0.0603,
      "step": 6790
    },
    {
      "epoch": 1.6745562130177514,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00016823507530759328,
      "loss": 0.0601,
      "step": 6792
    },
    {
      "epoch": 1.6750493096646943,
      "grad_norm": 0.34375,
      "learning_rate": 0.00016820990599483252,
      "loss": 0.0589,
      "step": 6794
    },
    {
      "epoch": 1.675542406311637,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001681847285988171,
      "loss": 0.0535,
      "step": 6796
    },
    {
      "epoch": 1.6760355029585798,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00016815954312253065,
      "loss": 0.0569,
      "step": 6798
    },
    {
      "epoch": 1.6765285996055228,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00016813434956895775,
      "loss": 0.0498,
      "step": 6800
    },
    {
      "epoch": 1.6770216962524653,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00016810914794108402,
      "loss": 0.0552,
      "step": 6802
    },
    {
      "epoch": 1.6775147928994083,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00016808393824189604,
      "loss": 0.0575,
      "step": 6804
    },
    {
      "epoch": 1.678007889546351,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00016805872047438117,
      "loss": 0.0586,
      "step": 6806
    },
    {
      "epoch": 1.6785009861932938,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00016803349464152798,
      "loss": 0.0589,
      "step": 6808
    },
    {
      "epoch": 1.6789940828402368,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001680082607463258,
      "loss": 0.06,
      "step": 6810
    },
    {
      "epoch": 1.6794871794871795,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00016798301879176504,
      "loss": 0.0547,
      "step": 6812
    },
    {
      "epoch": 1.6799802761341223,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00016795776878083694,
      "loss": 0.0594,
      "step": 6814
    },
    {
      "epoch": 1.6804733727810652,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00016793251071653387,
      "loss": 0.0655,
      "step": 6816
    },
    {
      "epoch": 1.6809664694280078,
      "grad_norm": 0.30859375,
      "learning_rate": 0.000167907244601849,
      "loss": 0.0533,
      "step": 6818
    },
    {
      "epoch": 1.6814595660749507,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00016788197043977646,
      "loss": 0.0599,
      "step": 6820
    },
    {
      "epoch": 1.6819526627218935,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00016785668823331138,
      "loss": 0.0547,
      "step": 6822
    },
    {
      "epoch": 1.6824457593688362,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00016783139798544995,
      "loss": 0.0564,
      "step": 6824
    },
    {
      "epoch": 1.6829388560157792,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001678060996991891,
      "loss": 0.0544,
      "step": 6826
    },
    {
      "epoch": 1.683431952662722,
      "grad_norm": 0.34375,
      "learning_rate": 0.00016778079337752687,
      "loss": 0.053,
      "step": 6828
    },
    {
      "epoch": 1.6839250493096647,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00016775547902346215,
      "loss": 0.0501,
      "step": 6830
    },
    {
      "epoch": 1.6844181459566077,
      "grad_norm": 0.59375,
      "learning_rate": 0.00016773015663999492,
      "loss": 0.0579,
      "step": 6832
    },
    {
      "epoch": 1.6849112426035502,
      "grad_norm": 0.453125,
      "learning_rate": 0.00016770482623012595,
      "loss": 0.0516,
      "step": 6834
    },
    {
      "epoch": 1.6854043392504932,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00016767948779685706,
      "loss": 0.0544,
      "step": 6836
    },
    {
      "epoch": 1.685897435897436,
      "grad_norm": 0.30078125,
      "learning_rate": 0.000167654141343191,
      "loss": 0.0548,
      "step": 6838
    },
    {
      "epoch": 1.6863905325443787,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00016762878687213143,
      "loss": 0.0489,
      "step": 6840
    },
    {
      "epoch": 1.6868836291913216,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00016760342438668305,
      "loss": 0.0544,
      "step": 6842
    },
    {
      "epoch": 1.6873767258382641,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00016757805388985142,
      "loss": 0.0553,
      "step": 6844
    },
    {
      "epoch": 1.6878698224852071,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00016755267538464308,
      "loss": 0.0546,
      "step": 6846
    },
    {
      "epoch": 1.6883629191321499,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001675272888740656,
      "loss": 0.0537,
      "step": 6848
    },
    {
      "epoch": 1.6888560157790926,
      "grad_norm": 0.578125,
      "learning_rate": 0.00016750189436112728,
      "loss": 0.0555,
      "step": 6850
    },
    {
      "epoch": 1.6893491124260356,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00016747649184883768,
      "loss": 0.0519,
      "step": 6852
    },
    {
      "epoch": 1.6898422090729783,
      "grad_norm": 0.40625,
      "learning_rate": 0.000167451081340207,
      "loss": 0.0571,
      "step": 6854
    },
    {
      "epoch": 1.690335305719921,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001674256628382466,
      "loss": 0.0471,
      "step": 6856
    },
    {
      "epoch": 1.690828402366864,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00016740023634596868,
      "loss": 0.0516,
      "step": 6858
    },
    {
      "epoch": 1.6913214990138066,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00016737480186638644,
      "loss": 0.0527,
      "step": 6860
    },
    {
      "epoch": 1.6918145956607495,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00016734935940251403,
      "loss": 0.0538,
      "step": 6862
    },
    {
      "epoch": 1.6923076923076923,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001673239089573665,
      "loss": 0.0538,
      "step": 6864
    },
    {
      "epoch": 1.692800788954635,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016729845053395988,
      "loss": 0.0513,
      "step": 6866
    },
    {
      "epoch": 1.693293885601578,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001672729841353111,
      "loss": 0.0619,
      "step": 6868
    },
    {
      "epoch": 1.6937869822485208,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00016724750976443808,
      "loss": 0.057,
      "step": 6870
    },
    {
      "epoch": 1.6942800788954635,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001672220274243597,
      "loss": 0.055,
      "step": 6872
    },
    {
      "epoch": 1.6947731755424065,
      "grad_norm": 0.5,
      "learning_rate": 0.00016719653711809579,
      "loss": 0.0609,
      "step": 6874
    },
    {
      "epoch": 1.695266272189349,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00016717103884866702,
      "loss": 0.0559,
      "step": 6876
    },
    {
      "epoch": 1.695759368836292,
      "grad_norm": 0.46875,
      "learning_rate": 0.00016714553261909512,
      "loss": 0.0577,
      "step": 6878
    },
    {
      "epoch": 1.6962524654832347,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001671200184324027,
      "loss": 0.0564,
      "step": 6880
    },
    {
      "epoch": 1.6967455621301775,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00016709449629161337,
      "loss": 0.0595,
      "step": 6882
    },
    {
      "epoch": 1.6972386587771204,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001670689661997516,
      "loss": 0.0545,
      "step": 6884
    },
    {
      "epoch": 1.697731755424063,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001670434281598429,
      "loss": 0.0531,
      "step": 6886
    },
    {
      "epoch": 1.698224852071006,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001670178821749136,
      "loss": 0.0558,
      "step": 6888
    },
    {
      "epoch": 1.6987179487179487,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001669923282479911,
      "loss": 0.0558,
      "step": 6890
    },
    {
      "epoch": 1.6992110453648914,
      "grad_norm": 0.4375,
      "learning_rate": 0.00016696676638210364,
      "loss": 0.0533,
      "step": 6892
    },
    {
      "epoch": 1.6997041420118344,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016694119658028052,
      "loss": 0.0556,
      "step": 6894
    },
    {
      "epoch": 1.7001972386587771,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001669156188455518,
      "loss": 0.056,
      "step": 6896
    },
    {
      "epoch": 1.70069033530572,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00016689003318094868,
      "loss": 0.0603,
      "step": 6898
    },
    {
      "epoch": 1.7011834319526629,
      "grad_norm": 0.46875,
      "learning_rate": 0.00016686443958950317,
      "loss": 0.0579,
      "step": 6900
    },
    {
      "epoch": 1.7016765285996054,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001668388380742482,
      "loss": 0.0544,
      "step": 6902
    },
    {
      "epoch": 1.7021696252465484,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00016681322863821776,
      "loss": 0.0546,
      "step": 6904
    },
    {
      "epoch": 1.702662721893491,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001667876112844467,
      "loss": 0.0569,
      "step": 6906
    },
    {
      "epoch": 1.7031558185404339,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001667619860159708,
      "loss": 0.0553,
      "step": 6908
    },
    {
      "epoch": 1.7036489151873768,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00016673635283582678,
      "loss": 0.0537,
      "step": 6910
    },
    {
      "epoch": 1.7041420118343196,
      "grad_norm": 0.546875,
      "learning_rate": 0.00016671071174705236,
      "loss": 0.0592,
      "step": 6912
    },
    {
      "epoch": 1.7046351084812623,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00016668506275268618,
      "loss": 0.0549,
      "step": 6914
    },
    {
      "epoch": 1.7051282051282053,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00016665940585576769,
      "loss": 0.0525,
      "step": 6916
    },
    {
      "epoch": 1.7056213017751478,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00016663374105933745,
      "loss": 0.0507,
      "step": 6918
    },
    {
      "epoch": 1.7061143984220908,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00016660806836643683,
      "loss": 0.0513,
      "step": 6920
    },
    {
      "epoch": 1.7066074950690335,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001665823877801082,
      "loss": 0.0538,
      "step": 6922
    },
    {
      "epoch": 1.7071005917159763,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00016655669930339492,
      "loss": 0.0535,
      "step": 6924
    },
    {
      "epoch": 1.7075936883629192,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00016653100293934116,
      "loss": 0.0513,
      "step": 6926
    },
    {
      "epoch": 1.7080867850098618,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00016650529869099208,
      "loss": 0.0503,
      "step": 6928
    },
    {
      "epoch": 1.7085798816568047,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00016647958656139378,
      "loss": 0.0562,
      "step": 6930
    },
    {
      "epoch": 1.7090729783037475,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001664538665535933,
      "loss": 0.0628,
      "step": 6932
    },
    {
      "epoch": 1.7095660749506902,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00016642813867063865,
      "loss": 0.0554,
      "step": 6934
    },
    {
      "epoch": 1.7100591715976332,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00016640240291557862,
      "loss": 0.0556,
      "step": 6936
    },
    {
      "epoch": 1.710552268244576,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00016637665929146312,
      "loss": 0.0571,
      "step": 6938
    },
    {
      "epoch": 1.7110453648915187,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001663509078013429,
      "loss": 0.0618,
      "step": 6940
    },
    {
      "epoch": 1.7115384615384617,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016632514844826962,
      "loss": 0.0539,
      "step": 6942
    },
    {
      "epoch": 1.7120315581854042,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00016629938123529598,
      "loss": 0.0535,
      "step": 6944
    },
    {
      "epoch": 1.7125246548323472,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00016627360616547547,
      "loss": 0.0517,
      "step": 6946
    },
    {
      "epoch": 1.71301775147929,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00016624782324186263,
      "loss": 0.0588,
      "step": 6948
    },
    {
      "epoch": 1.7135108481262327,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00016622203246751283,
      "loss": 0.0588,
      "step": 6950
    },
    {
      "epoch": 1.7140039447731756,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00016619623384548244,
      "loss": 0.0593,
      "step": 6952
    },
    {
      "epoch": 1.7144970414201184,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00016617042737882874,
      "loss": 0.058,
      "step": 6954
    },
    {
      "epoch": 1.7149901380670611,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00016614461307061,
      "loss": 0.0575,
      "step": 6956
    },
    {
      "epoch": 1.715483234714004,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00016611879092388526,
      "loss": 0.0562,
      "step": 6958
    },
    {
      "epoch": 1.7159763313609466,
      "grad_norm": 0.40625,
      "learning_rate": 0.00016609296094171464,
      "loss": 0.0558,
      "step": 6960
    },
    {
      "epoch": 1.7164694280078896,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00016606712312715915,
      "loss": 0.0545,
      "step": 6962
    },
    {
      "epoch": 1.7169625246548323,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0001660412774832807,
      "loss": 0.0462,
      "step": 6964
    },
    {
      "epoch": 1.717455621301775,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001660154240131421,
      "loss": 0.0561,
      "step": 6966
    },
    {
      "epoch": 1.717948717948718,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00016598956271980723,
      "loss": 0.0575,
      "step": 6968
    },
    {
      "epoch": 1.7184418145956606,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001659636936063407,
      "loss": 0.0578,
      "step": 6970
    },
    {
      "epoch": 1.7189349112426036,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00016593781667580818,
      "loss": 0.0561,
      "step": 6972
    },
    {
      "epoch": 1.7194280078895463,
      "grad_norm": 0.390625,
      "learning_rate": 0.00016591193193127627,
      "loss": 0.0547,
      "step": 6974
    },
    {
      "epoch": 1.719921104536489,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00016588603937581236,
      "loss": 0.0564,
      "step": 6976
    },
    {
      "epoch": 1.720414201183432,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000165860139012485,
      "loss": 0.0626,
      "step": 6978
    },
    {
      "epoch": 1.7209072978303748,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001658342308443634,
      "loss": 0.0544,
      "step": 6980
    },
    {
      "epoch": 1.7214003944773175,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00016580831487451788,
      "loss": 0.0554,
      "step": 6982
    },
    {
      "epoch": 1.7218934911242605,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00016578239110601965,
      "loss": 0.0556,
      "step": 6984
    },
    {
      "epoch": 1.722386587771203,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00016575645954194077,
      "loss": 0.055,
      "step": 6986
    },
    {
      "epoch": 1.722879684418146,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0001657305201853543,
      "loss": 0.0509,
      "step": 6988
    },
    {
      "epoch": 1.7233727810650887,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001657045730393342,
      "loss": 0.0568,
      "step": 6990
    },
    {
      "epoch": 1.7238658777120315,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00016567861810695532,
      "loss": 0.0529,
      "step": 6992
    },
    {
      "epoch": 1.7243589743589745,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001656526553912935,
      "loss": 0.0517,
      "step": 6994
    },
    {
      "epoch": 1.7248520710059172,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001656266848954255,
      "loss": 0.0584,
      "step": 6996
    },
    {
      "epoch": 1.72534516765286,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001656007066224289,
      "loss": 0.0579,
      "step": 6998
    },
    {
      "epoch": 1.725838264299803,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00016557472057538228,
      "loss": 0.06,
      "step": 7000
    },
    {
      "epoch": 1.7263313609467454,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00016554872675736518,
      "loss": 0.0564,
      "step": 7002
    },
    {
      "epoch": 1.7268244575936884,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00016552272517145794,
      "loss": 0.0555,
      "step": 7004
    },
    {
      "epoch": 1.7273175542406312,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00016549671582074198,
      "loss": 0.0521,
      "step": 7006
    },
    {
      "epoch": 1.727810650887574,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001654706987082995,
      "loss": 0.0533,
      "step": 7008
    },
    {
      "epoch": 1.7283037475345169,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00016544467383721367,
      "loss": 0.0552,
      "step": 7010
    },
    {
      "epoch": 1.7287968441814594,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00016541864121056858,
      "loss": 0.0572,
      "step": 7012
    },
    {
      "epoch": 1.7292899408284024,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001653926008314493,
      "loss": 0.0551,
      "step": 7014
    },
    {
      "epoch": 1.7297830374753451,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00016536655270294167,
      "loss": 0.0517,
      "step": 7016
    },
    {
      "epoch": 1.7302761341222879,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00016534049682813263,
      "loss": 0.0548,
      "step": 7018
    },
    {
      "epoch": 1.7307692307692308,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00016531443321010983,
      "loss": 0.0537,
      "step": 7020
    },
    {
      "epoch": 1.7312623274161736,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001652883618519621,
      "loss": 0.0533,
      "step": 7022
    },
    {
      "epoch": 1.7317554240631163,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00016526228275677893,
      "loss": 0.0532,
      "step": 7024
    },
    {
      "epoch": 1.7322485207100593,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001652361959276509,
      "loss": 0.0541,
      "step": 7026
    },
    {
      "epoch": 1.7327416173570018,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00016521010136766938,
      "loss": 0.0571,
      "step": 7028
    },
    {
      "epoch": 1.7332347140039448,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0001651839990799268,
      "loss": 0.0502,
      "step": 7030
    },
    {
      "epoch": 1.7337278106508875,
      "grad_norm": 0.390625,
      "learning_rate": 0.00016515788906751636,
      "loss": 0.057,
      "step": 7032
    },
    {
      "epoch": 1.7342209072978303,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0001651317713335323,
      "loss": 0.0547,
      "step": 7034
    },
    {
      "epoch": 1.7347140039447733,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00016510564588106968,
      "loss": 0.0553,
      "step": 7036
    },
    {
      "epoch": 1.735207100591716,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001650795127132245,
      "loss": 0.0588,
      "step": 7038
    },
    {
      "epoch": 1.7357001972386588,
      "grad_norm": 0.359375,
      "learning_rate": 0.00016505337183309371,
      "loss": 0.0541,
      "step": 7040
    },
    {
      "epoch": 1.7361932938856017,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00016502722324377515,
      "loss": 0.0533,
      "step": 7042
    },
    {
      "epoch": 1.7366863905325443,
      "grad_norm": 0.453125,
      "learning_rate": 0.00016500106694836757,
      "loss": 0.0602,
      "step": 7044
    },
    {
      "epoch": 1.7371794871794872,
      "grad_norm": 0.34375,
      "learning_rate": 0.00016497490294997068,
      "loss": 0.0488,
      "step": 7046
    },
    {
      "epoch": 1.73767258382643,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00016494873125168496,
      "loss": 0.0505,
      "step": 7048
    },
    {
      "epoch": 1.7381656804733727,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00016492255185661194,
      "loss": 0.0546,
      "step": 7050
    },
    {
      "epoch": 1.7386587771203157,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0001648963647678541,
      "loss": 0.0475,
      "step": 7052
    },
    {
      "epoch": 1.7391518737672582,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00016487016998851468,
      "loss": 0.0556,
      "step": 7054
    },
    {
      "epoch": 1.7396449704142012,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001648439675216979,
      "loss": 0.058,
      "step": 7056
    },
    {
      "epoch": 1.740138067061144,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00016481775737050896,
      "loss": 0.0554,
      "step": 7058
    },
    {
      "epoch": 1.7406311637080867,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001647915395380539,
      "loss": 0.0521,
      "step": 7060
    },
    {
      "epoch": 1.7411242603550297,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001647653140274396,
      "loss": 0.058,
      "step": 7062
    },
    {
      "epoch": 1.7416173570019724,
      "grad_norm": 0.421875,
      "learning_rate": 0.000164739080841774,
      "loss": 0.0581,
      "step": 7064
    },
    {
      "epoch": 1.7421104536489151,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001647128399841659,
      "loss": 0.0615,
      "step": 7066
    },
    {
      "epoch": 1.7426035502958581,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001646865914577249,
      "loss": 0.0515,
      "step": 7068
    },
    {
      "epoch": 1.7430966469428006,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00016466033526556172,
      "loss": 0.0515,
      "step": 7070
    },
    {
      "epoch": 1.7435897435897436,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016463407141078772,
      "loss": 0.0568,
      "step": 7072
    },
    {
      "epoch": 1.7440828402366864,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00016460779989651542,
      "loss": 0.0564,
      "step": 7074
    },
    {
      "epoch": 1.744575936883629,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001645815207258581,
      "loss": 0.0555,
      "step": 7076
    },
    {
      "epoch": 1.745069033530572,
      "grad_norm": 0.390625,
      "learning_rate": 0.00016455523390192998,
      "loss": 0.0482,
      "step": 7078
    },
    {
      "epoch": 1.7455621301775148,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00016452893942784627,
      "loss": 0.0513,
      "step": 7080
    },
    {
      "epoch": 1.7460552268244576,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00016450263730672288,
      "loss": 0.058,
      "step": 7082
    },
    {
      "epoch": 1.7465483234714005,
      "grad_norm": 0.3125,
      "learning_rate": 0.00016447632754167688,
      "loss": 0.047,
      "step": 7084
    },
    {
      "epoch": 1.747041420118343,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00016445001013582608,
      "loss": 0.0529,
      "step": 7086
    },
    {
      "epoch": 1.747534516765286,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001644236850922892,
      "loss": 0.0535,
      "step": 7088
    },
    {
      "epoch": 1.7480276134122288,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00016439735241418596,
      "loss": 0.056,
      "step": 7090
    },
    {
      "epoch": 1.7485207100591715,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00016437101210463692,
      "loss": 0.057,
      "step": 7092
    },
    {
      "epoch": 1.7490138067061145,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00016434466416676355,
      "loss": 0.0511,
      "step": 7094
    },
    {
      "epoch": 1.7495069033530573,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016431830860368823,
      "loss": 0.0578,
      "step": 7096
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00016429194541853418,
      "loss": 0.0565,
      "step": 7098
    },
    {
      "epoch": 1.7504930966469427,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00016426557461442568,
      "loss": 0.0544,
      "step": 7100
    },
    {
      "epoch": 1.7509861932938855,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00016423919619448776,
      "loss": 0.0572,
      "step": 7102
    },
    {
      "epoch": 1.7514792899408285,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00016421281016184642,
      "loss": 0.0542,
      "step": 7104
    },
    {
      "epoch": 1.7519723865877712,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00016418641651962855,
      "loss": 0.0589,
      "step": 7106
    },
    {
      "epoch": 1.752465483234714,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00016416001527096199,
      "loss": 0.0546,
      "step": 7108
    },
    {
      "epoch": 1.752958579881657,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00016413360641897537,
      "loss": 0.0586,
      "step": 7110
    },
    {
      "epoch": 1.7534516765285995,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00016410718996679832,
      "loss": 0.0498,
      "step": 7112
    },
    {
      "epoch": 1.7539447731755424,
      "grad_norm": 0.296875,
      "learning_rate": 0.00016408076591756135,
      "loss": 0.0533,
      "step": 7114
    },
    {
      "epoch": 1.7544378698224852,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001640543342743958,
      "loss": 0.0527,
      "step": 7116
    },
    {
      "epoch": 1.754930966469428,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00016402789504043401,
      "loss": 0.0548,
      "step": 7118
    },
    {
      "epoch": 1.755424063116371,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001640014482188092,
      "loss": 0.0571,
      "step": 7120
    },
    {
      "epoch": 1.7559171597633136,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001639749938126554,
      "loss": 0.0566,
      "step": 7122
    },
    {
      "epoch": 1.7564102564102564,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00016394853182510768,
      "loss": 0.0557,
      "step": 7124
    },
    {
      "epoch": 1.7569033530571994,
      "grad_norm": 0.5,
      "learning_rate": 0.00016392206225930187,
      "loss": 0.0492,
      "step": 7126
    },
    {
      "epoch": 1.7573964497041419,
      "grad_norm": 0.5,
      "learning_rate": 0.0001638955851183748,
      "loss": 0.0568,
      "step": 7128
    },
    {
      "epoch": 1.7578895463510849,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00016386910040546416,
      "loss": 0.0561,
      "step": 7130
    },
    {
      "epoch": 1.7583826429980276,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001638426081237085,
      "loss": 0.0551,
      "step": 7132
    },
    {
      "epoch": 1.7588757396449703,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00016381610827624732,
      "loss": 0.0483,
      "step": 7134
    },
    {
      "epoch": 1.7593688362919133,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00016378960086622102,
      "loss": 0.0503,
      "step": 7136
    },
    {
      "epoch": 1.759861932938856,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00016376308589677083,
      "loss": 0.0555,
      "step": 7138
    },
    {
      "epoch": 1.7603550295857988,
      "grad_norm": 0.40234375,
      "learning_rate": 0.000163736563371039,
      "loss": 0.0525,
      "step": 7140
    },
    {
      "epoch": 1.7608481262327418,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00016371003329216852,
      "loss": 0.0542,
      "step": 7142
    },
    {
      "epoch": 1.7613412228796843,
      "grad_norm": 0.625,
      "learning_rate": 0.00016368349566330342,
      "loss": 0.0533,
      "step": 7144
    },
    {
      "epoch": 1.7618343195266273,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001636569504875885,
      "loss": 0.0507,
      "step": 7146
    },
    {
      "epoch": 1.76232741617357,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001636303977681695,
      "loss": 0.0549,
      "step": 7148
    },
    {
      "epoch": 1.7628205128205128,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00016360383750819315,
      "loss": 0.0547,
      "step": 7150
    },
    {
      "epoch": 1.7633136094674557,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001635772697108069,
      "loss": 0.0557,
      "step": 7152
    },
    {
      "epoch": 1.7638067061143983,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00016355069437915925,
      "loss": 0.0516,
      "step": 7154
    },
    {
      "epoch": 1.7642998027613412,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00016352411151639946,
      "loss": 0.0568,
      "step": 7156
    },
    {
      "epoch": 1.764792899408284,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001634975211256778,
      "loss": 0.0564,
      "step": 7158
    },
    {
      "epoch": 1.7652859960552267,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00016347092321014535,
      "loss": 0.0489,
      "step": 7160
    },
    {
      "epoch": 1.7657790927021697,
      "grad_norm": 0.3125,
      "learning_rate": 0.00016344431777295414,
      "loss": 0.0561,
      "step": 7162
    },
    {
      "epoch": 1.7662721893491125,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00016341770481725706,
      "loss": 0.0563,
      "step": 7164
    },
    {
      "epoch": 1.7667652859960552,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00016339108434620787,
      "loss": 0.0479,
      "step": 7166
    },
    {
      "epoch": 1.7672583826429982,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00016336445636296122,
      "loss": 0.0551,
      "step": 7168
    },
    {
      "epoch": 1.7677514792899407,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00016333782087067276,
      "loss": 0.0557,
      "step": 7170
    },
    {
      "epoch": 1.7682445759368837,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00016331117787249884,
      "loss": 0.0517,
      "step": 7172
    },
    {
      "epoch": 1.7687376725838264,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00016328452737159692,
      "loss": 0.0533,
      "step": 7174
    },
    {
      "epoch": 1.7692307692307692,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00016325786937112512,
      "loss": 0.0467,
      "step": 7176
    },
    {
      "epoch": 1.7697238658777121,
      "grad_norm": 0.4375,
      "learning_rate": 0.00016323120387424264,
      "loss": 0.0584,
      "step": 7178
    },
    {
      "epoch": 1.7702169625246549,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00016320453088410942,
      "loss": 0.0522,
      "step": 7180
    },
    {
      "epoch": 1.7707100591715976,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00016317785040388647,
      "loss": 0.0517,
      "step": 7182
    },
    {
      "epoch": 1.7712031558185406,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00016315116243673545,
      "loss": 0.0534,
      "step": 7184
    },
    {
      "epoch": 1.7716962524654831,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00016312446698581913,
      "loss": 0.0554,
      "step": 7186
    },
    {
      "epoch": 1.772189349112426,
      "grad_norm": 0.345703125,
      "learning_rate": 0.000163097764054301,
      "loss": 0.0564,
      "step": 7188
    },
    {
      "epoch": 1.7726824457593688,
      "grad_norm": 0.625,
      "learning_rate": 0.00016307105364534556,
      "loss": 0.0532,
      "step": 7190
    },
    {
      "epoch": 1.7731755424063116,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00016304433576211808,
      "loss": 0.054,
      "step": 7192
    },
    {
      "epoch": 1.7736686390532546,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00016301761040778487,
      "loss": 0.0527,
      "step": 7194
    },
    {
      "epoch": 1.774161735700197,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00016299087758551298,
      "loss": 0.0631,
      "step": 7196
    },
    {
      "epoch": 1.77465483234714,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00016296413729847034,
      "loss": 0.0598,
      "step": 7198
    },
    {
      "epoch": 1.7751479289940828,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00016293738954982592,
      "loss": 0.0553,
      "step": 7200
    },
    {
      "epoch": 1.7756410256410255,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00016291063434274947,
      "loss": 0.0581,
      "step": 7202
    },
    {
      "epoch": 1.7761341222879685,
      "grad_norm": 0.515625,
      "learning_rate": 0.00016288387168041156,
      "loss": 0.0534,
      "step": 7204
    },
    {
      "epoch": 1.7766272189349113,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00016285710156598376,
      "loss": 0.0544,
      "step": 7206
    },
    {
      "epoch": 1.777120315581854,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00016283032400263848,
      "loss": 0.0538,
      "step": 7208
    },
    {
      "epoch": 1.777613412228797,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00016280353899354903,
      "loss": 0.0551,
      "step": 7210
    },
    {
      "epoch": 1.7781065088757395,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00016277674654188953,
      "loss": 0.0563,
      "step": 7212
    },
    {
      "epoch": 1.7785996055226825,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001627499466508351,
      "loss": 0.0556,
      "step": 7214
    },
    {
      "epoch": 1.7790927021696252,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00016272313932356162,
      "loss": 0.0529,
      "step": 7216
    },
    {
      "epoch": 1.779585798816568,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00016269632456324592,
      "loss": 0.0562,
      "step": 7218
    },
    {
      "epoch": 1.780078895463511,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001626695023730657,
      "loss": 0.0526,
      "step": 7220
    },
    {
      "epoch": 1.7805719921104537,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00016264267275619958,
      "loss": 0.0523,
      "step": 7222
    },
    {
      "epoch": 1.7810650887573964,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00016261583571582696,
      "loss": 0.0526,
      "step": 7224
    },
    {
      "epoch": 1.7815581854043394,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00016258899125512821,
      "loss": 0.0541,
      "step": 7226
    },
    {
      "epoch": 1.782051282051282,
      "grad_norm": 0.296875,
      "learning_rate": 0.00016256213937728455,
      "loss": 0.0575,
      "step": 7228
    },
    {
      "epoch": 1.782544378698225,
      "grad_norm": 0.421875,
      "learning_rate": 0.00016253528008547806,
      "loss": 0.0579,
      "step": 7230
    },
    {
      "epoch": 1.7830374753451677,
      "grad_norm": 0.34375,
      "learning_rate": 0.00016250841338289176,
      "loss": 0.0507,
      "step": 7232
    },
    {
      "epoch": 1.7835305719921104,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00016248153927270943,
      "loss": 0.0594,
      "step": 7234
    },
    {
      "epoch": 1.7840236686390534,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00016245465775811585,
      "loss": 0.0491,
      "step": 7236
    },
    {
      "epoch": 1.784516765285996,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00016242776884229664,
      "loss": 0.0524,
      "step": 7238
    },
    {
      "epoch": 1.7850098619329389,
      "grad_norm": 0.40625,
      "learning_rate": 0.00016240087252843827,
      "loss": 0.0568,
      "step": 7240
    },
    {
      "epoch": 1.7855029585798816,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001623739688197281,
      "loss": 0.0498,
      "step": 7242
    },
    {
      "epoch": 1.7859960552268244,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00016234705771935438,
      "loss": 0.0505,
      "step": 7244
    },
    {
      "epoch": 1.7864891518737673,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00016232013923050622,
      "loss": 0.0521,
      "step": 7246
    },
    {
      "epoch": 1.78698224852071,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001622932133563736,
      "loss": 0.0533,
      "step": 7248
    },
    {
      "epoch": 1.7874753451676528,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00016226628010014742,
      "loss": 0.0549,
      "step": 7250
    },
    {
      "epoch": 1.7879684418145958,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00016223933946501936,
      "loss": 0.0573,
      "step": 7252
    },
    {
      "epoch": 1.7884615384615383,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00016221239145418207,
      "loss": 0.0592,
      "step": 7254
    },
    {
      "epoch": 1.7889546351084813,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00016218543607082907,
      "loss": 0.053,
      "step": 7256
    },
    {
      "epoch": 1.789447731755424,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0001621584733181547,
      "loss": 0.0565,
      "step": 7258
    },
    {
      "epoch": 1.7899408284023668,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00016213150319935418,
      "loss": 0.0565,
      "step": 7260
    },
    {
      "epoch": 1.7904339250493098,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00016210452571762364,
      "loss": 0.0513,
      "step": 7262
    },
    {
      "epoch": 1.7909270216962525,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00016207754087616005,
      "loss": 0.0571,
      "step": 7264
    },
    {
      "epoch": 1.7914201183431953,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00016205054867816126,
      "loss": 0.0555,
      "step": 7266
    },
    {
      "epoch": 1.7919132149901382,
      "grad_norm": 0.439453125,
      "learning_rate": 0.000162023549126826,
      "loss": 0.055,
      "step": 7268
    },
    {
      "epoch": 1.7924063116370808,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001619965422253539,
      "loss": 0.0547,
      "step": 7270
    },
    {
      "epoch": 1.7928994082840237,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001619695279769454,
      "loss": 0.052,
      "step": 7272
    },
    {
      "epoch": 1.7933925049309665,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001619425063848018,
      "loss": 0.0524,
      "step": 7274
    },
    {
      "epoch": 1.7938856015779092,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001619154774521254,
      "loss": 0.054,
      "step": 7276
    },
    {
      "epoch": 1.7943786982248522,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00016188844118211924,
      "loss": 0.052,
      "step": 7278
    },
    {
      "epoch": 1.7948717948717947,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00016186139757798723,
      "loss": 0.0561,
      "step": 7280
    },
    {
      "epoch": 1.7953648915187377,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00016183434664293421,
      "loss": 0.0546,
      "step": 7282
    },
    {
      "epoch": 1.7958579881656804,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001618072883801659,
      "loss": 0.0549,
      "step": 7284
    },
    {
      "epoch": 1.7963510848126232,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001617802227928888,
      "loss": 0.0559,
      "step": 7286
    },
    {
      "epoch": 1.7968441814595661,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00016175314988431036,
      "loss": 0.0566,
      "step": 7288
    },
    {
      "epoch": 1.797337278106509,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001617260696576389,
      "loss": 0.0541,
      "step": 7290
    },
    {
      "epoch": 1.7978303747534516,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00016169898211608352,
      "loss": 0.0547,
      "step": 7292
    },
    {
      "epoch": 1.7983234714003946,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00016167188726285434,
      "loss": 0.0549,
      "step": 7294
    },
    {
      "epoch": 1.7988165680473371,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00016164478510116213,
      "loss": 0.0509,
      "step": 7296
    },
    {
      "epoch": 1.79930966469428,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001616176756342187,
      "loss": 0.0551,
      "step": 7298
    },
    {
      "epoch": 1.7998027613412229,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001615905588652367,
      "loss": 0.0517,
      "step": 7300
    },
    {
      "epoch": 1.8002958579881656,
      "grad_norm": 0.40625,
      "learning_rate": 0.00016156343479742963,
      "loss": 0.0568,
      "step": 7302
    },
    {
      "epoch": 1.8007889546351086,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00016153630343401174,
      "loss": 0.0578,
      "step": 7304
    },
    {
      "epoch": 1.8012820512820513,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00016150916477819836,
      "loss": 0.0573,
      "step": 7306
    },
    {
      "epoch": 1.801775147928994,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00016148201883320554,
      "loss": 0.0543,
      "step": 7308
    },
    {
      "epoch": 1.802268244575937,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00016145486560225017,
      "loss": 0.0515,
      "step": 7310
    },
    {
      "epoch": 1.8027613412228796,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00016142770508855017,
      "loss": 0.0515,
      "step": 7312
    },
    {
      "epoch": 1.8032544378698225,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00016140053729532408,
      "loss": 0.0571,
      "step": 7314
    },
    {
      "epoch": 1.8037475345167653,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00016137336222579154,
      "loss": 0.0512,
      "step": 7316
    },
    {
      "epoch": 1.804240631163708,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00016134617988317293,
      "loss": 0.0549,
      "step": 7318
    },
    {
      "epoch": 1.804733727810651,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00016131899027068945,
      "loss": 0.0492,
      "step": 7320
    },
    {
      "epoch": 1.8052268244575935,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001612917933915633,
      "loss": 0.0548,
      "step": 7322
    },
    {
      "epoch": 1.8057199211045365,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00016126458924901738,
      "loss": 0.0591,
      "step": 7324
    },
    {
      "epoch": 1.8062130177514792,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00016123737784627562,
      "loss": 0.0581,
      "step": 7326
    },
    {
      "epoch": 1.806706114398422,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00016121015918656268,
      "loss": 0.0568,
      "step": 7328
    },
    {
      "epoch": 1.807199211045365,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00016118293327310406,
      "loss": 0.0536,
      "step": 7330
    },
    {
      "epoch": 1.8076923076923077,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001611557001091263,
      "loss": 0.0558,
      "step": 7332
    },
    {
      "epoch": 1.8081854043392505,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00016112845969785665,
      "loss": 0.0497,
      "step": 7334
    },
    {
      "epoch": 1.8086785009861934,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00016110121204252316,
      "loss": 0.0527,
      "step": 7336
    },
    {
      "epoch": 1.809171597633136,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00016107395714635493,
      "loss": 0.0545,
      "step": 7338
    },
    {
      "epoch": 1.809664694280079,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001610466950125818,
      "loss": 0.0558,
      "step": 7340
    },
    {
      "epoch": 1.8101577909270217,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00016101942564443445,
      "loss": 0.0518,
      "step": 7342
    },
    {
      "epoch": 1.8106508875739644,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001609921490451445,
      "loss": 0.0549,
      "step": 7344
    },
    {
      "epoch": 1.8111439842209074,
      "grad_norm": 0.359375,
      "learning_rate": 0.00016096486521794434,
      "loss": 0.0522,
      "step": 7346
    },
    {
      "epoch": 1.8116370808678501,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00016093757416606725,
      "loss": 0.0521,
      "step": 7348
    },
    {
      "epoch": 1.8121301775147929,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001609102758927474,
      "loss": 0.0524,
      "step": 7350
    },
    {
      "epoch": 1.8126232741617359,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00016088297040121977,
      "loss": 0.0514,
      "step": 7352
    },
    {
      "epoch": 1.8131163708086784,
      "grad_norm": 0.328125,
      "learning_rate": 0.00016085565769472022,
      "loss": 0.0528,
      "step": 7354
    },
    {
      "epoch": 1.8136094674556213,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00016082833777648548,
      "loss": 0.0533,
      "step": 7356
    },
    {
      "epoch": 1.814102564102564,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00016080101064975305,
      "loss": 0.0531,
      "step": 7358
    },
    {
      "epoch": 1.8145956607495068,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001607736763177614,
      "loss": 0.0567,
      "step": 7360
    },
    {
      "epoch": 1.8150887573964498,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00016074633478374982,
      "loss": 0.0576,
      "step": 7362
    },
    {
      "epoch": 1.8155818540433923,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001607189860509584,
      "loss": 0.051,
      "step": 7364
    },
    {
      "epoch": 1.8160749506903353,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001606916301226281,
      "loss": 0.059,
      "step": 7366
    },
    {
      "epoch": 1.816568047337278,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00016066426700200078,
      "loss": 0.0538,
      "step": 7368
    },
    {
      "epoch": 1.8170611439842208,
      "grad_norm": 0.5,
      "learning_rate": 0.0001606368966923191,
      "loss": 0.0524,
      "step": 7370
    },
    {
      "epoch": 1.8175542406311638,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00016060951919682665,
      "loss": 0.055,
      "step": 7372
    },
    {
      "epoch": 1.8180473372781065,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00016058213451876772,
      "loss": 0.0557,
      "step": 7374
    },
    {
      "epoch": 1.8185404339250493,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001605547426613877,
      "loss": 0.054,
      "step": 7376
    },
    {
      "epoch": 1.8190335305719922,
      "grad_norm": 0.453125,
      "learning_rate": 0.00016052734362793247,
      "loss": 0.0579,
      "step": 7378
    },
    {
      "epoch": 1.8195266272189348,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001604999374216491,
      "loss": 0.0515,
      "step": 7380
    },
    {
      "epoch": 1.8200197238658777,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00016047252404578538,
      "loss": 0.0553,
      "step": 7382
    },
    {
      "epoch": 1.8205128205128205,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00016044510350358995,
      "loss": 0.0514,
      "step": 7384
    },
    {
      "epoch": 1.8210059171597632,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00016041767579831224,
      "loss": 0.0586,
      "step": 7386
    },
    {
      "epoch": 1.8214990138067062,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016039024093320262,
      "loss": 0.0482,
      "step": 7388
    },
    {
      "epoch": 1.821992110453649,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00016036279891151227,
      "loss": 0.0568,
      "step": 7390
    },
    {
      "epoch": 1.8224852071005917,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001603353497364932,
      "loss": 0.0579,
      "step": 7392
    },
    {
      "epoch": 1.8229783037475347,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001603078934113984,
      "loss": 0.0526,
      "step": 7394
    },
    {
      "epoch": 1.8234714003944772,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00016028042993948148,
      "loss": 0.0552,
      "step": 7396
    },
    {
      "epoch": 1.8239644970414202,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00016025295932399703,
      "loss": 0.0534,
      "step": 7398
    },
    {
      "epoch": 1.824457593688363,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00016022548156820047,
      "loss": 0.0572,
      "step": 7400
    },
    {
      "epoch": 1.8249506903353057,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00016019799667534813,
      "loss": 0.0586,
      "step": 7402
    },
    {
      "epoch": 1.8254437869822486,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00016017050464869705,
      "loss": 0.0546,
      "step": 7404
    },
    {
      "epoch": 1.8259368836291914,
      "grad_norm": 0.375,
      "learning_rate": 0.00016014300549150521,
      "loss": 0.0551,
      "step": 7406
    },
    {
      "epoch": 1.8264299802761341,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0001601154992070315,
      "loss": 0.0562,
      "step": 7408
    },
    {
      "epoch": 1.8269230769230769,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00016008798579853544,
      "loss": 0.052,
      "step": 7410
    },
    {
      "epoch": 1.8274161735700196,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00016006046526927757,
      "loss": 0.0508,
      "step": 7412
    },
    {
      "epoch": 1.8279092702169626,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00016003293762251923,
      "loss": 0.0507,
      "step": 7414
    },
    {
      "epoch": 1.8284023668639053,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00016000540286152258,
      "loss": 0.051,
      "step": 7416
    },
    {
      "epoch": 1.828895463510848,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001599778609895507,
      "loss": 0.056,
      "step": 7418
    },
    {
      "epoch": 1.829388560157791,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00015995031200986742,
      "loss": 0.0543,
      "step": 7420
    },
    {
      "epoch": 1.8298816568047336,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00015992275592573743,
      "loss": 0.0519,
      "step": 7422
    },
    {
      "epoch": 1.8303747534516766,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001598951927404263,
      "loss": 0.0558,
      "step": 7424
    },
    {
      "epoch": 1.8308678500986193,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00015986762245720043,
      "loss": 0.0542,
      "step": 7426
    },
    {
      "epoch": 1.831360946745562,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00015984004507932703,
      "loss": 0.0566,
      "step": 7428
    },
    {
      "epoch": 1.831854043392505,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001598124606100742,
      "loss": 0.0507,
      "step": 7430
    },
    {
      "epoch": 1.8323471400394478,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00015978486905271082,
      "loss": 0.0595,
      "step": 7432
    },
    {
      "epoch": 1.8328402366863905,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015975727041050669,
      "loss": 0.0486,
      "step": 7434
    },
    {
      "epoch": 1.8333333333333335,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00015972966468673233,
      "loss": 0.0541,
      "step": 7436
    },
    {
      "epoch": 1.833826429980276,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00015970205188465923,
      "loss": 0.0563,
      "step": 7438
    },
    {
      "epoch": 1.834319526627219,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00015967443200755964,
      "loss": 0.0521,
      "step": 7440
    },
    {
      "epoch": 1.8348126232741617,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00015964680505870673,
      "loss": 0.055,
      "step": 7442
    },
    {
      "epoch": 1.8353057199211045,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00015961917104137437,
      "loss": 0.0581,
      "step": 7444
    },
    {
      "epoch": 1.8357988165680474,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00015959152995883734,
      "loss": 0.0554,
      "step": 7446
    },
    {
      "epoch": 1.8362919132149902,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001595638818143713,
      "loss": 0.0562,
      "step": 7448
    },
    {
      "epoch": 1.836785009861933,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00015953622661125274,
      "loss": 0.0576,
      "step": 7450
    },
    {
      "epoch": 1.8372781065088757,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00015950856435275888,
      "loss": 0.0545,
      "step": 7452
    },
    {
      "epoch": 1.8377712031558184,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001594808950421679,
      "loss": 0.0521,
      "step": 7454
    },
    {
      "epoch": 1.8382642998027614,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00015945321868275877,
      "loss": 0.0578,
      "step": 7456
    },
    {
      "epoch": 1.8387573964497042,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015942553527781126,
      "loss": 0.0532,
      "step": 7458
    },
    {
      "epoch": 1.839250493096647,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00015939784483060606,
      "loss": 0.0558,
      "step": 7460
    },
    {
      "epoch": 1.8397435897435899,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001593701473444246,
      "loss": 0.0551,
      "step": 7462
    },
    {
      "epoch": 1.8402366863905324,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001593424428225492,
      "loss": 0.0519,
      "step": 7464
    },
    {
      "epoch": 1.8407297830374754,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00015931473126826298,
      "loss": 0.0511,
      "step": 7466
    },
    {
      "epoch": 1.8412228796844181,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00015928701268484995,
      "loss": 0.0582,
      "step": 7468
    },
    {
      "epoch": 1.8417159763313609,
      "grad_norm": 0.375,
      "learning_rate": 0.0001592592870755949,
      "loss": 0.0529,
      "step": 7470
    },
    {
      "epoch": 1.8422090729783038,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001592315544437835,
      "loss": 0.0539,
      "step": 7472
    },
    {
      "epoch": 1.8427021696252466,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00015920381479270215,
      "loss": 0.0581,
      "step": 7474
    },
    {
      "epoch": 1.8431952662721893,
      "grad_norm": 0.375,
      "learning_rate": 0.0001591760681256382,
      "loss": 0.0537,
      "step": 7476
    },
    {
      "epoch": 1.8436883629191323,
      "grad_norm": 0.375,
      "learning_rate": 0.0001591483144458798,
      "loss": 0.0546,
      "step": 7478
    },
    {
      "epoch": 1.8441814595660748,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001591205537567159,
      "loss": 0.0518,
      "step": 7480
    },
    {
      "epoch": 1.8446745562130178,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015909278606143627,
      "loss": 0.0539,
      "step": 7482
    },
    {
      "epoch": 1.8451676528599605,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00015906501136333156,
      "loss": 0.0519,
      "step": 7484
    },
    {
      "epoch": 1.8456607495069033,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00015903722966569323,
      "loss": 0.0532,
      "step": 7486
    },
    {
      "epoch": 1.8461538461538463,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00015900944097181356,
      "loss": 0.0502,
      "step": 7488
    },
    {
      "epoch": 1.846646942800789,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015898164528498567,
      "loss": 0.0564,
      "step": 7490
    },
    {
      "epoch": 1.8471400394477318,
      "grad_norm": 0.4375,
      "learning_rate": 0.00015895384260850354,
      "loss": 0.0521,
      "step": 7492
    },
    {
      "epoch": 1.8476331360946747,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00015892603294566187,
      "loss": 0.0505,
      "step": 7494
    },
    {
      "epoch": 1.8481262327416172,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00015889821629975627,
      "loss": 0.0551,
      "step": 7496
    },
    {
      "epoch": 1.8486193293885602,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00015887039267408325,
      "loss": 0.0513,
      "step": 7498
    },
    {
      "epoch": 1.849112426035503,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00015884256207193996,
      "loss": 0.0549,
      "step": 7500
    },
    {
      "epoch": 1.8496055226824457,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00015881472449662455,
      "loss": 0.0523,
      "step": 7502
    },
    {
      "epoch": 1.8500986193293887,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00015878687995143592,
      "loss": 0.0517,
      "step": 7504
    },
    {
      "epoch": 1.8505917159763312,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00015875902843967375,
      "loss": 0.0506,
      "step": 7506
    },
    {
      "epoch": 1.8510848126232742,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00015873116996463867,
      "loss": 0.0536,
      "step": 7508
    },
    {
      "epoch": 1.851577909270217,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00015870330452963204,
      "loss": 0.0568,
      "step": 7510
    },
    {
      "epoch": 1.8520710059171597,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00015867543213795608,
      "loss": 0.0525,
      "step": 7512
    },
    {
      "epoch": 1.8525641025641026,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00015864755279291378,
      "loss": 0.0558,
      "step": 7514
    },
    {
      "epoch": 1.8530571992110454,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00015861966649780907,
      "loss": 0.0482,
      "step": 7516
    },
    {
      "epoch": 1.8535502958579881,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00015859177325594655,
      "loss": 0.0519,
      "step": 7518
    },
    {
      "epoch": 1.854043392504931,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015856387307063176,
      "loss": 0.0579,
      "step": 7520
    },
    {
      "epoch": 1.8545364891518736,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00015853596594517103,
      "loss": 0.0613,
      "step": 7522
    },
    {
      "epoch": 1.8550295857988166,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00015850805188287156,
      "loss": 0.0558,
      "step": 7524
    },
    {
      "epoch": 1.8555226824457594,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00015848013088704126,
      "loss": 0.0575,
      "step": 7526
    },
    {
      "epoch": 1.856015779092702,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001584522029609889,
      "loss": 0.0604,
      "step": 7528
    },
    {
      "epoch": 1.856508875739645,
      "grad_norm": 0.390625,
      "learning_rate": 0.00015842426810802414,
      "loss": 0.0529,
      "step": 7530
    },
    {
      "epoch": 1.8570019723865878,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00015839632633145745,
      "loss": 0.0524,
      "step": 7532
    },
    {
      "epoch": 1.8574950690335306,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001583683776346,
      "loss": 0.0586,
      "step": 7534
    },
    {
      "epoch": 1.8579881656804735,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00015834042202076393,
      "loss": 0.0546,
      "step": 7536
    },
    {
      "epoch": 1.858481262327416,
      "grad_norm": 0.375,
      "learning_rate": 0.00015831245949326212,
      "loss": 0.0511,
      "step": 7538
    },
    {
      "epoch": 1.858974358974359,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001582844900554083,
      "loss": 0.0502,
      "step": 7540
    },
    {
      "epoch": 1.8594674556213018,
      "grad_norm": 0.287109375,
      "learning_rate": 0.000158256513710517,
      "loss": 0.0516,
      "step": 7542
    },
    {
      "epoch": 1.8599605522682445,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00015822853046190355,
      "loss": 0.0532,
      "step": 7544
    },
    {
      "epoch": 1.8604536489151875,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001582005403128841,
      "loss": 0.0539,
      "step": 7546
    },
    {
      "epoch": 1.86094674556213,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001581725432667757,
      "loss": 0.0553,
      "step": 7548
    },
    {
      "epoch": 1.861439842209073,
      "grad_norm": 0.40625,
      "learning_rate": 0.00015814453932689614,
      "loss": 0.0476,
      "step": 7550
    },
    {
      "epoch": 1.8619329388560157,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000158116528496564,
      "loss": 0.0523,
      "step": 7552
    },
    {
      "epoch": 1.8624260355029585,
      "grad_norm": 0.453125,
      "learning_rate": 0.00015808851077909878,
      "loss": 0.0514,
      "step": 7554
    },
    {
      "epoch": 1.8629191321499015,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001580604861778207,
      "loss": 0.0551,
      "step": 7556
    },
    {
      "epoch": 1.8634122287968442,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00015803245469605087,
      "loss": 0.049,
      "step": 7558
    },
    {
      "epoch": 1.863905325443787,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00015800441633711108,
      "loss": 0.0517,
      "step": 7560
    },
    {
      "epoch": 1.86439842209073,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00015797637110432416,
      "loss": 0.059,
      "step": 7562
    },
    {
      "epoch": 1.8648915187376724,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00015794831900101352,
      "loss": 0.0518,
      "step": 7564
    },
    {
      "epoch": 1.8653846153846154,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00015792026003050356,
      "loss": 0.0538,
      "step": 7566
    },
    {
      "epoch": 1.8658777120315582,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00015789219419611942,
      "loss": 0.0568,
      "step": 7568
    },
    {
      "epoch": 1.866370808678501,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00015786412150118698,
      "loss": 0.0534,
      "step": 7570
    },
    {
      "epoch": 1.8668639053254439,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00015783604194903313,
      "loss": 0.0549,
      "step": 7572
    },
    {
      "epoch": 1.8673570019723866,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00015780795554298536,
      "loss": 0.0509,
      "step": 7574
    },
    {
      "epoch": 1.8678500986193294,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00015777986228637212,
      "loss": 0.054,
      "step": 7576
    },
    {
      "epoch": 1.8683431952662723,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00015775176218252258,
      "loss": 0.0569,
      "step": 7578
    },
    {
      "epoch": 1.8688362919132149,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00015772365523476677,
      "loss": 0.0463,
      "step": 7580
    },
    {
      "epoch": 1.8693293885601578,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00015769554144643558,
      "loss": 0.0593,
      "step": 7582
    },
    {
      "epoch": 1.8698224852071006,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00015766742082086053,
      "loss": 0.0541,
      "step": 7584
    },
    {
      "epoch": 1.8703155818540433,
      "grad_norm": 0.3125,
      "learning_rate": 0.00015763929336137417,
      "loss": 0.0475,
      "step": 7586
    },
    {
      "epoch": 1.8708086785009863,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00015761115907130975,
      "loss": 0.0554,
      "step": 7588
    },
    {
      "epoch": 1.8713017751479288,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001575830179540013,
      "loss": 0.0566,
      "step": 7590
    },
    {
      "epoch": 1.8717948717948718,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001575548700127837,
      "loss": 0.0533,
      "step": 7592
    },
    {
      "epoch": 1.8722879684418146,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001575267152509927,
      "loss": 0.0487,
      "step": 7594
    },
    {
      "epoch": 1.8727810650887573,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00015749855367196472,
      "loss": 0.0505,
      "step": 7596
    },
    {
      "epoch": 1.8732741617357003,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00015747038527903713,
      "loss": 0.0541,
      "step": 7598
    },
    {
      "epoch": 1.873767258382643,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00015744221007554798,
      "loss": 0.0552,
      "step": 7600
    },
    {
      "epoch": 1.8742603550295858,
      "grad_norm": 0.4375,
      "learning_rate": 0.00015741402806483622,
      "loss": 0.0519,
      "step": 7602
    },
    {
      "epoch": 1.8747534516765287,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001573858392502416,
      "loss": 0.0499,
      "step": 7604
    },
    {
      "epoch": 1.8752465483234713,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001573576436351046,
      "loss": 0.0565,
      "step": 7606
    },
    {
      "epoch": 1.8757396449704142,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00015732944122276663,
      "loss": 0.0594,
      "step": 7608
    },
    {
      "epoch": 1.876232741617357,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00015730123201656974,
      "loss": 0.052,
      "step": 7610
    },
    {
      "epoch": 1.8767258382642997,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00015727301601985695,
      "loss": 0.0505,
      "step": 7612
    },
    {
      "epoch": 1.8772189349112427,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015724479323597198,
      "loss": 0.0549,
      "step": 7614
    },
    {
      "epoch": 1.8777120315581854,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00015721656366825943,
      "loss": 0.051,
      "step": 7616
    },
    {
      "epoch": 1.8782051282051282,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00015718832732006458,
      "loss": 0.0562,
      "step": 7618
    },
    {
      "epoch": 1.8786982248520712,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00015716008419473367,
      "loss": 0.0566,
      "step": 7620
    },
    {
      "epoch": 1.8791913214990137,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00015713183429561366,
      "loss": 0.0513,
      "step": 7622
    },
    {
      "epoch": 1.8796844181459567,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00015710357762605229,
      "loss": 0.054,
      "step": 7624
    },
    {
      "epoch": 1.8801775147928994,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00015707531418939811,
      "loss": 0.0527,
      "step": 7626
    },
    {
      "epoch": 1.8806706114398422,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00015704704398900057,
      "loss": 0.052,
      "step": 7628
    },
    {
      "epoch": 1.8811637080867851,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00015701876702820978,
      "loss": 0.0524,
      "step": 7630
    },
    {
      "epoch": 1.8816568047337277,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00015699048331037675,
      "loss": 0.0493,
      "step": 7632
    },
    {
      "epoch": 1.8821499013806706,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00015696219283885328,
      "loss": 0.0542,
      "step": 7634
    },
    {
      "epoch": 1.8826429980276134,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001569338956169919,
      "loss": 0.0501,
      "step": 7636
    },
    {
      "epoch": 1.8831360946745561,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015690559164814602,
      "loss": 0.0595,
      "step": 7638
    },
    {
      "epoch": 1.883629191321499,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001568772809356698,
      "loss": 0.0526,
      "step": 7640
    },
    {
      "epoch": 1.8841222879684418,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00015684896348291827,
      "loss": 0.0484,
      "step": 7642
    },
    {
      "epoch": 1.8846153846153846,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00015682063929324712,
      "loss": 0.053,
      "step": 7644
    },
    {
      "epoch": 1.8851084812623276,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00015679230837001295,
      "loss": 0.058,
      "step": 7646
    },
    {
      "epoch": 1.88560157790927,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015676397071657318,
      "loss": 0.0526,
      "step": 7648
    },
    {
      "epoch": 1.886094674556213,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00015673562633628595,
      "loss": 0.0493,
      "step": 7650
    },
    {
      "epoch": 1.8865877712031558,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00015670727523251022,
      "loss": 0.0476,
      "step": 7652
    },
    {
      "epoch": 1.8870808678500985,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00015667891740860575,
      "loss": 0.0578,
      "step": 7654
    },
    {
      "epoch": 1.8875739644970415,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00015665055286793314,
      "loss": 0.0543,
      "step": 7656
    },
    {
      "epoch": 1.8880670611439843,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015662218161385374,
      "loss": 0.0503,
      "step": 7658
    },
    {
      "epoch": 1.888560157790927,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00015659380364972965,
      "loss": 0.0553,
      "step": 7660
    },
    {
      "epoch": 1.88905325443787,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00015656541897892382,
      "loss": 0.0517,
      "step": 7662
    },
    {
      "epoch": 1.8895463510848125,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00015653702760480006,
      "loss": 0.0521,
      "step": 7664
    },
    {
      "epoch": 1.8900394477317555,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00015650862953072285,
      "loss": 0.0559,
      "step": 7666
    },
    {
      "epoch": 1.8905325443786982,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00015648022476005754,
      "loss": 0.0555,
      "step": 7668
    },
    {
      "epoch": 1.891025641025641,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00015645181329617023,
      "loss": 0.0525,
      "step": 7670
    },
    {
      "epoch": 1.891518737672584,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001564233951424279,
      "loss": 0.0516,
      "step": 7672
    },
    {
      "epoch": 1.8920118343195265,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001563949703021982,
      "loss": 0.0563,
      "step": 7674
    },
    {
      "epoch": 1.8925049309664694,
      "grad_norm": 0.359375,
      "learning_rate": 0.00015636653877884965,
      "loss": 0.053,
      "step": 7676
    },
    {
      "epoch": 1.8929980276134122,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00015633810057575156,
      "loss": 0.0518,
      "step": 7678
    },
    {
      "epoch": 1.893491124260355,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00015630965569627395,
      "loss": 0.0495,
      "step": 7680
    },
    {
      "epoch": 1.893984220907298,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00015628120414378776,
      "loss": 0.0537,
      "step": 7682
    },
    {
      "epoch": 1.8944773175542406,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00015625274592166467,
      "loss": 0.0576,
      "step": 7684
    },
    {
      "epoch": 1.8949704142011834,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001562242810332771,
      "loss": 0.0489,
      "step": 7686
    },
    {
      "epoch": 1.8954635108481264,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00015619580948199834,
      "loss": 0.0555,
      "step": 7688
    },
    {
      "epoch": 1.895956607495069,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0001561673312712024,
      "loss": 0.055,
      "step": 7690
    },
    {
      "epoch": 1.8964497041420119,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001561388464042641,
      "loss": 0.0568,
      "step": 7692
    },
    {
      "epoch": 1.8969428007889546,
      "grad_norm": 0.421875,
      "learning_rate": 0.00015611035488455902,
      "loss": 0.0521,
      "step": 7694
    },
    {
      "epoch": 1.8974358974358974,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001560818567154636,
      "loss": 0.0484,
      "step": 7696
    },
    {
      "epoch": 1.8979289940828403,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001560533519003551,
      "loss": 0.0528,
      "step": 7698
    },
    {
      "epoch": 1.898422090729783,
      "grad_norm": 0.484375,
      "learning_rate": 0.00015602484044261142,
      "loss": 0.0539,
      "step": 7700
    },
    {
      "epoch": 1.8989151873767258,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001559963223456113,
      "loss": 0.0515,
      "step": 7702
    },
    {
      "epoch": 1.8994082840236688,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00015596779761273437,
      "loss": 0.0578,
      "step": 7704
    },
    {
      "epoch": 1.8999013806706113,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00015593926624736098,
      "loss": 0.0549,
      "step": 7706
    },
    {
      "epoch": 1.9003944773175543,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015591072825287214,
      "loss": 0.0571,
      "step": 7708
    },
    {
      "epoch": 1.900887573964497,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001558821836326499,
      "loss": 0.0474,
      "step": 7710
    },
    {
      "epoch": 1.9013806706114398,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00015585363239007684,
      "loss": 0.0504,
      "step": 7712
    },
    {
      "epoch": 1.9018737672583828,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0001558250745285365,
      "loss": 0.0508,
      "step": 7714
    },
    {
      "epoch": 1.9023668639053253,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00015579651005141317,
      "loss": 0.044,
      "step": 7716
    },
    {
      "epoch": 1.9028599605522682,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00015576793896209188,
      "loss": 0.0532,
      "step": 7718
    },
    {
      "epoch": 1.903353057199211,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015573936126395843,
      "loss": 0.0528,
      "step": 7720
    },
    {
      "epoch": 1.9038461538461537,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00015571077696039946,
      "loss": 0.0556,
      "step": 7722
    },
    {
      "epoch": 1.9043392504930967,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00015568218605480237,
      "loss": 0.0629,
      "step": 7724
    },
    {
      "epoch": 1.9048323471400395,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00015565358855055536,
      "loss": 0.0501,
      "step": 7726
    },
    {
      "epoch": 1.9053254437869822,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001556249844510474,
      "loss": 0.0527,
      "step": 7728
    },
    {
      "epoch": 1.9058185404339252,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00015559637375966816,
      "loss": 0.0541,
      "step": 7730
    },
    {
      "epoch": 1.9063116370808677,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00015556775647980825,
      "loss": 0.0517,
      "step": 7732
    },
    {
      "epoch": 1.9068047337278107,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00015553913261485893,
      "loss": 0.0541,
      "step": 7734
    },
    {
      "epoch": 1.9072978303747534,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001555105021682123,
      "loss": 0.0548,
      "step": 7736
    },
    {
      "epoch": 1.9077909270216962,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015548186514326124,
      "loss": 0.0513,
      "step": 7738
    },
    {
      "epoch": 1.9082840236686391,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001554532215433994,
      "loss": 0.0497,
      "step": 7740
    },
    {
      "epoch": 1.9087771203155819,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0001554245713720212,
      "loss": 0.052,
      "step": 7742
    },
    {
      "epoch": 1.9092702169625246,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001553959146325218,
      "loss": 0.0511,
      "step": 7744
    },
    {
      "epoch": 1.9097633136094676,
      "grad_norm": 0.375,
      "learning_rate": 0.0001553672513282973,
      "loss": 0.0555,
      "step": 7746
    },
    {
      "epoch": 1.9102564102564101,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001553385814627443,
      "loss": 0.0505,
      "step": 7748
    },
    {
      "epoch": 1.910749506903353,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00015530990503926048,
      "loss": 0.0546,
      "step": 7750
    },
    {
      "epoch": 1.9112426035502958,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0001552812220612441,
      "loss": 0.0532,
      "step": 7752
    },
    {
      "epoch": 1.9117357001972386,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00015525253253209425,
      "loss": 0.0462,
      "step": 7754
    },
    {
      "epoch": 1.9122287968441816,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001552238364552108,
      "loss": 0.055,
      "step": 7756
    },
    {
      "epoch": 1.9127218934911243,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00015519513383399438,
      "loss": 0.0494,
      "step": 7758
    },
    {
      "epoch": 1.913214990138067,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00015516642467184645,
      "loss": 0.0538,
      "step": 7760
    },
    {
      "epoch": 1.9137080867850098,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015513770897216918,
      "loss": 0.0501,
      "step": 7762
    },
    {
      "epoch": 1.9142011834319526,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00015510898673836554,
      "loss": 0.0525,
      "step": 7764
    },
    {
      "epoch": 1.9146942800788955,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001550802579738393,
      "loss": 0.0547,
      "step": 7766
    },
    {
      "epoch": 1.9151873767258383,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00015505152268199495,
      "loss": 0.0473,
      "step": 7768
    },
    {
      "epoch": 1.915680473372781,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001550227808662378,
      "loss": 0.0501,
      "step": 7770
    },
    {
      "epoch": 1.916173570019724,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001549940325299739,
      "loss": 0.0525,
      "step": 7772
    },
    {
      "epoch": 1.9166666666666665,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00015496527767661013,
      "loss": 0.0485,
      "step": 7774
    },
    {
      "epoch": 1.9171597633136095,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00015493651630955405,
      "loss": 0.0521,
      "step": 7776
    },
    {
      "epoch": 1.9176528599605522,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00015490774843221404,
      "loss": 0.0525,
      "step": 7778
    },
    {
      "epoch": 1.918145956607495,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00015487897404799933,
      "loss": 0.0484,
      "step": 7780
    },
    {
      "epoch": 1.918639053254438,
      "grad_norm": 0.265625,
      "learning_rate": 0.0001548501931603197,
      "loss": 0.0479,
      "step": 7782
    },
    {
      "epoch": 1.9191321499013807,
      "grad_norm": 0.486328125,
      "learning_rate": 0.000154821405772586,
      "loss": 0.0543,
      "step": 7784
    },
    {
      "epoch": 1.9196252465483234,
      "grad_norm": 0.46875,
      "learning_rate": 0.00015479261188820962,
      "loss": 0.0558,
      "step": 7786
    },
    {
      "epoch": 1.9201183431952664,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00015476381151060278,
      "loss": 0.0529,
      "step": 7788
    },
    {
      "epoch": 1.920611439842209,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00015473500464317858,
      "loss": 0.0515,
      "step": 7790
    },
    {
      "epoch": 1.921104536489152,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00015470619128935068,
      "loss": 0.0513,
      "step": 7792
    },
    {
      "epoch": 1.9215976331360947,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00015467737145253373,
      "loss": 0.054,
      "step": 7794
    },
    {
      "epoch": 1.9220907297830374,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001546485451361429,
      "loss": 0.0512,
      "step": 7796
    },
    {
      "epoch": 1.9225838264299804,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001546197123435944,
      "loss": 0.054,
      "step": 7798
    },
    {
      "epoch": 1.9230769230769231,
      "grad_norm": 0.412109375,
      "learning_rate": 0.000154590873078305,
      "loss": 0.05,
      "step": 7800
    },
    {
      "epoch": 1.9235700197238659,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00015456202734369238,
      "loss": 0.0535,
      "step": 7802
    },
    {
      "epoch": 1.9240631163708086,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00015453317514317486,
      "loss": 0.0556,
      "step": 7804
    },
    {
      "epoch": 1.9245562130177514,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00015450431648017163,
      "loss": 0.0597,
      "step": 7806
    },
    {
      "epoch": 1.9250493096646943,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001544754513581026,
      "loss": 0.0495,
      "step": 7808
    },
    {
      "epoch": 1.925542406311637,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00015444657978038836,
      "loss": 0.0539,
      "step": 7810
    },
    {
      "epoch": 1.9260355029585798,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015441770175045048,
      "loss": 0.0538,
      "step": 7812
    },
    {
      "epoch": 1.9265285996055228,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001543888172717111,
      "loss": 0.0565,
      "step": 7814
    },
    {
      "epoch": 1.9270216962524653,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00015435992634759321,
      "loss": 0.0557,
      "step": 7816
    },
    {
      "epoch": 1.9275147928994083,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001543310289815205,
      "loss": 0.0514,
      "step": 7818
    },
    {
      "epoch": 1.928007889546351,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00015430212517691757,
      "loss": 0.0563,
      "step": 7820
    },
    {
      "epoch": 1.9285009861932938,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00015427321493720962,
      "loss": 0.054,
      "step": 7822
    },
    {
      "epoch": 1.9289940828402368,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00015424429826582268,
      "loss": 0.0547,
      "step": 7824
    },
    {
      "epoch": 1.9294871794871795,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001542153751661835,
      "loss": 0.0492,
      "step": 7826
    },
    {
      "epoch": 1.9299802761341223,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00015418644564171968,
      "loss": 0.0501,
      "step": 7828
    },
    {
      "epoch": 1.9304733727810652,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001541575096958595,
      "loss": 0.0542,
      "step": 7830
    },
    {
      "epoch": 1.9309664694280078,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00015412856733203206,
      "loss": 0.0515,
      "step": 7832
    },
    {
      "epoch": 1.9314595660749507,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00015409961855366718,
      "loss": 0.0611,
      "step": 7834
    },
    {
      "epoch": 1.9319526627218935,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001540706633641954,
      "loss": 0.0547,
      "step": 7836
    },
    {
      "epoch": 1.9324457593688362,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00015404170176704817,
      "loss": 0.0504,
      "step": 7838
    },
    {
      "epoch": 1.9329388560157792,
      "grad_norm": 0.3125,
      "learning_rate": 0.00015401273376565756,
      "loss": 0.0527,
      "step": 7840
    },
    {
      "epoch": 1.933431952662722,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001539837593634564,
      "loss": 0.0521,
      "step": 7842
    },
    {
      "epoch": 1.9339250493096647,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00015395477856387834,
      "loss": 0.0554,
      "step": 7844
    },
    {
      "epoch": 1.9344181459566077,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001539257913703578,
      "loss": 0.0542,
      "step": 7846
    },
    {
      "epoch": 1.9349112426035502,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015389679778632988,
      "loss": 0.0566,
      "step": 7848
    },
    {
      "epoch": 1.9354043392504932,
      "grad_norm": 0.59375,
      "learning_rate": 0.00015386779781523052,
      "loss": 0.0528,
      "step": 7850
    },
    {
      "epoch": 1.935897435897436,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00015383879146049634,
      "loss": 0.0492,
      "step": 7852
    },
    {
      "epoch": 1.9363905325443787,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001538097787255648,
      "loss": 0.0571,
      "step": 7854
    },
    {
      "epoch": 1.9368836291913216,
      "grad_norm": 0.40625,
      "learning_rate": 0.00015378075961387403,
      "loss": 0.0496,
      "step": 7856
    },
    {
      "epoch": 1.9373767258382641,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00015375173412886296,
      "loss": 0.053,
      "step": 7858
    },
    {
      "epoch": 1.9378698224852071,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001537227022739713,
      "loss": 0.0478,
      "step": 7860
    },
    {
      "epoch": 1.9383629191321499,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00015369366405263948,
      "loss": 0.05,
      "step": 7862
    },
    {
      "epoch": 1.9388560157790926,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00015366461946830865,
      "loss": 0.0566,
      "step": 7864
    },
    {
      "epoch": 1.9393491124260356,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00015363556852442085,
      "loss": 0.0483,
      "step": 7866
    },
    {
      "epoch": 1.9398422090729783,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00015360651122441868,
      "loss": 0.0529,
      "step": 7868
    },
    {
      "epoch": 1.940335305719921,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00015357744757174563,
      "loss": 0.0556,
      "step": 7870
    },
    {
      "epoch": 1.940828402366864,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001535483775698459,
      "loss": 0.0537,
      "step": 7872
    },
    {
      "epoch": 1.9413214990138066,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00015351930122216442,
      "loss": 0.0544,
      "step": 7874
    },
    {
      "epoch": 1.9418145956607495,
      "grad_norm": 0.318359375,
      "learning_rate": 0.000153490218532147,
      "loss": 0.0497,
      "step": 7876
    },
    {
      "epoch": 1.9423076923076923,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00015346112950323997,
      "loss": 0.0538,
      "step": 7878
    },
    {
      "epoch": 1.942800788954635,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00015343203413889064,
      "loss": 0.0533,
      "step": 7880
    },
    {
      "epoch": 1.943293885601578,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00015340293244254694,
      "loss": 0.053,
      "step": 7882
    },
    {
      "epoch": 1.9437869822485208,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00015337382441765756,
      "loss": 0.0554,
      "step": 7884
    },
    {
      "epoch": 1.9442800788954635,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00015334471006767198,
      "loss": 0.0515,
      "step": 7886
    },
    {
      "epoch": 1.9447731755424065,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00015331558939604047,
      "loss": 0.0561,
      "step": 7888
    },
    {
      "epoch": 1.945266272189349,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001532864624062139,
      "loss": 0.0521,
      "step": 7890
    },
    {
      "epoch": 1.945759368836292,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000153257329101644,
      "loss": 0.0545,
      "step": 7892
    },
    {
      "epoch": 1.9462524654832347,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0001532281894857833,
      "loss": 0.0499,
      "step": 7894
    },
    {
      "epoch": 1.9467455621301775,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00015319904356208498,
      "loss": 0.0516,
      "step": 7896
    },
    {
      "epoch": 1.9472386587771204,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001531698913340029,
      "loss": 0.0522,
      "step": 7898
    },
    {
      "epoch": 1.947731755424063,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00015314073280499186,
      "loss": 0.0499,
      "step": 7900
    },
    {
      "epoch": 1.948224852071006,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00015311156797850726,
      "loss": 0.0609,
      "step": 7902
    },
    {
      "epoch": 1.9487179487179487,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00015308239685800538,
      "loss": 0.0496,
      "step": 7904
    },
    {
      "epoch": 1.9492110453648914,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00015305321944694304,
      "loss": 0.0518,
      "step": 7906
    },
    {
      "epoch": 1.9497041420118344,
      "grad_norm": 0.296875,
      "learning_rate": 0.00015302403574877805,
      "loss": 0.0513,
      "step": 7908
    },
    {
      "epoch": 1.9501972386587771,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00015299484576696871,
      "loss": 0.052,
      "step": 7910
    },
    {
      "epoch": 1.95069033530572,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00015296564950497427,
      "loss": 0.0575,
      "step": 7912
    },
    {
      "epoch": 1.9511834319526629,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00015293644696625464,
      "loss": 0.0576,
      "step": 7914
    },
    {
      "epoch": 1.9516765285996054,
      "grad_norm": 0.421875,
      "learning_rate": 0.00015290723815427053,
      "loss": 0.0561,
      "step": 7916
    },
    {
      "epoch": 1.9521696252465484,
      "grad_norm": 0.328125,
      "learning_rate": 0.00015287802307248325,
      "loss": 0.0534,
      "step": 7918
    },
    {
      "epoch": 1.952662721893491,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00015284880172435506,
      "loss": 0.0575,
      "step": 7920
    },
    {
      "epoch": 1.9531558185404339,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00015281957411334874,
      "loss": 0.0514,
      "step": 7922
    },
    {
      "epoch": 1.9536489151873768,
      "grad_norm": 0.361328125,
      "learning_rate": 0.000152790340242928,
      "loss": 0.0558,
      "step": 7924
    },
    {
      "epoch": 1.9541420118343196,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001527611001165572,
      "loss": 0.048,
      "step": 7926
    },
    {
      "epoch": 1.9546351084812623,
      "grad_norm": 0.5,
      "learning_rate": 0.00015273185373770146,
      "loss": 0.0514,
      "step": 7928
    },
    {
      "epoch": 1.9551282051282053,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001527026011098266,
      "loss": 0.0538,
      "step": 7930
    },
    {
      "epoch": 1.9556213017751478,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001526733422363993,
      "loss": 0.0523,
      "step": 7932
    },
    {
      "epoch": 1.9561143984220908,
      "grad_norm": 0.453125,
      "learning_rate": 0.00015264407712088684,
      "loss": 0.0449,
      "step": 7934
    },
    {
      "epoch": 1.9566074950690335,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00015261480576675733,
      "loss": 0.0527,
      "step": 7936
    },
    {
      "epoch": 1.9571005917159763,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00015258552817747954,
      "loss": 0.0544,
      "step": 7938
    },
    {
      "epoch": 1.9575936883629192,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00015255624435652311,
      "loss": 0.0519,
      "step": 7940
    },
    {
      "epoch": 1.9580867850098618,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00015252695430735823,
      "loss": 0.0543,
      "step": 7942
    },
    {
      "epoch": 1.9585798816568047,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000152497658033456,
      "loss": 0.0628,
      "step": 7944
    },
    {
      "epoch": 1.9590729783037475,
      "grad_norm": 0.421875,
      "learning_rate": 0.00015246835553828821,
      "loss": 0.0511,
      "step": 7946
    },
    {
      "epoch": 1.9595660749506902,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00015243904682532734,
      "loss": 0.0544,
      "step": 7948
    },
    {
      "epoch": 1.9600591715976332,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00015240973189804664,
      "loss": 0.0582,
      "step": 7950
    },
    {
      "epoch": 1.960552268244576,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00015238041075992008,
      "loss": 0.0539,
      "step": 7952
    },
    {
      "epoch": 1.9610453648915187,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001523510834144224,
      "loss": 0.0543,
      "step": 7954
    },
    {
      "epoch": 1.9615384615384617,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00015232174986502906,
      "loss": 0.0569,
      "step": 7956
    },
    {
      "epoch": 1.9620315581854042,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001522924101152162,
      "loss": 0.0505,
      "step": 7958
    },
    {
      "epoch": 1.9625246548323472,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00015226306416846078,
      "loss": 0.0554,
      "step": 7960
    },
    {
      "epoch": 1.96301775147929,
      "grad_norm": 0.40625,
      "learning_rate": 0.00015223371202824048,
      "loss": 0.0538,
      "step": 7962
    },
    {
      "epoch": 1.9635108481262327,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00015220435369803367,
      "loss": 0.0508,
      "step": 7964
    },
    {
      "epoch": 1.9640039447731756,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00015217498918131944,
      "loss": 0.0564,
      "step": 7966
    },
    {
      "epoch": 1.9644970414201184,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001521456184815777,
      "loss": 0.0532,
      "step": 7968
    },
    {
      "epoch": 1.9649901380670611,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00015211624160228906,
      "loss": 0.0526,
      "step": 7970
    },
    {
      "epoch": 1.965483234714004,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00015208685854693475,
      "loss": 0.05,
      "step": 7972
    },
    {
      "epoch": 1.9659763313609466,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00015205746931899693,
      "loss": 0.0535,
      "step": 7974
    },
    {
      "epoch": 1.9664694280078896,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00015202807392195836,
      "loss": 0.0529,
      "step": 7976
    },
    {
      "epoch": 1.9669625246548323,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001519986723593025,
      "loss": 0.051,
      "step": 7978
    },
    {
      "epoch": 1.967455621301775,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00015196926463451367,
      "loss": 0.0524,
      "step": 7980
    },
    {
      "epoch": 1.967948717948718,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001519398507510768,
      "loss": 0.0521,
      "step": 7982
    },
    {
      "epoch": 1.9684418145956606,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001519104307124776,
      "loss": 0.047,
      "step": 7984
    },
    {
      "epoch": 1.9689349112426036,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00015188100452220255,
      "loss": 0.0531,
      "step": 7986
    },
    {
      "epoch": 1.9694280078895463,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00015185157218373883,
      "loss": 0.0612,
      "step": 7988
    },
    {
      "epoch": 1.969921104536489,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001518221337005743,
      "loss": 0.0502,
      "step": 7990
    },
    {
      "epoch": 1.970414201183432,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00015179268907619754,
      "loss": 0.0491,
      "step": 7992
    },
    {
      "epoch": 1.9709072978303748,
      "grad_norm": 0.44140625,
      "learning_rate": 0.000151763238314098,
      "loss": 0.0486,
      "step": 7994
    },
    {
      "epoch": 1.9714003944773175,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00015173378141776568,
      "loss": 0.0542,
      "step": 7996
    },
    {
      "epoch": 1.9718934911242605,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00015170431839069143,
      "loss": 0.0484,
      "step": 7998
    },
    {
      "epoch": 1.972386587771203,
      "grad_norm": 0.375,
      "learning_rate": 0.00015167484923636677,
      "loss": 0.054,
      "step": 8000
    },
    {
      "epoch": 1.972879684418146,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00015164537395828396,
      "loss": 0.0456,
      "step": 8002
    },
    {
      "epoch": 1.9733727810650887,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00015161589255993597,
      "loss": 0.0554,
      "step": 8004
    },
    {
      "epoch": 1.9738658777120315,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00015158640504481658,
      "loss": 0.0566,
      "step": 8006
    },
    {
      "epoch": 1.9743589743589745,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00015155691141642013,
      "loss": 0.0578,
      "step": 8008
    },
    {
      "epoch": 1.9748520710059172,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001515274116782418,
      "loss": 0.0505,
      "step": 8010
    },
    {
      "epoch": 1.97534516765286,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0001514979058337775,
      "loss": 0.0516,
      "step": 8012
    },
    {
      "epoch": 1.975838264299803,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00015146839388652388,
      "loss": 0.0531,
      "step": 8014
    },
    {
      "epoch": 1.9763313609467454,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0001514388758399782,
      "loss": 0.0517,
      "step": 8016
    },
    {
      "epoch": 1.9768244575936884,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00015140935169763851,
      "loss": 0.0477,
      "step": 8018
    },
    {
      "epoch": 1.9773175542406312,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00015137982146300365,
      "loss": 0.0521,
      "step": 8020
    },
    {
      "epoch": 1.977810650887574,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00015135028513957302,
      "loss": 0.0488,
      "step": 8022
    },
    {
      "epoch": 1.9783037475345169,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00015132074273084694,
      "loss": 0.0481,
      "step": 8024
    },
    {
      "epoch": 1.9787968441814594,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00015129119424032628,
      "loss": 0.0567,
      "step": 8026
    },
    {
      "epoch": 1.9792899408284024,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015126163967151275,
      "loss": 0.0529,
      "step": 8028
    },
    {
      "epoch": 1.9797830374753451,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001512320790279087,
      "loss": 0.0536,
      "step": 8030
    },
    {
      "epoch": 1.9802761341222879,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015120251231301728,
      "loss": 0.0536,
      "step": 8032
    },
    {
      "epoch": 1.9807692307692308,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00015117293953034223,
      "loss": 0.0569,
      "step": 8034
    },
    {
      "epoch": 1.9812623274161736,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00015114336068338817,
      "loss": 0.0506,
      "step": 8036
    },
    {
      "epoch": 1.9817554240631163,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00015111377577566027,
      "loss": 0.0506,
      "step": 8038
    },
    {
      "epoch": 1.9822485207100593,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001510841848106646,
      "loss": 0.0545,
      "step": 8040
    },
    {
      "epoch": 1.9827416173570018,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00015105458779190778,
      "loss": 0.0497,
      "step": 8042
    },
    {
      "epoch": 1.9832347140039448,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00015102498472289728,
      "loss": 0.057,
      "step": 8044
    },
    {
      "epoch": 1.9837278106508875,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001509953756071412,
      "loss": 0.0515,
      "step": 8046
    },
    {
      "epoch": 1.9842209072978303,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00015096576044814838,
      "loss": 0.0593,
      "step": 8048
    },
    {
      "epoch": 1.9847140039447733,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00015093613924942837,
      "loss": 0.0551,
      "step": 8050
    },
    {
      "epoch": 1.985207100591716,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001509065120144915,
      "loss": 0.0538,
      "step": 8052
    },
    {
      "epoch": 1.9857001972386588,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00015087687874684873,
      "loss": 0.0523,
      "step": 8054
    },
    {
      "epoch": 1.9861932938856017,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00015084723945001175,
      "loss": 0.0475,
      "step": 8056
    },
    {
      "epoch": 1.9866863905325443,
      "grad_norm": 0.427734375,
      "learning_rate": 0.000150817594127493,
      "loss": 0.0478,
      "step": 8058
    },
    {
      "epoch": 1.9871794871794872,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00015078794278280565,
      "loss": 0.0543,
      "step": 8060
    },
    {
      "epoch": 1.98767258382643,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00015075828541946348,
      "loss": 0.0481,
      "step": 8062
    },
    {
      "epoch": 1.9881656804733727,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001507286220409811,
      "loss": 0.0521,
      "step": 8064
    },
    {
      "epoch": 1.9886587771203157,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001506989526508738,
      "loss": 0.0552,
      "step": 8066
    },
    {
      "epoch": 1.9891518737672582,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0001506692772526575,
      "loss": 0.0509,
      "step": 8068
    },
    {
      "epoch": 1.9896449704142012,
      "grad_norm": 0.392578125,
      "learning_rate": 0.000150639595849849,
      "loss": 0.055,
      "step": 8070
    },
    {
      "epoch": 1.990138067061144,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001506099084459656,
      "loss": 0.0515,
      "step": 8072
    },
    {
      "epoch": 1.9906311637080867,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00015058021504452552,
      "loss": 0.0537,
      "step": 8074
    },
    {
      "epoch": 1.9911242603550297,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00015055051564904755,
      "loss": 0.0539,
      "step": 8076
    },
    {
      "epoch": 1.9916173570019724,
      "grad_norm": 0.34375,
      "learning_rate": 0.00015052081026305122,
      "loss": 0.0519,
      "step": 8078
    },
    {
      "epoch": 1.9921104536489151,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00015049109889005678,
      "loss": 0.0487,
      "step": 8080
    },
    {
      "epoch": 1.9926035502958581,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00015046138153358528,
      "loss": 0.0495,
      "step": 8082
    },
    {
      "epoch": 1.9930966469428006,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00015043165819715829,
      "loss": 0.0525,
      "step": 8084
    },
    {
      "epoch": 1.9935897435897436,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001504019288842982,
      "loss": 0.056,
      "step": 8086
    },
    {
      "epoch": 1.9940828402366864,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00015037219359852818,
      "loss": 0.0561,
      "step": 8088
    },
    {
      "epoch": 1.994575936883629,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00015034245234337197,
      "loss": 0.0471,
      "step": 8090
    },
    {
      "epoch": 1.995069033530572,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00015031270512235405,
      "loss": 0.0588,
      "step": 8092
    },
    {
      "epoch": 1.9955621301775148,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00015028295193899967,
      "loss": 0.0509,
      "step": 8094
    },
    {
      "epoch": 1.9960552268244576,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00015025319279683474,
      "loss": 0.0495,
      "step": 8096
    },
    {
      "epoch": 1.9965483234714005,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00015022342769938592,
      "loss": 0.0525,
      "step": 8098
    },
    {
      "epoch": 1.997041420118343,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00015019365665018048,
      "loss": 0.054,
      "step": 8100
    },
    {
      "epoch": 1.997534516765286,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001501638796527465,
      "loss": 0.0524,
      "step": 8102
    },
    {
      "epoch": 1.9980276134122288,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00015013409671061267,
      "loss": 0.0548,
      "step": 8104
    },
    {
      "epoch": 1.9985207100591715,
      "grad_norm": 0.4375,
      "learning_rate": 0.00015010430782730848,
      "loss": 0.0585,
      "step": 8106
    },
    {
      "epoch": 1.9990138067061145,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00015007451300636405,
      "loss": 0.0493,
      "step": 8108
    },
    {
      "epoch": 1.9995069033530573,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00015004471225131026,
      "loss": 0.0511,
      "step": 8110
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00015001490556567867,
      "loss": 0.0497,
      "step": 8112
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.06181851029396057,
      "eval_runtime": 59.6015,
      "eval_samples_per_second": 267.258,
      "eval_steps_per_second": 2.097,
      "step": 8112
    },
    {
      "epoch": 2.000493096646943,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00014998509295300155,
      "loss": 0.0378,
      "step": 8114
    },
    {
      "epoch": 2.0009861932938855,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001499552744168118,
      "loss": 0.0324,
      "step": 8116
    },
    {
      "epoch": 2.0014792899408285,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00014992544996064307,
      "loss": 0.0336,
      "step": 8118
    },
    {
      "epoch": 2.001972386587771,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00014989561958802985,
      "loss": 0.0389,
      "step": 8120
    },
    {
      "epoch": 2.002465483234714,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0001498657833025071,
      "loss": 0.035,
      "step": 8122
    },
    {
      "epoch": 2.002958579881657,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0001498359411076106,
      "loss": 0.0316,
      "step": 8124
    },
    {
      "epoch": 2.0034516765285995,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00014980609300687683,
      "loss": 0.0417,
      "step": 8126
    },
    {
      "epoch": 2.0039447731755424,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00014977623900384299,
      "loss": 0.043,
      "step": 8128
    },
    {
      "epoch": 2.0044378698224854,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00014974637910204687,
      "loss": 0.039,
      "step": 8130
    },
    {
      "epoch": 2.004930966469428,
      "grad_norm": 0.271484375,
      "learning_rate": 0.0001497165133050271,
      "loss": 0.0374,
      "step": 8132
    },
    {
      "epoch": 2.005424063116371,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001496866416163229,
      "loss": 0.0362,
      "step": 8134
    },
    {
      "epoch": 2.0059171597633134,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00014965676403947428,
      "loss": 0.0391,
      "step": 8136
    },
    {
      "epoch": 2.0064102564102564,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00014962688057802188,
      "loss": 0.0361,
      "step": 8138
    },
    {
      "epoch": 2.0069033530571994,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00014959699123550703,
      "loss": 0.035,
      "step": 8140
    },
    {
      "epoch": 2.007396449704142,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001495670960154718,
      "loss": 0.0358,
      "step": 8142
    },
    {
      "epoch": 2.007889546351085,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00014953719492145898,
      "loss": 0.0364,
      "step": 8144
    },
    {
      "epoch": 2.008382642998028,
      "grad_norm": 0.2578125,
      "learning_rate": 0.00014950728795701194,
      "loss": 0.0374,
      "step": 8146
    },
    {
      "epoch": 2.0088757396449703,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00014947737512567487,
      "loss": 0.0358,
      "step": 8148
    },
    {
      "epoch": 2.0093688362919133,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00014944745643099262,
      "loss": 0.0381,
      "step": 8150
    },
    {
      "epoch": 2.009861932938856,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001494175318765107,
      "loss": 0.0387,
      "step": 8152
    },
    {
      "epoch": 2.010355029585799,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014938760146577534,
      "loss": 0.0358,
      "step": 8154
    },
    {
      "epoch": 2.010848126232742,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00014935766520233348,
      "loss": 0.0327,
      "step": 8156
    },
    {
      "epoch": 2.0113412228796843,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00014932772308973266,
      "loss": 0.0398,
      "step": 8158
    },
    {
      "epoch": 2.0118343195266273,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001492977751315213,
      "loss": 0.0347,
      "step": 8160
    },
    {
      "epoch": 2.01232741617357,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0001492678213312483,
      "loss": 0.0337,
      "step": 8162
    },
    {
      "epoch": 2.0128205128205128,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0001492378616924634,
      "loss": 0.036,
      "step": 8164
    },
    {
      "epoch": 2.0133136094674557,
      "grad_norm": 0.46875,
      "learning_rate": 0.000149207896218717,
      "loss": 0.0348,
      "step": 8166
    },
    {
      "epoch": 2.0138067061143983,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014917792491356017,
      "loss": 0.0355,
      "step": 8168
    },
    {
      "epoch": 2.0142998027613412,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.00014914794778054467,
      "loss": 0.0337,
      "step": 8170
    },
    {
      "epoch": 2.014792899408284,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0001491179648232229,
      "loss": 0.0357,
      "step": 8172
    },
    {
      "epoch": 2.0152859960552267,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0001490879760451481,
      "loss": 0.0416,
      "step": 8174
    },
    {
      "epoch": 2.0157790927021697,
      "grad_norm": 0.265625,
      "learning_rate": 0.00014905798144987403,
      "loss": 0.0351,
      "step": 8176
    },
    {
      "epoch": 2.0162721893491122,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00014902798104095527,
      "loss": 0.0654,
      "step": 8178
    },
    {
      "epoch": 2.016765285996055,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00014899797482194704,
      "loss": 0.0301,
      "step": 8180
    },
    {
      "epoch": 2.017258382642998,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00014896796279640522,
      "loss": 0.0332,
      "step": 8182
    },
    {
      "epoch": 2.0177514792899407,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001489379449678864,
      "loss": 0.0392,
      "step": 8184
    },
    {
      "epoch": 2.0182445759368837,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00014890792133994787,
      "loss": 0.0356,
      "step": 8186
    },
    {
      "epoch": 2.0187376725838266,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00014887789191614758,
      "loss": 0.0341,
      "step": 8188
    },
    {
      "epoch": 2.019230769230769,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0001488478567000442,
      "loss": 0.0351,
      "step": 8190
    },
    {
      "epoch": 2.019723865877712,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00014881781569519705,
      "loss": 0.0428,
      "step": 8192
    },
    {
      "epoch": 2.0202169625246547,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001487877689051662,
      "loss": 0.0348,
      "step": 8194
    },
    {
      "epoch": 2.0207100591715976,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00014875771633351238,
      "loss": 0.0396,
      "step": 8196
    },
    {
      "epoch": 2.0212031558185406,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0001487276579837969,
      "loss": 0.0322,
      "step": 8198
    },
    {
      "epoch": 2.021696252465483,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001486975938595819,
      "loss": 0.035,
      "step": 8200
    },
    {
      "epoch": 2.022189349112426,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00014866752396443014,
      "loss": 0.0351,
      "step": 8202
    },
    {
      "epoch": 2.0226824457593686,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00014863744830190503,
      "loss": 0.032,
      "step": 8204
    },
    {
      "epoch": 2.0231755424063116,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00014860736687557076,
      "loss": 0.0421,
      "step": 8206
    },
    {
      "epoch": 2.0236686390532546,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00014857727968899212,
      "loss": 0.0378,
      "step": 8208
    },
    {
      "epoch": 2.024161735700197,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00014854718674573464,
      "loss": 0.0352,
      "step": 8210
    },
    {
      "epoch": 2.02465483234714,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00014851708804936449,
      "loss": 0.0362,
      "step": 8212
    },
    {
      "epoch": 2.025147928994083,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014848698360344847,
      "loss": 0.0346,
      "step": 8214
    },
    {
      "epoch": 2.0256410256410255,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001484568734115542,
      "loss": 0.0335,
      "step": 8216
    },
    {
      "epoch": 2.0261341222879685,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001484267574772499,
      "loss": 0.0368,
      "step": 8218
    },
    {
      "epoch": 2.026627218934911,
      "grad_norm": 0.40625,
      "learning_rate": 0.00014839663580410445,
      "loss": 0.0349,
      "step": 8220
    },
    {
      "epoch": 2.027120315581854,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00014836650839568743,
      "loss": 0.0318,
      "step": 8222
    },
    {
      "epoch": 2.027613412228797,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00014833637525556913,
      "loss": 0.0367,
      "step": 8224
    },
    {
      "epoch": 2.0281065088757395,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00014830623638732048,
      "loss": 0.0344,
      "step": 8226
    },
    {
      "epoch": 2.0285996055226825,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00014827609179451313,
      "loss": 0.041,
      "step": 8228
    },
    {
      "epoch": 2.0290927021696255,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00014824594148071934,
      "loss": 0.031,
      "step": 8230
    },
    {
      "epoch": 2.029585798816568,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014821578544951212,
      "loss": 0.0333,
      "step": 8232
    },
    {
      "epoch": 2.030078895463511,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00014818562370446511,
      "loss": 0.0339,
      "step": 8234
    },
    {
      "epoch": 2.0305719921104535,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001481554562491527,
      "loss": 0.0373,
      "step": 8236
    },
    {
      "epoch": 2.0310650887573964,
      "grad_norm": 0.28125,
      "learning_rate": 0.00014812528308714984,
      "loss": 0.0379,
      "step": 8238
    },
    {
      "epoch": 2.0315581854043394,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00014809510422203223,
      "loss": 0.0314,
      "step": 8240
    },
    {
      "epoch": 2.032051282051282,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00014806491965737624,
      "loss": 0.0351,
      "step": 8242
    },
    {
      "epoch": 2.032544378698225,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001480347293967589,
      "loss": 0.0367,
      "step": 8244
    },
    {
      "epoch": 2.033037475345168,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00014800453344375794,
      "loss": 0.0337,
      "step": 8246
    },
    {
      "epoch": 2.0335305719921104,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00014797433180195176,
      "loss": 0.0353,
      "step": 8248
    },
    {
      "epoch": 2.0340236686390534,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001479441244749194,
      "loss": 0.0332,
      "step": 8250
    },
    {
      "epoch": 2.034516765285996,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001479139114662406,
      "loss": 0.0368,
      "step": 8252
    },
    {
      "epoch": 2.035009861932939,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00014788369277949576,
      "loss": 0.035,
      "step": 8254
    },
    {
      "epoch": 2.035502958579882,
      "grad_norm": 0.31640625,
      "learning_rate": 0.000147853468418266,
      "loss": 0.032,
      "step": 8256
    },
    {
      "epoch": 2.0359960552268244,
      "grad_norm": 0.2578125,
      "learning_rate": 0.00014782323838613304,
      "loss": 0.0345,
      "step": 8258
    },
    {
      "epoch": 2.0364891518737673,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00014779300268667932,
      "loss": 0.0333,
      "step": 8260
    },
    {
      "epoch": 2.03698224852071,
      "grad_norm": 0.2578125,
      "learning_rate": 0.00014776276132348793,
      "loss": 0.0367,
      "step": 8262
    },
    {
      "epoch": 2.037475345167653,
      "grad_norm": 0.248046875,
      "learning_rate": 0.00014773251430014268,
      "loss": 0.032,
      "step": 8264
    },
    {
      "epoch": 2.037968441814596,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00014770226162022797,
      "loss": 0.0293,
      "step": 8266
    },
    {
      "epoch": 2.0384615384615383,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00014767200328732894,
      "loss": 0.0335,
      "step": 8268
    },
    {
      "epoch": 2.0389546351084813,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014764173930503132,
      "loss": 0.0318,
      "step": 8270
    },
    {
      "epoch": 2.0394477317554243,
      "grad_norm": 0.296875,
      "learning_rate": 0.00014761146967692162,
      "loss": 0.0295,
      "step": 8272
    },
    {
      "epoch": 2.039940828402367,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001475811944065869,
      "loss": 0.0364,
      "step": 8274
    },
    {
      "epoch": 2.0404339250493098,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00014755091349761502,
      "loss": 0.032,
      "step": 8276
    },
    {
      "epoch": 2.0409270216962523,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001475206269535944,
      "loss": 0.0306,
      "step": 8278
    },
    {
      "epoch": 2.0414201183431953,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00014749033477811412,
      "loss": 0.0375,
      "step": 8280
    },
    {
      "epoch": 2.0419132149901382,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00014746003697476404,
      "loss": 0.0317,
      "step": 8282
    },
    {
      "epoch": 2.0424063116370808,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00014742973354713463,
      "loss": 0.0333,
      "step": 8284
    },
    {
      "epoch": 2.0428994082840237,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014739942449881695,
      "loss": 0.0311,
      "step": 8286
    },
    {
      "epoch": 2.0433925049309662,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00014736910983340282,
      "loss": 0.0369,
      "step": 8288
    },
    {
      "epoch": 2.043885601577909,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001473387895544847,
      "loss": 0.0339,
      "step": 8290
    },
    {
      "epoch": 2.044378698224852,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00014730846366565575,
      "loss": 0.0338,
      "step": 8292
    },
    {
      "epoch": 2.0448717948717947,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00014727813217050965,
      "loss": 0.0343,
      "step": 8294
    },
    {
      "epoch": 2.0453648915187377,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00014724779507264098,
      "loss": 0.0346,
      "step": 8296
    },
    {
      "epoch": 2.0458579881656807,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00014721745237564475,
      "loss": 0.0353,
      "step": 8298
    },
    {
      "epoch": 2.046351084812623,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0001471871040831168,
      "loss": 0.0341,
      "step": 8300
    },
    {
      "epoch": 2.046844181459566,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001471567501986536,
      "loss": 0.035,
      "step": 8302
    },
    {
      "epoch": 2.0473372781065087,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00014712639072585215,
      "loss": 0.0326,
      "step": 8304
    },
    {
      "epoch": 2.0478303747534516,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00014709602566831032,
      "loss": 0.0309,
      "step": 8306
    },
    {
      "epoch": 2.0483234714003946,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00014706565502962648,
      "loss": 0.0363,
      "step": 8308
    },
    {
      "epoch": 2.048816568047337,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014703527881339973,
      "loss": 0.0298,
      "step": 8310
    },
    {
      "epoch": 2.04930966469428,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00014700489702322984,
      "loss": 0.0329,
      "step": 8312
    },
    {
      "epoch": 2.049802761341223,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00014697450966271721,
      "loss": 0.0342,
      "step": 8314
    },
    {
      "epoch": 2.0502958579881656,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001469441167354629,
      "loss": 0.0327,
      "step": 8316
    },
    {
      "epoch": 2.0507889546351086,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0001469137182450687,
      "loss": 0.0342,
      "step": 8318
    },
    {
      "epoch": 2.051282051282051,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00014688331419513692,
      "loss": 0.034,
      "step": 8320
    },
    {
      "epoch": 2.051775147928994,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00014685290458927065,
      "loss": 0.0329,
      "step": 8322
    },
    {
      "epoch": 2.052268244575937,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001468224894310736,
      "loss": 0.0337,
      "step": 8324
    },
    {
      "epoch": 2.0527613412228796,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00014679206872415012,
      "loss": 0.0369,
      "step": 8326
    },
    {
      "epoch": 2.0532544378698225,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00014676164247210525,
      "loss": 0.0337,
      "step": 8328
    },
    {
      "epoch": 2.0537475345167655,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00014673121067854464,
      "loss": 0.0311,
      "step": 8330
    },
    {
      "epoch": 2.054240631163708,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00014670077334707468,
      "loss": 0.033,
      "step": 8332
    },
    {
      "epoch": 2.054733727810651,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00014667033048130232,
      "loss": 0.0346,
      "step": 8334
    },
    {
      "epoch": 2.0552268244575935,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00014663988208483524,
      "loss": 0.0306,
      "step": 8336
    },
    {
      "epoch": 2.0557199211045365,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0001466094281612817,
      "loss": 0.0332,
      "step": 8338
    },
    {
      "epoch": 2.0562130177514795,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0001465789687142507,
      "loss": 0.0363,
      "step": 8340
    },
    {
      "epoch": 2.056706114398422,
      "grad_norm": 0.23046875,
      "learning_rate": 0.00014654850374735186,
      "loss": 0.034,
      "step": 8342
    },
    {
      "epoch": 2.057199211045365,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00014651803326419543,
      "loss": 0.0351,
      "step": 8344
    },
    {
      "epoch": 2.0576923076923075,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00014648755726839232,
      "loss": 0.0333,
      "step": 8346
    },
    {
      "epoch": 2.0581854043392505,
      "grad_norm": 0.228515625,
      "learning_rate": 0.00014645707576355413,
      "loss": 0.0312,
      "step": 8348
    },
    {
      "epoch": 2.0586785009861934,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001464265887532931,
      "loss": 0.0313,
      "step": 8350
    },
    {
      "epoch": 2.059171597633136,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00014639609624122205,
      "loss": 0.0305,
      "step": 8352
    },
    {
      "epoch": 2.059664694280079,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00014636559823095457,
      "loss": 0.0327,
      "step": 8354
    },
    {
      "epoch": 2.060157790927022,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00014633509472610482,
      "loss": 0.0339,
      "step": 8356
    },
    {
      "epoch": 2.0606508875739644,
      "grad_norm": 0.248046875,
      "learning_rate": 0.00014630458573028764,
      "loss": 0.0289,
      "step": 8358
    },
    {
      "epoch": 2.0611439842209074,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00014627407124711853,
      "loss": 0.0342,
      "step": 8360
    },
    {
      "epoch": 2.06163708086785,
      "grad_norm": 0.28125,
      "learning_rate": 0.00014624355128021362,
      "loss": 0.0339,
      "step": 8362
    },
    {
      "epoch": 2.062130177514793,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001462130258331897,
      "loss": 0.0366,
      "step": 8364
    },
    {
      "epoch": 2.062623274161736,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00014618249490966418,
      "loss": 0.0336,
      "step": 8366
    },
    {
      "epoch": 2.0631163708086784,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0001461519585132552,
      "loss": 0.0309,
      "step": 8368
    },
    {
      "epoch": 2.0636094674556213,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00014612141664758138,
      "loss": 0.0319,
      "step": 8370
    },
    {
      "epoch": 2.064102564102564,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014609086931626224,
      "loss": 0.0345,
      "step": 8372
    },
    {
      "epoch": 2.064595660749507,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001460603165229177,
      "loss": 0.03,
      "step": 8374
    },
    {
      "epoch": 2.06508875739645,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00014602975827116853,
      "loss": 0.0357,
      "step": 8376
    },
    {
      "epoch": 2.0655818540433923,
      "grad_norm": 0.26953125,
      "learning_rate": 0.000145999194564636,
      "loss": 0.0339,
      "step": 8378
    },
    {
      "epoch": 2.0660749506903353,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00014596862540694206,
      "loss": 0.0287,
      "step": 8380
    },
    {
      "epoch": 2.0665680473372783,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00014593805080170938,
      "loss": 0.0323,
      "step": 8382
    },
    {
      "epoch": 2.067061143984221,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00014590747075256122,
      "loss": 0.0356,
      "step": 8384
    },
    {
      "epoch": 2.0675542406311638,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.00014587688526312143,
      "loss": 0.0299,
      "step": 8386
    },
    {
      "epoch": 2.0680473372781063,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00014584629433701455,
      "loss": 0.0333,
      "step": 8388
    },
    {
      "epoch": 2.0685404339250493,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0001458156979778659,
      "loss": 0.0365,
      "step": 8390
    },
    {
      "epoch": 2.0690335305719922,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00014578509618930118,
      "loss": 0.0309,
      "step": 8392
    },
    {
      "epoch": 2.0695266272189348,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00014575448897494693,
      "loss": 0.0314,
      "step": 8394
    },
    {
      "epoch": 2.0700197238658777,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0001457238763384303,
      "loss": 0.0299,
      "step": 8396
    },
    {
      "epoch": 2.0705128205128207,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00014569325828337902,
      "loss": 0.0316,
      "step": 8398
    },
    {
      "epoch": 2.0710059171597632,
      "grad_norm": 0.26953125,
      "learning_rate": 0.00014566263481342154,
      "loss": 0.0321,
      "step": 8400
    },
    {
      "epoch": 2.071499013806706,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00014563200593218687,
      "loss": 0.0353,
      "step": 8402
    },
    {
      "epoch": 2.0719921104536487,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00014560137164330473,
      "loss": 0.0328,
      "step": 8404
    },
    {
      "epoch": 2.0724852071005917,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00014557073195040543,
      "loss": 0.0329,
      "step": 8406
    },
    {
      "epoch": 2.0729783037475347,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00014554008685711999,
      "loss": 0.0358,
      "step": 8408
    },
    {
      "epoch": 2.073471400394477,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.00014550943636707994,
      "loss": 0.0311,
      "step": 8410
    },
    {
      "epoch": 2.07396449704142,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0001454787804839176,
      "loss": 0.0324,
      "step": 8412
    },
    {
      "epoch": 2.074457593688363,
      "grad_norm": 0.25,
      "learning_rate": 0.00014544811921126587,
      "loss": 0.0312,
      "step": 8414
    },
    {
      "epoch": 2.0749506903353057,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0001454174525527583,
      "loss": 0.0338,
      "step": 8416
    },
    {
      "epoch": 2.0754437869822486,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00014538678051202898,
      "loss": 0.0289,
      "step": 8418
    },
    {
      "epoch": 2.075936883629191,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00014535610309271277,
      "loss": 0.0338,
      "step": 8420
    },
    {
      "epoch": 2.076429980276134,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0001453254202984451,
      "loss": 0.0341,
      "step": 8422
    },
    {
      "epoch": 2.076923076923077,
      "grad_norm": 0.328125,
      "learning_rate": 0.00014529473213286206,
      "loss": 0.0315,
      "step": 8424
    },
    {
      "epoch": 2.0774161735700196,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0001452640385996004,
      "loss": 0.0354,
      "step": 8426
    },
    {
      "epoch": 2.0779092702169626,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001452333397022974,
      "loss": 0.0313,
      "step": 8428
    },
    {
      "epoch": 2.078402366863905,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0001452026354445911,
      "loss": 0.0337,
      "step": 8430
    },
    {
      "epoch": 2.078895463510848,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00014517192583012013,
      "loss": 0.0325,
      "step": 8432
    },
    {
      "epoch": 2.079388560157791,
      "grad_norm": 0.25390625,
      "learning_rate": 0.00014514121086252376,
      "loss": 0.0325,
      "step": 8434
    },
    {
      "epoch": 2.0798816568047336,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00014511049054544186,
      "loss": 0.032,
      "step": 8436
    },
    {
      "epoch": 2.0803747534516766,
      "grad_norm": 0.25,
      "learning_rate": 0.00014507976488251496,
      "loss": 0.032,
      "step": 8438
    },
    {
      "epoch": 2.0808678500986195,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00014504903387738423,
      "loss": 0.0298,
      "step": 8440
    },
    {
      "epoch": 2.081360946745562,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00014501829753369144,
      "loss": 0.0314,
      "step": 8442
    },
    {
      "epoch": 2.081854043392505,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014498755585507907,
      "loss": 0.0339,
      "step": 8444
    },
    {
      "epoch": 2.0823471400394475,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00014495680884519017,
      "loss": 0.0318,
      "step": 8446
    },
    {
      "epoch": 2.0828402366863905,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014492605650766834,
      "loss": 0.0316,
      "step": 8448
    },
    {
      "epoch": 2.0833333333333335,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014489529884615808,
      "loss": 0.0314,
      "step": 8450
    },
    {
      "epoch": 2.083826429980276,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014486453586430418,
      "loss": 0.0352,
      "step": 8452
    },
    {
      "epoch": 2.084319526627219,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00014483376756575228,
      "loss": 0.0322,
      "step": 8454
    },
    {
      "epoch": 2.084812623274162,
      "grad_norm": 0.25390625,
      "learning_rate": 0.00014480299395414862,
      "loss": 0.0353,
      "step": 8456
    },
    {
      "epoch": 2.0853057199211045,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00014477221503314005,
      "loss": 0.0348,
      "step": 8458
    },
    {
      "epoch": 2.0857988165680474,
      "grad_norm": 0.34375,
      "learning_rate": 0.00014474143080637402,
      "loss": 0.0335,
      "step": 8460
    },
    {
      "epoch": 2.08629191321499,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0001447106412774986,
      "loss": 0.0342,
      "step": 8462
    },
    {
      "epoch": 2.086785009861933,
      "grad_norm": 0.328125,
      "learning_rate": 0.00014467984645016258,
      "loss": 0.0313,
      "step": 8464
    },
    {
      "epoch": 2.087278106508876,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0001446490463280153,
      "loss": 0.0306,
      "step": 8466
    },
    {
      "epoch": 2.0877712031558184,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00014461824091470673,
      "loss": 0.0332,
      "step": 8468
    },
    {
      "epoch": 2.0882642998027614,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0001445874302138875,
      "loss": 0.0281,
      "step": 8470
    },
    {
      "epoch": 2.088757396449704,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014455661422920887,
      "loss": 0.0347,
      "step": 8472
    },
    {
      "epoch": 2.089250493096647,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00014452579296432265,
      "loss": 0.0354,
      "step": 8474
    },
    {
      "epoch": 2.08974358974359,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00014449496642288137,
      "loss": 0.0309,
      "step": 8476
    },
    {
      "epoch": 2.0902366863905324,
      "grad_norm": 0.25,
      "learning_rate": 0.00014446413460853812,
      "loss": 0.0304,
      "step": 8478
    },
    {
      "epoch": 2.0907297830374754,
      "grad_norm": 0.296875,
      "learning_rate": 0.00014443329752494667,
      "loss": 0.033,
      "step": 8480
    },
    {
      "epoch": 2.0912228796844183,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001444024551757614,
      "loss": 0.0321,
      "step": 8482
    },
    {
      "epoch": 2.091715976331361,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001443716075646373,
      "loss": 0.0288,
      "step": 8484
    },
    {
      "epoch": 2.092209072978304,
      "grad_norm": 0.390625,
      "learning_rate": 0.00014434075469522994,
      "loss": 0.0299,
      "step": 8486
    },
    {
      "epoch": 2.0927021696252464,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014430989657119558,
      "loss": 0.0302,
      "step": 8488
    },
    {
      "epoch": 2.0931952662721893,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014427903319619108,
      "loss": 0.0314,
      "step": 8490
    },
    {
      "epoch": 2.0936883629191323,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0001442481645738739,
      "loss": 0.0322,
      "step": 8492
    },
    {
      "epoch": 2.094181459566075,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001442172907079022,
      "loss": 0.0335,
      "step": 8494
    },
    {
      "epoch": 2.094674556213018,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00014418641160193467,
      "loss": 0.035,
      "step": 8496
    },
    {
      "epoch": 2.0951676528599608,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00014415552725963065,
      "loss": 0.0309,
      "step": 8498
    },
    {
      "epoch": 2.0956607495069033,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.00014412463768465018,
      "loss": 0.0308,
      "step": 8500
    },
    {
      "epoch": 2.0961538461538463,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001440937428806537,
      "loss": 0.0347,
      "step": 8502
    },
    {
      "epoch": 2.096646942800789,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0001440628428513025,
      "loss": 0.0417,
      "step": 8504
    },
    {
      "epoch": 2.0971400394477318,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00014403193760025845,
      "loss": 0.0313,
      "step": 8506
    },
    {
      "epoch": 2.0976331360946747,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00014400102713118393,
      "loss": 0.0321,
      "step": 8508
    },
    {
      "epoch": 2.0981262327416172,
      "grad_norm": 0.326171875,
      "learning_rate": 0.000143970111447742,
      "loss": 0.0371,
      "step": 8510
    },
    {
      "epoch": 2.09861932938856,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00014393919055359644,
      "loss": 0.0382,
      "step": 8512
    },
    {
      "epoch": 2.0991124260355027,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0001439082644524114,
      "loss": 0.0305,
      "step": 8514
    },
    {
      "epoch": 2.0996055226824457,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00014387733314785193,
      "loss": 0.0294,
      "step": 8516
    },
    {
      "epoch": 2.1000986193293887,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00014384639664358345,
      "loss": 0.0363,
      "step": 8518
    },
    {
      "epoch": 2.100591715976331,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001438154549432722,
      "loss": 0.0334,
      "step": 8520
    },
    {
      "epoch": 2.101084812623274,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00014378450805058488,
      "loss": 0.0302,
      "step": 8522
    },
    {
      "epoch": 2.101577909270217,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001437535559691889,
      "loss": 0.0297,
      "step": 8524
    },
    {
      "epoch": 2.1020710059171597,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00014372259870275227,
      "loss": 0.0325,
      "step": 8526
    },
    {
      "epoch": 2.1025641025641026,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00014369163625494355,
      "loss": 0.0382,
      "step": 8528
    },
    {
      "epoch": 2.103057199211045,
      "grad_norm": 0.267578125,
      "learning_rate": 0.000143660668629432,
      "loss": 0.0319,
      "step": 8530
    },
    {
      "epoch": 2.103550295857988,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00014362969582988746,
      "loss": 0.0308,
      "step": 8532
    },
    {
      "epoch": 2.104043392504931,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00014359871785998035,
      "loss": 0.0425,
      "step": 8534
    },
    {
      "epoch": 2.1045364891518736,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00014356773472338173,
      "loss": 0.0419,
      "step": 8536
    },
    {
      "epoch": 2.1050295857988166,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.00014353674642376334,
      "loss": 0.0312,
      "step": 8538
    },
    {
      "epoch": 2.1055226824457596,
      "grad_norm": 0.248046875,
      "learning_rate": 0.0001435057529647974,
      "loss": 0.028,
      "step": 8540
    },
    {
      "epoch": 2.106015779092702,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00014347475435015685,
      "loss": 0.0394,
      "step": 8542
    },
    {
      "epoch": 2.106508875739645,
      "grad_norm": 0.359375,
      "learning_rate": 0.00014344375058351514,
      "loss": 0.0409,
      "step": 8544
    },
    {
      "epoch": 2.1070019723865876,
      "grad_norm": 0.296875,
      "learning_rate": 0.00014341274166854644,
      "loss": 0.0337,
      "step": 8546
    },
    {
      "epoch": 2.1074950690335306,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00014338172760892546,
      "loss": 0.0309,
      "step": 8548
    },
    {
      "epoch": 2.1079881656804735,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00014335070840832756,
      "loss": 0.0413,
      "step": 8550
    },
    {
      "epoch": 2.108481262327416,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00014331968407042868,
      "loss": 0.0404,
      "step": 8552
    },
    {
      "epoch": 2.108974358974359,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00014328865459890537,
      "loss": 0.0336,
      "step": 8554
    },
    {
      "epoch": 2.109467455621302,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00014325761999743478,
      "loss": 0.0353,
      "step": 8556
    },
    {
      "epoch": 2.1099605522682445,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00014322658026969472,
      "loss": 0.0424,
      "step": 8558
    },
    {
      "epoch": 2.1104536489151875,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00014319553541936354,
      "loss": 0.0496,
      "step": 8560
    },
    {
      "epoch": 2.11094674556213,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00014316448545012022,
      "loss": 0.0349,
      "step": 8562
    },
    {
      "epoch": 2.111439842209073,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001431334303656444,
      "loss": 0.0352,
      "step": 8564
    },
    {
      "epoch": 2.111932938856016,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00014310237016961627,
      "loss": 0.0425,
      "step": 8566
    },
    {
      "epoch": 2.1124260355029585,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001430713048657166,
      "loss": 0.045,
      "step": 8568
    },
    {
      "epoch": 2.1129191321499015,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00014304023445762686,
      "loss": 0.0325,
      "step": 8570
    },
    {
      "epoch": 2.113412228796844,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00014300915894902895,
      "loss": 0.03,
      "step": 8572
    },
    {
      "epoch": 2.113905325443787,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00014297807834360565,
      "loss": 0.0454,
      "step": 8574
    },
    {
      "epoch": 2.11439842209073,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00014294699264504006,
      "loss": 0.0431,
      "step": 8576
    },
    {
      "epoch": 2.1148915187376724,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00014291590185701607,
      "loss": 0.0329,
      "step": 8578
    },
    {
      "epoch": 2.1153846153846154,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0001428848059832181,
      "loss": 0.0335,
      "step": 8580
    },
    {
      "epoch": 2.1158777120315584,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00014285370502733116,
      "loss": 0.0422,
      "step": 8582
    },
    {
      "epoch": 2.116370808678501,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00014282259899304098,
      "loss": 0.0468,
      "step": 8584
    },
    {
      "epoch": 2.116863905325444,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00014279148788403368,
      "loss": 0.0294,
      "step": 8586
    },
    {
      "epoch": 2.1173570019723864,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00014276037170399614,
      "loss": 0.0418,
      "step": 8588
    },
    {
      "epoch": 2.1178500986193294,
      "grad_norm": 2.796875,
      "learning_rate": 0.00014272925045661584,
      "loss": 0.047,
      "step": 8590
    },
    {
      "epoch": 2.1183431952662723,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00014269812414558075,
      "loss": 0.0417,
      "step": 8592
    },
    {
      "epoch": 2.118836291913215,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00014266699277457956,
      "loss": 0.0435,
      "step": 8594
    },
    {
      "epoch": 2.119329388560158,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00014263585634730153,
      "loss": 0.0352,
      "step": 8596
    },
    {
      "epoch": 2.1198224852071004,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0001426047148674365,
      "loss": 0.0445,
      "step": 8598
    },
    {
      "epoch": 2.1203155818540433,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00014257356833867485,
      "loss": 0.0418,
      "step": 8600
    },
    {
      "epoch": 2.1208086785009863,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00014254241676470763,
      "loss": 0.0351,
      "step": 8602
    },
    {
      "epoch": 2.121301775147929,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00014251126014922652,
      "loss": 0.0375,
      "step": 8604
    },
    {
      "epoch": 2.121794871794872,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00014248009849592372,
      "loss": 0.0469,
      "step": 8606
    },
    {
      "epoch": 2.1222879684418148,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00014244893180849206,
      "loss": 0.0452,
      "step": 8608
    },
    {
      "epoch": 2.1227810650887573,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00014241776009062503,
      "loss": 0.0407,
      "step": 8610
    },
    {
      "epoch": 2.1232741617357003,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00014238658334601655,
      "loss": 0.0381,
      "step": 8612
    },
    {
      "epoch": 2.123767258382643,
      "grad_norm": 0.46875,
      "learning_rate": 0.00014235540157836132,
      "loss": 0.0459,
      "step": 8614
    },
    {
      "epoch": 2.1242603550295858,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00014232421479135448,
      "loss": 0.0445,
      "step": 8616
    },
    {
      "epoch": 2.1247534516765287,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00014229302298869188,
      "loss": 0.0451,
      "step": 8618
    },
    {
      "epoch": 2.1252465483234713,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00014226182617406996,
      "loss": 0.0404,
      "step": 8620
    },
    {
      "epoch": 2.1257396449704142,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00014223062435118563,
      "loss": 0.0436,
      "step": 8622
    },
    {
      "epoch": 2.126232741617357,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00014219941752373658,
      "loss": 0.0462,
      "step": 8624
    },
    {
      "epoch": 2.1267258382642997,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001421682056954209,
      "loss": 0.0451,
      "step": 8626
    },
    {
      "epoch": 2.1272189349112427,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00014213698886993744,
      "loss": 0.0483,
      "step": 8628
    },
    {
      "epoch": 2.1277120315581852,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00014210576705098554,
      "loss": 0.0488,
      "step": 8630
    },
    {
      "epoch": 2.128205128205128,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00014207454024226513,
      "loss": 0.0529,
      "step": 8632
    },
    {
      "epoch": 2.128698224852071,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001420433084474768,
      "loss": 0.0448,
      "step": 8634
    },
    {
      "epoch": 2.1291913214990137,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00014201207167032168,
      "loss": 0.0425,
      "step": 8636
    },
    {
      "epoch": 2.1296844181459567,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001419808299145015,
      "loss": 0.0443,
      "step": 8638
    },
    {
      "epoch": 2.1301775147928996,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0001419495831837186,
      "loss": 0.0466,
      "step": 8640
    },
    {
      "epoch": 2.130670611439842,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00014191833148167584,
      "loss": 0.0417,
      "step": 8642
    },
    {
      "epoch": 2.131163708086785,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00014188707481207677,
      "loss": 0.0381,
      "step": 8644
    },
    {
      "epoch": 2.1316568047337277,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00014185581317862546,
      "loss": 0.0454,
      "step": 8646
    },
    {
      "epoch": 2.1321499013806706,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00014182454658502662,
      "loss": 0.0491,
      "step": 8648
    },
    {
      "epoch": 2.1326429980276136,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00014179327503498548,
      "loss": 0.0435,
      "step": 8650
    },
    {
      "epoch": 2.133136094674556,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00014176199853220794,
      "loss": 0.0434,
      "step": 8652
    },
    {
      "epoch": 2.133629191321499,
      "grad_norm": 0.421875,
      "learning_rate": 0.00014173071708040036,
      "loss": 0.0487,
      "step": 8654
    },
    {
      "epoch": 2.1341222879684416,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001416994306832698,
      "loss": 0.0487,
      "step": 8656
    },
    {
      "epoch": 2.1346153846153846,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001416681393445239,
      "loss": 0.0548,
      "step": 8658
    },
    {
      "epoch": 2.1351084812623276,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00014163684306787083,
      "loss": 0.0469,
      "step": 8660
    },
    {
      "epoch": 2.13560157790927,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00014160554185701943,
      "loss": 0.048,
      "step": 8662
    },
    {
      "epoch": 2.136094674556213,
      "grad_norm": 0.4453125,
      "learning_rate": 0.000141574235715679,
      "loss": 0.0456,
      "step": 8664
    },
    {
      "epoch": 2.136587771203156,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001415429246475595,
      "loss": 0.0456,
      "step": 8666
    },
    {
      "epoch": 2.1370808678500985,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001415116086563715,
      "loss": 0.0443,
      "step": 8668
    },
    {
      "epoch": 2.1375739644970415,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00014148028774582612,
      "loss": 0.0523,
      "step": 8670
    },
    {
      "epoch": 2.138067061143984,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000141448961919635,
      "loss": 0.0485,
      "step": 8672
    },
    {
      "epoch": 2.138560157790927,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001414176311815105,
      "loss": 0.043,
      "step": 8674
    },
    {
      "epoch": 2.13905325443787,
      "grad_norm": 0.34375,
      "learning_rate": 0.00014138629553516548,
      "loss": 0.0444,
      "step": 8676
    },
    {
      "epoch": 2.1395463510848125,
      "grad_norm": 0.515625,
      "learning_rate": 0.00014135495498431334,
      "loss": 0.0471,
      "step": 8678
    },
    {
      "epoch": 2.1400394477317555,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00014132360953266816,
      "loss": 0.0544,
      "step": 8680
    },
    {
      "epoch": 2.140532544378698,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00014129225918394452,
      "loss": 0.0468,
      "step": 8682
    },
    {
      "epoch": 2.141025641025641,
      "grad_norm": 0.375,
      "learning_rate": 0.00014126090394185758,
      "loss": 0.0498,
      "step": 8684
    },
    {
      "epoch": 2.141518737672584,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001412295438101232,
      "loss": 0.0592,
      "step": 8686
    },
    {
      "epoch": 2.1420118343195265,
      "grad_norm": 0.453125,
      "learning_rate": 0.00014119817879245766,
      "loss": 0.0503,
      "step": 8688
    },
    {
      "epoch": 2.1425049309664694,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001411668088925779,
      "loss": 0.0491,
      "step": 8690
    },
    {
      "epoch": 2.1429980276134124,
      "grad_norm": 0.3125,
      "learning_rate": 0.00014113543411420148,
      "loss": 0.0517,
      "step": 8692
    },
    {
      "epoch": 2.143491124260355,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00014110405446104643,
      "loss": 0.0523,
      "step": 8694
    },
    {
      "epoch": 2.143984220907298,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00014107266993683139,
      "loss": 0.0537,
      "step": 8696
    },
    {
      "epoch": 2.1444773175542404,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001410412805452757,
      "loss": 0.0475,
      "step": 8698
    },
    {
      "epoch": 2.1449704142011834,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00014100988629009908,
      "loss": 0.0438,
      "step": 8700
    },
    {
      "epoch": 2.1454635108481264,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00014097848717502197,
      "loss": 0.0523,
      "step": 8702
    },
    {
      "epoch": 2.145956607495069,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00014094708320376535,
      "loss": 0.0487,
      "step": 8704
    },
    {
      "epoch": 2.146449704142012,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00014091567438005077,
      "loss": 0.0489,
      "step": 8706
    },
    {
      "epoch": 2.146942800788955,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001408842607076003,
      "loss": 0.0474,
      "step": 8708
    },
    {
      "epoch": 2.1474358974358974,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00014085284219013668,
      "loss": 0.0523,
      "step": 8710
    },
    {
      "epoch": 2.1479289940828403,
      "grad_norm": 0.3125,
      "learning_rate": 0.00014082141883138316,
      "loss": 0.0521,
      "step": 8712
    },
    {
      "epoch": 2.148422090729783,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001407899906350636,
      "loss": 0.0547,
      "step": 8714
    },
    {
      "epoch": 2.148915187376726,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001407585576049024,
      "loss": 0.0502,
      "step": 8716
    },
    {
      "epoch": 2.149408284023669,
      "grad_norm": 0.421875,
      "learning_rate": 0.00014072711974462456,
      "loss": 0.0522,
      "step": 8718
    },
    {
      "epoch": 2.1499013806706113,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00014069567705795566,
      "loss": 0.0561,
      "step": 8720
    },
    {
      "epoch": 2.1503944773175543,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001406642295486218,
      "loss": 0.0517,
      "step": 8722
    },
    {
      "epoch": 2.1508875739644973,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00014063277722034968,
      "loss": 0.0483,
      "step": 8724
    },
    {
      "epoch": 2.15138067061144,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00014060132007686659,
      "loss": 0.053,
      "step": 8726
    },
    {
      "epoch": 2.1518737672583828,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001405698581219004,
      "loss": 0.0485,
      "step": 8728
    },
    {
      "epoch": 2.1523668639053253,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00014053839135917952,
      "loss": 0.0516,
      "step": 8730
    },
    {
      "epoch": 2.1528599605522682,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001405069197924329,
      "loss": 0.0485,
      "step": 8732
    },
    {
      "epoch": 2.153353057199211,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00014047544342539017,
      "loss": 0.0529,
      "step": 8734
    },
    {
      "epoch": 2.1538461538461537,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00014044396226178135,
      "loss": 0.0553,
      "step": 8736
    },
    {
      "epoch": 2.1543392504930967,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00014041247630533724,
      "loss": 0.0474,
      "step": 8738
    },
    {
      "epoch": 2.1548323471400392,
      "grad_norm": 0.4140625,
      "learning_rate": 0.000140380985559789,
      "loss": 0.0506,
      "step": 8740
    },
    {
      "epoch": 2.155325443786982,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00014034949002886854,
      "loss": 0.0506,
      "step": 8742
    },
    {
      "epoch": 2.155818540433925,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00014031798971630822,
      "loss": 0.057,
      "step": 8744
    },
    {
      "epoch": 2.1563116370808677,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00014028648462584108,
      "loss": 0.0452,
      "step": 8746
    },
    {
      "epoch": 2.1568047337278107,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001402549747612005,
      "loss": 0.0509,
      "step": 8748
    },
    {
      "epoch": 2.1572978303747536,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001402234601261207,
      "loss": 0.0484,
      "step": 8750
    },
    {
      "epoch": 2.157790927021696,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00014019194072433625,
      "loss": 0.0535,
      "step": 8752
    },
    {
      "epoch": 2.158284023668639,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00014016041655958244,
      "loss": 0.0519,
      "step": 8754
    },
    {
      "epoch": 2.1587771203155817,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014012888763559507,
      "loss": 0.0539,
      "step": 8756
    },
    {
      "epoch": 2.1592702169625246,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00014009735395611043,
      "loss": 0.0581,
      "step": 8758
    },
    {
      "epoch": 2.1597633136094676,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00014006581552486547,
      "loss": 0.0562,
      "step": 8760
    },
    {
      "epoch": 2.16025641025641,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001400342723455977,
      "loss": 0.0488,
      "step": 8762
    },
    {
      "epoch": 2.160749506903353,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00014000272442204513,
      "loss": 0.0532,
      "step": 8764
    },
    {
      "epoch": 2.1612426035502956,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00013997117175794637,
      "loss": 0.0554,
      "step": 8766
    },
    {
      "epoch": 2.1617357001972386,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00013993961435704056,
      "loss": 0.0637,
      "step": 8768
    },
    {
      "epoch": 2.1622287968441816,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00013990805222306745,
      "loss": 0.0505,
      "step": 8770
    },
    {
      "epoch": 2.162721893491124,
      "grad_norm": 0.390625,
      "learning_rate": 0.00013987648535976737,
      "loss": 0.048,
      "step": 8772
    },
    {
      "epoch": 2.163214990138067,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00013984491377088114,
      "loss": 0.0532,
      "step": 8774
    },
    {
      "epoch": 2.16370808678501,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001398133374601501,
      "loss": 0.0571,
      "step": 8776
    },
    {
      "epoch": 2.1642011834319526,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00013978175643131633,
      "loss": 0.0506,
      "step": 8778
    },
    {
      "epoch": 2.1646942800788955,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001397501706881223,
      "loss": 0.0469,
      "step": 8780
    },
    {
      "epoch": 2.1651873767258385,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001397185802343111,
      "loss": 0.0514,
      "step": 8782
    },
    {
      "epoch": 2.165680473372781,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001396869850736264,
      "loss": 0.0523,
      "step": 8784
    },
    {
      "epoch": 2.166173570019724,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00013965538520981235,
      "loss": 0.0481,
      "step": 8786
    },
    {
      "epoch": 2.1666666666666665,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001396237806466138,
      "loss": 0.0528,
      "step": 8788
    },
    {
      "epoch": 2.1671597633136095,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00013959217138777597,
      "loss": 0.0556,
      "step": 8790
    },
    {
      "epoch": 2.1676528599605525,
      "grad_norm": 0.375,
      "learning_rate": 0.0001395605574370448,
      "loss": 0.0546,
      "step": 8792
    },
    {
      "epoch": 2.168145956607495,
      "grad_norm": 0.359375,
      "learning_rate": 0.00013952893879816668,
      "loss": 0.0463,
      "step": 8794
    },
    {
      "epoch": 2.168639053254438,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00013949731547488863,
      "loss": 0.0515,
      "step": 8796
    },
    {
      "epoch": 2.1691321499013805,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00013946568747095817,
      "loss": 0.0549,
      "step": 8798
    },
    {
      "epoch": 2.1696252465483234,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00013943405479012336,
      "loss": 0.0572,
      "step": 8800
    },
    {
      "epoch": 2.1701183431952664,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00013940241743613295,
      "loss": 0.0526,
      "step": 8802
    },
    {
      "epoch": 2.170611439842209,
      "grad_norm": 0.46875,
      "learning_rate": 0.00013937077541273602,
      "loss": 0.0564,
      "step": 8804
    },
    {
      "epoch": 2.171104536489152,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001393391287236824,
      "loss": 0.053,
      "step": 8806
    },
    {
      "epoch": 2.171597633136095,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013930747737272235,
      "loss": 0.0577,
      "step": 8808
    },
    {
      "epoch": 2.1720907297830374,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001392758213636068,
      "loss": 0.0569,
      "step": 8810
    },
    {
      "epoch": 2.1725838264299804,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00013924416070008717,
      "loss": 0.0484,
      "step": 8812
    },
    {
      "epoch": 2.173076923076923,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013921249538591533,
      "loss": 0.06,
      "step": 8814
    },
    {
      "epoch": 2.173570019723866,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001391808254248439,
      "loss": 0.0606,
      "step": 8816
    },
    {
      "epoch": 2.174063116370809,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00013914915082062588,
      "loss": 0.0539,
      "step": 8818
    },
    {
      "epoch": 2.1745562130177514,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001391174715770149,
      "loss": 0.0479,
      "step": 8820
    },
    {
      "epoch": 2.1750493096646943,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00013908578769776512,
      "loss": 0.0632,
      "step": 8822
    },
    {
      "epoch": 2.175542406311637,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001390540991866313,
      "loss": 0.0554,
      "step": 8824
    },
    {
      "epoch": 2.17603550295858,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013902240604736862,
      "loss": 0.0579,
      "step": 8826
    },
    {
      "epoch": 2.176528599605523,
      "grad_norm": 0.328125,
      "learning_rate": 0.000138990708283733,
      "loss": 0.0552,
      "step": 8828
    },
    {
      "epoch": 2.1770216962524653,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00013895900589948077,
      "loss": 0.0577,
      "step": 8830
    },
    {
      "epoch": 2.1775147928994083,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001389272988983688,
      "loss": 0.0582,
      "step": 8832
    },
    {
      "epoch": 2.1780078895463513,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00013889558728415454,
      "loss": 0.0503,
      "step": 8834
    },
    {
      "epoch": 2.178500986193294,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013886387106059605,
      "loss": 0.0486,
      "step": 8836
    },
    {
      "epoch": 2.1789940828402368,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013883215023145185,
      "loss": 0.0513,
      "step": 8838
    },
    {
      "epoch": 2.1794871794871793,
      "grad_norm": 0.359375,
      "learning_rate": 0.00013880042480048102,
      "loss": 0.0559,
      "step": 8840
    },
    {
      "epoch": 2.1799802761341223,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013876869477144323,
      "loss": 0.0509,
      "step": 8842
    },
    {
      "epoch": 2.1804733727810652,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00013873696014809865,
      "loss": 0.0533,
      "step": 8844
    },
    {
      "epoch": 2.1809664694280078,
      "grad_norm": 0.390625,
      "learning_rate": 0.00013870522093420803,
      "loss": 0.0526,
      "step": 8846
    },
    {
      "epoch": 2.1814595660749507,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001386734771335326,
      "loss": 0.054,
      "step": 8848
    },
    {
      "epoch": 2.1819526627218937,
      "grad_norm": 0.421875,
      "learning_rate": 0.00013864172874983424,
      "loss": 0.0503,
      "step": 8850
    },
    {
      "epoch": 2.1824457593688362,
      "grad_norm": 0.4375,
      "learning_rate": 0.00013860997578687522,
      "loss": 0.0502,
      "step": 8852
    },
    {
      "epoch": 2.182938856015779,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013857821824841854,
      "loss": 0.0565,
      "step": 8854
    },
    {
      "epoch": 2.1834319526627217,
      "grad_norm": 0.5,
      "learning_rate": 0.0001385464561382276,
      "loss": 0.0546,
      "step": 8856
    },
    {
      "epoch": 2.1839250493096647,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00013851468946006639,
      "loss": 0.0498,
      "step": 8858
    },
    {
      "epoch": 2.1844181459566077,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00013848291821769943,
      "loss": 0.0544,
      "step": 8860
    },
    {
      "epoch": 2.18491124260355,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00013845114241489183,
      "loss": 0.06,
      "step": 8862
    },
    {
      "epoch": 2.185404339250493,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00013841936205540915,
      "loss": 0.0537,
      "step": 8864
    },
    {
      "epoch": 2.185897435897436,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00013838757714301756,
      "loss": 0.0564,
      "step": 8866
    },
    {
      "epoch": 2.1863905325443787,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001383557876814837,
      "loss": 0.0517,
      "step": 8868
    },
    {
      "epoch": 2.1868836291913216,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013832399367457492,
      "loss": 0.0551,
      "step": 8870
    },
    {
      "epoch": 2.187376725838264,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001382921951260589,
      "loss": 0.0592,
      "step": 8872
    },
    {
      "epoch": 2.187869822485207,
      "grad_norm": 0.375,
      "learning_rate": 0.00013826039203970394,
      "loss": 0.055,
      "step": 8874
    },
    {
      "epoch": 2.18836291913215,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00013822858441927893,
      "loss": 0.0524,
      "step": 8876
    },
    {
      "epoch": 2.1888560157790926,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013819677226855324,
      "loss": 0.0577,
      "step": 8878
    },
    {
      "epoch": 2.1893491124260356,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001381649555912967,
      "loss": 0.0613,
      "step": 8880
    },
    {
      "epoch": 2.189842209072978,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001381331343912799,
      "loss": 0.0634,
      "step": 8882
    },
    {
      "epoch": 2.190335305719921,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00013810130867227373,
      "loss": 0.0543,
      "step": 8884
    },
    {
      "epoch": 2.190828402366864,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001380694784380498,
      "loss": 0.0596,
      "step": 8886
    },
    {
      "epoch": 2.1913214990138066,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00013803764369238008,
      "loss": 0.0575,
      "step": 8888
    },
    {
      "epoch": 2.1918145956607495,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00013800580443903724,
      "loss": 0.0549,
      "step": 8890
    },
    {
      "epoch": 2.1923076923076925,
      "grad_norm": 0.453125,
      "learning_rate": 0.00013797396068179434,
      "loss": 0.0584,
      "step": 8892
    },
    {
      "epoch": 2.192800788954635,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001379421124244251,
      "loss": 0.0553,
      "step": 8894
    },
    {
      "epoch": 2.193293885601578,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001379102596707037,
      "loss": 0.0548,
      "step": 8896
    },
    {
      "epoch": 2.1937869822485205,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013787840242440486,
      "loss": 0.0564,
      "step": 8898
    },
    {
      "epoch": 2.1942800788954635,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00013784654068930386,
      "loss": 0.0549,
      "step": 8900
    },
    {
      "epoch": 2.1947731755424065,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00013781467446917644,
      "loss": 0.0588,
      "step": 8902
    },
    {
      "epoch": 2.195266272189349,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00013778280376779898,
      "loss": 0.0596,
      "step": 8904
    },
    {
      "epoch": 2.195759368836292,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00013775092858894837,
      "loss": 0.0564,
      "step": 8906
    },
    {
      "epoch": 2.1962524654832345,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00013771904893640187,
      "loss": 0.0612,
      "step": 8908
    },
    {
      "epoch": 2.1967455621301775,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00013768716481393756,
      "loss": 0.0622,
      "step": 8910
    },
    {
      "epoch": 2.1972386587771204,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001376552762253337,
      "loss": 0.0624,
      "step": 8912
    },
    {
      "epoch": 2.197731755424063,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013762338317436948,
      "loss": 0.051,
      "step": 8914
    },
    {
      "epoch": 2.198224852071006,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013759148566482423,
      "loss": 0.0473,
      "step": 8916
    },
    {
      "epoch": 2.198717948717949,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00013755958370047802,
      "loss": 0.0544,
      "step": 8918
    },
    {
      "epoch": 2.1992110453648914,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00013752767728511146,
      "loss": 0.0545,
      "step": 8920
    },
    {
      "epoch": 2.1997041420118344,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00013749576642250563,
      "loss": 0.0524,
      "step": 8922
    },
    {
      "epoch": 2.200197238658777,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001374638511164421,
      "loss": 0.056,
      "step": 8924
    },
    {
      "epoch": 2.20069033530572,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001374319313707031,
      "loss": 0.0607,
      "step": 8926
    },
    {
      "epoch": 2.201183431952663,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001374000071890712,
      "loss": 0.051,
      "step": 8928
    },
    {
      "epoch": 2.2016765285996054,
      "grad_norm": 0.3125,
      "learning_rate": 0.00013736807857532964,
      "loss": 0.0522,
      "step": 8930
    },
    {
      "epoch": 2.2021696252465484,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013733614553326212,
      "loss": 0.0593,
      "step": 8932
    },
    {
      "epoch": 2.2026627218934913,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001373042080666529,
      "loss": 0.0631,
      "step": 8934
    },
    {
      "epoch": 2.203155818540434,
      "grad_norm": 0.46875,
      "learning_rate": 0.00013727226617928672,
      "loss": 0.0566,
      "step": 8936
    },
    {
      "epoch": 2.203648915187377,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001372403198749489,
      "loss": 0.051,
      "step": 8938
    },
    {
      "epoch": 2.2041420118343193,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00013720836915742532,
      "loss": 0.0477,
      "step": 8940
    },
    {
      "epoch": 2.2046351084812623,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001371764140305022,
      "loss": 0.0559,
      "step": 8942
    },
    {
      "epoch": 2.2051282051282053,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00013714445449796646,
      "loss": 0.0552,
      "step": 8944
    },
    {
      "epoch": 2.205621301775148,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00013711249056360547,
      "loss": 0.0552,
      "step": 8946
    },
    {
      "epoch": 2.206114398422091,
      "grad_norm": 0.40625,
      "learning_rate": 0.00013708052223120713,
      "loss": 0.059,
      "step": 8948
    },
    {
      "epoch": 2.2066074950690338,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013704854950455988,
      "loss": 0.0573,
      "step": 8950
    },
    {
      "epoch": 2.2071005917159763,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00013701657238745265,
      "loss": 0.0595,
      "step": 8952
    },
    {
      "epoch": 2.2075936883629192,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00013698459088367496,
      "loss": 0.0557,
      "step": 8954
    },
    {
      "epoch": 2.2080867850098618,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001369526049970167,
      "loss": 0.054,
      "step": 8956
    },
    {
      "epoch": 2.2085798816568047,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013692061473126845,
      "loss": 0.0564,
      "step": 8958
    },
    {
      "epoch": 2.2090729783037477,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00013688862009022122,
      "loss": 0.0624,
      "step": 8960
    },
    {
      "epoch": 2.2095660749506902,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013685662107766657,
      "loss": 0.0593,
      "step": 8962
    },
    {
      "epoch": 2.210059171597633,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001368246176973965,
      "loss": 0.0606,
      "step": 8964
    },
    {
      "epoch": 2.2105522682445757,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00013679260995320367,
      "loss": 0.0643,
      "step": 8966
    },
    {
      "epoch": 2.2110453648915187,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013676059784888112,
      "loss": 0.0636,
      "step": 8968
    },
    {
      "epoch": 2.2115384615384617,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00013672858138822247,
      "loss": 0.0539,
      "step": 8970
    },
    {
      "epoch": 2.212031558185404,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013669656057502187,
      "loss": 0.0619,
      "step": 8972
    },
    {
      "epoch": 2.212524654832347,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00013666453541307395,
      "loss": 0.0586,
      "step": 8974
    },
    {
      "epoch": 2.21301775147929,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013663250590617386,
      "loss": 0.0641,
      "step": 8976
    },
    {
      "epoch": 2.2135108481262327,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00013660047205811735,
      "loss": 0.057,
      "step": 8978
    },
    {
      "epoch": 2.2140039447731756,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001365684338727005,
      "loss": 0.0565,
      "step": 8980
    },
    {
      "epoch": 2.214497041420118,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013653639135372008,
      "loss": 0.0547,
      "step": 8982
    },
    {
      "epoch": 2.214990138067061,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00013650434450497326,
      "loss": 0.0595,
      "step": 8984
    },
    {
      "epoch": 2.215483234714004,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013647229333025785,
      "loss": 0.0558,
      "step": 8986
    },
    {
      "epoch": 2.2159763313609466,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00013644023783337202,
      "loss": 0.0615,
      "step": 8988
    },
    {
      "epoch": 2.2164694280078896,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013640817801811454,
      "loss": 0.0576,
      "step": 8990
    },
    {
      "epoch": 2.216962524654832,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001363761138882847,
      "loss": 0.0565,
      "step": 8992
    },
    {
      "epoch": 2.217455621301775,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00013634404544768226,
      "loss": 0.0554,
      "step": 8994
    },
    {
      "epoch": 2.217948717948718,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00013631197270010757,
      "loss": 0.0579,
      "step": 8996
    },
    {
      "epoch": 2.2184418145956606,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001362798956493613,
      "loss": 0.058,
      "step": 8998
    },
    {
      "epoch": 2.2189349112426036,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00013624781429924485,
      "loss": 0.0601,
      "step": 9000
    },
    {
      "epoch": 2.2194280078895465,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00013621572865356004,
      "loss": 0.0605,
      "step": 9002
    },
    {
      "epoch": 2.219921104536489,
      "grad_norm": 0.453125,
      "learning_rate": 0.00013618363871610917,
      "loss": 0.0558,
      "step": 9004
    },
    {
      "epoch": 2.220414201183432,
      "grad_norm": 0.625,
      "learning_rate": 0.0001361515444906951,
      "loss": 0.0549,
      "step": 9006
    },
    {
      "epoch": 2.2209072978303745,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00013611944598112118,
      "loss": 0.0567,
      "step": 9008
    },
    {
      "epoch": 2.2214003944773175,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00013608734319119122,
      "loss": 0.0591,
      "step": 9010
    },
    {
      "epoch": 2.2218934911242605,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001360552361247096,
      "loss": 0.0554,
      "step": 9012
    },
    {
      "epoch": 2.222386587771203,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001360231247854812,
      "loss": 0.0555,
      "step": 9014
    },
    {
      "epoch": 2.222879684418146,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00013599100917731138,
      "loss": 0.0582,
      "step": 9016
    },
    {
      "epoch": 2.223372781065089,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00013595888930400605,
      "loss": 0.0579,
      "step": 9018
    },
    {
      "epoch": 2.2238658777120315,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00013592676516937152,
      "loss": 0.0548,
      "step": 9020
    },
    {
      "epoch": 2.2243589743589745,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001358946367772148,
      "loss": 0.0603,
      "step": 9022
    },
    {
      "epoch": 2.224852071005917,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00013586250413134316,
      "loss": 0.0501,
      "step": 9024
    },
    {
      "epoch": 2.22534516765286,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00013583036723556455,
      "loss": 0.0584,
      "step": 9026
    },
    {
      "epoch": 2.225838264299803,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001357982260936874,
      "loss": 0.0583,
      "step": 9028
    },
    {
      "epoch": 2.2263313609467454,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00013576608070952058,
      "loss": 0.0582,
      "step": 9030
    },
    {
      "epoch": 2.2268244575936884,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00013573393108687348,
      "loss": 0.0563,
      "step": 9032
    },
    {
      "epoch": 2.2273175542406314,
      "grad_norm": 0.390625,
      "learning_rate": 0.00013570177722955603,
      "loss": 0.0615,
      "step": 9034
    },
    {
      "epoch": 2.227810650887574,
      "grad_norm": 0.421875,
      "learning_rate": 0.00013566961914137867,
      "loss": 0.059,
      "step": 9036
    },
    {
      "epoch": 2.228303747534517,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00013563745682615227,
      "loss": 0.0599,
      "step": 9038
    },
    {
      "epoch": 2.2287968441814594,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013560529028768827,
      "loss": 0.0574,
      "step": 9040
    },
    {
      "epoch": 2.2292899408284024,
      "grad_norm": 0.5,
      "learning_rate": 0.0001355731195297986,
      "loss": 0.0588,
      "step": 9042
    },
    {
      "epoch": 2.2297830374753453,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001355409445562956,
      "loss": 0.057,
      "step": 9044
    },
    {
      "epoch": 2.230276134122288,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013550876537099228,
      "loss": 0.0629,
      "step": 9046
    },
    {
      "epoch": 2.230769230769231,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000135476581977702,
      "loss": 0.0595,
      "step": 9048
    },
    {
      "epoch": 2.2312623274161734,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00013544439438023867,
      "loss": 0.0574,
      "step": 9050
    },
    {
      "epoch": 2.2317554240631163,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00013541220258241673,
      "loss": 0.0588,
      "step": 9052
    },
    {
      "epoch": 2.2322485207100593,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013538000658805104,
      "loss": 0.0548,
      "step": 9054
    },
    {
      "epoch": 2.232741617357002,
      "grad_norm": 0.369140625,
      "learning_rate": 0.000135347806400957,
      "loss": 0.0597,
      "step": 9056
    },
    {
      "epoch": 2.233234714003945,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001353156020249506,
      "loss": 0.0584,
      "step": 9058
    },
    {
      "epoch": 2.2337278106508878,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013528339346384819,
      "loss": 0.0534,
      "step": 9060
    },
    {
      "epoch": 2.2342209072978303,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00013525118072146662,
      "loss": 0.0563,
      "step": 9062
    },
    {
      "epoch": 2.2347140039447733,
      "grad_norm": 0.34375,
      "learning_rate": 0.00013521896380162335,
      "loss": 0.0559,
      "step": 9064
    },
    {
      "epoch": 2.235207100591716,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013518674270813623,
      "loss": 0.056,
      "step": 9066
    },
    {
      "epoch": 2.2357001972386588,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001351545174448236,
      "loss": 0.0585,
      "step": 9068
    },
    {
      "epoch": 2.2361932938856017,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001351222880155044,
      "loss": 0.0546,
      "step": 9070
    },
    {
      "epoch": 2.2366863905325443,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013509005442399798,
      "loss": 0.0622,
      "step": 9072
    },
    {
      "epoch": 2.2371794871794872,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001350578166741242,
      "loss": 0.0595,
      "step": 9074
    },
    {
      "epoch": 2.2376725838264298,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001350255747697034,
      "loss": 0.0589,
      "step": 9076
    },
    {
      "epoch": 2.2381656804733727,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00013499332871455644,
      "loss": 0.0575,
      "step": 9078
    },
    {
      "epoch": 2.2386587771203157,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00013496107851250461,
      "loss": 0.0625,
      "step": 9080
    },
    {
      "epoch": 2.239151873767258,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001349288241673698,
      "loss": 0.0561,
      "step": 9082
    },
    {
      "epoch": 2.239644970414201,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00013489656568297433,
      "loss": 0.0579,
      "step": 9084
    },
    {
      "epoch": 2.240138067061144,
      "grad_norm": 0.5,
      "learning_rate": 0.00013486430306314096,
      "loss": 0.0588,
      "step": 9086
    },
    {
      "epoch": 2.2406311637080867,
      "grad_norm": 0.388671875,
      "learning_rate": 0.000134832036311693,
      "loss": 0.0571,
      "step": 9088
    },
    {
      "epoch": 2.2411242603550297,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013479976543245433,
      "loss": 0.0581,
      "step": 9090
    },
    {
      "epoch": 2.2416173570019726,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013476749042924912,
      "loss": 0.0577,
      "step": 9092
    },
    {
      "epoch": 2.242110453648915,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001347352113059022,
      "loss": 0.0571,
      "step": 9094
    },
    {
      "epoch": 2.242603550295858,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00013470292806623877,
      "loss": 0.0604,
      "step": 9096
    },
    {
      "epoch": 2.2430966469428006,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001346706407140846,
      "loss": 0.055,
      "step": 9098
    },
    {
      "epoch": 2.2435897435897436,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00013463834925326595,
      "loss": 0.0567,
      "step": 9100
    },
    {
      "epoch": 2.2440828402366866,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001346060536876095,
      "loss": 0.0599,
      "step": 9102
    },
    {
      "epoch": 2.244575936883629,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001345737540209425,
      "loss": 0.0558,
      "step": 9104
    },
    {
      "epoch": 2.245069033530572,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001345414502570926,
      "loss": 0.0583,
      "step": 9106
    },
    {
      "epoch": 2.2455621301775146,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000134509142399888,
      "loss": 0.0535,
      "step": 9108
    },
    {
      "epoch": 2.2460552268244576,
      "grad_norm": 0.359375,
      "learning_rate": 0.00013447683045315735,
      "loss": 0.0601,
      "step": 9110
    },
    {
      "epoch": 2.2465483234714005,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013444451442072978,
      "loss": 0.0529,
      "step": 9112
    },
    {
      "epoch": 2.247041420118343,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00013441219430643494,
      "loss": 0.0575,
      "step": 9114
    },
    {
      "epoch": 2.247534516765286,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00013437987011410296,
      "loss": 0.066,
      "step": 9116
    },
    {
      "epoch": 2.248027613412229,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00013434754184756442,
      "loss": 0.0561,
      "step": 9118
    },
    {
      "epoch": 2.2485207100591715,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001343152095106504,
      "loss": 0.0627,
      "step": 9120
    },
    {
      "epoch": 2.2490138067061145,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013428287310719242,
      "loss": 0.0578,
      "step": 9122
    },
    {
      "epoch": 2.249506903353057,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001342505326410226,
      "loss": 0.064,
      "step": 9124
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00013421818811597345,
      "loss": 0.0594,
      "step": 9126
    },
    {
      "epoch": 2.250493096646943,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001341858395358779,
      "loss": 0.0577,
      "step": 9128
    },
    {
      "epoch": 2.2509861932938855,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00013415348690456953,
      "loss": 0.0546,
      "step": 9130
    },
    {
      "epoch": 2.2514792899408285,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001341211302258823,
      "loss": 0.0554,
      "step": 9132
    },
    {
      "epoch": 2.251972386587771,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001340887695036506,
      "loss": 0.0627,
      "step": 9134
    },
    {
      "epoch": 2.252465483234714,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001340564047417094,
      "loss": 0.0567,
      "step": 9136
    },
    {
      "epoch": 2.252958579881657,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00013402403594389405,
      "loss": 0.0562,
      "step": 9138
    },
    {
      "epoch": 2.2534516765285995,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001339916631140405,
      "loss": 0.0529,
      "step": 9140
    },
    {
      "epoch": 2.2539447731755424,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00013395928625598513,
      "loss": 0.0598,
      "step": 9142
    },
    {
      "epoch": 2.2544378698224854,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001339269053735647,
      "loss": 0.0591,
      "step": 9144
    },
    {
      "epoch": 2.254930966469428,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00013389452047061652,
      "loss": 0.0589,
      "step": 9146
    },
    {
      "epoch": 2.255424063116371,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001338621315509785,
      "loss": 0.0625,
      "step": 9148
    },
    {
      "epoch": 2.2559171597633134,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00013382973861848874,
      "loss": 0.0572,
      "step": 9150
    },
    {
      "epoch": 2.2564102564102564,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013379734167698614,
      "loss": 0.0652,
      "step": 9152
    },
    {
      "epoch": 2.2569033530571994,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013376494073030984,
      "loss": 0.0606,
      "step": 9154
    },
    {
      "epoch": 2.257396449704142,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001337325357822995,
      "loss": 0.0614,
      "step": 9156
    },
    {
      "epoch": 2.257889546351085,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001337001268367954,
      "loss": 0.0649,
      "step": 9158
    },
    {
      "epoch": 2.2583826429980274,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00013366771389763805,
      "loss": 0.0616,
      "step": 9160
    },
    {
      "epoch": 2.2588757396449703,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001336352969686687,
      "loss": 0.0559,
      "step": 9162
    },
    {
      "epoch": 2.2593688362919133,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001336028760537288,
      "loss": 0.0638,
      "step": 9164
    },
    {
      "epoch": 2.259861932938856,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001335704511566605,
      "loss": 0.0561,
      "step": 9166
    },
    {
      "epoch": 2.260355029585799,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001335380222813063,
      "loss": 0.0616,
      "step": 9168
    },
    {
      "epoch": 2.260848126232742,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00013350558943150918,
      "loss": 0.0635,
      "step": 9170
    },
    {
      "epoch": 2.2613412228796843,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00013347315261111269,
      "loss": 0.0545,
      "step": 9172
    },
    {
      "epoch": 2.2618343195266273,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001334407118239607,
      "loss": 0.0605,
      "step": 9174
    },
    {
      "epoch": 2.2623274161735702,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00013340826707389767,
      "loss": 0.0596,
      "step": 9176
    },
    {
      "epoch": 2.2628205128205128,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013337581836476842,
      "loss": 0.0568,
      "step": 9178
    },
    {
      "epoch": 2.2633136094674557,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013334336570041837,
      "loss": 0.055,
      "step": 9180
    },
    {
      "epoch": 2.2638067061143983,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001333109090846933,
      "loss": 0.0607,
      "step": 9182
    },
    {
      "epoch": 2.2642998027613412,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013327844852143955,
      "loss": 0.0606,
      "step": 9184
    },
    {
      "epoch": 2.264792899408284,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00013324598401450382,
      "loss": 0.0612,
      "step": 9186
    },
    {
      "epoch": 2.2652859960552267,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00013321351556773342,
      "loss": 0.0571,
      "step": 9188
    },
    {
      "epoch": 2.2657790927021697,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00013318104318497592,
      "loss": 0.0638,
      "step": 9190
    },
    {
      "epoch": 2.2662721893491122,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00013314856687007958,
      "loss": 0.0591,
      "step": 9192
    },
    {
      "epoch": 2.266765285996055,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00013311608662689296,
      "loss": 0.0582,
      "step": 9194
    },
    {
      "epoch": 2.267258382642998,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013308360245926521,
      "loss": 0.0626,
      "step": 9196
    },
    {
      "epoch": 2.2677514792899407,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00013305111437104583,
      "loss": 0.0603,
      "step": 9198
    },
    {
      "epoch": 2.2682445759368837,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013301862236608488,
      "loss": 0.0691,
      "step": 9200
    },
    {
      "epoch": 2.2687376725838266,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013298612644823283,
      "loss": 0.0587,
      "step": 9202
    },
    {
      "epoch": 2.269230769230769,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001329536266213406,
      "loss": 0.0578,
      "step": 9204
    },
    {
      "epoch": 2.269723865877712,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013292112288925962,
      "loss": 0.0623,
      "step": 9206
    },
    {
      "epoch": 2.2702169625246547,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00013288861525584183,
      "loss": 0.0538,
      "step": 9208
    },
    {
      "epoch": 2.2707100591715976,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00013285610372493943,
      "loss": 0.062,
      "step": 9210
    },
    {
      "epoch": 2.2712031558185406,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013282358830040534,
      "loss": 0.0579,
      "step": 9212
    },
    {
      "epoch": 2.271696252465483,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00013279106898609274,
      "loss": 0.0524,
      "step": 9214
    },
    {
      "epoch": 2.272189349112426,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001327585457858554,
      "loss": 0.0638,
      "step": 9216
    },
    {
      "epoch": 2.2726824457593686,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001327260187035475,
      "loss": 0.0584,
      "step": 9218
    },
    {
      "epoch": 2.2731755424063116,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00013269348774302365,
      "loss": 0.0625,
      "step": 9220
    },
    {
      "epoch": 2.2736686390532546,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00013266095290813895,
      "loss": 0.0606,
      "step": 9222
    },
    {
      "epoch": 2.274161735700197,
      "grad_norm": 0.404296875,
      "learning_rate": 0.000132628414202749,
      "loss": 0.0611,
      "step": 9224
    },
    {
      "epoch": 2.27465483234714,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001325958716307098,
      "loss": 0.0528,
      "step": 9226
    },
    {
      "epoch": 2.275147928994083,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001325633251958778,
      "loss": 0.0563,
      "step": 9228
    },
    {
      "epoch": 2.2756410256410255,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00013253077490211,
      "loss": 0.0607,
      "step": 9230
    },
    {
      "epoch": 2.2761341222879685,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00013249822075326367,
      "loss": 0.0565,
      "step": 9232
    },
    {
      "epoch": 2.2766272189349115,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001324656627531968,
      "loss": 0.0595,
      "step": 9234
    },
    {
      "epoch": 2.277120315581854,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013243310090576757,
      "loss": 0.0574,
      "step": 9236
    },
    {
      "epoch": 2.277613412228797,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00013240053521483482,
      "loss": 0.0588,
      "step": 9238
    },
    {
      "epoch": 2.2781065088757395,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00013236796568425775,
      "loss": 0.0605,
      "step": 9240
    },
    {
      "epoch": 2.2785996055226825,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00013233539231789605,
      "loss": 0.0507,
      "step": 9242
    },
    {
      "epoch": 2.279092702169625,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001323028151196098,
      "loss": 0.0527,
      "step": 9244
    },
    {
      "epoch": 2.279585798816568,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001322702340932596,
      "loss": 0.054,
      "step": 9246
    },
    {
      "epoch": 2.280078895463511,
      "grad_norm": 0.375,
      "learning_rate": 0.0001322376492427065,
      "loss": 0.0573,
      "step": 9248
    },
    {
      "epoch": 2.2805719921104535,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00013220506057181196,
      "loss": 0.0578,
      "step": 9250
    },
    {
      "epoch": 2.2810650887573964,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00013217246808443792,
      "loss": 0.0588,
      "step": 9252
    },
    {
      "epoch": 2.2815581854043394,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001321398717844468,
      "loss": 0.0562,
      "step": 9254
    },
    {
      "epoch": 2.282051282051282,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00013210727167570143,
      "loss": 0.0592,
      "step": 9256
    },
    {
      "epoch": 2.282544378698225,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001320746677620651,
      "loss": 0.0674,
      "step": 9258
    },
    {
      "epoch": 2.283037475345168,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001320420600474016,
      "loss": 0.0601,
      "step": 9260
    },
    {
      "epoch": 2.2835305719921104,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000132009448535575,
      "loss": 0.0594,
      "step": 9262
    },
    {
      "epoch": 2.2840236686390534,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00013197683323045006,
      "loss": 0.0581,
      "step": 9264
    },
    {
      "epoch": 2.284516765285996,
      "grad_norm": 0.5,
      "learning_rate": 0.00013194421413589188,
      "loss": 0.0612,
      "step": 9266
    },
    {
      "epoch": 2.285009861932939,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00013191159125576594,
      "loss": 0.0564,
      "step": 9268
    },
    {
      "epoch": 2.285502958579882,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001318789645939383,
      "loss": 0.0597,
      "step": 9270
    },
    {
      "epoch": 2.2859960552268244,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00013184633415427532,
      "loss": 0.0544,
      "step": 9272
    },
    {
      "epoch": 2.2864891518737673,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00013181369994064398,
      "loss": 0.0609,
      "step": 9274
    },
    {
      "epoch": 2.28698224852071,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00013178106195691153,
      "loss": 0.0618,
      "step": 9276
    },
    {
      "epoch": 2.287475345167653,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013174842020694582,
      "loss": 0.0583,
      "step": 9278
    },
    {
      "epoch": 2.287968441814596,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00013171577469461504,
      "loss": 0.0624,
      "step": 9280
    },
    {
      "epoch": 2.2884615384615383,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001316831254237879,
      "loss": 0.0574,
      "step": 9282
    },
    {
      "epoch": 2.2889546351084813,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013165047239833348,
      "loss": 0.0564,
      "step": 9284
    },
    {
      "epoch": 2.2894477317554243,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00013161781562212138,
      "loss": 0.0548,
      "step": 9286
    },
    {
      "epoch": 2.289940828402367,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001315851550990216,
      "loss": 0.0563,
      "step": 9288
    },
    {
      "epoch": 2.2904339250493098,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00013155249083290457,
      "loss": 0.0547,
      "step": 9290
    },
    {
      "epoch": 2.2909270216962523,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00013151982282764125,
      "loss": 0.0603,
      "step": 9292
    },
    {
      "epoch": 2.2914201183431953,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00013148715108710294,
      "loss": 0.0574,
      "step": 9294
    },
    {
      "epoch": 2.2919132149901382,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013145447561516138,
      "loss": 0.0569,
      "step": 9296
    },
    {
      "epoch": 2.2924063116370808,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00013142179641568888,
      "loss": 0.0571,
      "step": 9298
    },
    {
      "epoch": 2.2928994082840237,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00013138911349255806,
      "loss": 0.0558,
      "step": 9300
    },
    {
      "epoch": 2.2933925049309662,
      "grad_norm": 0.578125,
      "learning_rate": 0.000131356426849642,
      "loss": 0.0592,
      "step": 9302
    },
    {
      "epoch": 2.293885601577909,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001313237364908143,
      "loss": 0.0549,
      "step": 9304
    },
    {
      "epoch": 2.294378698224852,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013129104241994894,
      "loss": 0.06,
      "step": 9306
    },
    {
      "epoch": 2.2948717948717947,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00013125834464092039,
      "loss": 0.0572,
      "step": 9308
    },
    {
      "epoch": 2.2953648915187377,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00013122564315760343,
      "loss": 0.0593,
      "step": 9310
    },
    {
      "epoch": 2.2958579881656807,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013119293797387343,
      "loss": 0.0581,
      "step": 9312
    },
    {
      "epoch": 2.296351084812623,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013116022909360613,
      "loss": 0.0607,
      "step": 9314
    },
    {
      "epoch": 2.296844181459566,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0001311275165206777,
      "loss": 0.0581,
      "step": 9316
    },
    {
      "epoch": 2.297337278106509,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00013109480025896476,
      "loss": 0.0535,
      "step": 9318
    },
    {
      "epoch": 2.2978303747534516,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00013106208031234435,
      "loss": 0.0603,
      "step": 9320
    },
    {
      "epoch": 2.2983234714003946,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00013102935668469402,
      "loss": 0.0576,
      "step": 9322
    },
    {
      "epoch": 2.298816568047337,
      "grad_norm": 0.375,
      "learning_rate": 0.0001309966293798917,
      "loss": 0.0559,
      "step": 9324
    },
    {
      "epoch": 2.29930966469428,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013096389840181572,
      "loss": 0.0573,
      "step": 9326
    },
    {
      "epoch": 2.2998027613412226,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00013093116375434492,
      "loss": 0.058,
      "step": 9328
    },
    {
      "epoch": 2.3002958579881656,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00013089842544135847,
      "loss": 0.0542,
      "step": 9330
    },
    {
      "epoch": 2.3007889546351086,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00013086568346673612,
      "loss": 0.0574,
      "step": 9332
    },
    {
      "epoch": 2.301282051282051,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00013083293783435794,
      "loss": 0.0604,
      "step": 9334
    },
    {
      "epoch": 2.301775147928994,
      "grad_norm": 0.421875,
      "learning_rate": 0.00013080018854810446,
      "loss": 0.0565,
      "step": 9336
    },
    {
      "epoch": 2.302268244575937,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001307674356118567,
      "loss": 0.0537,
      "step": 9338
    },
    {
      "epoch": 2.3027613412228796,
      "grad_norm": 0.34375,
      "learning_rate": 0.00013073467902949604,
      "loss": 0.0617,
      "step": 9340
    },
    {
      "epoch": 2.3032544378698225,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00013070191880490433,
      "loss": 0.0592,
      "step": 9342
    },
    {
      "epoch": 2.3037475345167655,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00013066915494196378,
      "loss": 0.0666,
      "step": 9344
    },
    {
      "epoch": 2.304240631163708,
      "grad_norm": 0.390625,
      "learning_rate": 0.00013063638744455712,
      "loss": 0.0615,
      "step": 9346
    },
    {
      "epoch": 2.304733727810651,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00013060361631656755,
      "loss": 0.0576,
      "step": 9348
    },
    {
      "epoch": 2.3052268244575935,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013057084156187853,
      "loss": 0.0593,
      "step": 9350
    },
    {
      "epoch": 2.3057199211045365,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00013053806318437413,
      "loss": 0.0612,
      "step": 9352
    },
    {
      "epoch": 2.3062130177514795,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013050528118793874,
      "loss": 0.0578,
      "step": 9354
    },
    {
      "epoch": 2.306706114398422,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001304724955764572,
      "loss": 0.054,
      "step": 9356
    },
    {
      "epoch": 2.307199211045365,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001304397063538148,
      "loss": 0.0569,
      "step": 9358
    },
    {
      "epoch": 2.3076923076923075,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00013040691352389727,
      "loss": 0.0555,
      "step": 9360
    },
    {
      "epoch": 2.3081854043392505,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001303741170905907,
      "loss": 0.0616,
      "step": 9362
    },
    {
      "epoch": 2.3086785009861934,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00013034131705778163,
      "loss": 0.0549,
      "step": 9364
    },
    {
      "epoch": 2.309171597633136,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00013030851342935712,
      "loss": 0.0544,
      "step": 9366
    },
    {
      "epoch": 2.309664694280079,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00013027570620920454,
      "loss": 0.0608,
      "step": 9368
    },
    {
      "epoch": 2.310157790927022,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00013024289540121173,
      "loss": 0.0622,
      "step": 9370
    },
    {
      "epoch": 2.3106508875739644,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00013021008100926696,
      "loss": 0.0579,
      "step": 9372
    },
    {
      "epoch": 2.3111439842209074,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00013017726303725895,
      "loss": 0.0611,
      "step": 9374
    },
    {
      "epoch": 2.31163708086785,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00013014444148907678,
      "loss": 0.0612,
      "step": 9376
    },
    {
      "epoch": 2.312130177514793,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00013011161636860993,
      "loss": 0.0665,
      "step": 9378
    },
    {
      "epoch": 2.312623274161736,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001300787876797485,
      "loss": 0.0554,
      "step": 9380
    },
    {
      "epoch": 2.3131163708086784,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00013004595542638272,
      "loss": 0.058,
      "step": 9382
    },
    {
      "epoch": 2.3136094674556213,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013001311961240348,
      "loss": 0.052,
      "step": 9384
    },
    {
      "epoch": 2.314102564102564,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00012998028024170202,
      "loss": 0.0556,
      "step": 9386
    },
    {
      "epoch": 2.314595660749507,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012994743731816995,
      "loss": 0.0604,
      "step": 9388
    },
    {
      "epoch": 2.31508875739645,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00012991459084569934,
      "loss": 0.0581,
      "step": 9390
    },
    {
      "epoch": 2.3155818540433923,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00012988174082818272,
      "loss": 0.0554,
      "step": 9392
    },
    {
      "epoch": 2.3160749506903353,
      "grad_norm": 0.375,
      "learning_rate": 0.00012984888726951297,
      "loss": 0.0546,
      "step": 9394
    },
    {
      "epoch": 2.3165680473372783,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001298160301735834,
      "loss": 0.0572,
      "step": 9396
    },
    {
      "epoch": 2.317061143984221,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001297831695442878,
      "loss": 0.0539,
      "step": 9398
    },
    {
      "epoch": 2.3175542406311638,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00012975030538552032,
      "loss": 0.0537,
      "step": 9400
    },
    {
      "epoch": 2.3180473372781067,
      "grad_norm": 0.75,
      "learning_rate": 0.00012971743770117552,
      "loss": 0.0587,
      "step": 9402
    },
    {
      "epoch": 2.3185404339250493,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012968456649514845,
      "loss": 0.0549,
      "step": 9404
    },
    {
      "epoch": 2.3190335305719922,
      "grad_norm": 0.40625,
      "learning_rate": 0.00012965169177133455,
      "loss": 0.0572,
      "step": 9406
    },
    {
      "epoch": 2.3195266272189348,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001296188135336296,
      "loss": 0.0574,
      "step": 9408
    },
    {
      "epoch": 2.3200197238658777,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012958593178592987,
      "loss": 0.0615,
      "step": 9410
    },
    {
      "epoch": 2.3205128205128207,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012955304653213201,
      "loss": 0.0576,
      "step": 9412
    },
    {
      "epoch": 2.3210059171597632,
      "grad_norm": 0.40625,
      "learning_rate": 0.00012952015777613316,
      "loss": 0.055,
      "step": 9414
    },
    {
      "epoch": 2.321499013806706,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001294872655218308,
      "loss": 0.0561,
      "step": 9416
    },
    {
      "epoch": 2.3219921104536487,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001294543697731228,
      "loss": 0.055,
      "step": 9418
    },
    {
      "epoch": 2.3224852071005917,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00012942147053390758,
      "loss": 0.0579,
      "step": 9420
    },
    {
      "epoch": 2.3229783037475347,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001293885678080838,
      "loss": 0.0595,
      "step": 9422
    },
    {
      "epoch": 2.323471400394477,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001293556615995507,
      "loss": 0.0569,
      "step": 9424
    },
    {
      "epoch": 2.32396449704142,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00012932275191220776,
      "loss": 0.0546,
      "step": 9426
    },
    {
      "epoch": 2.324457593688363,
      "grad_norm": 0.390625,
      "learning_rate": 0.000129289838749955,
      "loss": 0.0507,
      "step": 9428
    },
    {
      "epoch": 2.3249506903353057,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001292569221166928,
      "loss": 0.0589,
      "step": 9430
    },
    {
      "epoch": 2.3254437869822486,
      "grad_norm": 0.36328125,
      "learning_rate": 0.000129224002016322,
      "loss": 0.0547,
      "step": 9432
    },
    {
      "epoch": 2.325936883629191,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012919107845274378,
      "loss": 0.0567,
      "step": 9434
    },
    {
      "epoch": 2.326429980276134,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012915815142985982,
      "loss": 0.0606,
      "step": 9436
    },
    {
      "epoch": 2.326923076923077,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012912522095157207,
      "loss": 0.0536,
      "step": 9438
    },
    {
      "epoch": 2.3274161735700196,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012909228702178305,
      "loss": 0.0529,
      "step": 9440
    },
    {
      "epoch": 2.3279092702169626,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001290593496443956,
      "loss": 0.0572,
      "step": 9442
    },
    {
      "epoch": 2.328402366863905,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00012902640882331295,
      "loss": 0.0599,
      "step": 9444
    },
    {
      "epoch": 2.328895463510848,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012899346456243877,
      "loss": 0.0572,
      "step": 9446
    },
    {
      "epoch": 2.329388560157791,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001289605168656772,
      "loss": 0.053,
      "step": 9448
    },
    {
      "epoch": 2.3298816568047336,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00012892756573693265,
      "loss": 0.0585,
      "step": 9450
    },
    {
      "epoch": 2.3303747534516766,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012889461118011008,
      "loss": 0.0537,
      "step": 9452
    },
    {
      "epoch": 2.3308678500986195,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00012886165319911475,
      "loss": 0.0577,
      "step": 9454
    },
    {
      "epoch": 2.331360946745562,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00012882869179785237,
      "loss": 0.0515,
      "step": 9456
    },
    {
      "epoch": 2.331854043392505,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012879572698022907,
      "loss": 0.061,
      "step": 9458
    },
    {
      "epoch": 2.3323471400394475,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00012876275875015132,
      "loss": 0.0578,
      "step": 9460
    },
    {
      "epoch": 2.3328402366863905,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012872978711152608,
      "loss": 0.0556,
      "step": 9462
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012869681206826067,
      "loss": 0.0588,
      "step": 9464
    },
    {
      "epoch": 2.333826429980276,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001286638336242628,
      "loss": 0.0516,
      "step": 9466
    },
    {
      "epoch": 2.334319526627219,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001286308517834406,
      "loss": 0.0491,
      "step": 9468
    },
    {
      "epoch": 2.3348126232741615,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00012859786654970262,
      "loss": 0.0535,
      "step": 9470
    },
    {
      "epoch": 2.3353057199211045,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00012856487792695776,
      "loss": 0.0543,
      "step": 9472
    },
    {
      "epoch": 2.3357988165680474,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001285318859191154,
      "loss": 0.0509,
      "step": 9474
    },
    {
      "epoch": 2.33629191321499,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00012849889053008525,
      "loss": 0.0564,
      "step": 9476
    },
    {
      "epoch": 2.336785009861933,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012846589176377748,
      "loss": 0.0534,
      "step": 9478
    },
    {
      "epoch": 2.337278106508876,
      "grad_norm": 0.390625,
      "learning_rate": 0.0001284328896241026,
      "loss": 0.057,
      "step": 9480
    },
    {
      "epoch": 2.3377712031558184,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00012839988411497152,
      "loss": 0.0547,
      "step": 9482
    },
    {
      "epoch": 2.3382642998027614,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00012836687524029561,
      "loss": 0.0589,
      "step": 9484
    },
    {
      "epoch": 2.3387573964497044,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001283338630039866,
      "loss": 0.0536,
      "step": 9486
    },
    {
      "epoch": 2.339250493096647,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00012830084740995666,
      "loss": 0.0528,
      "step": 9488
    },
    {
      "epoch": 2.33974358974359,
      "grad_norm": 0.375,
      "learning_rate": 0.0001282678284621183,
      "loss": 0.056,
      "step": 9490
    },
    {
      "epoch": 2.3402366863905324,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012823480616438443,
      "loss": 0.0529,
      "step": 9492
    },
    {
      "epoch": 2.3407297830374754,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012820178052066832,
      "loss": 0.0526,
      "step": 9494
    },
    {
      "epoch": 2.3412228796844183,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012816875153488381,
      "loss": 0.0584,
      "step": 9496
    },
    {
      "epoch": 2.341715976331361,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00012813571921094498,
      "loss": 0.0522,
      "step": 9498
    },
    {
      "epoch": 2.342209072978304,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001281026835527663,
      "loss": 0.0567,
      "step": 9500
    },
    {
      "epoch": 2.3427021696252464,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012806964456426275,
      "loss": 0.0511,
      "step": 9502
    },
    {
      "epoch": 2.3431952662721893,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012803660224934958,
      "loss": 0.0543,
      "step": 9504
    },
    {
      "epoch": 2.3436883629191323,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00012800355661194253,
      "loss": 0.0627,
      "step": 9506
    },
    {
      "epoch": 2.344181459566075,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012797050765595767,
      "loss": 0.0577,
      "step": 9508
    },
    {
      "epoch": 2.344674556213018,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012793745538531144,
      "loss": 0.0588,
      "step": 9510
    },
    {
      "epoch": 2.3451676528599608,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001279043998039208,
      "loss": 0.0572,
      "step": 9512
    },
    {
      "epoch": 2.3456607495069033,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012787134091570297,
      "loss": 0.0578,
      "step": 9514
    },
    {
      "epoch": 2.3461538461538463,
      "grad_norm": 0.375,
      "learning_rate": 0.00012783827872457564,
      "loss": 0.0589,
      "step": 9516
    },
    {
      "epoch": 2.346646942800789,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012780521323445685,
      "loss": 0.0579,
      "step": 9518
    },
    {
      "epoch": 2.3471400394477318,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00012777214444926503,
      "loss": 0.0548,
      "step": 9520
    },
    {
      "epoch": 2.3476331360946747,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00012773907237291908,
      "loss": 0.0572,
      "step": 9522
    },
    {
      "epoch": 2.3481262327416172,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001277059970093381,
      "loss": 0.0499,
      "step": 9524
    },
    {
      "epoch": 2.34861932938856,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00012767291836244185,
      "loss": 0.0545,
      "step": 9526
    },
    {
      "epoch": 2.3491124260355027,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012763983643615022,
      "loss": 0.052,
      "step": 9528
    },
    {
      "epoch": 2.3496055226824457,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012760675123438367,
      "loss": 0.0505,
      "step": 9530
    },
    {
      "epoch": 2.3500986193293887,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00012757366276106298,
      "loss": 0.0512,
      "step": 9532
    },
    {
      "epoch": 2.350591715976331,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0001275405710201093,
      "loss": 0.054,
      "step": 9534
    },
    {
      "epoch": 2.351084812623274,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00012750747601544417,
      "loss": 0.0561,
      "step": 9536
    },
    {
      "epoch": 2.351577909270217,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001274743777509896,
      "loss": 0.0558,
      "step": 9538
    },
    {
      "epoch": 2.3520710059171597,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012744127623066782,
      "loss": 0.0529,
      "step": 9540
    },
    {
      "epoch": 2.3525641025641026,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001274081714584016,
      "loss": 0.0534,
      "step": 9542
    },
    {
      "epoch": 2.3530571992110456,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012737506343811402,
      "loss": 0.0554,
      "step": 9544
    },
    {
      "epoch": 2.353550295857988,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0001273419521737286,
      "loss": 0.0543,
      "step": 9546
    },
    {
      "epoch": 2.354043392504931,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012730883766916919,
      "loss": 0.0516,
      "step": 9548
    },
    {
      "epoch": 2.3545364891518736,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00012727571992836003,
      "loss": 0.0509,
      "step": 9550
    },
    {
      "epoch": 2.3550295857988166,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012724259895522576,
      "loss": 0.0526,
      "step": 9552
    },
    {
      "epoch": 2.355522682445759,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012720947475369142,
      "loss": 0.0515,
      "step": 9554
    },
    {
      "epoch": 2.356015779092702,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012717634732768243,
      "loss": 0.0535,
      "step": 9556
    },
    {
      "epoch": 2.356508875739645,
      "grad_norm": 0.296875,
      "learning_rate": 0.00012714321668112454,
      "loss": 0.0509,
      "step": 9558
    },
    {
      "epoch": 2.3570019723865876,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00012711008281794393,
      "loss": 0.0541,
      "step": 9560
    },
    {
      "epoch": 2.3574950690335306,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012707694574206713,
      "loss": 0.0604,
      "step": 9562
    },
    {
      "epoch": 2.3579881656804735,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00012704380545742108,
      "loss": 0.0491,
      "step": 9564
    },
    {
      "epoch": 2.358481262327416,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00012701066196793307,
      "loss": 0.0558,
      "step": 9566
    },
    {
      "epoch": 2.358974358974359,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012697751527753084,
      "loss": 0.0542,
      "step": 9568
    },
    {
      "epoch": 2.359467455621302,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00012694436539014241,
      "loss": 0.0468,
      "step": 9570
    },
    {
      "epoch": 2.3599605522682445,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012691121230969628,
      "loss": 0.0481,
      "step": 9572
    },
    {
      "epoch": 2.3604536489151875,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00012687805604012122,
      "loss": 0.0572,
      "step": 9574
    },
    {
      "epoch": 2.36094674556213,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012684489658534642,
      "loss": 0.0571,
      "step": 9576
    },
    {
      "epoch": 2.361439842209073,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001268117339493015,
      "loss": 0.0547,
      "step": 9578
    },
    {
      "epoch": 2.361932938856016,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012677856813591643,
      "loss": 0.0459,
      "step": 9580
    },
    {
      "epoch": 2.3624260355029585,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001267453991491215,
      "loss": 0.0554,
      "step": 9582
    },
    {
      "epoch": 2.3629191321499015,
      "grad_norm": 0.390625,
      "learning_rate": 0.00012671222699284744,
      "loss": 0.0514,
      "step": 9584
    },
    {
      "epoch": 2.363412228796844,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00012667905167102538,
      "loss": 0.053,
      "step": 9586
    },
    {
      "epoch": 2.363905325443787,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012664587318758673,
      "loss": 0.0537,
      "step": 9588
    },
    {
      "epoch": 2.36439842209073,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00012661269154646334,
      "loss": 0.055,
      "step": 9590
    },
    {
      "epoch": 2.3648915187376724,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00012657950675158737,
      "loss": 0.052,
      "step": 9592
    },
    {
      "epoch": 2.3653846153846154,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00012654631880689148,
      "loss": 0.058,
      "step": 9594
    },
    {
      "epoch": 2.3658777120315584,
      "grad_norm": 0.46875,
      "learning_rate": 0.00012651312771630858,
      "loss": 0.055,
      "step": 9596
    },
    {
      "epoch": 2.366370808678501,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012647993348377203,
      "loss": 0.0575,
      "step": 9598
    },
    {
      "epoch": 2.366863905325444,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00012644673611321552,
      "loss": 0.0536,
      "step": 9600
    },
    {
      "epoch": 2.3673570019723864,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012641353560857312,
      "loss": 0.0522,
      "step": 9602
    },
    {
      "epoch": 2.3678500986193294,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001263803319737793,
      "loss": 0.0481,
      "step": 9604
    },
    {
      "epoch": 2.3683431952662723,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012634712521276883,
      "loss": 0.0501,
      "step": 9606
    },
    {
      "epoch": 2.368836291913215,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012631391532947687,
      "loss": 0.0497,
      "step": 9608
    },
    {
      "epoch": 2.369329388560158,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012628070232783908,
      "loss": 0.0499,
      "step": 9610
    },
    {
      "epoch": 2.3698224852071004,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00012624748621179133,
      "loss": 0.0518,
      "step": 9612
    },
    {
      "epoch": 2.3703155818540433,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0001262142669852699,
      "loss": 0.0503,
      "step": 9614
    },
    {
      "epoch": 2.3708086785009863,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00012618104465221148,
      "loss": 0.052,
      "step": 9616
    },
    {
      "epoch": 2.371301775147929,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00012614781921655308,
      "loss": 0.0552,
      "step": 9618
    },
    {
      "epoch": 2.371794871794872,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00012611459068223214,
      "loss": 0.0541,
      "step": 9620
    },
    {
      "epoch": 2.3722879684418148,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00012608135905318635,
      "loss": 0.0518,
      "step": 9622
    },
    {
      "epoch": 2.3727810650887573,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00012604812433335394,
      "loss": 0.0552,
      "step": 9624
    },
    {
      "epoch": 2.3732741617357003,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00012601488652667334,
      "loss": 0.0568,
      "step": 9626
    },
    {
      "epoch": 2.3737672583826432,
      "grad_norm": 0.359375,
      "learning_rate": 0.00012598164563708345,
      "loss": 0.0546,
      "step": 9628
    },
    {
      "epoch": 2.3742603550295858,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001259484016685235,
      "loss": 0.0537,
      "step": 9630
    },
    {
      "epoch": 2.3747534516765287,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012591515462493305,
      "loss": 0.0476,
      "step": 9632
    },
    {
      "epoch": 2.3752465483234713,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00012588190451025207,
      "loss": 0.0547,
      "step": 9634
    },
    {
      "epoch": 2.3757396449704142,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00012584865132842093,
      "loss": 0.0539,
      "step": 9636
    },
    {
      "epoch": 2.3762327416173568,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001258153950833803,
      "loss": 0.0561,
      "step": 9638
    },
    {
      "epoch": 2.3767258382642997,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012578213577907118,
      "loss": 0.0493,
      "step": 9640
    },
    {
      "epoch": 2.3772189349112427,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00012574887341943507,
      "loss": 0.0489,
      "step": 9642
    },
    {
      "epoch": 2.3777120315581852,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00012571560800841365,
      "loss": 0.045,
      "step": 9644
    },
    {
      "epoch": 2.378205128205128,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012568233954994912,
      "loss": 0.0499,
      "step": 9646
    },
    {
      "epoch": 2.378698224852071,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00012564906804798394,
      "loss": 0.0519,
      "step": 9648
    },
    {
      "epoch": 2.3791913214990137,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00012561579350646101,
      "loss": 0.0565,
      "step": 9650
    },
    {
      "epoch": 2.3796844181459567,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012558251592932352,
      "loss": 0.0502,
      "step": 9652
    },
    {
      "epoch": 2.3801775147928996,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012554923532051508,
      "loss": 0.0542,
      "step": 9654
    },
    {
      "epoch": 2.380670611439842,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00012551595168397962,
      "loss": 0.0545,
      "step": 9656
    },
    {
      "epoch": 2.381163708086785,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012548266502366139,
      "loss": 0.0523,
      "step": 9658
    },
    {
      "epoch": 2.3816568047337277,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00012544937534350508,
      "loss": 0.0508,
      "step": 9660
    },
    {
      "epoch": 2.3821499013806706,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001254160826474557,
      "loss": 0.0503,
      "step": 9662
    },
    {
      "epoch": 2.3826429980276136,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001253827869394586,
      "loss": 0.0537,
      "step": 9664
    },
    {
      "epoch": 2.383136094674556,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00012534948822345957,
      "loss": 0.0481,
      "step": 9666
    },
    {
      "epoch": 2.383629191321499,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00012531618650340464,
      "loss": 0.0492,
      "step": 9668
    },
    {
      "epoch": 2.3841222879684416,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00012528288178324028,
      "loss": 0.0528,
      "step": 9670
    },
    {
      "epoch": 2.3846153846153846,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012524957406691325,
      "loss": 0.0566,
      "step": 9672
    },
    {
      "epoch": 2.3851084812623276,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00012521626335837072,
      "loss": 0.052,
      "step": 9674
    },
    {
      "epoch": 2.38560157790927,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00012518294966156016,
      "loss": 0.0503,
      "step": 9676
    },
    {
      "epoch": 2.386094674556213,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00012514963298042948,
      "loss": 0.0452,
      "step": 9678
    },
    {
      "epoch": 2.386587771203156,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001251163133189269,
      "loss": 0.0467,
      "step": 9680
    },
    {
      "epoch": 2.3870808678500985,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012508299068100093,
      "loss": 0.0537,
      "step": 9682
    },
    {
      "epoch": 2.3875739644970415,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012504966507060052,
      "loss": 0.0481,
      "step": 9684
    },
    {
      "epoch": 2.388067061143984,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00012501633649167495,
      "loss": 0.0534,
      "step": 9686
    },
    {
      "epoch": 2.388560157790927,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00012498300494817385,
      "loss": 0.0578,
      "step": 9688
    },
    {
      "epoch": 2.38905325443787,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00012494967044404716,
      "loss": 0.0525,
      "step": 9690
    },
    {
      "epoch": 2.3895463510848125,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001249163329832452,
      "loss": 0.0501,
      "step": 9692
    },
    {
      "epoch": 2.3900394477317555,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001248829925697187,
      "loss": 0.0496,
      "step": 9694
    },
    {
      "epoch": 2.390532544378698,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012484964920741863,
      "loss": 0.0558,
      "step": 9696
    },
    {
      "epoch": 2.391025641025641,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001248163029002964,
      "loss": 0.0508,
      "step": 9698
    },
    {
      "epoch": 2.391518737672584,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001247829536523037,
      "loss": 0.0509,
      "step": 9700
    },
    {
      "epoch": 2.3920118343195265,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00012474960146739264,
      "loss": 0.0494,
      "step": 9702
    },
    {
      "epoch": 2.3925049309664694,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00012471624634951563,
      "loss": 0.0484,
      "step": 9704
    },
    {
      "epoch": 2.3929980276134124,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012468288830262542,
      "loss": 0.0501,
      "step": 9706
    },
    {
      "epoch": 2.393491124260355,
      "grad_norm": 0.375,
      "learning_rate": 0.00012464952733067512,
      "loss": 0.0529,
      "step": 9708
    },
    {
      "epoch": 2.393984220907298,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012461616343761823,
      "loss": 0.0555,
      "step": 9710
    },
    {
      "epoch": 2.394477317554241,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00012458279662740852,
      "loss": 0.0531,
      "step": 9712
    },
    {
      "epoch": 2.3949704142011834,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00012454942690400016,
      "loss": 0.0485,
      "step": 9714
    },
    {
      "epoch": 2.3954635108481264,
      "grad_norm": 0.34375,
      "learning_rate": 0.00012451605427134761,
      "loss": 0.046,
      "step": 9716
    },
    {
      "epoch": 2.395956607495069,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0001244826787334058,
      "loss": 0.0533,
      "step": 9718
    },
    {
      "epoch": 2.396449704142012,
      "grad_norm": 0.375,
      "learning_rate": 0.00012444930029412987,
      "loss": 0.052,
      "step": 9720
    },
    {
      "epoch": 2.3969428007889544,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001244159189574753,
      "loss": 0.0524,
      "step": 9722
    },
    {
      "epoch": 2.3974358974358974,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00012438253472739805,
      "loss": 0.0511,
      "step": 9724
    },
    {
      "epoch": 2.3979289940828403,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001243491476078543,
      "loss": 0.0434,
      "step": 9726
    },
    {
      "epoch": 2.398422090729783,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001243157576028006,
      "loss": 0.0586,
      "step": 9728
    },
    {
      "epoch": 2.398915187376726,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012428236471619385,
      "loss": 0.052,
      "step": 9730
    },
    {
      "epoch": 2.399408284023669,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012424896895199132,
      "loss": 0.0503,
      "step": 9732
    },
    {
      "epoch": 2.3999013806706113,
      "grad_norm": 0.34375,
      "learning_rate": 0.00012421557031415056,
      "loss": 0.0487,
      "step": 9734
    },
    {
      "epoch": 2.4003944773175543,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012418216880662956,
      "loss": 0.0478,
      "step": 9736
    },
    {
      "epoch": 2.4008875739644973,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001241487644333865,
      "loss": 0.0515,
      "step": 9738
    },
    {
      "epoch": 2.40138067061144,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012411535719838006,
      "loss": 0.0551,
      "step": 9740
    },
    {
      "epoch": 2.4018737672583828,
      "grad_norm": 0.453125,
      "learning_rate": 0.00012408194710556914,
      "loss": 0.0555,
      "step": 9742
    },
    {
      "epoch": 2.4023668639053253,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012404853415891299,
      "loss": 0.0564,
      "step": 9744
    },
    {
      "epoch": 2.4028599605522682,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0001240151183623713,
      "loss": 0.0494,
      "step": 9746
    },
    {
      "epoch": 2.403353057199211,
      "grad_norm": 0.455078125,
      "learning_rate": 0.000123981699719904,
      "loss": 0.0496,
      "step": 9748
    },
    {
      "epoch": 2.4038461538461537,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00012394827823547142,
      "loss": 0.0505,
      "step": 9750
    },
    {
      "epoch": 2.4043392504930967,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001239148539130341,
      "loss": 0.0468,
      "step": 9752
    },
    {
      "epoch": 2.4048323471400392,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001238814267565531,
      "loss": 0.0475,
      "step": 9754
    },
    {
      "epoch": 2.405325443786982,
      "grad_norm": 0.5625,
      "learning_rate": 0.00012384799676998972,
      "loss": 0.0472,
      "step": 9756
    },
    {
      "epoch": 2.405818540433925,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00012381456395730553,
      "loss": 0.048,
      "step": 9758
    },
    {
      "epoch": 2.4063116370808677,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012378112832246255,
      "loss": 0.0515,
      "step": 9760
    },
    {
      "epoch": 2.4068047337278107,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001237476898694231,
      "loss": 0.0522,
      "step": 9762
    },
    {
      "epoch": 2.4072978303747536,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00012371424860214977,
      "loss": 0.0486,
      "step": 9764
    },
    {
      "epoch": 2.407790927021696,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00012368080452460564,
      "loss": 0.049,
      "step": 9766
    },
    {
      "epoch": 2.408284023668639,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012364735764075393,
      "loss": 0.0468,
      "step": 9768
    },
    {
      "epoch": 2.4087771203155817,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001236139079545583,
      "loss": 0.0493,
      "step": 9770
    },
    {
      "epoch": 2.4092702169625246,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00012358045546998275,
      "loss": 0.0521,
      "step": 9772
    },
    {
      "epoch": 2.4097633136094676,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00012354700019099157,
      "loss": 0.046,
      "step": 9774
    },
    {
      "epoch": 2.41025641025641,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001235135421215494,
      "loss": 0.0517,
      "step": 9776
    },
    {
      "epoch": 2.410749506903353,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001234800812656212,
      "loss": 0.0498,
      "step": 9778
    },
    {
      "epoch": 2.4112426035502956,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012344661762717226,
      "loss": 0.0423,
      "step": 9780
    },
    {
      "epoch": 2.4117357001972386,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012341315121016823,
      "loss": 0.0457,
      "step": 9782
    },
    {
      "epoch": 2.4122287968441816,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00012337968201857506,
      "loss": 0.0488,
      "step": 9784
    },
    {
      "epoch": 2.412721893491124,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00012334621005635906,
      "loss": 0.0519,
      "step": 9786
    },
    {
      "epoch": 2.413214990138067,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001233127353274868,
      "loss": 0.0536,
      "step": 9788
    },
    {
      "epoch": 2.41370808678501,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00012327925783592523,
      "loss": 0.0496,
      "step": 9790
    },
    {
      "epoch": 2.4142011834319526,
      "grad_norm": 0.5625,
      "learning_rate": 0.00012324577758564164,
      "loss": 0.0491,
      "step": 9792
    },
    {
      "epoch": 2.4146942800788955,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012321229458060363,
      "loss": 0.0565,
      "step": 9794
    },
    {
      "epoch": 2.4151873767258385,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001231788088247791,
      "loss": 0.054,
      "step": 9796
    },
    {
      "epoch": 2.415680473372781,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00012314532032213632,
      "loss": 0.0488,
      "step": 9798
    },
    {
      "epoch": 2.416173570019724,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0001231118290766438,
      "loss": 0.0523,
      "step": 9800
    },
    {
      "epoch": 2.4166666666666665,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00012307833509227057,
      "loss": 0.0506,
      "step": 9802
    },
    {
      "epoch": 2.4171597633136095,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012304483837298578,
      "loss": 0.0504,
      "step": 9804
    },
    {
      "epoch": 2.4176528599605525,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00012301133892275894,
      "loss": 0.0462,
      "step": 9806
    },
    {
      "epoch": 2.418145956607495,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012297783674555996,
      "loss": 0.0518,
      "step": 9808
    },
    {
      "epoch": 2.418639053254438,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00012294433184535905,
      "loss": 0.0445,
      "step": 9810
    },
    {
      "epoch": 2.4191321499013805,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001229108242261267,
      "loss": 0.0442,
      "step": 9812
    },
    {
      "epoch": 2.4196252465483234,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00012287731389183375,
      "loss": 0.0464,
      "step": 9814
    },
    {
      "epoch": 2.4201183431952664,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012284380084645139,
      "loss": 0.0419,
      "step": 9816
    },
    {
      "epoch": 2.420611439842209,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00012281028509395108,
      "loss": 0.0505,
      "step": 9818
    },
    {
      "epoch": 2.421104536489152,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00012277676663830463,
      "loss": 0.0487,
      "step": 9820
    },
    {
      "epoch": 2.421597633136095,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00012274324548348418,
      "loss": 0.051,
      "step": 9822
    },
    {
      "epoch": 2.4220907297830374,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00012270972163346214,
      "loss": 0.0509,
      "step": 9824
    },
    {
      "epoch": 2.4225838264299804,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001226761950922113,
      "loss": 0.0474,
      "step": 9826
    },
    {
      "epoch": 2.423076923076923,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012264266586370473,
      "loss": 0.0521,
      "step": 9828
    },
    {
      "epoch": 2.423570019723866,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00012260913395191583,
      "loss": 0.0506,
      "step": 9830
    },
    {
      "epoch": 2.424063116370809,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00012257559936081834,
      "loss": 0.0507,
      "step": 9832
    },
    {
      "epoch": 2.4245562130177514,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00012254206209438625,
      "loss": 0.05,
      "step": 9834
    },
    {
      "epoch": 2.4250493096646943,
      "grad_norm": 0.3359375,
      "learning_rate": 0.000122508522156594,
      "loss": 0.0458,
      "step": 9836
    },
    {
      "epoch": 2.425542406311637,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00012247497955141613,
      "loss": 0.052,
      "step": 9838
    },
    {
      "epoch": 2.42603550295858,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00012244143428282776,
      "loss": 0.0535,
      "step": 9840
    },
    {
      "epoch": 2.426528599605523,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012240788635480408,
      "loss": 0.0499,
      "step": 9842
    },
    {
      "epoch": 2.4270216962524653,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00012237433577132079,
      "loss": 0.0508,
      "step": 9844
    },
    {
      "epoch": 2.4275147928994083,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00012234078253635377,
      "loss": 0.0465,
      "step": 9846
    },
    {
      "epoch": 2.4280078895463513,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001223072266538793,
      "loss": 0.0489,
      "step": 9848
    },
    {
      "epoch": 2.428500986193294,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001222736681278739,
      "loss": 0.0469,
      "step": 9850
    },
    {
      "epoch": 2.4289940828402368,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00012224010696231452,
      "loss": 0.0475,
      "step": 9852
    },
    {
      "epoch": 2.4294871794871793,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012220654316117825,
      "loss": 0.0496,
      "step": 9854
    },
    {
      "epoch": 2.4299802761341223,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00012217297672844263,
      "loss": 0.0479,
      "step": 9856
    },
    {
      "epoch": 2.4304733727810652,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00012213940766808546,
      "loss": 0.0444,
      "step": 9858
    },
    {
      "epoch": 2.4309664694280078,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001221058359840849,
      "loss": 0.0472,
      "step": 9860
    },
    {
      "epoch": 2.4314595660749507,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012207226168041934,
      "loss": 0.0485,
      "step": 9862
    },
    {
      "epoch": 2.4319526627218933,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00012203868476106755,
      "loss": 0.0424,
      "step": 9864
    },
    {
      "epoch": 2.4324457593688362,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00012200510523000854,
      "loss": 0.0514,
      "step": 9866
    },
    {
      "epoch": 2.432938856015779,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012197152309122173,
      "loss": 0.0473,
      "step": 9868
    },
    {
      "epoch": 2.4334319526627217,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00012193793834868674,
      "loss": 0.0507,
      "step": 9870
    },
    {
      "epoch": 2.4339250493096647,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00012190435100638355,
      "loss": 0.0467,
      "step": 9872
    },
    {
      "epoch": 2.4344181459566077,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00012187076106829249,
      "loss": 0.0536,
      "step": 9874
    },
    {
      "epoch": 2.43491124260355,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012183716853839416,
      "loss": 0.0529,
      "step": 9876
    },
    {
      "epoch": 2.435404339250493,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012180357342066939,
      "loss": 0.0478,
      "step": 9878
    },
    {
      "epoch": 2.435897435897436,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00012176997571909945,
      "loss": 0.0515,
      "step": 9880
    },
    {
      "epoch": 2.4363905325443787,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00012173637543766583,
      "loss": 0.0521,
      "step": 9882
    },
    {
      "epoch": 2.4368836291913216,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00012170277258035038,
      "loss": 0.0501,
      "step": 9884
    },
    {
      "epoch": 2.437376725838264,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012166916715113521,
      "loss": 0.0453,
      "step": 9886
    },
    {
      "epoch": 2.437869822485207,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00012163555915400273,
      "loss": 0.0436,
      "step": 9888
    },
    {
      "epoch": 2.43836291913215,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012160194859293572,
      "loss": 0.0475,
      "step": 9890
    },
    {
      "epoch": 2.4388560157790926,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00012156833547191717,
      "loss": 0.049,
      "step": 9892
    },
    {
      "epoch": 2.4393491124260356,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00012153471979493043,
      "loss": 0.0478,
      "step": 9894
    },
    {
      "epoch": 2.439842209072978,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012150110156595918,
      "loss": 0.0474,
      "step": 9896
    },
    {
      "epoch": 2.440335305719921,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012146748078898733,
      "loss": 0.0489,
      "step": 9898
    },
    {
      "epoch": 2.440828402366864,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00012143385746799915,
      "loss": 0.0473,
      "step": 9900
    },
    {
      "epoch": 2.4413214990138066,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012140023160697922,
      "loss": 0.0456,
      "step": 9902
    },
    {
      "epoch": 2.4418145956607495,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012136660320991236,
      "loss": 0.0473,
      "step": 9904
    },
    {
      "epoch": 2.4423076923076925,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00012133297228078369,
      "loss": 0.0492,
      "step": 9906
    },
    {
      "epoch": 2.442800788954635,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012129933882357871,
      "loss": 0.044,
      "step": 9908
    },
    {
      "epoch": 2.443293885601578,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00012126570284228314,
      "loss": 0.0473,
      "step": 9910
    },
    {
      "epoch": 2.4437869822485205,
      "grad_norm": 0.390625,
      "learning_rate": 0.00012123206434088308,
      "loss": 0.0437,
      "step": 9912
    },
    {
      "epoch": 2.4442800788954635,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012119842332336482,
      "loss": 0.0489,
      "step": 9914
    },
    {
      "epoch": 2.4447731755424065,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012116477979371508,
      "loss": 0.0453,
      "step": 9916
    },
    {
      "epoch": 2.445266272189349,
      "grad_norm": 0.46875,
      "learning_rate": 0.00012113113375592076,
      "loss": 0.0497,
      "step": 9918
    },
    {
      "epoch": 2.445759368836292,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00012109748521396908,
      "loss": 0.0475,
      "step": 9920
    },
    {
      "epoch": 2.4462524654832345,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012106383417184763,
      "loss": 0.0473,
      "step": 9922
    },
    {
      "epoch": 2.4467455621301775,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00012103018063354421,
      "loss": 0.0432,
      "step": 9924
    },
    {
      "epoch": 2.4472386587771204,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00012099652460304698,
      "loss": 0.0541,
      "step": 9926
    },
    {
      "epoch": 2.447731755424063,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00012096286608434432,
      "loss": 0.0447,
      "step": 9928
    },
    {
      "epoch": 2.448224852071006,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00012092920508142505,
      "loss": 0.0449,
      "step": 9930
    },
    {
      "epoch": 2.448717948717949,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00012089554159827807,
      "loss": 0.0438,
      "step": 9932
    },
    {
      "epoch": 2.4492110453648914,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001208618756388928,
      "loss": 0.0472,
      "step": 9934
    },
    {
      "epoch": 2.4497041420118344,
      "grad_norm": 0.296875,
      "learning_rate": 0.00012082820720725876,
      "loss": 0.0474,
      "step": 9936
    },
    {
      "epoch": 2.4501972386587774,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012079453630736586,
      "loss": 0.0477,
      "step": 9938
    },
    {
      "epoch": 2.45069033530572,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001207608629432043,
      "loss": 0.0531,
      "step": 9940
    },
    {
      "epoch": 2.451183431952663,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00012072718711876459,
      "loss": 0.0427,
      "step": 9942
    },
    {
      "epoch": 2.4516765285996054,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00012069350883803749,
      "loss": 0.0483,
      "step": 9944
    },
    {
      "epoch": 2.4521696252465484,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012065982810501404,
      "loss": 0.0532,
      "step": 9946
    },
    {
      "epoch": 2.452662721893491,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00012062614492368561,
      "loss": 0.051,
      "step": 9948
    },
    {
      "epoch": 2.453155818540434,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00012059245929804384,
      "loss": 0.0441,
      "step": 9950
    },
    {
      "epoch": 2.453648915187377,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00012055877123208067,
      "loss": 0.0478,
      "step": 9952
    },
    {
      "epoch": 2.4541420118343193,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00012052508072978832,
      "loss": 0.0487,
      "step": 9954
    },
    {
      "epoch": 2.4546351084812623,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00012049138779515932,
      "loss": 0.0519,
      "step": 9956
    },
    {
      "epoch": 2.4551282051282053,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012045769243218647,
      "loss": 0.0468,
      "step": 9958
    },
    {
      "epoch": 2.455621301775148,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00012042399464486282,
      "loss": 0.0467,
      "step": 9960
    },
    {
      "epoch": 2.456114398422091,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00012039029443718177,
      "loss": 0.0446,
      "step": 9962
    },
    {
      "epoch": 2.4566074950690338,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012035659181313699,
      "loss": 0.0486,
      "step": 9964
    },
    {
      "epoch": 2.4571005917159763,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012032288677672244,
      "loss": 0.0489,
      "step": 9966
    },
    {
      "epoch": 2.4575936883629192,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00012028917933193233,
      "loss": 0.0496,
      "step": 9968
    },
    {
      "epoch": 2.4580867850098618,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00012025546948276116,
      "loss": 0.0498,
      "step": 9970
    },
    {
      "epoch": 2.4585798816568047,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00012022175723320381,
      "loss": 0.0456,
      "step": 9972
    },
    {
      "epoch": 2.4590729783037477,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00012018804258725532,
      "loss": 0.0425,
      "step": 9974
    },
    {
      "epoch": 2.4595660749506902,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00012015432554891105,
      "loss": 0.047,
      "step": 9976
    },
    {
      "epoch": 2.460059171597633,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012012060612216664,
      "loss": 0.0457,
      "step": 9978
    },
    {
      "epoch": 2.4605522682445757,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001200868843110181,
      "loss": 0.0472,
      "step": 9980
    },
    {
      "epoch": 2.4610453648915187,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012005316011946162,
      "loss": 0.0452,
      "step": 9982
    },
    {
      "epoch": 2.4615384615384617,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00012001943355149371,
      "loss": 0.0457,
      "step": 9984
    },
    {
      "epoch": 2.462031558185404,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00011998570461111115,
      "loss": 0.048,
      "step": 9986
    },
    {
      "epoch": 2.462524654832347,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011995197330231101,
      "loss": 0.0495,
      "step": 9988
    },
    {
      "epoch": 2.46301775147929,
      "grad_norm": 0.390625,
      "learning_rate": 0.00011991823962909062,
      "loss": 0.045,
      "step": 9990
    },
    {
      "epoch": 2.4635108481262327,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00011988450359544763,
      "loss": 0.0561,
      "step": 9992
    },
    {
      "epoch": 2.4640039447731756,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00011985076520537995,
      "loss": 0.0471,
      "step": 9994
    },
    {
      "epoch": 2.464497041420118,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011981702446288575,
      "loss": 0.0506,
      "step": 9996
    },
    {
      "epoch": 2.464990138067061,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001197832813719635,
      "loss": 0.0489,
      "step": 9998
    },
    {
      "epoch": 2.465483234714004,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00011974953593661198,
      "loss": 0.0502,
      "step": 10000
    },
    {
      "epoch": 2.4659763313609466,
      "grad_norm": 0.375,
      "learning_rate": 0.00011971578816083018,
      "loss": 0.0475,
      "step": 10002
    },
    {
      "epoch": 2.4664694280078896,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00011968203804861737,
      "loss": 0.0492,
      "step": 10004
    },
    {
      "epoch": 2.466962524654832,
      "grad_norm": 0.390625,
      "learning_rate": 0.00011964828560397319,
      "loss": 0.0476,
      "step": 10006
    },
    {
      "epoch": 2.467455621301775,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011961453083089744,
      "loss": 0.0494,
      "step": 10008
    },
    {
      "epoch": 2.467948717948718,
      "grad_norm": 0.375,
      "learning_rate": 0.0001195807737333903,
      "loss": 0.0511,
      "step": 10010
    },
    {
      "epoch": 2.4684418145956606,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011954701431545214,
      "loss": 0.0525,
      "step": 10012
    },
    {
      "epoch": 2.4689349112426036,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00011951325258108363,
      "loss": 0.0463,
      "step": 10014
    },
    {
      "epoch": 2.4694280078895465,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00011947948853428576,
      "loss": 0.0439,
      "step": 10016
    },
    {
      "epoch": 2.469921104536489,
      "grad_norm": 0.328125,
      "learning_rate": 0.00011944572217905976,
      "loss": 0.049,
      "step": 10018
    },
    {
      "epoch": 2.470414201183432,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00011941195351940708,
      "loss": 0.048,
      "step": 10020
    },
    {
      "epoch": 2.470907297830375,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001193781825593295,
      "loss": 0.0463,
      "step": 10022
    },
    {
      "epoch": 2.4714003944773175,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011934440930282913,
      "loss": 0.0472,
      "step": 10024
    },
    {
      "epoch": 2.4718934911242605,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00011931063375390825,
      "loss": 0.0447,
      "step": 10026
    },
    {
      "epoch": 2.472386587771203,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00011927685591656947,
      "loss": 0.0458,
      "step": 10028
    },
    {
      "epoch": 2.472879684418146,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001192430757948156,
      "loss": 0.05,
      "step": 10030
    },
    {
      "epoch": 2.4733727810650885,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011920929339264982,
      "loss": 0.0448,
      "step": 10032
    },
    {
      "epoch": 2.4738658777120315,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00011917550871407554,
      "loss": 0.0456,
      "step": 10034
    },
    {
      "epoch": 2.4743589743589745,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0001191417217630964,
      "loss": 0.0452,
      "step": 10036
    },
    {
      "epoch": 2.474852071005917,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00011910793254371634,
      "loss": 0.049,
      "step": 10038
    },
    {
      "epoch": 2.47534516765286,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011907414105993961,
      "loss": 0.0486,
      "step": 10040
    },
    {
      "epoch": 2.475838264299803,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011904034731577068,
      "loss": 0.0438,
      "step": 10042
    },
    {
      "epoch": 2.4763313609467454,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011900655131521427,
      "loss": 0.0461,
      "step": 10044
    },
    {
      "epoch": 2.4768244575936884,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001189727530622754,
      "loss": 0.0475,
      "step": 10046
    },
    {
      "epoch": 2.4773175542406314,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011893895256095937,
      "loss": 0.0476,
      "step": 10048
    },
    {
      "epoch": 2.477810650887574,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00011890514981527175,
      "loss": 0.0447,
      "step": 10050
    },
    {
      "epoch": 2.478303747534517,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00011887134482921829,
      "loss": 0.0496,
      "step": 10052
    },
    {
      "epoch": 2.4787968441814594,
      "grad_norm": 0.546875,
      "learning_rate": 0.00011883753760680509,
      "loss": 0.0464,
      "step": 10054
    },
    {
      "epoch": 2.4792899408284024,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011880372815203855,
      "loss": 0.0436,
      "step": 10056
    },
    {
      "epoch": 2.4797830374753453,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00011876991646892522,
      "loss": 0.0466,
      "step": 10058
    },
    {
      "epoch": 2.480276134122288,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00011873610256147197,
      "loss": 0.0483,
      "step": 10060
    },
    {
      "epoch": 2.480769230769231,
      "grad_norm": 0.4375,
      "learning_rate": 0.00011870228643368598,
      "loss": 0.048,
      "step": 10062
    },
    {
      "epoch": 2.4812623274161734,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001186684680895746,
      "loss": 0.0498,
      "step": 10064
    },
    {
      "epoch": 2.4817554240631163,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00011863464753314556,
      "loss": 0.0471,
      "step": 10066
    },
    {
      "epoch": 2.4822485207100593,
      "grad_norm": 0.4375,
      "learning_rate": 0.00011860082476840672,
      "loss": 0.046,
      "step": 10068
    },
    {
      "epoch": 2.482741617357002,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00011856699979936632,
      "loss": 0.0425,
      "step": 10070
    },
    {
      "epoch": 2.483234714003945,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00011853317263003274,
      "loss": 0.0485,
      "step": 10072
    },
    {
      "epoch": 2.4837278106508878,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00011849934326441474,
      "loss": 0.0455,
      "step": 10074
    },
    {
      "epoch": 2.4842209072978303,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00011846551170652127,
      "loss": 0.0456,
      "step": 10076
    },
    {
      "epoch": 2.4847140039447733,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00011843167796036156,
      "loss": 0.047,
      "step": 10078
    },
    {
      "epoch": 2.485207100591716,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011839784202994513,
      "loss": 0.0467,
      "step": 10080
    },
    {
      "epoch": 2.4857001972386588,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011836400391928168,
      "loss": 0.0466,
      "step": 10082
    },
    {
      "epoch": 2.4861932938856017,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00011833016363238127,
      "loss": 0.0413,
      "step": 10084
    },
    {
      "epoch": 2.4866863905325443,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00011829632117325408,
      "loss": 0.0466,
      "step": 10086
    },
    {
      "epoch": 2.4871794871794872,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001182624765459107,
      "loss": 0.0474,
      "step": 10088
    },
    {
      "epoch": 2.4876725838264298,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011822862975436187,
      "loss": 0.0483,
      "step": 10090
    },
    {
      "epoch": 2.4881656804733727,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011819478080261863,
      "loss": 0.0447,
      "step": 10092
    },
    {
      "epoch": 2.4886587771203157,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001181609296946923,
      "loss": 0.0478,
      "step": 10094
    },
    {
      "epoch": 2.489151873767258,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00011812707643459441,
      "loss": 0.0461,
      "step": 10096
    },
    {
      "epoch": 2.489644970414201,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00011809322102633673,
      "loss": 0.0518,
      "step": 10098
    },
    {
      "epoch": 2.490138067061144,
      "grad_norm": 0.46875,
      "learning_rate": 0.00011805936347393139,
      "loss": 0.0438,
      "step": 10100
    },
    {
      "epoch": 2.4906311637080867,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0001180255037813906,
      "loss": 0.0462,
      "step": 10102
    },
    {
      "epoch": 2.4911242603550297,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00011799164195272697,
      "loss": 0.0416,
      "step": 10104
    },
    {
      "epoch": 2.4916173570019726,
      "grad_norm": 0.40625,
      "learning_rate": 0.00011795777799195331,
      "loss": 0.0446,
      "step": 10106
    },
    {
      "epoch": 2.492110453648915,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011792391190308272,
      "loss": 0.0452,
      "step": 10108
    },
    {
      "epoch": 2.492603550295858,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011789004369012848,
      "loss": 0.046,
      "step": 10110
    },
    {
      "epoch": 2.4930966469428006,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011785617335710419,
      "loss": 0.0442,
      "step": 10112
    },
    {
      "epoch": 2.4935897435897436,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011782230090802365,
      "loss": 0.0479,
      "step": 10114
    },
    {
      "epoch": 2.4940828402366866,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00011778842634690097,
      "loss": 0.038,
      "step": 10116
    },
    {
      "epoch": 2.494575936883629,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00011775454967775042,
      "loss": 0.0451,
      "step": 10118
    },
    {
      "epoch": 2.495069033530572,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00011772067090458661,
      "loss": 0.0455,
      "step": 10120
    },
    {
      "epoch": 2.4955621301775146,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011768679003142435,
      "loss": 0.0534,
      "step": 10122
    },
    {
      "epoch": 2.4960552268244576,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011765290706227877,
      "loss": 0.046,
      "step": 10124
    },
    {
      "epoch": 2.4965483234714005,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011761902200116508,
      "loss": 0.0508,
      "step": 10126
    },
    {
      "epoch": 2.497041420118343,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00011758513485209892,
      "loss": 0.0466,
      "step": 10128
    },
    {
      "epoch": 2.497534516765286,
      "grad_norm": 0.296875,
      "learning_rate": 0.00011755124561909612,
      "loss": 0.0455,
      "step": 10130
    },
    {
      "epoch": 2.498027613412229,
      "grad_norm": 0.390625,
      "learning_rate": 0.00011751735430617272,
      "loss": 0.0477,
      "step": 10132
    },
    {
      "epoch": 2.4985207100591715,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00011748346091734504,
      "loss": 0.0475,
      "step": 10134
    },
    {
      "epoch": 2.4990138067061145,
      "grad_norm": 0.34375,
      "learning_rate": 0.00011744956545662959,
      "loss": 0.0482,
      "step": 10136
    },
    {
      "epoch": 2.499506903353057,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011741566792804324,
      "loss": 0.056,
      "step": 10138
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00011738176833560299,
      "loss": 0.055,
      "step": 10140
    },
    {
      "epoch": 2.5004930966469425,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011734786668332614,
      "loss": 0.0466,
      "step": 10142
    },
    {
      "epoch": 2.5009861932938855,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00011731396297523023,
      "loss": 0.0449,
      "step": 10144
    },
    {
      "epoch": 2.5014792899408285,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011728005721533305,
      "loss": 0.0478,
      "step": 10146
    },
    {
      "epoch": 2.501972386587771,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001172461494076526,
      "loss": 0.0497,
      "step": 10148
    },
    {
      "epoch": 2.502465483234714,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00011721223955620718,
      "loss": 0.0463,
      "step": 10150
    },
    {
      "epoch": 2.502958579881657,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00011717832766501526,
      "loss": 0.0463,
      "step": 10152
    },
    {
      "epoch": 2.5034516765285995,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011714441373809557,
      "loss": 0.0513,
      "step": 10154
    },
    {
      "epoch": 2.5039447731755424,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00011711049777946716,
      "loss": 0.0443,
      "step": 10156
    },
    {
      "epoch": 2.5044378698224854,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00011707657979314922,
      "loss": 0.0466,
      "step": 10158
    },
    {
      "epoch": 2.504930966469428,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00011704265978316122,
      "loss": 0.0456,
      "step": 10160
    },
    {
      "epoch": 2.505424063116371,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001170087377535229,
      "loss": 0.0448,
      "step": 10162
    },
    {
      "epoch": 2.505917159763314,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0001169748137082542,
      "loss": 0.049,
      "step": 10164
    },
    {
      "epoch": 2.5064102564102564,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001169408876513753,
      "loss": 0.0472,
      "step": 10166
    },
    {
      "epoch": 2.5069033530571994,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011690695958690661,
      "loss": 0.0483,
      "step": 10168
    },
    {
      "epoch": 2.507396449704142,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001168730295188688,
      "loss": 0.0487,
      "step": 10170
    },
    {
      "epoch": 2.507889546351085,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00011683909745128281,
      "loss": 0.0435,
      "step": 10172
    },
    {
      "epoch": 2.5083826429980274,
      "grad_norm": 0.34375,
      "learning_rate": 0.00011680516338816974,
      "loss": 0.0448,
      "step": 10174
    },
    {
      "epoch": 2.5088757396449703,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011677122733355097,
      "loss": 0.0462,
      "step": 10176
    },
    {
      "epoch": 2.5093688362919133,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011673728929144815,
      "loss": 0.046,
      "step": 10178
    },
    {
      "epoch": 2.509861932938856,
      "grad_norm": 0.3125,
      "learning_rate": 0.00011670334926588309,
      "loss": 0.0397,
      "step": 10180
    },
    {
      "epoch": 2.510355029585799,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00011666940726087791,
      "loss": 0.0448,
      "step": 10182
    },
    {
      "epoch": 2.510848126232742,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011663546328045488,
      "loss": 0.0484,
      "step": 10184
    },
    {
      "epoch": 2.5113412228796843,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011660151732863653,
      "loss": 0.0458,
      "step": 10186
    },
    {
      "epoch": 2.5118343195266273,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00011656756940944573,
      "loss": 0.0475,
      "step": 10188
    },
    {
      "epoch": 2.5123274161735702,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00011653361952690542,
      "loss": 0.0462,
      "step": 10190
    },
    {
      "epoch": 2.5128205128205128,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011649966768503893,
      "loss": 0.0469,
      "step": 10192
    },
    {
      "epoch": 2.5133136094674557,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00011646571388786967,
      "loss": 0.0508,
      "step": 10194
    },
    {
      "epoch": 2.5138067061143983,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011643175813942137,
      "loss": 0.0474,
      "step": 10196
    },
    {
      "epoch": 2.5142998027613412,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00011639780044371803,
      "loss": 0.0438,
      "step": 10198
    },
    {
      "epoch": 2.5147928994082838,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00011636384080478376,
      "loss": 0.0468,
      "step": 10200
    },
    {
      "epoch": 2.5152859960552267,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011632987922664295,
      "loss": 0.0449,
      "step": 10202
    },
    {
      "epoch": 2.5157790927021697,
      "grad_norm": 0.375,
      "learning_rate": 0.00011629591571332033,
      "loss": 0.0444,
      "step": 10204
    },
    {
      "epoch": 2.5162721893491122,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001162619502688407,
      "loss": 0.0461,
      "step": 10206
    },
    {
      "epoch": 2.516765285996055,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011622798289722915,
      "loss": 0.0471,
      "step": 10208
    },
    {
      "epoch": 2.517258382642998,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011619401360251104,
      "loss": 0.0448,
      "step": 10210
    },
    {
      "epoch": 2.5177514792899407,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001161600423887119,
      "loss": 0.0455,
      "step": 10212
    },
    {
      "epoch": 2.5182445759368837,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001161260692598575,
      "loss": 0.0489,
      "step": 10214
    },
    {
      "epoch": 2.5187376725838266,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00011609209421997386,
      "loss": 0.0436,
      "step": 10216
    },
    {
      "epoch": 2.519230769230769,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011605811727308722,
      "loss": 0.0478,
      "step": 10218
    },
    {
      "epoch": 2.519723865877712,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011602413842322404,
      "loss": 0.049,
      "step": 10220
    },
    {
      "epoch": 2.520216962524655,
      "grad_norm": 0.40625,
      "learning_rate": 0.00011599015767441097,
      "loss": 0.0469,
      "step": 10222
    },
    {
      "epoch": 2.5207100591715976,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011595617503067493,
      "loss": 0.046,
      "step": 10224
    },
    {
      "epoch": 2.5212031558185406,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011592219049604305,
      "loss": 0.0437,
      "step": 10226
    },
    {
      "epoch": 2.521696252465483,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00011588820407454273,
      "loss": 0.0434,
      "step": 10228
    },
    {
      "epoch": 2.522189349112426,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011585421577020148,
      "loss": 0.0435,
      "step": 10230
    },
    {
      "epoch": 2.5226824457593686,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0001158202255870472,
      "loss": 0.0465,
      "step": 10232
    },
    {
      "epoch": 2.5231755424063116,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00011578623352910783,
      "loss": 0.0431,
      "step": 10234
    },
    {
      "epoch": 2.5236686390532546,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011575223960041164,
      "loss": 0.0476,
      "step": 10236
    },
    {
      "epoch": 2.524161735700197,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011571824380498711,
      "loss": 0.0492,
      "step": 10238
    },
    {
      "epoch": 2.52465483234714,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011568424614686293,
      "loss": 0.0473,
      "step": 10240
    },
    {
      "epoch": 2.525147928994083,
      "grad_norm": 0.283203125,
      "learning_rate": 0.000115650246630068,
      "loss": 0.0443,
      "step": 10242
    },
    {
      "epoch": 2.5256410256410255,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00011561624525863148,
      "loss": 0.0474,
      "step": 10244
    },
    {
      "epoch": 2.5261341222879685,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00011558224203658272,
      "loss": 0.0457,
      "step": 10246
    },
    {
      "epoch": 2.5266272189349115,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00011554823696795127,
      "loss": 0.0477,
      "step": 10248
    },
    {
      "epoch": 2.527120315581854,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00011551423005676691,
      "loss": 0.0486,
      "step": 10250
    },
    {
      "epoch": 2.527613412228797,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00011548022130705969,
      "loss": 0.0427,
      "step": 10252
    },
    {
      "epoch": 2.5281065088757395,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001154462107228598,
      "loss": 0.0433,
      "step": 10254
    },
    {
      "epoch": 2.5285996055226825,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00011541219830819769,
      "loss": 0.0461,
      "step": 10256
    },
    {
      "epoch": 2.529092702169625,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00011537818406710405,
      "loss": 0.0465,
      "step": 10258
    },
    {
      "epoch": 2.529585798816568,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00011534416800360975,
      "loss": 0.041,
      "step": 10260
    },
    {
      "epoch": 2.530078895463511,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011531015012174584,
      "loss": 0.0506,
      "step": 10262
    },
    {
      "epoch": 2.5305719921104535,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011527613042554368,
      "loss": 0.0486,
      "step": 10264
    },
    {
      "epoch": 2.5310650887573964,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00011524210891903478,
      "loss": 0.0459,
      "step": 10266
    },
    {
      "epoch": 2.5315581854043394,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00011520808560625085,
      "loss": 0.0508,
      "step": 10268
    },
    {
      "epoch": 2.532051282051282,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011517406049122385,
      "loss": 0.0462,
      "step": 10270
    },
    {
      "epoch": 2.532544378698225,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011514003357798597,
      "loss": 0.0457,
      "step": 10272
    },
    {
      "epoch": 2.533037475345168,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011510600487056961,
      "loss": 0.0451,
      "step": 10274
    },
    {
      "epoch": 2.5335305719921104,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00011507197437300732,
      "loss": 0.044,
      "step": 10276
    },
    {
      "epoch": 2.5340236686390534,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001150379420893319,
      "loss": 0.0445,
      "step": 10278
    },
    {
      "epoch": 2.534516765285996,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00011500390802357638,
      "loss": 0.0423,
      "step": 10280
    },
    {
      "epoch": 2.535009861932939,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00011496987217977402,
      "loss": 0.0398,
      "step": 10282
    },
    {
      "epoch": 2.5355029585798814,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001149358345619582,
      "loss": 0.0488,
      "step": 10284
    },
    {
      "epoch": 2.5359960552268244,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001149017951741626,
      "loss": 0.0464,
      "step": 10286
    },
    {
      "epoch": 2.5364891518737673,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00011486775402042109,
      "loss": 0.0456,
      "step": 10288
    },
    {
      "epoch": 2.53698224852071,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011483371110476768,
      "loss": 0.0425,
      "step": 10290
    },
    {
      "epoch": 2.537475345167653,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001147996664312367,
      "loss": 0.0431,
      "step": 10292
    },
    {
      "epoch": 2.537968441814596,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011476562000386261,
      "loss": 0.0443,
      "step": 10294
    },
    {
      "epoch": 2.5384615384615383,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001147315718266801,
      "loss": 0.0438,
      "step": 10296
    },
    {
      "epoch": 2.5389546351084813,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011469752190372411,
      "loss": 0.0466,
      "step": 10298
    },
    {
      "epoch": 2.5394477317554243,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001146634702390297,
      "loss": 0.0486,
      "step": 10300
    },
    {
      "epoch": 2.539940828402367,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0001146294168366322,
      "loss": 0.0474,
      "step": 10302
    },
    {
      "epoch": 2.5404339250493098,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011459536170056713,
      "loss": 0.0441,
      "step": 10304
    },
    {
      "epoch": 2.5409270216962527,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001145613048348702,
      "loss": 0.0493,
      "step": 10306
    },
    {
      "epoch": 2.5414201183431953,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011452724624357734,
      "loss": 0.0451,
      "step": 10308
    },
    {
      "epoch": 2.5419132149901382,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011449318593072466,
      "loss": 0.0456,
      "step": 10310
    },
    {
      "epoch": 2.5424063116370808,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00011445912390034857,
      "loss": 0.0464,
      "step": 10312
    },
    {
      "epoch": 2.5428994082840237,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011442506015648558,
      "loss": 0.0484,
      "step": 10314
    },
    {
      "epoch": 2.5433925049309662,
      "grad_norm": 0.31640625,
      "learning_rate": 0.0001143909947031724,
      "loss": 0.0448,
      "step": 10316
    },
    {
      "epoch": 2.543885601577909,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011435692754444602,
      "loss": 0.0428,
      "step": 10318
    },
    {
      "epoch": 2.544378698224852,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011432285868434354,
      "loss": 0.0421,
      "step": 10320
    },
    {
      "epoch": 2.5448717948717947,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00011428878812690234,
      "loss": 0.0425,
      "step": 10322
    },
    {
      "epoch": 2.5453648915187377,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00011425471587615997,
      "loss": 0.0472,
      "step": 10324
    },
    {
      "epoch": 2.5458579881656807,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011422064193615418,
      "loss": 0.0404,
      "step": 10326
    },
    {
      "epoch": 2.546351084812623,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00011418656631092295,
      "loss": 0.0486,
      "step": 10328
    },
    {
      "epoch": 2.546844181459566,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011415248900450437,
      "loss": 0.0471,
      "step": 10330
    },
    {
      "epoch": 2.547337278106509,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011411841002093686,
      "loss": 0.0512,
      "step": 10332
    },
    {
      "epoch": 2.5478303747534516,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001140843293642589,
      "loss": 0.0439,
      "step": 10334
    },
    {
      "epoch": 2.5483234714003946,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00011405024703850929,
      "loss": 0.0492,
      "step": 10336
    },
    {
      "epoch": 2.548816568047337,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011401616304772696,
      "loss": 0.046,
      "step": 10338
    },
    {
      "epoch": 2.54930966469428,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00011398207739595104,
      "loss": 0.0436,
      "step": 10340
    },
    {
      "epoch": 2.5498027613412226,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00011394799008722088,
      "loss": 0.0456,
      "step": 10342
    },
    {
      "epoch": 2.5502958579881656,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011391390112557605,
      "loss": 0.0498,
      "step": 10344
    },
    {
      "epoch": 2.5507889546351086,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001138798105150562,
      "loss": 0.0441,
      "step": 10346
    },
    {
      "epoch": 2.551282051282051,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00011384571825970138,
      "loss": 0.042,
      "step": 10348
    },
    {
      "epoch": 2.551775147928994,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001138116243635516,
      "loss": 0.0448,
      "step": 10350
    },
    {
      "epoch": 2.552268244575937,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0001137775288306472,
      "loss": 0.0464,
      "step": 10352
    },
    {
      "epoch": 2.5527613412228796,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011374343166502873,
      "loss": 0.0446,
      "step": 10354
    },
    {
      "epoch": 2.5532544378698225,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011370933287073687,
      "loss": 0.0497,
      "step": 10356
    },
    {
      "epoch": 2.5537475345167655,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00011367523245181256,
      "loss": 0.0425,
      "step": 10358
    },
    {
      "epoch": 2.554240631163708,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001136411304122968,
      "loss": 0.0433,
      "step": 10360
    },
    {
      "epoch": 2.554733727810651,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00011360702675623094,
      "loss": 0.0432,
      "step": 10362
    },
    {
      "epoch": 2.5552268244575935,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00011357292148765646,
      "loss": 0.0468,
      "step": 10364
    },
    {
      "epoch": 2.5557199211045365,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00011353881461061498,
      "loss": 0.0454,
      "step": 10366
    },
    {
      "epoch": 2.556213017751479,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00011350470612914836,
      "loss": 0.0449,
      "step": 10368
    },
    {
      "epoch": 2.556706114398422,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00011347059604729869,
      "loss": 0.0471,
      "step": 10370
    },
    {
      "epoch": 2.557199211045365,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001134364843691082,
      "loss": 0.045,
      "step": 10372
    },
    {
      "epoch": 2.5576923076923075,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011340237109861924,
      "loss": 0.0455,
      "step": 10374
    },
    {
      "epoch": 2.5581854043392505,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011336825623987448,
      "loss": 0.0474,
      "step": 10376
    },
    {
      "epoch": 2.5586785009861934,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00011333413979691674,
      "loss": 0.0438,
      "step": 10378
    },
    {
      "epoch": 2.559171597633136,
      "grad_norm": 0.330078125,
      "learning_rate": 0.000113300021773789,
      "loss": 0.0484,
      "step": 10380
    },
    {
      "epoch": 2.559664694280079,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011326590217453439,
      "loss": 0.0418,
      "step": 10382
    },
    {
      "epoch": 2.560157790927022,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00011323178100319632,
      "loss": 0.0433,
      "step": 10384
    },
    {
      "epoch": 2.5606508875739644,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011319765826381834,
      "loss": 0.0484,
      "step": 10386
    },
    {
      "epoch": 2.5611439842209074,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00011316353396044415,
      "loss": 0.0464,
      "step": 10388
    },
    {
      "epoch": 2.5616370808678504,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0001131294080971177,
      "loss": 0.0428,
      "step": 10390
    },
    {
      "epoch": 2.562130177514793,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011309528067788306,
      "loss": 0.0426,
      "step": 10392
    },
    {
      "epoch": 2.562623274161736,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00011306115170678455,
      "loss": 0.0471,
      "step": 10394
    },
    {
      "epoch": 2.5631163708086784,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00011302702118786667,
      "loss": 0.0485,
      "step": 10396
    },
    {
      "epoch": 2.5636094674556213,
      "grad_norm": 0.375,
      "learning_rate": 0.00011299288912517403,
      "loss": 0.038,
      "step": 10398
    },
    {
      "epoch": 2.564102564102564,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0001129587555227515,
      "loss": 0.0432,
      "step": 10400
    },
    {
      "epoch": 2.564595660749507,
      "grad_norm": 0.359375,
      "learning_rate": 0.00011292462038464407,
      "loss": 0.0438,
      "step": 10402
    },
    {
      "epoch": 2.56508875739645,
      "grad_norm": 0.28125,
      "learning_rate": 0.00011289048371489694,
      "loss": 0.0459,
      "step": 10404
    },
    {
      "epoch": 2.5655818540433923,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011285634551755554,
      "loss": 0.0424,
      "step": 10406
    },
    {
      "epoch": 2.5660749506903353,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011282220579666542,
      "loss": 0.044,
      "step": 10408
    },
    {
      "epoch": 2.5665680473372783,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001127880645562723,
      "loss": 0.044,
      "step": 10410
    },
    {
      "epoch": 2.567061143984221,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00011275392180042216,
      "loss": 0.0446,
      "step": 10412
    },
    {
      "epoch": 2.5675542406311638,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0001127197775331611,
      "loss": 0.0447,
      "step": 10414
    },
    {
      "epoch": 2.5680473372781067,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00011268563175853535,
      "loss": 0.0455,
      "step": 10416
    },
    {
      "epoch": 2.5685404339250493,
      "grad_norm": 0.296875,
      "learning_rate": 0.00011265148448059139,
      "loss": 0.0456,
      "step": 10418
    },
    {
      "epoch": 2.5690335305719922,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00011261733570337591,
      "loss": 0.0429,
      "step": 10420
    },
    {
      "epoch": 2.5695266272189348,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00011258318543093569,
      "loss": 0.0434,
      "step": 10422
    },
    {
      "epoch": 2.5700197238658777,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011254903366731775,
      "loss": 0.0474,
      "step": 10424
    },
    {
      "epoch": 2.5705128205128203,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00011251488041656928,
      "loss": 0.0476,
      "step": 10426
    },
    {
      "epoch": 2.5710059171597632,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011248072568273755,
      "loss": 0.0516,
      "step": 10428
    },
    {
      "epoch": 2.571499013806706,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001124465694698702,
      "loss": 0.0516,
      "step": 10430
    },
    {
      "epoch": 2.5719921104536487,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00011241241178201485,
      "loss": 0.0433,
      "step": 10432
    },
    {
      "epoch": 2.5724852071005917,
      "grad_norm": 0.390625,
      "learning_rate": 0.0001123782526232194,
      "loss": 0.0452,
      "step": 10434
    },
    {
      "epoch": 2.5729783037475347,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011234409199753189,
      "loss": 0.0454,
      "step": 10436
    },
    {
      "epoch": 2.573471400394477,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00011230992990900055,
      "loss": 0.0473,
      "step": 10438
    },
    {
      "epoch": 2.57396449704142,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011227576636167385,
      "loss": 0.0447,
      "step": 10440
    },
    {
      "epoch": 2.574457593688363,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011224160135960024,
      "loss": 0.047,
      "step": 10442
    },
    {
      "epoch": 2.5749506903353057,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00011220743490682852,
      "loss": 0.0473,
      "step": 10444
    },
    {
      "epoch": 2.5754437869822486,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00011217326700740765,
      "loss": 0.0419,
      "step": 10446
    },
    {
      "epoch": 2.5759368836291916,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011213909766538663,
      "loss": 0.0458,
      "step": 10448
    },
    {
      "epoch": 2.576429980276134,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00011210492688481477,
      "loss": 0.0465,
      "step": 10450
    },
    {
      "epoch": 2.5769230769230766,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011207075466974148,
      "loss": 0.0488,
      "step": 10452
    },
    {
      "epoch": 2.5774161735700196,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00011203658102421642,
      "loss": 0.0419,
      "step": 10454
    },
    {
      "epoch": 2.5779092702169626,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00011200240595228923,
      "loss": 0.0486,
      "step": 10456
    },
    {
      "epoch": 2.578402366863905,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011196822945800995,
      "loss": 0.0503,
      "step": 10458
    },
    {
      "epoch": 2.578895463510848,
      "grad_norm": 0.296875,
      "learning_rate": 0.00011193405154542867,
      "loss": 0.0483,
      "step": 10460
    },
    {
      "epoch": 2.579388560157791,
      "grad_norm": 0.34375,
      "learning_rate": 0.00011189987221859566,
      "loss": 0.0494,
      "step": 10462
    },
    {
      "epoch": 2.5798816568047336,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011186569148156134,
      "loss": 0.0434,
      "step": 10464
    },
    {
      "epoch": 2.5803747534516766,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011183150933837632,
      "loss": 0.0482,
      "step": 10466
    },
    {
      "epoch": 2.5808678500986195,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011179732579309141,
      "loss": 0.0448,
      "step": 10468
    },
    {
      "epoch": 2.581360946745562,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001117631408497575,
      "loss": 0.0478,
      "step": 10470
    },
    {
      "epoch": 2.581854043392505,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011172895451242571,
      "loss": 0.0431,
      "step": 10472
    },
    {
      "epoch": 2.582347140039448,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011169476678514734,
      "loss": 0.0456,
      "step": 10474
    },
    {
      "epoch": 2.5828402366863905,
      "grad_norm": 0.296875,
      "learning_rate": 0.00011166057767197383,
      "loss": 0.0438,
      "step": 10476
    },
    {
      "epoch": 2.5833333333333335,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011162638717695672,
      "loss": 0.0416,
      "step": 10478
    },
    {
      "epoch": 2.583826429980276,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00011159219530414787,
      "loss": 0.0427,
      "step": 10480
    },
    {
      "epoch": 2.584319526627219,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00011155800205759915,
      "loss": 0.047,
      "step": 10482
    },
    {
      "epoch": 2.5848126232741615,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011152380744136261,
      "loss": 0.0494,
      "step": 10484
    },
    {
      "epoch": 2.5853057199211045,
      "grad_norm": 0.359375,
      "learning_rate": 0.00011148961145949055,
      "loss": 0.0406,
      "step": 10486
    },
    {
      "epoch": 2.5857988165680474,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00011145541411603539,
      "loss": 0.0459,
      "step": 10488
    },
    {
      "epoch": 2.58629191321499,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001114212154150497,
      "loss": 0.0511,
      "step": 10490
    },
    {
      "epoch": 2.586785009861933,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00011138701536058624,
      "loss": 0.05,
      "step": 10492
    },
    {
      "epoch": 2.587278106508876,
      "grad_norm": 0.34375,
      "learning_rate": 0.00011135281395669787,
      "loss": 0.0422,
      "step": 10494
    },
    {
      "epoch": 2.5877712031558184,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00011131861120743767,
      "loss": 0.0459,
      "step": 10496
    },
    {
      "epoch": 2.5882642998027614,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011128440711685883,
      "loss": 0.047,
      "step": 10498
    },
    {
      "epoch": 2.5887573964497044,
      "grad_norm": 0.328125,
      "learning_rate": 0.00011125020168901474,
      "loss": 0.0416,
      "step": 10500
    },
    {
      "epoch": 2.589250493096647,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011121599492795896,
      "loss": 0.0494,
      "step": 10502
    },
    {
      "epoch": 2.58974358974359,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00011118178683774511,
      "loss": 0.0471,
      "step": 10504
    },
    {
      "epoch": 2.5902366863905324,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011114757742242712,
      "loss": 0.0475,
      "step": 10506
    },
    {
      "epoch": 2.5907297830374754,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00011111336668605899,
      "loss": 0.0441,
      "step": 10508
    },
    {
      "epoch": 2.591222879684418,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00011107915463269483,
      "loss": 0.04,
      "step": 10510
    },
    {
      "epoch": 2.591715976331361,
      "grad_norm": 0.3125,
      "learning_rate": 0.000111044941266389,
      "loss": 0.0448,
      "step": 10512
    },
    {
      "epoch": 2.592209072978304,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00011101072659119595,
      "loss": 0.0478,
      "step": 10514
    },
    {
      "epoch": 2.5927021696252464,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00011097651061117033,
      "loss": 0.0442,
      "step": 10516
    },
    {
      "epoch": 2.5931952662721893,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001109422933303669,
      "loss": 0.0466,
      "step": 10518
    },
    {
      "epoch": 2.5936883629191323,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011090807475284064,
      "loss": 0.0482,
      "step": 10520
    },
    {
      "epoch": 2.594181459566075,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00011087385488264663,
      "loss": 0.0461,
      "step": 10522
    },
    {
      "epoch": 2.594674556213018,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011083963372384007,
      "loss": 0.0435,
      "step": 10524
    },
    {
      "epoch": 2.5951676528599608,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00011080541128047641,
      "loss": 0.0484,
      "step": 10526
    },
    {
      "epoch": 2.5956607495069033,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001107711875566112,
      "loss": 0.0478,
      "step": 10528
    },
    {
      "epoch": 2.5961538461538463,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0001107369625563001,
      "loss": 0.044,
      "step": 10530
    },
    {
      "epoch": 2.5966469428007892,
      "grad_norm": 0.318359375,
      "learning_rate": 0.000110702736283599,
      "loss": 0.048,
      "step": 10532
    },
    {
      "epoch": 2.5971400394477318,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00011066850874256387,
      "loss": 0.0447,
      "step": 10534
    },
    {
      "epoch": 2.5976331360946747,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00011063427993725092,
      "loss": 0.0454,
      "step": 10536
    },
    {
      "epoch": 2.5981262327416172,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00011060004987171642,
      "loss": 0.0408,
      "step": 10538
    },
    {
      "epoch": 2.59861932938856,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011056581855001683,
      "loss": 0.0433,
      "step": 10540
    },
    {
      "epoch": 2.5991124260355027,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00011053158597620872,
      "loss": 0.0435,
      "step": 10542
    },
    {
      "epoch": 2.5996055226824457,
      "grad_norm": 0.375,
      "learning_rate": 0.00011049735215434889,
      "loss": 0.0417,
      "step": 10544
    },
    {
      "epoch": 2.6000986193293887,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00011046311708849424,
      "loss": 0.0456,
      "step": 10546
    },
    {
      "epoch": 2.600591715976331,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00011042888078270177,
      "loss": 0.0458,
      "step": 10548
    },
    {
      "epoch": 2.601084812623274,
      "grad_norm": 0.3125,
      "learning_rate": 0.00011039464324102873,
      "loss": 0.0461,
      "step": 10550
    },
    {
      "epoch": 2.601577909270217,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011036040446753239,
      "loss": 0.0451,
      "step": 10552
    },
    {
      "epoch": 2.6020710059171597,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00011032616446627029,
      "loss": 0.0477,
      "step": 10554
    },
    {
      "epoch": 2.6025641025641026,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011029192324130004,
      "loss": 0.0466,
      "step": 10556
    },
    {
      "epoch": 2.6030571992110456,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00011025768079667942,
      "loss": 0.05,
      "step": 10558
    },
    {
      "epoch": 2.603550295857988,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00011022343713646637,
      "loss": 0.0457,
      "step": 10560
    },
    {
      "epoch": 2.604043392504931,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00011018919226471896,
      "loss": 0.0449,
      "step": 10562
    },
    {
      "epoch": 2.6045364891518736,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011015494618549533,
      "loss": 0.0438,
      "step": 10564
    },
    {
      "epoch": 2.6050295857988166,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011012069890285392,
      "loss": 0.0459,
      "step": 10566
    },
    {
      "epoch": 2.605522682445759,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011008645042085317,
      "loss": 0.0459,
      "step": 10568
    },
    {
      "epoch": 2.606015779092702,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011005220074355171,
      "loss": 0.0482,
      "step": 10570
    },
    {
      "epoch": 2.606508875739645,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00011001794987500835,
      "loss": 0.0443,
      "step": 10572
    },
    {
      "epoch": 2.6070019723865876,
      "grad_norm": 0.3046875,
      "learning_rate": 0.000109983697819282,
      "loss": 0.0433,
      "step": 10574
    },
    {
      "epoch": 2.6074950690335306,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010994944458043173,
      "loss": 0.0443,
      "step": 10576
    },
    {
      "epoch": 2.6079881656804735,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001099151901625167,
      "loss": 0.0444,
      "step": 10578
    },
    {
      "epoch": 2.608481262327416,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010988093456959631,
      "loss": 0.0454,
      "step": 10580
    },
    {
      "epoch": 2.608974358974359,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00010984667780572999,
      "loss": 0.0424,
      "step": 10582
    },
    {
      "epoch": 2.609467455621302,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00010981241987497734,
      "loss": 0.0463,
      "step": 10584
    },
    {
      "epoch": 2.6099605522682445,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010977816078139817,
      "loss": 0.0461,
      "step": 10586
    },
    {
      "epoch": 2.6104536489151875,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010974390052905237,
      "loss": 0.0486,
      "step": 10588
    },
    {
      "epoch": 2.61094674556213,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00010970963912199997,
      "loss": 0.0398,
      "step": 10590
    },
    {
      "epoch": 2.611439842209073,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00010967537656430109,
      "loss": 0.0438,
      "step": 10592
    },
    {
      "epoch": 2.6119329388560155,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00010964111286001608,
      "loss": 0.0458,
      "step": 10594
    },
    {
      "epoch": 2.6124260355029585,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010960684801320536,
      "loss": 0.0474,
      "step": 10596
    },
    {
      "epoch": 2.6129191321499015,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001095725820279295,
      "loss": 0.0441,
      "step": 10598
    },
    {
      "epoch": 2.613412228796844,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010953831490824924,
      "loss": 0.0451,
      "step": 10600
    },
    {
      "epoch": 2.613905325443787,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0001095040466582254,
      "loss": 0.0421,
      "step": 10602
    },
    {
      "epoch": 2.61439842209073,
      "grad_norm": 0.421875,
      "learning_rate": 0.00010946977728191896,
      "loss": 0.0473,
      "step": 10604
    },
    {
      "epoch": 2.6148915187376724,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00010943550678339108,
      "loss": 0.0441,
      "step": 10606
    },
    {
      "epoch": 2.6153846153846154,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010940123516670293,
      "loss": 0.0466,
      "step": 10608
    },
    {
      "epoch": 2.6158777120315584,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00010936696243591591,
      "loss": 0.0418,
      "step": 10610
    },
    {
      "epoch": 2.616370808678501,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010933268859509158,
      "loss": 0.0467,
      "step": 10612
    },
    {
      "epoch": 2.616863905325444,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010929841364829152,
      "loss": 0.0422,
      "step": 10614
    },
    {
      "epoch": 2.617357001972387,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00010926413759957754,
      "loss": 0.0422,
      "step": 10616
    },
    {
      "epoch": 2.6178500986193294,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00010922986045301152,
      "loss": 0.0442,
      "step": 10618
    },
    {
      "epoch": 2.6183431952662723,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010919558221265552,
      "loss": 0.0475,
      "step": 10620
    },
    {
      "epoch": 2.618836291913215,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010916130288257166,
      "loss": 0.0441,
      "step": 10622
    },
    {
      "epoch": 2.619329388560158,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010912702246682227,
      "loss": 0.0393,
      "step": 10624
    },
    {
      "epoch": 2.6198224852071004,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00010909274096946976,
      "loss": 0.0425,
      "step": 10626
    },
    {
      "epoch": 2.6203155818540433,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010905845839457669,
      "loss": 0.0436,
      "step": 10628
    },
    {
      "epoch": 2.6208086785009863,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001090241747462057,
      "loss": 0.0505,
      "step": 10630
    },
    {
      "epoch": 2.621301775147929,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010898989002841964,
      "loss": 0.043,
      "step": 10632
    },
    {
      "epoch": 2.621794871794872,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00010895560424528142,
      "loss": 0.0503,
      "step": 10634
    },
    {
      "epoch": 2.6222879684418148,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010892131740085408,
      "loss": 0.045,
      "step": 10636
    },
    {
      "epoch": 2.6227810650887573,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010888702949920081,
      "loss": 0.0448,
      "step": 10638
    },
    {
      "epoch": 2.6232741617357003,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010885274054438495,
      "loss": 0.0381,
      "step": 10640
    },
    {
      "epoch": 2.6237672583826432,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010881845054046988,
      "loss": 0.0436,
      "step": 10642
    },
    {
      "epoch": 2.6242603550295858,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00010878415949151922,
      "loss": 0.0457,
      "step": 10644
    },
    {
      "epoch": 2.6247534516765287,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00010874986740159658,
      "loss": 0.0418,
      "step": 10646
    },
    {
      "epoch": 2.6252465483234713,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010871557427476583,
      "loss": 0.0396,
      "step": 10648
    },
    {
      "epoch": 2.6257396449704142,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00010868128011509087,
      "loss": 0.0463,
      "step": 10650
    },
    {
      "epoch": 2.6262327416173568,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00010864698492663572,
      "loss": 0.043,
      "step": 10652
    },
    {
      "epoch": 2.6267258382642997,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00010861268871346458,
      "loss": 0.0458,
      "step": 10654
    },
    {
      "epoch": 2.6272189349112427,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00010857839147964173,
      "loss": 0.0448,
      "step": 10656
    },
    {
      "epoch": 2.6277120315581852,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010854409322923161,
      "loss": 0.0427,
      "step": 10658
    },
    {
      "epoch": 2.628205128205128,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00010850979396629877,
      "loss": 0.0446,
      "step": 10660
    },
    {
      "epoch": 2.628698224852071,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010847549369490781,
      "loss": 0.0486,
      "step": 10662
    },
    {
      "epoch": 2.6291913214990137,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010844119241912352,
      "loss": 0.0482,
      "step": 10664
    },
    {
      "epoch": 2.6296844181459567,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010840689014301078,
      "loss": 0.046,
      "step": 10666
    },
    {
      "epoch": 2.6301775147928996,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00010837258687063463,
      "loss": 0.044,
      "step": 10668
    },
    {
      "epoch": 2.630670611439842,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00010833828260606021,
      "loss": 0.0418,
      "step": 10670
    },
    {
      "epoch": 2.631163708086785,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010830397735335272,
      "loss": 0.0432,
      "step": 10672
    },
    {
      "epoch": 2.6316568047337277,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010826967111657757,
      "loss": 0.0459,
      "step": 10674
    },
    {
      "epoch": 2.6321499013806706,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0001082353638998002,
      "loss": 0.0445,
      "step": 10676
    },
    {
      "epoch": 2.632642998027613,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00010820105570708627,
      "loss": 0.0455,
      "step": 10678
    },
    {
      "epoch": 2.633136094674556,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010816674654250141,
      "loss": 0.0461,
      "step": 10680
    },
    {
      "epoch": 2.633629191321499,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001081324364101115,
      "loss": 0.0395,
      "step": 10682
    },
    {
      "epoch": 2.6341222879684416,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00010809812531398247,
      "loss": 0.0442,
      "step": 10684
    },
    {
      "epoch": 2.6346153846153846,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00010806381325818036,
      "loss": 0.0463,
      "step": 10686
    },
    {
      "epoch": 2.6351084812623276,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00010802950024677138,
      "loss": 0.043,
      "step": 10688
    },
    {
      "epoch": 2.63560157790927,
      "grad_norm": 0.37109375,
      "learning_rate": 0.0001079951862838218,
      "loss": 0.0443,
      "step": 10690
    },
    {
      "epoch": 2.636094674556213,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00010796087137339796,
      "loss": 0.046,
      "step": 10692
    },
    {
      "epoch": 2.636587771203156,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010792655551956645,
      "loss": 0.0409,
      "step": 10694
    },
    {
      "epoch": 2.6370808678500985,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010789223872639384,
      "loss": 0.0451,
      "step": 10696
    },
    {
      "epoch": 2.6375739644970415,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00010785792099794689,
      "loss": 0.0481,
      "step": 10698
    },
    {
      "epoch": 2.6380670611439845,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001078236023382924,
      "loss": 0.0449,
      "step": 10700
    },
    {
      "epoch": 2.638560157790927,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0001077892827514974,
      "loss": 0.0455,
      "step": 10702
    },
    {
      "epoch": 2.63905325443787,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00010775496224162886,
      "loss": 0.0469,
      "step": 10704
    },
    {
      "epoch": 2.6395463510848125,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010772064081275402,
      "loss": 0.0437,
      "step": 10706
    },
    {
      "epoch": 2.6400394477317555,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00010768631846894013,
      "loss": 0.0403,
      "step": 10708
    },
    {
      "epoch": 2.640532544378698,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001076519952142546,
      "loss": 0.0448,
      "step": 10710
    },
    {
      "epoch": 2.641025641025641,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0001076176710527649,
      "loss": 0.0393,
      "step": 10712
    },
    {
      "epoch": 2.641518737672584,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010758334598853863,
      "loss": 0.0458,
      "step": 10714
    },
    {
      "epoch": 2.6420118343195265,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010754902002564355,
      "loss": 0.0454,
      "step": 10716
    },
    {
      "epoch": 2.6425049309664694,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00010751469316814743,
      "loss": 0.0429,
      "step": 10718
    },
    {
      "epoch": 2.6429980276134124,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010748036542011823,
      "loss": 0.0445,
      "step": 10720
    },
    {
      "epoch": 2.643491124260355,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00010744603678562395,
      "loss": 0.0412,
      "step": 10722
    },
    {
      "epoch": 2.643984220907298,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010741170726873274,
      "loss": 0.0439,
      "step": 10724
    },
    {
      "epoch": 2.644477317554241,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00010737737687351284,
      "loss": 0.044,
      "step": 10726
    },
    {
      "epoch": 2.6449704142011834,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00010734304560403259,
      "loss": 0.043,
      "step": 10728
    },
    {
      "epoch": 2.6454635108481264,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00010730871346436045,
      "loss": 0.0391,
      "step": 10730
    },
    {
      "epoch": 2.645956607495069,
      "grad_norm": 0.421875,
      "learning_rate": 0.00010727438045856492,
      "loss": 0.0405,
      "step": 10732
    },
    {
      "epoch": 2.646449704142012,
      "grad_norm": 0.26953125,
      "learning_rate": 0.00010724004659071472,
      "loss": 0.0426,
      "step": 10734
    },
    {
      "epoch": 2.6469428007889544,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010720571186487858,
      "loss": 0.0471,
      "step": 10736
    },
    {
      "epoch": 2.6474358974358974,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00010717137628512533,
      "loss": 0.0429,
      "step": 10738
    },
    {
      "epoch": 2.6479289940828403,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00010713703985552395,
      "loss": 0.0437,
      "step": 10740
    },
    {
      "epoch": 2.648422090729783,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00010710270258014354,
      "loss": 0.044,
      "step": 10742
    },
    {
      "epoch": 2.648915187376726,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001070683644630532,
      "loss": 0.0381,
      "step": 10744
    },
    {
      "epoch": 2.649408284023669,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001070340255083222,
      "loss": 0.0449,
      "step": 10746
    },
    {
      "epoch": 2.6499013806706113,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00010699968572001988,
      "loss": 0.0469,
      "step": 10748
    },
    {
      "epoch": 2.6503944773175543,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00010696534510221574,
      "loss": 0.0414,
      "step": 10750
    },
    {
      "epoch": 2.6508875739644973,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010693100365897933,
      "loss": 0.0423,
      "step": 10752
    },
    {
      "epoch": 2.65138067061144,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010689666139438026,
      "loss": 0.0472,
      "step": 10754
    },
    {
      "epoch": 2.6518737672583828,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010686231831248834,
      "loss": 0.0443,
      "step": 10756
    },
    {
      "epoch": 2.6523668639053253,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00010682797441737335,
      "loss": 0.0463,
      "step": 10758
    },
    {
      "epoch": 2.6528599605522682,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00010679362971310531,
      "loss": 0.0512,
      "step": 10760
    },
    {
      "epoch": 2.6533530571992108,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00010675928420375421,
      "loss": 0.0452,
      "step": 10762
    },
    {
      "epoch": 2.6538461538461537,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00010672493789339015,
      "loss": 0.0452,
      "step": 10764
    },
    {
      "epoch": 2.6543392504930967,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010669059078608344,
      "loss": 0.043,
      "step": 10766
    },
    {
      "epoch": 2.6548323471400392,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010665624288590437,
      "loss": 0.0463,
      "step": 10768
    },
    {
      "epoch": 2.655325443786982,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010662189419692334,
      "loss": 0.0455,
      "step": 10770
    },
    {
      "epoch": 2.655818540433925,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010658754472321089,
      "loss": 0.0447,
      "step": 10772
    },
    {
      "epoch": 2.6563116370808677,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00010655319446883758,
      "loss": 0.0435,
      "step": 10774
    },
    {
      "epoch": 2.6568047337278107,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010651884343787418,
      "loss": 0.0479,
      "step": 10776
    },
    {
      "epoch": 2.6572978303747536,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0001064844916343914,
      "loss": 0.0438,
      "step": 10778
    },
    {
      "epoch": 2.657790927021696,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010645013906246017,
      "loss": 0.0427,
      "step": 10780
    },
    {
      "epoch": 2.658284023668639,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00010641578572615145,
      "loss": 0.0419,
      "step": 10782
    },
    {
      "epoch": 2.658777120315582,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010638143162953631,
      "loss": 0.0404,
      "step": 10784
    },
    {
      "epoch": 2.6592702169625246,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010634707677668587,
      "loss": 0.0437,
      "step": 10786
    },
    {
      "epoch": 2.6597633136094676,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010631272117167139,
      "loss": 0.0438,
      "step": 10788
    },
    {
      "epoch": 2.66025641025641,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010627836481856421,
      "loss": 0.0406,
      "step": 10790
    },
    {
      "epoch": 2.660749506903353,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010624400772143579,
      "loss": 0.0444,
      "step": 10792
    },
    {
      "epoch": 2.6612426035502956,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010620964988435755,
      "loss": 0.0432,
      "step": 10794
    },
    {
      "epoch": 2.6617357001972386,
      "grad_norm": 0.375,
      "learning_rate": 0.00010617529131140114,
      "loss": 0.0399,
      "step": 10796
    },
    {
      "epoch": 2.6622287968441816,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00010614093200663824,
      "loss": 0.0436,
      "step": 10798
    },
    {
      "epoch": 2.662721893491124,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010610657197414058,
      "loss": 0.0431,
      "step": 10800
    },
    {
      "epoch": 2.663214990138067,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00010607221121798006,
      "loss": 0.0452,
      "step": 10802
    },
    {
      "epoch": 2.66370808678501,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010603784974222861,
      "loss": 0.0486,
      "step": 10804
    },
    {
      "epoch": 2.6642011834319526,
      "grad_norm": 0.375,
      "learning_rate": 0.00010600348755095826,
      "loss": 0.047,
      "step": 10806
    },
    {
      "epoch": 2.6646942800788955,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001059691246482411,
      "loss": 0.0462,
      "step": 10808
    },
    {
      "epoch": 2.6651873767258385,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010593476103814936,
      "loss": 0.0476,
      "step": 10810
    },
    {
      "epoch": 2.665680473372781,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010590039672475534,
      "loss": 0.0469,
      "step": 10812
    },
    {
      "epoch": 2.666173570019724,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001058660317121313,
      "loss": 0.0552,
      "step": 10814
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010583166600434977,
      "loss": 0.0422,
      "step": 10816
    },
    {
      "epoch": 2.6671597633136095,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010579729960548326,
      "loss": 0.0463,
      "step": 10818
    },
    {
      "epoch": 2.667652859960552,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00010576293251960436,
      "loss": 0.0412,
      "step": 10820
    },
    {
      "epoch": 2.668145956607495,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00010572856475078578,
      "loss": 0.0476,
      "step": 10822
    },
    {
      "epoch": 2.668639053254438,
      "grad_norm": 0.296875,
      "learning_rate": 0.00010569419630310036,
      "loss": 0.0403,
      "step": 10824
    },
    {
      "epoch": 2.6691321499013805,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00010565982718062085,
      "loss": 0.0426,
      "step": 10826
    },
    {
      "epoch": 2.6696252465483234,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010562545738742021,
      "loss": 0.0422,
      "step": 10828
    },
    {
      "epoch": 2.6701183431952664,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00010559108692757142,
      "loss": 0.0432,
      "step": 10830
    },
    {
      "epoch": 2.670611439842209,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00010555671580514765,
      "loss": 0.0402,
      "step": 10832
    },
    {
      "epoch": 2.671104536489152,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00010552234402422203,
      "loss": 0.0494,
      "step": 10834
    },
    {
      "epoch": 2.671597633136095,
      "grad_norm": 0.3125,
      "learning_rate": 0.0001054879715888678,
      "loss": 0.0479,
      "step": 10836
    },
    {
      "epoch": 2.6720907297830374,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00010545359850315835,
      "loss": 0.0425,
      "step": 10838
    },
    {
      "epoch": 2.6725838264299804,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010541922477116697,
      "loss": 0.0466,
      "step": 10840
    },
    {
      "epoch": 2.6730769230769234,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00010538485039696722,
      "loss": 0.0446,
      "step": 10842
    },
    {
      "epoch": 2.673570019723866,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010535047538463263,
      "loss": 0.0416,
      "step": 10844
    },
    {
      "epoch": 2.6740631163708084,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00010531609973823681,
      "loss": 0.0471,
      "step": 10846
    },
    {
      "epoch": 2.6745562130177514,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00010528172346185352,
      "loss": 0.0471,
      "step": 10848
    },
    {
      "epoch": 2.6750493096646943,
      "grad_norm": 0.390625,
      "learning_rate": 0.0001052473465595565,
      "loss": 0.0464,
      "step": 10850
    },
    {
      "epoch": 2.675542406311637,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010521296903541962,
      "loss": 0.0416,
      "step": 10852
    },
    {
      "epoch": 2.67603550295858,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001051785908935168,
      "loss": 0.0442,
      "step": 10854
    },
    {
      "epoch": 2.676528599605523,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010514421213792205,
      "loss": 0.0391,
      "step": 10856
    },
    {
      "epoch": 2.6770216962524653,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010510983277270945,
      "loss": 0.0443,
      "step": 10858
    },
    {
      "epoch": 2.6775147928994083,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001050754528019531,
      "loss": 0.0446,
      "step": 10860
    },
    {
      "epoch": 2.6780078895463513,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00010504107222972727,
      "loss": 0.0452,
      "step": 10862
    },
    {
      "epoch": 2.678500986193294,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010500669106010625,
      "loss": 0.0456,
      "step": 10864
    },
    {
      "epoch": 2.6789940828402368,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00010497230929716439,
      "loss": 0.0453,
      "step": 10866
    },
    {
      "epoch": 2.6794871794871797,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001049379269449761,
      "loss": 0.0421,
      "step": 10868
    },
    {
      "epoch": 2.6799802761341223,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001049035440076159,
      "loss": 0.0465,
      "step": 10870
    },
    {
      "epoch": 2.6804733727810652,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00010486916048915836,
      "loss": 0.0515,
      "step": 10872
    },
    {
      "epoch": 2.6809664694280078,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00010483477639367811,
      "loss": 0.0409,
      "step": 10874
    },
    {
      "epoch": 2.6814595660749507,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00010480039172524987,
      "loss": 0.0468,
      "step": 10876
    },
    {
      "epoch": 2.6819526627218933,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00010476600648794841,
      "loss": 0.0425,
      "step": 10878
    },
    {
      "epoch": 2.6824457593688362,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010473162068584857,
      "loss": 0.0447,
      "step": 10880
    },
    {
      "epoch": 2.682938856015779,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010469723432302527,
      "loss": 0.043,
      "step": 10882
    },
    {
      "epoch": 2.6834319526627217,
      "grad_norm": 0.375,
      "learning_rate": 0.00010466284740355345,
      "loss": 0.0426,
      "step": 10884
    },
    {
      "epoch": 2.6839250493096647,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0001046284599315082,
      "loss": 0.0391,
      "step": 10886
    },
    {
      "epoch": 2.6844181459566077,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00010459407191096459,
      "loss": 0.0449,
      "step": 10888
    },
    {
      "epoch": 2.68491124260355,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00010455968334599782,
      "loss": 0.0401,
      "step": 10890
    },
    {
      "epoch": 2.685404339250493,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00010452529424068314,
      "loss": 0.0429,
      "step": 10892
    },
    {
      "epoch": 2.685897435897436,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010449090459909578,
      "loss": 0.0432,
      "step": 10894
    },
    {
      "epoch": 2.6863905325443787,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00010445651442531119,
      "loss": 0.0377,
      "step": 10896
    },
    {
      "epoch": 2.6868836291913216,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00010442212372340474,
      "loss": 0.0421,
      "step": 10898
    },
    {
      "epoch": 2.687376725838264,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00010438773249745192,
      "loss": 0.0433,
      "step": 10900
    },
    {
      "epoch": 2.687869822485207,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010435334075152832,
      "loss": 0.0424,
      "step": 10902
    },
    {
      "epoch": 2.6883629191321496,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00010431894848970953,
      "loss": 0.0417,
      "step": 10904
    },
    {
      "epoch": 2.6888560157790926,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00010428455571607123,
      "loss": 0.0432,
      "step": 10906
    },
    {
      "epoch": 2.6893491124260356,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0001042501624346892,
      "loss": 0.0395,
      "step": 10908
    },
    {
      "epoch": 2.689842209072978,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00010421576864963915,
      "loss": 0.0455,
      "step": 10910
    },
    {
      "epoch": 2.690335305719921,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00010418137436499698,
      "loss": 0.0364,
      "step": 10912
    },
    {
      "epoch": 2.690828402366864,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001041469795848386,
      "loss": 0.0398,
      "step": 10914
    },
    {
      "epoch": 2.6913214990138066,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00010411258431323999,
      "loss": 0.0403,
      "step": 10916
    },
    {
      "epoch": 2.6918145956607495,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00010407818855427716,
      "loss": 0.0414,
      "step": 10918
    },
    {
      "epoch": 2.6923076923076925,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010404379231202623,
      "loss": 0.0415,
      "step": 10920
    },
    {
      "epoch": 2.692800788954635,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010400939559056336,
      "loss": 0.0389,
      "step": 10922
    },
    {
      "epoch": 2.693293885601578,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001039749983939647,
      "loss": 0.048,
      "step": 10924
    },
    {
      "epoch": 2.693786982248521,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010394060072630653,
      "loss": 0.0437,
      "step": 10926
    },
    {
      "epoch": 2.6942800788954635,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001039062025916652,
      "loss": 0.0427,
      "step": 10928
    },
    {
      "epoch": 2.6947731755424065,
      "grad_norm": 0.33984375,
      "learning_rate": 0.000103871803994117,
      "loss": 0.0474,
      "step": 10930
    },
    {
      "epoch": 2.695266272189349,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00010383740493773845,
      "loss": 0.0442,
      "step": 10932
    },
    {
      "epoch": 2.695759368836292,
      "grad_norm": 0.375,
      "learning_rate": 0.00010380300542660596,
      "loss": 0.0448,
      "step": 10934
    },
    {
      "epoch": 2.6962524654832345,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010376860546479614,
      "loss": 0.0438,
      "step": 10936
    },
    {
      "epoch": 2.6967455621301775,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010373420505638552,
      "loss": 0.0462,
      "step": 10938
    },
    {
      "epoch": 2.6972386587771204,
      "grad_norm": 0.40625,
      "learning_rate": 0.00010369980420545074,
      "loss": 0.0428,
      "step": 10940
    },
    {
      "epoch": 2.697731755424063,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001036654029160685,
      "loss": 0.0421,
      "step": 10942
    },
    {
      "epoch": 2.698224852071006,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010363100119231554,
      "loss": 0.0441,
      "step": 10944
    },
    {
      "epoch": 2.698717948717949,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00010359659903826869,
      "loss": 0.043,
      "step": 10946
    },
    {
      "epoch": 2.6992110453648914,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010356219645800475,
      "loss": 0.0422,
      "step": 10948
    },
    {
      "epoch": 2.6997041420118344,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00010352779345560069,
      "loss": 0.0438,
      "step": 10950
    },
    {
      "epoch": 2.7001972386587774,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010349339003513336,
      "loss": 0.0441,
      "step": 10952
    },
    {
      "epoch": 2.70069033530572,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010345898620067979,
      "loss": 0.0483,
      "step": 10954
    },
    {
      "epoch": 2.701183431952663,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010342458195631708,
      "loss": 0.046,
      "step": 10956
    },
    {
      "epoch": 2.7016765285996054,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00010339017730612231,
      "loss": 0.0423,
      "step": 10958
    },
    {
      "epoch": 2.7021696252465484,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00010335577225417255,
      "loss": 0.0418,
      "step": 10960
    },
    {
      "epoch": 2.702662721893491,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00010332136680454505,
      "loss": 0.0436,
      "step": 10962
    },
    {
      "epoch": 2.703155818540434,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010328696096131707,
      "loss": 0.042,
      "step": 10964
    },
    {
      "epoch": 2.703648915187377,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00010325255472856585,
      "loss": 0.0418,
      "step": 10966
    },
    {
      "epoch": 2.7041420118343193,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010321814811036872,
      "loss": 0.0469,
      "step": 10968
    },
    {
      "epoch": 2.7046351084812623,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010318374111080306,
      "loss": 0.0435,
      "step": 10970
    },
    {
      "epoch": 2.7051282051282053,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010314933373394633,
      "loss": 0.0399,
      "step": 10972
    },
    {
      "epoch": 2.705621301775148,
      "grad_norm": 0.296875,
      "learning_rate": 0.00010311492598387598,
      "loss": 0.0379,
      "step": 10974
    },
    {
      "epoch": 2.706114398422091,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001030805178646695,
      "loss": 0.039,
      "step": 10976
    },
    {
      "epoch": 2.7066074950690338,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00010304610938040446,
      "loss": 0.0429,
      "step": 10978
    },
    {
      "epoch": 2.7071005917159763,
      "grad_norm": 0.28125,
      "learning_rate": 0.00010301170053515843,
      "loss": 0.0414,
      "step": 10980
    },
    {
      "epoch": 2.7075936883629192,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00010297729133300907,
      "loss": 0.0389,
      "step": 10982
    },
    {
      "epoch": 2.7080867850098618,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010294288177803406,
      "loss": 0.0391,
      "step": 10984
    },
    {
      "epoch": 2.7085798816568047,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00010290847187431113,
      "loss": 0.0451,
      "step": 10986
    },
    {
      "epoch": 2.7090729783037473,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00010287406162591806,
      "loss": 0.0491,
      "step": 10988
    },
    {
      "epoch": 2.7095660749506902,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00010283965103693265,
      "loss": 0.0439,
      "step": 10990
    },
    {
      "epoch": 2.710059171597633,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00010280524011143273,
      "loss": 0.0437,
      "step": 10992
    },
    {
      "epoch": 2.7105522682445757,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010277082885349621,
      "loss": 0.0436,
      "step": 10994
    },
    {
      "epoch": 2.7110453648915187,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00010273641726720096,
      "loss": 0.0475,
      "step": 10996
    },
    {
      "epoch": 2.7115384615384617,
      "grad_norm": 0.33203125,
      "learning_rate": 0.000102702005356625,
      "loss": 0.0412,
      "step": 10998
    },
    {
      "epoch": 2.712031558185404,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001026675931258463,
      "loss": 0.0416,
      "step": 11000
    },
    {
      "epoch": 2.712524654832347,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00010263318057894296,
      "loss": 0.0394,
      "step": 11002
    },
    {
      "epoch": 2.71301775147929,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010259876771999302,
      "loss": 0.045,
      "step": 11004
    },
    {
      "epoch": 2.7135108481262327,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00010256435455307457,
      "loss": 0.0444,
      "step": 11006
    },
    {
      "epoch": 2.7140039447731756,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001025299410822658,
      "loss": 0.0463,
      "step": 11008
    },
    {
      "epoch": 2.7144970414201186,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00010249552731164488,
      "loss": 0.0455,
      "step": 11010
    },
    {
      "epoch": 2.714990138067061,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00010246111324529004,
      "loss": 0.0454,
      "step": 11012
    },
    {
      "epoch": 2.715483234714004,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010242669888727954,
      "loss": 0.0442,
      "step": 11014
    },
    {
      "epoch": 2.7159763313609466,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010239228424169167,
      "loss": 0.0438,
      "step": 11016
    },
    {
      "epoch": 2.7164694280078896,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010235786931260478,
      "loss": 0.0424,
      "step": 11018
    },
    {
      "epoch": 2.716962524654832,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00010232345410409718,
      "loss": 0.0352,
      "step": 11020
    },
    {
      "epoch": 2.717455621301775,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00010228903862024732,
      "loss": 0.0434,
      "step": 11022
    },
    {
      "epoch": 2.717948717948718,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0001022546228651336,
      "loss": 0.0453,
      "step": 11024
    },
    {
      "epoch": 2.7184418145956606,
      "grad_norm": 0.3515625,
      "learning_rate": 0.0001022202068428345,
      "loss": 0.0447,
      "step": 11026
    },
    {
      "epoch": 2.7189349112426036,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010218579055742845,
      "loss": 0.0446,
      "step": 11028
    },
    {
      "epoch": 2.7194280078895465,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00010215137401299404,
      "loss": 0.0422,
      "step": 11030
    },
    {
      "epoch": 2.719921104536489,
      "grad_norm": 0.333984375,
      "learning_rate": 0.0001021169572136098,
      "loss": 0.0439,
      "step": 11032
    },
    {
      "epoch": 2.720414201183432,
      "grad_norm": 0.328125,
      "learning_rate": 0.0001020825401633543,
      "loss": 0.0489,
      "step": 11034
    },
    {
      "epoch": 2.720907297830375,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00010204812286630616,
      "loss": 0.0435,
      "step": 11036
    },
    {
      "epoch": 2.7214003944773175,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010201370532654404,
      "loss": 0.0431,
      "step": 11038
    },
    {
      "epoch": 2.7218934911242605,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010197928754814661,
      "loss": 0.0436,
      "step": 11040
    },
    {
      "epoch": 2.722386587771203,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0001019448695351925,
      "loss": 0.0429,
      "step": 11042
    },
    {
      "epoch": 2.722879684418146,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00010191045129176053,
      "loss": 0.0402,
      "step": 11044
    },
    {
      "epoch": 2.7233727810650885,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00010187603282192943,
      "loss": 0.0442,
      "step": 11046
    },
    {
      "epoch": 2.7238658777120315,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00010184161412977795,
      "loss": 0.0411,
      "step": 11048
    },
    {
      "epoch": 2.7243589743589745,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010180719521938491,
      "loss": 0.0408,
      "step": 11050
    },
    {
      "epoch": 2.724852071005917,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00010177277609482914,
      "loss": 0.0443,
      "step": 11052
    },
    {
      "epoch": 2.72534516765286,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001017383567601895,
      "loss": 0.0448,
      "step": 11054
    },
    {
      "epoch": 2.725838264299803,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0001017039372195449,
      "loss": 0.0466,
      "step": 11056
    },
    {
      "epoch": 2.7263313609467454,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00010166951747697421,
      "loss": 0.0453,
      "step": 11058
    },
    {
      "epoch": 2.7268244575936884,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0001016350975365564,
      "loss": 0.0434,
      "step": 11060
    },
    {
      "epoch": 2.7273175542406314,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010160067740237038,
      "loss": 0.0391,
      "step": 11062
    },
    {
      "epoch": 2.727810650887574,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010156625707849517,
      "loss": 0.0428,
      "step": 11064
    },
    {
      "epoch": 2.728303747534517,
      "grad_norm": 0.375,
      "learning_rate": 0.00010153183656900973,
      "loss": 0.0436,
      "step": 11066
    },
    {
      "epoch": 2.7287968441814594,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001014974158779931,
      "loss": 0.0443,
      "step": 11068
    },
    {
      "epoch": 2.7292899408284024,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010146299500952434,
      "loss": 0.0416,
      "step": 11070
    },
    {
      "epoch": 2.729783037475345,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00010142857396768251,
      "loss": 0.0414,
      "step": 11072
    },
    {
      "epoch": 2.730276134122288,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010139415275654671,
      "loss": 0.044,
      "step": 11074
    },
    {
      "epoch": 2.730769230769231,
      "grad_norm": 0.296875,
      "learning_rate": 0.00010135973138019599,
      "loss": 0.0429,
      "step": 11076
    },
    {
      "epoch": 2.7312623274161734,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001013253098427095,
      "loss": 0.0416,
      "step": 11078
    },
    {
      "epoch": 2.7317554240631163,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010129088814816641,
      "loss": 0.0409,
      "step": 11080
    },
    {
      "epoch": 2.7322485207100593,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010125646630064586,
      "loss": 0.0421,
      "step": 11082
    },
    {
      "epoch": 2.732741617357002,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010122204430422704,
      "loss": 0.0443,
      "step": 11084
    },
    {
      "epoch": 2.733234714003945,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00010118762216298916,
      "loss": 0.0403,
      "step": 11086
    },
    {
      "epoch": 2.7337278106508878,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010115319988101141,
      "loss": 0.0439,
      "step": 11088
    },
    {
      "epoch": 2.7342209072978303,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010111877746237306,
      "loss": 0.0424,
      "step": 11090
    },
    {
      "epoch": 2.7347140039447733,
      "grad_norm": 0.302734375,
      "learning_rate": 0.0001010843549111533,
      "loss": 0.0428,
      "step": 11092
    },
    {
      "epoch": 2.7352071005917162,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00010104993223143144,
      "loss": 0.0471,
      "step": 11094
    },
    {
      "epoch": 2.7357001972386588,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010101550942728675,
      "loss": 0.0425,
      "step": 11096
    },
    {
      "epoch": 2.7361932938856017,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010098108650279851,
      "loss": 0.0414,
      "step": 11098
    },
    {
      "epoch": 2.7366863905325443,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00010094666346204609,
      "loss": 0.0463,
      "step": 11100
    },
    {
      "epoch": 2.7371794871794872,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00010091224030910872,
      "loss": 0.0383,
      "step": 11102
    },
    {
      "epoch": 2.7376725838264298,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00010087781704806578,
      "loss": 0.0397,
      "step": 11104
    },
    {
      "epoch": 2.7381656804733727,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00010084339368299667,
      "loss": 0.0429,
      "step": 11106
    },
    {
      "epoch": 2.7386587771203157,
      "grad_norm": 0.236328125,
      "learning_rate": 0.00010080897021798064,
      "loss": 0.0371,
      "step": 11108
    },
    {
      "epoch": 2.739151873767258,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010077454665709713,
      "loss": 0.0426,
      "step": 11110
    },
    {
      "epoch": 2.739644970414201,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00010074012300442552,
      "loss": 0.0452,
      "step": 11112
    },
    {
      "epoch": 2.740138067061144,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010070569926404523,
      "loss": 0.0433,
      "step": 11114
    },
    {
      "epoch": 2.7406311637080867,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00010067127544003563,
      "loss": 0.0414,
      "step": 11116
    },
    {
      "epoch": 2.7411242603550297,
      "grad_norm": 0.375,
      "learning_rate": 0.0001006368515364761,
      "loss": 0.0456,
      "step": 11118
    },
    {
      "epoch": 2.7416173570019726,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00010060242755744614,
      "loss": 0.0452,
      "step": 11120
    },
    {
      "epoch": 2.742110453648915,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010056800350702517,
      "loss": 0.0472,
      "step": 11122
    },
    {
      "epoch": 2.742603550295858,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010053357938929257,
      "loss": 0.0386,
      "step": 11124
    },
    {
      "epoch": 2.7430966469428006,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00010049915520832787,
      "loss": 0.0382,
      "step": 11126
    },
    {
      "epoch": 2.7435897435897436,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010046473096821047,
      "loss": 0.0437,
      "step": 11128
    },
    {
      "epoch": 2.744082840236686,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010043030667301986,
      "loss": 0.0435,
      "step": 11130
    },
    {
      "epoch": 2.744575936883629,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0001003958823268355,
      "loss": 0.0447,
      "step": 11132
    },
    {
      "epoch": 2.745069033530572,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010036145793373688,
      "loss": 0.0383,
      "step": 11134
    },
    {
      "epoch": 2.7455621301775146,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010032703349780348,
      "loss": 0.0397,
      "step": 11136
    },
    {
      "epoch": 2.7460552268244576,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00010029260902311477,
      "loss": 0.0464,
      "step": 11138
    },
    {
      "epoch": 2.7465483234714005,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00010025818451375028,
      "loss": 0.0367,
      "step": 11140
    },
    {
      "epoch": 2.747041420118343,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010022375997378947,
      "loss": 0.041,
      "step": 11142
    },
    {
      "epoch": 2.747534516765286,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00010018933540731188,
      "loss": 0.042,
      "step": 11144
    },
    {
      "epoch": 2.748027613412229,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00010015491081839696,
      "loss": 0.0444,
      "step": 11146
    },
    {
      "epoch": 2.7485207100591715,
      "grad_norm": 0.375,
      "learning_rate": 0.00010012048621112424,
      "loss": 0.045,
      "step": 11148
    },
    {
      "epoch": 2.7490138067061145,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010008606158957323,
      "loss": 0.0404,
      "step": 11150
    },
    {
      "epoch": 2.7495069033530575,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010005163695782345,
      "loss": 0.0462,
      "step": 11152
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.328125,
      "learning_rate": 0.00010001721231995443,
      "loss": 0.0438,
      "step": 11154
    },
    {
      "epoch": 2.7504930966469425,
      "grad_norm": 0.337890625,
      "learning_rate": 9.998278768004561e-05,
      "loss": 0.043,
      "step": 11156
    },
    {
      "epoch": 2.7509861932938855,
      "grad_norm": 0.287109375,
      "learning_rate": 9.994836304217657e-05,
      "loss": 0.0448,
      "step": 11158
    },
    {
      "epoch": 2.7514792899408285,
      "grad_norm": 0.359375,
      "learning_rate": 9.991393841042676e-05,
      "loss": 0.0417,
      "step": 11160
    },
    {
      "epoch": 2.751972386587771,
      "grad_norm": 0.314453125,
      "learning_rate": 9.987951378887578e-05,
      "loss": 0.0465,
      "step": 11162
    },
    {
      "epoch": 2.752465483234714,
      "grad_norm": 0.318359375,
      "learning_rate": 9.984508918160308e-05,
      "loss": 0.0425,
      "step": 11164
    },
    {
      "epoch": 2.752958579881657,
      "grad_norm": 0.302734375,
      "learning_rate": 9.981066459268814e-05,
      "loss": 0.0451,
      "step": 11166
    },
    {
      "epoch": 2.7534516765285995,
      "grad_norm": 0.341796875,
      "learning_rate": 9.977624002621054e-05,
      "loss": 0.04,
      "step": 11168
    },
    {
      "epoch": 2.7539447731755424,
      "grad_norm": 0.330078125,
      "learning_rate": 9.974181548624973e-05,
      "loss": 0.0425,
      "step": 11170
    },
    {
      "epoch": 2.7544378698224854,
      "grad_norm": 0.341796875,
      "learning_rate": 9.970739097688525e-05,
      "loss": 0.0415,
      "step": 11172
    },
    {
      "epoch": 2.754930966469428,
      "grad_norm": 0.400390625,
      "learning_rate": 9.967296650219652e-05,
      "loss": 0.042,
      "step": 11174
    },
    {
      "epoch": 2.755424063116371,
      "grad_norm": 0.423828125,
      "learning_rate": 9.963854206626314e-05,
      "loss": 0.0447,
      "step": 11176
    },
    {
      "epoch": 2.755917159763314,
      "grad_norm": 0.3359375,
      "learning_rate": 9.960411767316448e-05,
      "loss": 0.0442,
      "step": 11178
    },
    {
      "epoch": 2.7564102564102564,
      "grad_norm": 0.455078125,
      "learning_rate": 9.956969332698015e-05,
      "loss": 0.0433,
      "step": 11180
    },
    {
      "epoch": 2.7569033530571994,
      "grad_norm": 0.40234375,
      "learning_rate": 9.953526903178956e-05,
      "loss": 0.0367,
      "step": 11182
    },
    {
      "epoch": 2.757396449704142,
      "grad_norm": 0.32421875,
      "learning_rate": 9.950084479167217e-05,
      "loss": 0.0434,
      "step": 11184
    },
    {
      "epoch": 2.757889546351085,
      "grad_norm": 0.41015625,
      "learning_rate": 9.946642061070746e-05,
      "loss": 0.0423,
      "step": 11186
    },
    {
      "epoch": 2.7583826429980274,
      "grad_norm": 0.306640625,
      "learning_rate": 9.943199649297485e-05,
      "loss": 0.0421,
      "step": 11188
    },
    {
      "epoch": 2.7588757396449703,
      "grad_norm": 0.31640625,
      "learning_rate": 9.939757244255387e-05,
      "loss": 0.0376,
      "step": 11190
    },
    {
      "epoch": 2.7593688362919133,
      "grad_norm": 0.40234375,
      "learning_rate": 9.93631484635239e-05,
      "loss": 0.0408,
      "step": 11192
    },
    {
      "epoch": 2.759861932938856,
      "grad_norm": 0.34765625,
      "learning_rate": 9.93287245599644e-05,
      "loss": 0.0436,
      "step": 11194
    },
    {
      "epoch": 2.760355029585799,
      "grad_norm": 0.32421875,
      "learning_rate": 9.929430073595478e-05,
      "loss": 0.04,
      "step": 11196
    },
    {
      "epoch": 2.760848126232742,
      "grad_norm": 0.3828125,
      "learning_rate": 9.92598769955745e-05,
      "loss": 0.0416,
      "step": 11198
    },
    {
      "epoch": 2.7613412228796843,
      "grad_norm": 0.306640625,
      "learning_rate": 9.92254533429029e-05,
      "loss": 0.0416,
      "step": 11200
    },
    {
      "epoch": 2.7618343195266273,
      "grad_norm": 0.296875,
      "learning_rate": 9.919102978201939e-05,
      "loss": 0.0395,
      "step": 11202
    },
    {
      "epoch": 2.7623274161735702,
      "grad_norm": 0.35546875,
      "learning_rate": 9.915660631700338e-05,
      "loss": 0.0415,
      "step": 11204
    },
    {
      "epoch": 2.7628205128205128,
      "grad_norm": 0.322265625,
      "learning_rate": 9.91221829519342e-05,
      "loss": 0.0421,
      "step": 11206
    },
    {
      "epoch": 2.7633136094674557,
      "grad_norm": 0.33203125,
      "learning_rate": 9.90877596908913e-05,
      "loss": 0.0424,
      "step": 11208
    },
    {
      "epoch": 2.7638067061143983,
      "grad_norm": 0.333984375,
      "learning_rate": 9.905333653795393e-05,
      "loss": 0.0409,
      "step": 11210
    },
    {
      "epoch": 2.7642998027613412,
      "grad_norm": 0.625,
      "learning_rate": 9.90189134972015e-05,
      "loss": 0.0446,
      "step": 11212
    },
    {
      "epoch": 2.7647928994082838,
      "grad_norm": 0.421875,
      "learning_rate": 9.898449057271328e-05,
      "loss": 0.0443,
      "step": 11214
    },
    {
      "epoch": 2.7652859960552267,
      "grad_norm": 0.314453125,
      "learning_rate": 9.895006776856857e-05,
      "loss": 0.0367,
      "step": 11216
    },
    {
      "epoch": 2.7657790927021697,
      "grad_norm": 0.333984375,
      "learning_rate": 9.891564508884673e-05,
      "loss": 0.0451,
      "step": 11218
    },
    {
      "epoch": 2.7662721893491122,
      "grad_norm": 0.42578125,
      "learning_rate": 9.888122253762696e-05,
      "loss": 0.0449,
      "step": 11220
    },
    {
      "epoch": 2.766765285996055,
      "grad_norm": 0.318359375,
      "learning_rate": 9.88468001189886e-05,
      "loss": 0.0377,
      "step": 11222
    },
    {
      "epoch": 2.767258382642998,
      "grad_norm": 0.357421875,
      "learning_rate": 9.881237783701085e-05,
      "loss": 0.0438,
      "step": 11224
    },
    {
      "epoch": 2.7677514792899407,
      "grad_norm": 0.318359375,
      "learning_rate": 9.877795569577299e-05,
      "loss": 0.0433,
      "step": 11226
    },
    {
      "epoch": 2.7682445759368837,
      "grad_norm": 0.330078125,
      "learning_rate": 9.874353369935415e-05,
      "loss": 0.0413,
      "step": 11228
    },
    {
      "epoch": 2.7687376725838266,
      "grad_norm": 0.3203125,
      "learning_rate": 9.870911185183361e-05,
      "loss": 0.042,
      "step": 11230
    },
    {
      "epoch": 2.769230769230769,
      "grad_norm": 0.365234375,
      "learning_rate": 9.867469015729054e-05,
      "loss": 0.0365,
      "step": 11232
    },
    {
      "epoch": 2.769723865877712,
      "grad_norm": 0.294921875,
      "learning_rate": 9.864026861980404e-05,
      "loss": 0.0448,
      "step": 11234
    },
    {
      "epoch": 2.770216962524655,
      "grad_norm": 0.31640625,
      "learning_rate": 9.860584724345334e-05,
      "loss": 0.0399,
      "step": 11236
    },
    {
      "epoch": 2.7707100591715976,
      "grad_norm": 0.29296875,
      "learning_rate": 9.85714260323175e-05,
      "loss": 0.0402,
      "step": 11238
    },
    {
      "epoch": 2.7712031558185406,
      "grad_norm": 0.3203125,
      "learning_rate": 9.853700499047569e-05,
      "loss": 0.0414,
      "step": 11240
    },
    {
      "epoch": 2.771696252465483,
      "grad_norm": 0.4296875,
      "learning_rate": 9.850258412200691e-05,
      "loss": 0.0438,
      "step": 11242
    },
    {
      "epoch": 2.772189349112426,
      "grad_norm": 0.328125,
      "learning_rate": 9.84681634309903e-05,
      "loss": 0.0444,
      "step": 11244
    },
    {
      "epoch": 2.7726824457593686,
      "grad_norm": 0.298828125,
      "learning_rate": 9.843374292150488e-05,
      "loss": 0.0412,
      "step": 11246
    },
    {
      "epoch": 2.7731755424063116,
      "grad_norm": 0.294921875,
      "learning_rate": 9.839932259762963e-05,
      "loss": 0.0417,
      "step": 11248
    },
    {
      "epoch": 2.7736686390532546,
      "grad_norm": 0.474609375,
      "learning_rate": 9.836490246344364e-05,
      "loss": 0.0409,
      "step": 11250
    },
    {
      "epoch": 2.774161735700197,
      "grad_norm": 0.41015625,
      "learning_rate": 9.83304825230258e-05,
      "loss": 0.0489,
      "step": 11252
    },
    {
      "epoch": 2.77465483234714,
      "grad_norm": 0.330078125,
      "learning_rate": 9.829606278045513e-05,
      "loss": 0.048,
      "step": 11254
    },
    {
      "epoch": 2.775147928994083,
      "grad_norm": 0.3828125,
      "learning_rate": 9.82616432398105e-05,
      "loss": 0.0434,
      "step": 11256
    },
    {
      "epoch": 2.7756410256410255,
      "grad_norm": 0.365234375,
      "learning_rate": 9.822722390517089e-05,
      "loss": 0.0452,
      "step": 11258
    },
    {
      "epoch": 2.7761341222879685,
      "grad_norm": 0.33984375,
      "learning_rate": 9.81928047806151e-05,
      "loss": 0.0408,
      "step": 11260
    },
    {
      "epoch": 2.7766272189349115,
      "grad_norm": 0.337890625,
      "learning_rate": 9.815838587022206e-05,
      "loss": 0.0418,
      "step": 11262
    },
    {
      "epoch": 2.777120315581854,
      "grad_norm": 0.404296875,
      "learning_rate": 9.81239671780706e-05,
      "loss": 0.0417,
      "step": 11264
    },
    {
      "epoch": 2.777613412228797,
      "grad_norm": 0.40625,
      "learning_rate": 9.808954870823949e-05,
      "loss": 0.0421,
      "step": 11266
    },
    {
      "epoch": 2.7781065088757395,
      "grad_norm": 0.302734375,
      "learning_rate": 9.805513046480752e-05,
      "loss": 0.0451,
      "step": 11268
    },
    {
      "epoch": 2.7785996055226825,
      "grad_norm": 0.357421875,
      "learning_rate": 9.802071245185343e-05,
      "loss": 0.043,
      "step": 11270
    },
    {
      "epoch": 2.779092702169625,
      "grad_norm": 0.333984375,
      "learning_rate": 9.798629467345599e-05,
      "loss": 0.041,
      "step": 11272
    },
    {
      "epoch": 2.779585798816568,
      "grad_norm": 0.4375,
      "learning_rate": 9.795187713369384e-05,
      "loss": 0.0446,
      "step": 11274
    },
    {
      "epoch": 2.780078895463511,
      "grad_norm": 0.33984375,
      "learning_rate": 9.791745983664572e-05,
      "loss": 0.0404,
      "step": 11276
    },
    {
      "epoch": 2.7805719921104535,
      "grad_norm": 0.29296875,
      "learning_rate": 9.788304278639022e-05,
      "loss": 0.0407,
      "step": 11278
    },
    {
      "epoch": 2.7810650887573964,
      "grad_norm": 0.306640625,
      "learning_rate": 9.7848625987006e-05,
      "loss": 0.0414,
      "step": 11280
    },
    {
      "epoch": 2.7815581854043394,
      "grad_norm": 0.306640625,
      "learning_rate": 9.781420944257159e-05,
      "loss": 0.0414,
      "step": 11282
    },
    {
      "epoch": 2.782051282051282,
      "grad_norm": 0.30078125,
      "learning_rate": 9.777979315716553e-05,
      "loss": 0.045,
      "step": 11284
    },
    {
      "epoch": 2.782544378698225,
      "grad_norm": 0.439453125,
      "learning_rate": 9.774537713486643e-05,
      "loss": 0.0447,
      "step": 11286
    },
    {
      "epoch": 2.783037475345168,
      "grad_norm": 0.330078125,
      "learning_rate": 9.771096137975269e-05,
      "loss": 0.0391,
      "step": 11288
    },
    {
      "epoch": 2.7835305719921104,
      "grad_norm": 0.4375,
      "learning_rate": 9.767654589590284e-05,
      "loss": 0.0462,
      "step": 11290
    },
    {
      "epoch": 2.7840236686390534,
      "grad_norm": 0.435546875,
      "learning_rate": 9.764213068739526e-05,
      "loss": 0.0386,
      "step": 11292
    },
    {
      "epoch": 2.784516765285996,
      "grad_norm": 0.453125,
      "learning_rate": 9.760771575830837e-05,
      "loss": 0.0414,
      "step": 11294
    },
    {
      "epoch": 2.785009861932939,
      "grad_norm": 0.37890625,
      "learning_rate": 9.75733011127205e-05,
      "loss": 0.0463,
      "step": 11296
    },
    {
      "epoch": 2.7855029585798814,
      "grad_norm": 0.294921875,
      "learning_rate": 9.753888675470999e-05,
      "loss": 0.0384,
      "step": 11298
    },
    {
      "epoch": 2.7859960552268244,
      "grad_norm": 0.302734375,
      "learning_rate": 9.750447268835517e-05,
      "loss": 0.0389,
      "step": 11300
    },
    {
      "epoch": 2.7864891518737673,
      "grad_norm": 0.302734375,
      "learning_rate": 9.747005891773422e-05,
      "loss": 0.0397,
      "step": 11302
    },
    {
      "epoch": 2.78698224852071,
      "grad_norm": 0.294921875,
      "learning_rate": 9.743564544692546e-05,
      "loss": 0.0429,
      "step": 11304
    },
    {
      "epoch": 2.787475345167653,
      "grad_norm": 0.3046875,
      "learning_rate": 9.740123228000702e-05,
      "loss": 0.0421,
      "step": 11306
    },
    {
      "epoch": 2.787968441814596,
      "grad_norm": 0.349609375,
      "learning_rate": 9.736681942105708e-05,
      "loss": 0.0441,
      "step": 11308
    },
    {
      "epoch": 2.7884615384615383,
      "grad_norm": 0.318359375,
      "learning_rate": 9.73324068741537e-05,
      "loss": 0.0463,
      "step": 11310
    },
    {
      "epoch": 2.7889546351084813,
      "grad_norm": 0.314453125,
      "learning_rate": 9.729799464337503e-05,
      "loss": 0.0401,
      "step": 11312
    },
    {
      "epoch": 2.7894477317554243,
      "grad_norm": 0.28125,
      "learning_rate": 9.726358273279909e-05,
      "loss": 0.0442,
      "step": 11314
    },
    {
      "epoch": 2.789940828402367,
      "grad_norm": 0.337890625,
      "learning_rate": 9.722917114650383e-05,
      "loss": 0.0444,
      "step": 11316
    },
    {
      "epoch": 2.7904339250493098,
      "grad_norm": 0.3828125,
      "learning_rate": 9.71947598885673e-05,
      "loss": 0.0408,
      "step": 11318
    },
    {
      "epoch": 2.7909270216962527,
      "grad_norm": 0.388671875,
      "learning_rate": 9.716034896306737e-05,
      "loss": 0.0449,
      "step": 11320
    },
    {
      "epoch": 2.7914201183431953,
      "grad_norm": 0.38671875,
      "learning_rate": 9.712593837408196e-05,
      "loss": 0.0421,
      "step": 11322
    },
    {
      "epoch": 2.7919132149901382,
      "grad_norm": 0.314453125,
      "learning_rate": 9.709152812568886e-05,
      "loss": 0.0428,
      "step": 11324
    },
    {
      "epoch": 2.7924063116370808,
      "grad_norm": 0.32421875,
      "learning_rate": 9.705711822196597e-05,
      "loss": 0.0426,
      "step": 11326
    },
    {
      "epoch": 2.7928994082840237,
      "grad_norm": 0.341796875,
      "learning_rate": 9.702270866699094e-05,
      "loss": 0.04,
      "step": 11328
    },
    {
      "epoch": 2.7933925049309662,
      "grad_norm": 0.275390625,
      "learning_rate": 9.69882994648416e-05,
      "loss": 0.0402,
      "step": 11330
    },
    {
      "epoch": 2.793885601577909,
      "grad_norm": 0.3359375,
      "learning_rate": 9.695389061959559e-05,
      "loss": 0.0413,
      "step": 11332
    },
    {
      "epoch": 2.794378698224852,
      "grad_norm": 0.326171875,
      "learning_rate": 9.691948213533053e-05,
      "loss": 0.0391,
      "step": 11334
    },
    {
      "epoch": 2.7948717948717947,
      "grad_norm": 0.34375,
      "learning_rate": 9.688507401612406e-05,
      "loss": 0.0423,
      "step": 11336
    },
    {
      "epoch": 2.7953648915187377,
      "grad_norm": 0.302734375,
      "learning_rate": 9.685066626605366e-05,
      "loss": 0.0428,
      "step": 11338
    },
    {
      "epoch": 2.7958579881656807,
      "grad_norm": 0.341796875,
      "learning_rate": 9.681625888919695e-05,
      "loss": 0.0421,
      "step": 11340
    },
    {
      "epoch": 2.796351084812623,
      "grad_norm": 0.380859375,
      "learning_rate": 9.678185188963129e-05,
      "loss": 0.0429,
      "step": 11342
    },
    {
      "epoch": 2.796844181459566,
      "grad_norm": 0.32421875,
      "learning_rate": 9.674744527143417e-05,
      "loss": 0.0437,
      "step": 11344
    },
    {
      "epoch": 2.797337278106509,
      "grad_norm": 0.357421875,
      "learning_rate": 9.671303903868297e-05,
      "loss": 0.0407,
      "step": 11346
    },
    {
      "epoch": 2.7978303747534516,
      "grad_norm": 0.36328125,
      "learning_rate": 9.667863319545496e-05,
      "loss": 0.0421,
      "step": 11348
    },
    {
      "epoch": 2.7983234714003946,
      "grad_norm": 0.326171875,
      "learning_rate": 9.664422774582749e-05,
      "loss": 0.0431,
      "step": 11350
    },
    {
      "epoch": 2.798816568047337,
      "grad_norm": 0.3125,
      "learning_rate": 9.660982269387773e-05,
      "loss": 0.039,
      "step": 11352
    },
    {
      "epoch": 2.79930966469428,
      "grad_norm": 0.39453125,
      "learning_rate": 9.657541804368294e-05,
      "loss": 0.0429,
      "step": 11354
    },
    {
      "epoch": 2.7998027613412226,
      "grad_norm": 0.345703125,
      "learning_rate": 9.654101379932019e-05,
      "loss": 0.0392,
      "step": 11356
    },
    {
      "epoch": 2.8002958579881656,
      "grad_norm": 0.306640625,
      "learning_rate": 9.650660996486666e-05,
      "loss": 0.0437,
      "step": 11358
    },
    {
      "epoch": 2.8007889546351086,
      "grad_norm": 0.404296875,
      "learning_rate": 9.647220654439935e-05,
      "loss": 0.0451,
      "step": 11360
    },
    {
      "epoch": 2.801282051282051,
      "grad_norm": 0.357421875,
      "learning_rate": 9.643780354199527e-05,
      "loss": 0.0452,
      "step": 11362
    },
    {
      "epoch": 2.801775147928994,
      "grad_norm": 0.302734375,
      "learning_rate": 9.640340096173135e-05,
      "loss": 0.0415,
      "step": 11364
    },
    {
      "epoch": 2.802268244575937,
      "grad_norm": 0.302734375,
      "learning_rate": 9.636899880768446e-05,
      "loss": 0.0403,
      "step": 11366
    },
    {
      "epoch": 2.8027613412228796,
      "grad_norm": 0.310546875,
      "learning_rate": 9.633459708393154e-05,
      "loss": 0.0405,
      "step": 11368
    },
    {
      "epoch": 2.8032544378698225,
      "grad_norm": 0.306640625,
      "learning_rate": 9.630019579454927e-05,
      "loss": 0.046,
      "step": 11370
    },
    {
      "epoch": 2.8037475345167655,
      "grad_norm": 0.3828125,
      "learning_rate": 9.62657949436145e-05,
      "loss": 0.0405,
      "step": 11372
    },
    {
      "epoch": 2.804240631163708,
      "grad_norm": 0.298828125,
      "learning_rate": 9.623139453520387e-05,
      "loss": 0.0439,
      "step": 11374
    },
    {
      "epoch": 2.804733727810651,
      "grad_norm": 0.298828125,
      "learning_rate": 9.619699457339405e-05,
      "loss": 0.0387,
      "step": 11376
    },
    {
      "epoch": 2.8052268244575935,
      "grad_norm": 0.322265625,
      "learning_rate": 9.616259506226157e-05,
      "loss": 0.043,
      "step": 11378
    },
    {
      "epoch": 2.8057199211045365,
      "grad_norm": 0.345703125,
      "learning_rate": 9.6128196005883e-05,
      "loss": 0.0463,
      "step": 11380
    },
    {
      "epoch": 2.806213017751479,
      "grad_norm": 0.349609375,
      "learning_rate": 9.609379740833487e-05,
      "loss": 0.0464,
      "step": 11382
    },
    {
      "epoch": 2.806706114398422,
      "grad_norm": 0.365234375,
      "learning_rate": 9.605939927369349e-05,
      "loss": 0.0448,
      "step": 11384
    },
    {
      "epoch": 2.807199211045365,
      "grad_norm": 0.32421875,
      "learning_rate": 9.602500160603534e-05,
      "loss": 0.0417,
      "step": 11386
    },
    {
      "epoch": 2.8076923076923075,
      "grad_norm": 0.3671875,
      "learning_rate": 9.599060440943668e-05,
      "loss": 0.0431,
      "step": 11388
    },
    {
      "epoch": 2.8081854043392505,
      "grad_norm": 0.48046875,
      "learning_rate": 9.59562076879738e-05,
      "loss": 0.0387,
      "step": 11390
    },
    {
      "epoch": 2.8086785009861934,
      "grad_norm": 0.31640625,
      "learning_rate": 9.592181144572285e-05,
      "loss": 0.0403,
      "step": 11392
    },
    {
      "epoch": 2.809171597633136,
      "grad_norm": 0.296875,
      "learning_rate": 9.588741568676004e-05,
      "loss": 0.0425,
      "step": 11394
    },
    {
      "epoch": 2.809664694280079,
      "grad_norm": 0.353515625,
      "learning_rate": 9.585302041516144e-05,
      "loss": 0.0439,
      "step": 11396
    },
    {
      "epoch": 2.810157790927022,
      "grad_norm": 0.3125,
      "learning_rate": 9.581862563500304e-05,
      "loss": 0.0407,
      "step": 11398
    },
    {
      "epoch": 2.8106508875739644,
      "grad_norm": 0.33984375,
      "learning_rate": 9.578423135036088e-05,
      "loss": 0.0425,
      "step": 11400
    },
    {
      "epoch": 2.8111439842209074,
      "grad_norm": 0.353515625,
      "learning_rate": 9.574983756531084e-05,
      "loss": 0.0406,
      "step": 11402
    },
    {
      "epoch": 2.8116370808678504,
      "grad_norm": 0.41796875,
      "learning_rate": 9.57154442839288e-05,
      "loss": 0.041,
      "step": 11404
    },
    {
      "epoch": 2.812130177514793,
      "grad_norm": 0.3359375,
      "learning_rate": 9.568105151029047e-05,
      "loss": 0.0404,
      "step": 11406
    },
    {
      "epoch": 2.812623274161736,
      "grad_norm": 0.302734375,
      "learning_rate": 9.564665924847169e-05,
      "loss": 0.0414,
      "step": 11408
    },
    {
      "epoch": 2.8131163708086784,
      "grad_norm": 0.357421875,
      "learning_rate": 9.561226750254807e-05,
      "loss": 0.0408,
      "step": 11410
    },
    {
      "epoch": 2.8136094674556213,
      "grad_norm": 0.30078125,
      "learning_rate": 9.557787627659528e-05,
      "loss": 0.041,
      "step": 11412
    },
    {
      "epoch": 2.814102564102564,
      "grad_norm": 0.3828125,
      "learning_rate": 9.554348557468885e-05,
      "loss": 0.0422,
      "step": 11414
    },
    {
      "epoch": 2.814595660749507,
      "grad_norm": 0.494140625,
      "learning_rate": 9.550909540090424e-05,
      "loss": 0.0439,
      "step": 11416
    },
    {
      "epoch": 2.81508875739645,
      "grad_norm": 0.34765625,
      "learning_rate": 9.54747057593169e-05,
      "loss": 0.0434,
      "step": 11418
    },
    {
      "epoch": 2.8155818540433923,
      "grad_norm": 0.314453125,
      "learning_rate": 9.544031665400218e-05,
      "loss": 0.0403,
      "step": 11420
    },
    {
      "epoch": 2.8160749506903353,
      "grad_norm": 0.328125,
      "learning_rate": 9.540592808903543e-05,
      "loss": 0.0462,
      "step": 11422
    },
    {
      "epoch": 2.8165680473372783,
      "grad_norm": 0.400390625,
      "learning_rate": 9.53715400684918e-05,
      "loss": 0.0424,
      "step": 11424
    },
    {
      "epoch": 2.817061143984221,
      "grad_norm": 0.3046875,
      "learning_rate": 9.533715259644657e-05,
      "loss": 0.0407,
      "step": 11426
    },
    {
      "epoch": 2.8175542406311638,
      "grad_norm": 0.349609375,
      "learning_rate": 9.530276567697478e-05,
      "loss": 0.0429,
      "step": 11428
    },
    {
      "epoch": 2.8180473372781067,
      "grad_norm": 0.4296875,
      "learning_rate": 9.526837931415146e-05,
      "loss": 0.0439,
      "step": 11430
    },
    {
      "epoch": 2.8185404339250493,
      "grad_norm": 0.5078125,
      "learning_rate": 9.523399351205161e-05,
      "loss": 0.0423,
      "step": 11432
    },
    {
      "epoch": 2.8190335305719922,
      "grad_norm": 0.322265625,
      "learning_rate": 9.519960827475014e-05,
      "loss": 0.0449,
      "step": 11434
    },
    {
      "epoch": 2.8195266272189348,
      "grad_norm": 0.3125,
      "learning_rate": 9.516522360632192e-05,
      "loss": 0.0399,
      "step": 11436
    },
    {
      "epoch": 2.8200197238658777,
      "grad_norm": 0.52734375,
      "learning_rate": 9.513083951084165e-05,
      "loss": 0.0424,
      "step": 11438
    },
    {
      "epoch": 2.8205128205128203,
      "grad_norm": 0.359375,
      "learning_rate": 9.509645599238412e-05,
      "loss": 0.0402,
      "step": 11440
    },
    {
      "epoch": 2.8210059171597632,
      "grad_norm": 0.3359375,
      "learning_rate": 9.506207305502391e-05,
      "loss": 0.0454,
      "step": 11442
    },
    {
      "epoch": 2.821499013806706,
      "grad_norm": 0.35546875,
      "learning_rate": 9.502769070283562e-05,
      "loss": 0.0373,
      "step": 11444
    },
    {
      "epoch": 2.8219921104536487,
      "grad_norm": 0.5,
      "learning_rate": 9.499330893989377e-05,
      "loss": 0.045,
      "step": 11446
    },
    {
      "epoch": 2.8224852071005917,
      "grad_norm": 0.59375,
      "learning_rate": 9.495892777027274e-05,
      "loss": 0.045,
      "step": 11448
    },
    {
      "epoch": 2.8229783037475347,
      "grad_norm": 0.31640625,
      "learning_rate": 9.492454719804693e-05,
      "loss": 0.0414,
      "step": 11450
    },
    {
      "epoch": 2.823471400394477,
      "grad_norm": 0.349609375,
      "learning_rate": 9.489016722729058e-05,
      "loss": 0.0436,
      "step": 11452
    },
    {
      "epoch": 2.82396449704142,
      "grad_norm": 0.296875,
      "learning_rate": 9.485578786207799e-05,
      "loss": 0.0426,
      "step": 11454
    },
    {
      "epoch": 2.824457593688363,
      "grad_norm": 0.40234375,
      "learning_rate": 9.482140910648322e-05,
      "loss": 0.046,
      "step": 11456
    },
    {
      "epoch": 2.8249506903353057,
      "grad_norm": 0.32421875,
      "learning_rate": 9.478703096458039e-05,
      "loss": 0.0459,
      "step": 11458
    },
    {
      "epoch": 2.8254437869822486,
      "grad_norm": 0.37890625,
      "learning_rate": 9.475265344044351e-05,
      "loss": 0.0415,
      "step": 11460
    },
    {
      "epoch": 2.8259368836291916,
      "grad_norm": 0.294921875,
      "learning_rate": 9.47182765381465e-05,
      "loss": 0.0425,
      "step": 11462
    },
    {
      "epoch": 2.826429980276134,
      "grad_norm": 0.326171875,
      "learning_rate": 9.468390026176321e-05,
      "loss": 0.0432,
      "step": 11464
    },
    {
      "epoch": 2.8269230769230766,
      "grad_norm": 0.302734375,
      "learning_rate": 9.464952461536739e-05,
      "loss": 0.0407,
      "step": 11466
    },
    {
      "epoch": 2.8274161735700196,
      "grad_norm": 0.291015625,
      "learning_rate": 9.461514960303282e-05,
      "loss": 0.0401,
      "step": 11468
    },
    {
      "epoch": 2.8279092702169626,
      "grad_norm": 0.322265625,
      "learning_rate": 9.458077522883305e-05,
      "loss": 0.0399,
      "step": 11470
    },
    {
      "epoch": 2.828402366863905,
      "grad_norm": 0.27734375,
      "learning_rate": 9.454640149684169e-05,
      "loss": 0.0394,
      "step": 11472
    },
    {
      "epoch": 2.828895463510848,
      "grad_norm": 0.41796875,
      "learning_rate": 9.451202841113219e-05,
      "loss": 0.042,
      "step": 11474
    },
    {
      "epoch": 2.829388560157791,
      "grad_norm": 0.30859375,
      "learning_rate": 9.447765597577799e-05,
      "loss": 0.0414,
      "step": 11476
    },
    {
      "epoch": 2.8298816568047336,
      "grad_norm": 0.396484375,
      "learning_rate": 9.444328419485239e-05,
      "loss": 0.0399,
      "step": 11478
    },
    {
      "epoch": 2.8303747534516766,
      "grad_norm": 0.322265625,
      "learning_rate": 9.440891307242859e-05,
      "loss": 0.0416,
      "step": 11480
    },
    {
      "epoch": 2.8308678500986195,
      "grad_norm": 0.3828125,
      "learning_rate": 9.437454261257986e-05,
      "loss": 0.0419,
      "step": 11482
    },
    {
      "epoch": 2.831360946745562,
      "grad_norm": 0.302734375,
      "learning_rate": 9.434017281937918e-05,
      "loss": 0.0442,
      "step": 11484
    },
    {
      "epoch": 2.831854043392505,
      "grad_norm": 0.30078125,
      "learning_rate": 9.430580369689968e-05,
      "loss": 0.0398,
      "step": 11486
    },
    {
      "epoch": 2.832347140039448,
      "grad_norm": 0.4609375,
      "learning_rate": 9.42714352492142e-05,
      "loss": 0.0462,
      "step": 11488
    },
    {
      "epoch": 2.8328402366863905,
      "grad_norm": 0.3125,
      "learning_rate": 9.423706748039566e-05,
      "loss": 0.0378,
      "step": 11490
    },
    {
      "epoch": 2.8333333333333335,
      "grad_norm": 0.287109375,
      "learning_rate": 9.420270039451675e-05,
      "loss": 0.0432,
      "step": 11492
    },
    {
      "epoch": 2.833826429980276,
      "grad_norm": 0.361328125,
      "learning_rate": 9.416833399565027e-05,
      "loss": 0.0441,
      "step": 11494
    },
    {
      "epoch": 2.834319526627219,
      "grad_norm": 0.365234375,
      "learning_rate": 9.413396828786875e-05,
      "loss": 0.041,
      "step": 11496
    },
    {
      "epoch": 2.8348126232741615,
      "grad_norm": 0.3828125,
      "learning_rate": 9.40996032752447e-05,
      "loss": 0.0424,
      "step": 11498
    },
    {
      "epoch": 2.8353057199211045,
      "grad_norm": 0.318359375,
      "learning_rate": 9.406523896185066e-05,
      "loss": 0.0451,
      "step": 11500
    },
    {
      "epoch": 2.8357988165680474,
      "grad_norm": 0.3515625,
      "learning_rate": 9.40308753517589e-05,
      "loss": 0.042,
      "step": 11502
    },
    {
      "epoch": 2.83629191321499,
      "grad_norm": 0.326171875,
      "learning_rate": 9.399651244904178e-05,
      "loss": 0.0443,
      "step": 11504
    },
    {
      "epoch": 2.836785009861933,
      "grad_norm": 0.40234375,
      "learning_rate": 9.396215025777139e-05,
      "loss": 0.0453,
      "step": 11506
    },
    {
      "epoch": 2.837278106508876,
      "grad_norm": 0.279296875,
      "learning_rate": 9.392778878201995e-05,
      "loss": 0.0424,
      "step": 11508
    },
    {
      "epoch": 2.8377712031558184,
      "grad_norm": 0.376953125,
      "learning_rate": 9.389342802585946e-05,
      "loss": 0.0411,
      "step": 11510
    },
    {
      "epoch": 2.8382642998027614,
      "grad_norm": 0.353515625,
      "learning_rate": 9.385906799336179e-05,
      "loss": 0.046,
      "step": 11512
    },
    {
      "epoch": 2.8387573964497044,
      "grad_norm": 0.322265625,
      "learning_rate": 9.38247086885989e-05,
      "loss": 0.0425,
      "step": 11514
    },
    {
      "epoch": 2.839250493096647,
      "grad_norm": 0.337890625,
      "learning_rate": 9.379035011564247e-05,
      "loss": 0.0432,
      "step": 11516
    },
    {
      "epoch": 2.83974358974359,
      "grad_norm": 0.337890625,
      "learning_rate": 9.375599227856426e-05,
      "loss": 0.043,
      "step": 11518
    },
    {
      "epoch": 2.8402366863905324,
      "grad_norm": 0.4453125,
      "learning_rate": 9.372163518143578e-05,
      "loss": 0.0408,
      "step": 11520
    },
    {
      "epoch": 2.8407297830374754,
      "grad_norm": 0.3515625,
      "learning_rate": 9.368727882832863e-05,
      "loss": 0.0409,
      "step": 11522
    },
    {
      "epoch": 2.841222879684418,
      "grad_norm": 0.337890625,
      "learning_rate": 9.365292322331413e-05,
      "loss": 0.0455,
      "step": 11524
    },
    {
      "epoch": 2.841715976331361,
      "grad_norm": 0.3671875,
      "learning_rate": 9.361856837046373e-05,
      "loss": 0.0411,
      "step": 11526
    },
    {
      "epoch": 2.842209072978304,
      "grad_norm": 0.33203125,
      "learning_rate": 9.358421427384859e-05,
      "loss": 0.0422,
      "step": 11528
    },
    {
      "epoch": 2.8427021696252464,
      "grad_norm": 0.357421875,
      "learning_rate": 9.354986093753986e-05,
      "loss": 0.045,
      "step": 11530
    },
    {
      "epoch": 2.8431952662721893,
      "grad_norm": 0.435546875,
      "learning_rate": 9.351550836560864e-05,
      "loss": 0.0423,
      "step": 11532
    },
    {
      "epoch": 2.8436883629191323,
      "grad_norm": 0.3515625,
      "learning_rate": 9.348115656212585e-05,
      "loss": 0.0424,
      "step": 11534
    },
    {
      "epoch": 2.844181459566075,
      "grad_norm": 0.333984375,
      "learning_rate": 9.344680553116244e-05,
      "loss": 0.0395,
      "step": 11536
    },
    {
      "epoch": 2.844674556213018,
      "grad_norm": 0.294921875,
      "learning_rate": 9.341245527678912e-05,
      "loss": 0.0423,
      "step": 11538
    },
    {
      "epoch": 2.8451676528599608,
      "grad_norm": 0.310546875,
      "learning_rate": 9.337810580307668e-05,
      "loss": 0.0409,
      "step": 11540
    },
    {
      "epoch": 2.8456607495069033,
      "grad_norm": 0.34765625,
      "learning_rate": 9.334375711409564e-05,
      "loss": 0.0427,
      "step": 11542
    },
    {
      "epoch": 2.8461538461538463,
      "grad_norm": 0.28125,
      "learning_rate": 9.330940921391657e-05,
      "loss": 0.0386,
      "step": 11544
    },
    {
      "epoch": 2.8466469428007892,
      "grad_norm": 0.318359375,
      "learning_rate": 9.327506210660988e-05,
      "loss": 0.0438,
      "step": 11546
    },
    {
      "epoch": 2.8471400394477318,
      "grad_norm": 0.296875,
      "learning_rate": 9.324071579624583e-05,
      "loss": 0.0405,
      "step": 11548
    },
    {
      "epoch": 2.8476331360946747,
      "grad_norm": 0.34765625,
      "learning_rate": 9.320637028689474e-05,
      "loss": 0.0392,
      "step": 11550
    },
    {
      "epoch": 2.8481262327416172,
      "grad_norm": 0.359375,
      "learning_rate": 9.317202558262664e-05,
      "loss": 0.0421,
      "step": 11552
    },
    {
      "epoch": 2.84861932938856,
      "grad_norm": 0.380859375,
      "learning_rate": 9.31376816875117e-05,
      "loss": 0.038,
      "step": 11554
    },
    {
      "epoch": 2.8491124260355027,
      "grad_norm": 0.3125,
      "learning_rate": 9.310333860561975e-05,
      "loss": 0.042,
      "step": 11556
    },
    {
      "epoch": 2.8496055226824457,
      "grad_norm": 0.35546875,
      "learning_rate": 9.306899634102071e-05,
      "loss": 0.0393,
      "step": 11558
    },
    {
      "epoch": 2.8500986193293887,
      "grad_norm": 0.3203125,
      "learning_rate": 9.30346548977843e-05,
      "loss": 0.0397,
      "step": 11560
    },
    {
      "epoch": 2.850591715976331,
      "grad_norm": 0.2890625,
      "learning_rate": 9.300031427998014e-05,
      "loss": 0.0386,
      "step": 11562
    },
    {
      "epoch": 2.851084812623274,
      "grad_norm": 0.3515625,
      "learning_rate": 9.296597449167785e-05,
      "loss": 0.0418,
      "step": 11564
    },
    {
      "epoch": 2.851577909270217,
      "grad_norm": 0.357421875,
      "learning_rate": 9.293163553694682e-05,
      "loss": 0.0438,
      "step": 11566
    },
    {
      "epoch": 2.8520710059171597,
      "grad_norm": 0.39453125,
      "learning_rate": 9.289729741985649e-05,
      "loss": 0.0416,
      "step": 11568
    },
    {
      "epoch": 2.8525641025641026,
      "grad_norm": 0.333984375,
      "learning_rate": 9.286296014447604e-05,
      "loss": 0.0436,
      "step": 11570
    },
    {
      "epoch": 2.8530571992110456,
      "grad_norm": 0.263671875,
      "learning_rate": 9.28286237148747e-05,
      "loss": 0.0377,
      "step": 11572
    },
    {
      "epoch": 2.853550295857988,
      "grad_norm": 0.29296875,
      "learning_rate": 9.279428813512143e-05,
      "loss": 0.0415,
      "step": 11574
    },
    {
      "epoch": 2.854043392504931,
      "grad_norm": 0.33984375,
      "learning_rate": 9.275995340928529e-05,
      "loss": 0.0445,
      "step": 11576
    },
    {
      "epoch": 2.8545364891518736,
      "grad_norm": 0.443359375,
      "learning_rate": 9.27256195414351e-05,
      "loss": 0.0461,
      "step": 11578
    },
    {
      "epoch": 2.8550295857988166,
      "grad_norm": 0.302734375,
      "learning_rate": 9.269128653563957e-05,
      "loss": 0.0434,
      "step": 11580
    },
    {
      "epoch": 2.855522682445759,
      "grad_norm": 0.421875,
      "learning_rate": 9.265695439596744e-05,
      "loss": 0.0454,
      "step": 11582
    },
    {
      "epoch": 2.856015779092702,
      "grad_norm": 0.3359375,
      "learning_rate": 9.262262312648717e-05,
      "loss": 0.0468,
      "step": 11584
    },
    {
      "epoch": 2.856508875739645,
      "grad_norm": 0.328125,
      "learning_rate": 9.258829273126729e-05,
      "loss": 0.0411,
      "step": 11586
    },
    {
      "epoch": 2.8570019723865876,
      "grad_norm": 0.298828125,
      "learning_rate": 9.255396321437606e-05,
      "loss": 0.0412,
      "step": 11588
    },
    {
      "epoch": 2.8574950690335306,
      "grad_norm": 0.373046875,
      "learning_rate": 9.251963457988179e-05,
      "loss": 0.0462,
      "step": 11590
    },
    {
      "epoch": 2.8579881656804735,
      "grad_norm": 0.302734375,
      "learning_rate": 9.248530683185256e-05,
      "loss": 0.0434,
      "step": 11592
    },
    {
      "epoch": 2.858481262327416,
      "grad_norm": 0.47265625,
      "learning_rate": 9.245097997435646e-05,
      "loss": 0.0388,
      "step": 11594
    },
    {
      "epoch": 2.858974358974359,
      "grad_norm": 0.36328125,
      "learning_rate": 9.241665401146139e-05,
      "loss": 0.0385,
      "step": 11596
    },
    {
      "epoch": 2.859467455621302,
      "grad_norm": 0.275390625,
      "learning_rate": 9.238232894723513e-05,
      "loss": 0.039,
      "step": 11598
    },
    {
      "epoch": 2.8599605522682445,
      "grad_norm": 0.333984375,
      "learning_rate": 9.234800478574544e-05,
      "loss": 0.0404,
      "step": 11600
    },
    {
      "epoch": 2.8604536489151875,
      "grad_norm": 0.455078125,
      "learning_rate": 9.231368153105988e-05,
      "loss": 0.0421,
      "step": 11602
    },
    {
      "epoch": 2.86094674556213,
      "grad_norm": 0.30859375,
      "learning_rate": 9.2279359187246e-05,
      "loss": 0.0426,
      "step": 11604
    },
    {
      "epoch": 2.861439842209073,
      "grad_norm": 0.421875,
      "learning_rate": 9.224503775837112e-05,
      "loss": 0.0382,
      "step": 11606
    },
    {
      "epoch": 2.8619329388560155,
      "grad_norm": 0.345703125,
      "learning_rate": 9.221071724850262e-05,
      "loss": 0.041,
      "step": 11608
    },
    {
      "epoch": 2.8624260355029585,
      "grad_norm": 0.392578125,
      "learning_rate": 9.21763976617076e-05,
      "loss": 0.0395,
      "step": 11610
    },
    {
      "epoch": 2.8629191321499015,
      "grad_norm": 0.37109375,
      "learning_rate": 9.214207900205314e-05,
      "loss": 0.0422,
      "step": 11612
    },
    {
      "epoch": 2.863412228796844,
      "grad_norm": 0.30859375,
      "learning_rate": 9.21077612736062e-05,
      "loss": 0.0377,
      "step": 11614
    },
    {
      "epoch": 2.863905325443787,
      "grad_norm": 0.31640625,
      "learning_rate": 9.207344448043356e-05,
      "loss": 0.0397,
      "step": 11616
    },
    {
      "epoch": 2.86439842209073,
      "grad_norm": 0.546875,
      "learning_rate": 9.203912862660205e-05,
      "loss": 0.046,
      "step": 11618
    },
    {
      "epoch": 2.8648915187376724,
      "grad_norm": 0.33203125,
      "learning_rate": 9.200481371617821e-05,
      "loss": 0.0401,
      "step": 11620
    },
    {
      "epoch": 2.8653846153846154,
      "grad_norm": 0.3046875,
      "learning_rate": 9.197049975322863e-05,
      "loss": 0.0424,
      "step": 11622
    },
    {
      "epoch": 2.8658777120315584,
      "grad_norm": 0.333984375,
      "learning_rate": 9.193618674181963e-05,
      "loss": 0.0431,
      "step": 11624
    },
    {
      "epoch": 2.866370808678501,
      "grad_norm": 0.31640625,
      "learning_rate": 9.190187468601756e-05,
      "loss": 0.0401,
      "step": 11626
    },
    {
      "epoch": 2.866863905325444,
      "grad_norm": 0.375,
      "learning_rate": 9.186756358988855e-05,
      "loss": 0.0422,
      "step": 11628
    },
    {
      "epoch": 2.867357001972387,
      "grad_norm": 0.361328125,
      "learning_rate": 9.18332534574986e-05,
      "loss": 0.0394,
      "step": 11630
    },
    {
      "epoch": 2.8678500986193294,
      "grad_norm": 0.41015625,
      "learning_rate": 9.179894429291376e-05,
      "loss": 0.0415,
      "step": 11632
    },
    {
      "epoch": 2.8683431952662723,
      "grad_norm": 0.357421875,
      "learning_rate": 9.176463610019979e-05,
      "loss": 0.0432,
      "step": 11634
    },
    {
      "epoch": 2.868836291913215,
      "grad_norm": 0.34375,
      "learning_rate": 9.173032888342244e-05,
      "loss": 0.0346,
      "step": 11636
    },
    {
      "epoch": 2.869329388560158,
      "grad_norm": 0.341796875,
      "learning_rate": 9.169602264664728e-05,
      "loss": 0.0466,
      "step": 11638
    },
    {
      "epoch": 2.8698224852071004,
      "grad_norm": 0.3515625,
      "learning_rate": 9.166171739393983e-05,
      "loss": 0.0419,
      "step": 11640
    },
    {
      "epoch": 2.8703155818540433,
      "grad_norm": 0.283203125,
      "learning_rate": 9.16274131293654e-05,
      "loss": 0.0361,
      "step": 11642
    },
    {
      "epoch": 2.8708086785009863,
      "grad_norm": 0.333984375,
      "learning_rate": 9.159310985698923e-05,
      "loss": 0.0427,
      "step": 11644
    },
    {
      "epoch": 2.871301775147929,
      "grad_norm": 0.4453125,
      "learning_rate": 9.155880758087653e-05,
      "loss": 0.0431,
      "step": 11646
    },
    {
      "epoch": 2.871794871794872,
      "grad_norm": 0.3828125,
      "learning_rate": 9.152450630509221e-05,
      "loss": 0.038,
      "step": 11648
    },
    {
      "epoch": 2.8722879684418148,
      "grad_norm": 0.3984375,
      "learning_rate": 9.149020603370127e-05,
      "loss": 0.0373,
      "step": 11650
    },
    {
      "epoch": 2.8727810650887573,
      "grad_norm": 0.30078125,
      "learning_rate": 9.145590677076839e-05,
      "loss": 0.0388,
      "step": 11652
    },
    {
      "epoch": 2.8732741617357003,
      "grad_norm": 0.306640625,
      "learning_rate": 9.142160852035828e-05,
      "loss": 0.0412,
      "step": 11654
    },
    {
      "epoch": 2.8737672583826432,
      "grad_norm": 0.39453125,
      "learning_rate": 9.138731128653543e-05,
      "loss": 0.0434,
      "step": 11656
    },
    {
      "epoch": 2.8742603550295858,
      "grad_norm": 0.404296875,
      "learning_rate": 9.135301507336431e-05,
      "loss": 0.0412,
      "step": 11658
    },
    {
      "epoch": 2.8747534516765287,
      "grad_norm": 0.326171875,
      "learning_rate": 9.13187198849092e-05,
      "loss": 0.0396,
      "step": 11660
    },
    {
      "epoch": 2.8752465483234713,
      "grad_norm": 0.349609375,
      "learning_rate": 9.128442572523417e-05,
      "loss": 0.0435,
      "step": 11662
    },
    {
      "epoch": 2.8757396449704142,
      "grad_norm": 0.30859375,
      "learning_rate": 9.125013259840343e-05,
      "loss": 0.0452,
      "step": 11664
    },
    {
      "epoch": 2.8762327416173568,
      "grad_norm": 0.431640625,
      "learning_rate": 9.121584050848081e-05,
      "loss": 0.0397,
      "step": 11666
    },
    {
      "epoch": 2.8767258382642997,
      "grad_norm": 0.392578125,
      "learning_rate": 9.118154945953014e-05,
      "loss": 0.0399,
      "step": 11668
    },
    {
      "epoch": 2.8772189349112427,
      "grad_norm": 0.3515625,
      "learning_rate": 9.114725945561506e-05,
      "loss": 0.0428,
      "step": 11670
    },
    {
      "epoch": 2.8777120315581852,
      "grad_norm": 0.3515625,
      "learning_rate": 9.111297050079921e-05,
      "loss": 0.0398,
      "step": 11672
    },
    {
      "epoch": 2.878205128205128,
      "grad_norm": 0.31640625,
      "learning_rate": 9.107868259914592e-05,
      "loss": 0.0446,
      "step": 11674
    },
    {
      "epoch": 2.878698224852071,
      "grad_norm": 0.3984375,
      "learning_rate": 9.10443957547186e-05,
      "loss": 0.0438,
      "step": 11676
    },
    {
      "epoch": 2.8791913214990137,
      "grad_norm": 0.318359375,
      "learning_rate": 9.10101099715804e-05,
      "loss": 0.0403,
      "step": 11678
    },
    {
      "epoch": 2.8796844181459567,
      "grad_norm": 0.333984375,
      "learning_rate": 9.097582525379431e-05,
      "loss": 0.0416,
      "step": 11680
    },
    {
      "epoch": 2.8801775147928996,
      "grad_norm": 0.3203125,
      "learning_rate": 9.094154160542335e-05,
      "loss": 0.0405,
      "step": 11682
    },
    {
      "epoch": 2.880670611439842,
      "grad_norm": 0.41015625,
      "learning_rate": 9.090725903053025e-05,
      "loss": 0.0408,
      "step": 11684
    },
    {
      "epoch": 2.881163708086785,
      "grad_norm": 0.3046875,
      "learning_rate": 9.087297753317776e-05,
      "loss": 0.0402,
      "step": 11686
    },
    {
      "epoch": 2.8816568047337277,
      "grad_norm": 0.39453125,
      "learning_rate": 9.083869711742834e-05,
      "loss": 0.0376,
      "step": 11688
    },
    {
      "epoch": 2.8821499013806706,
      "grad_norm": 0.33203125,
      "learning_rate": 9.080441778734451e-05,
      "loss": 0.0404,
      "step": 11690
    },
    {
      "epoch": 2.882642998027613,
      "grad_norm": 0.455078125,
      "learning_rate": 9.077013954698851e-05,
      "loss": 0.0387,
      "step": 11692
    },
    {
      "epoch": 2.883136094674556,
      "grad_norm": 0.322265625,
      "learning_rate": 9.07358624004225e-05,
      "loss": 0.046,
      "step": 11694
    },
    {
      "epoch": 2.883629191321499,
      "grad_norm": 0.361328125,
      "learning_rate": 9.070158635170852e-05,
      "loss": 0.0403,
      "step": 11696
    },
    {
      "epoch": 2.8841222879684416,
      "grad_norm": 0.341796875,
      "learning_rate": 9.066731140490844e-05,
      "loss": 0.0371,
      "step": 11698
    },
    {
      "epoch": 2.8846153846153846,
      "grad_norm": 0.37109375,
      "learning_rate": 9.063303756408411e-05,
      "loss": 0.0416,
      "step": 11700
    },
    {
      "epoch": 2.8851084812623276,
      "grad_norm": 0.365234375,
      "learning_rate": 9.059876483329708e-05,
      "loss": 0.0458,
      "step": 11702
    },
    {
      "epoch": 2.88560157790927,
      "grad_norm": 0.390625,
      "learning_rate": 9.056449321660895e-05,
      "loss": 0.0406,
      "step": 11704
    },
    {
      "epoch": 2.886094674556213,
      "grad_norm": 0.296875,
      "learning_rate": 9.053022271808104e-05,
      "loss": 0.0393,
      "step": 11706
    },
    {
      "epoch": 2.886587771203156,
      "grad_norm": 0.361328125,
      "learning_rate": 9.049595334177463e-05,
      "loss": 0.0369,
      "step": 11708
    },
    {
      "epoch": 2.8870808678500985,
      "grad_norm": 0.349609375,
      "learning_rate": 9.04616850917508e-05,
      "loss": 0.0452,
      "step": 11710
    },
    {
      "epoch": 2.8875739644970415,
      "grad_norm": 0.3828125,
      "learning_rate": 9.042741797207052e-05,
      "loss": 0.0435,
      "step": 11712
    },
    {
      "epoch": 2.8880670611439845,
      "grad_norm": 0.33984375,
      "learning_rate": 9.039315198679469e-05,
      "loss": 0.0387,
      "step": 11714
    },
    {
      "epoch": 2.888560157790927,
      "grad_norm": 0.337890625,
      "learning_rate": 9.035888713998394e-05,
      "loss": 0.0431,
      "step": 11716
    },
    {
      "epoch": 2.88905325443787,
      "grad_norm": 0.380859375,
      "learning_rate": 9.032462343569894e-05,
      "loss": 0.0393,
      "step": 11718
    },
    {
      "epoch": 2.8895463510848125,
      "grad_norm": 0.310546875,
      "learning_rate": 9.029036087800005e-05,
      "loss": 0.0414,
      "step": 11720
    },
    {
      "epoch": 2.8900394477317555,
      "grad_norm": 0.3828125,
      "learning_rate": 9.025609947094766e-05,
      "loss": 0.0436,
      "step": 11722
    },
    {
      "epoch": 2.890532544378698,
      "grad_norm": 0.29296875,
      "learning_rate": 9.022183921860181e-05,
      "loss": 0.0436,
      "step": 11724
    },
    {
      "epoch": 2.891025641025641,
      "grad_norm": 0.330078125,
      "learning_rate": 9.018758012502267e-05,
      "loss": 0.0405,
      "step": 11726
    },
    {
      "epoch": 2.891518737672584,
      "grad_norm": 0.30078125,
      "learning_rate": 9.015332219427006e-05,
      "loss": 0.0411,
      "step": 11728
    },
    {
      "epoch": 2.8920118343195265,
      "grad_norm": 0.51171875,
      "learning_rate": 9.01190654304037e-05,
      "loss": 0.0439,
      "step": 11730
    },
    {
      "epoch": 2.8925049309664694,
      "grad_norm": 0.306640625,
      "learning_rate": 9.008480983748331e-05,
      "loss": 0.041,
      "step": 11732
    },
    {
      "epoch": 2.8929980276134124,
      "grad_norm": 0.296875,
      "learning_rate": 9.00505554195683e-05,
      "loss": 0.0408,
      "step": 11734
    },
    {
      "epoch": 2.893491124260355,
      "grad_norm": 0.31640625,
      "learning_rate": 9.001630218071803e-05,
      "loss": 0.0378,
      "step": 11736
    },
    {
      "epoch": 2.893984220907298,
      "grad_norm": 0.291015625,
      "learning_rate": 8.998205012499165e-05,
      "loss": 0.0413,
      "step": 11738
    },
    {
      "epoch": 2.894477317554241,
      "grad_norm": 0.32421875,
      "learning_rate": 8.994779925644831e-05,
      "loss": 0.0445,
      "step": 11740
    },
    {
      "epoch": 2.8949704142011834,
      "grad_norm": 0.28125,
      "learning_rate": 8.991354957914687e-05,
      "loss": 0.038,
      "step": 11742
    },
    {
      "epoch": 2.8954635108481264,
      "grad_norm": 0.41796875,
      "learning_rate": 8.98793010971461e-05,
      "loss": 0.0419,
      "step": 11744
    },
    {
      "epoch": 2.895956607495069,
      "grad_norm": 0.314453125,
      "learning_rate": 8.984505381450468e-05,
      "loss": 0.0411,
      "step": 11746
    },
    {
      "epoch": 2.896449704142012,
      "grad_norm": 0.359375,
      "learning_rate": 8.981080773528107e-05,
      "loss": 0.0431,
      "step": 11748
    },
    {
      "epoch": 2.8969428007889544,
      "grad_norm": 0.353515625,
      "learning_rate": 8.977656286353366e-05,
      "loss": 0.0394,
      "step": 11750
    },
    {
      "epoch": 2.8974358974358974,
      "grad_norm": 0.35546875,
      "learning_rate": 8.974231920332056e-05,
      "loss": 0.038,
      "step": 11752
    },
    {
      "epoch": 2.8979289940828403,
      "grad_norm": 0.35546875,
      "learning_rate": 8.970807675869997e-05,
      "loss": 0.0414,
      "step": 11754
    },
    {
      "epoch": 2.898422090729783,
      "grad_norm": 0.330078125,
      "learning_rate": 8.967383553372971e-05,
      "loss": 0.0416,
      "step": 11756
    },
    {
      "epoch": 2.898915187376726,
      "grad_norm": 0.322265625,
      "learning_rate": 8.963959553246763e-05,
      "loss": 0.039,
      "step": 11758
    },
    {
      "epoch": 2.899408284023669,
      "grad_norm": 0.283203125,
      "learning_rate": 8.960535675897133e-05,
      "loss": 0.044,
      "step": 11760
    },
    {
      "epoch": 2.8999013806706113,
      "grad_norm": 0.458984375,
      "learning_rate": 8.957111921729826e-05,
      "loss": 0.0421,
      "step": 11762
    },
    {
      "epoch": 2.9003944773175543,
      "grad_norm": 0.38671875,
      "learning_rate": 8.953688291150582e-05,
      "loss": 0.0449,
      "step": 11764
    },
    {
      "epoch": 2.9008875739644973,
      "grad_norm": 0.294921875,
      "learning_rate": 8.950264784565112e-05,
      "loss": 0.0363,
      "step": 11766
    },
    {
      "epoch": 2.90138067061144,
      "grad_norm": 0.28515625,
      "learning_rate": 8.94684140237913e-05,
      "loss": 0.0393,
      "step": 11768
    },
    {
      "epoch": 2.9018737672583828,
      "grad_norm": 0.484375,
      "learning_rate": 8.943418144998318e-05,
      "loss": 0.0397,
      "step": 11770
    },
    {
      "epoch": 2.9023668639053253,
      "grad_norm": 0.322265625,
      "learning_rate": 8.93999501282836e-05,
      "loss": 0.0341,
      "step": 11772
    },
    {
      "epoch": 2.9028599605522682,
      "grad_norm": 0.3125,
      "learning_rate": 8.936572006274911e-05,
      "loss": 0.0418,
      "step": 11774
    },
    {
      "epoch": 2.9033530571992108,
      "grad_norm": 0.33984375,
      "learning_rate": 8.933149125743615e-05,
      "loss": 0.0416,
      "step": 11776
    },
    {
      "epoch": 2.9038461538461537,
      "grad_norm": 0.482421875,
      "learning_rate": 8.929726371640106e-05,
      "loss": 0.0421,
      "step": 11778
    },
    {
      "epoch": 2.9043392504930967,
      "grad_norm": 0.39453125,
      "learning_rate": 8.926303744369993e-05,
      "loss": 0.0496,
      "step": 11780
    },
    {
      "epoch": 2.9048323471400392,
      "grad_norm": 0.306640625,
      "learning_rate": 8.922881244338886e-05,
      "loss": 0.039,
      "step": 11782
    },
    {
      "epoch": 2.905325443786982,
      "grad_norm": 0.296875,
      "learning_rate": 8.91945887195236e-05,
      "loss": 0.0414,
      "step": 11784
    },
    {
      "epoch": 2.905818540433925,
      "grad_norm": 0.298828125,
      "learning_rate": 8.916036627615995e-05,
      "loss": 0.0417,
      "step": 11786
    },
    {
      "epoch": 2.9063116370808677,
      "grad_norm": 0.330078125,
      "learning_rate": 8.91261451173534e-05,
      "loss": 0.041,
      "step": 11788
    },
    {
      "epoch": 2.9068047337278107,
      "grad_norm": 0.318359375,
      "learning_rate": 8.90919252471594e-05,
      "loss": 0.042,
      "step": 11790
    },
    {
      "epoch": 2.9072978303747536,
      "grad_norm": 0.32421875,
      "learning_rate": 8.905770666963314e-05,
      "loss": 0.043,
      "step": 11792
    },
    {
      "epoch": 2.907790927021696,
      "grad_norm": 0.302734375,
      "learning_rate": 8.902348938882969e-05,
      "loss": 0.0409,
      "step": 11794
    },
    {
      "epoch": 2.908284023668639,
      "grad_norm": 0.3203125,
      "learning_rate": 8.898927340880409e-05,
      "loss": 0.0375,
      "step": 11796
    },
    {
      "epoch": 2.908777120315582,
      "grad_norm": 0.291015625,
      "learning_rate": 8.895505873361102e-05,
      "loss": 0.0412,
      "step": 11798
    },
    {
      "epoch": 2.9092702169625246,
      "grad_norm": 0.30859375,
      "learning_rate": 8.892084536730519e-05,
      "loss": 0.0397,
      "step": 11800
    },
    {
      "epoch": 2.9097633136094676,
      "grad_norm": 0.318359375,
      "learning_rate": 8.888663331394103e-05,
      "loss": 0.0431,
      "step": 11802
    },
    {
      "epoch": 2.91025641025641,
      "grad_norm": 0.3046875,
      "learning_rate": 8.885242257757289e-05,
      "loss": 0.0398,
      "step": 11804
    },
    {
      "epoch": 2.910749506903353,
      "grad_norm": 0.2734375,
      "learning_rate": 8.881821316225489e-05,
      "loss": 0.0417,
      "step": 11806
    },
    {
      "epoch": 2.9112426035502956,
      "grad_norm": 0.2734375,
      "learning_rate": 8.878400507204108e-05,
      "loss": 0.041,
      "step": 11808
    },
    {
      "epoch": 2.9117357001972386,
      "grad_norm": 0.28125,
      "learning_rate": 8.87497983109853e-05,
      "loss": 0.0347,
      "step": 11810
    },
    {
      "epoch": 2.9122287968441816,
      "grad_norm": 0.30859375,
      "learning_rate": 8.871559288314119e-05,
      "loss": 0.0425,
      "step": 11812
    },
    {
      "epoch": 2.912721893491124,
      "grad_norm": 0.32421875,
      "learning_rate": 8.868138879256235e-05,
      "loss": 0.0377,
      "step": 11814
    },
    {
      "epoch": 2.913214990138067,
      "grad_norm": 0.330078125,
      "learning_rate": 8.864718604330215e-05,
      "loss": 0.0411,
      "step": 11816
    },
    {
      "epoch": 2.91370808678501,
      "grad_norm": 0.310546875,
      "learning_rate": 8.861298463941378e-05,
      "loss": 0.039,
      "step": 11818
    },
    {
      "epoch": 2.9142011834319526,
      "grad_norm": 0.333984375,
      "learning_rate": 8.857878458495028e-05,
      "loss": 0.0408,
      "step": 11820
    },
    {
      "epoch": 2.9146942800788955,
      "grad_norm": 0.3359375,
      "learning_rate": 8.854458588396462e-05,
      "loss": 0.0415,
      "step": 11822
    },
    {
      "epoch": 2.9151873767258385,
      "grad_norm": 0.287109375,
      "learning_rate": 8.851038854050948e-05,
      "loss": 0.0373,
      "step": 11824
    },
    {
      "epoch": 2.915680473372781,
      "grad_norm": 0.34375,
      "learning_rate": 8.847619255863742e-05,
      "loss": 0.0392,
      "step": 11826
    },
    {
      "epoch": 2.916173570019724,
      "grad_norm": 0.328125,
      "learning_rate": 8.84419979424009e-05,
      "loss": 0.0412,
      "step": 11828
    },
    {
      "epoch": 2.9166666666666665,
      "grad_norm": 0.34765625,
      "learning_rate": 8.840780469585216e-05,
      "loss": 0.0383,
      "step": 11830
    },
    {
      "epoch": 2.9171597633136095,
      "grad_norm": 0.283203125,
      "learning_rate": 8.837361282304329e-05,
      "loss": 0.0402,
      "step": 11832
    },
    {
      "epoch": 2.917652859960552,
      "grad_norm": 0.3203125,
      "learning_rate": 8.833942232802618e-05,
      "loss": 0.0415,
      "step": 11834
    },
    {
      "epoch": 2.918145956607495,
      "grad_norm": 0.349609375,
      "learning_rate": 8.830523321485268e-05,
      "loss": 0.0384,
      "step": 11836
    },
    {
      "epoch": 2.918639053254438,
      "grad_norm": 0.345703125,
      "learning_rate": 8.827104548757427e-05,
      "loss": 0.0381,
      "step": 11838
    },
    {
      "epoch": 2.9191321499013805,
      "grad_norm": 0.279296875,
      "learning_rate": 8.823685915024253e-05,
      "loss": 0.0419,
      "step": 11840
    },
    {
      "epoch": 2.9196252465483234,
      "grad_norm": 0.3203125,
      "learning_rate": 8.820267420690864e-05,
      "loss": 0.0432,
      "step": 11842
    },
    {
      "epoch": 2.9201183431952664,
      "grad_norm": 0.296875,
      "learning_rate": 8.81684906616237e-05,
      "loss": 0.0405,
      "step": 11844
    },
    {
      "epoch": 2.920611439842209,
      "grad_norm": 0.28515625,
      "learning_rate": 8.813430851843871e-05,
      "loss": 0.0398,
      "step": 11846
    },
    {
      "epoch": 2.921104536489152,
      "grad_norm": 0.33984375,
      "learning_rate": 8.810012778140437e-05,
      "loss": 0.0395,
      "step": 11848
    },
    {
      "epoch": 2.921597633136095,
      "grad_norm": 0.287109375,
      "learning_rate": 8.806594845457134e-05,
      "loss": 0.0417,
      "step": 11850
    },
    {
      "epoch": 2.9220907297830374,
      "grad_norm": 0.283203125,
      "learning_rate": 8.803177054199003e-05,
      "loss": 0.0391,
      "step": 11852
    },
    {
      "epoch": 2.9225838264299804,
      "grad_norm": 0.33203125,
      "learning_rate": 8.799759404771078e-05,
      "loss": 0.0418,
      "step": 11854
    },
    {
      "epoch": 2.9230769230769234,
      "grad_norm": 0.3125,
      "learning_rate": 8.796341897578362e-05,
      "loss": 0.0372,
      "step": 11856
    },
    {
      "epoch": 2.923570019723866,
      "grad_norm": 0.353515625,
      "learning_rate": 8.792924533025854e-05,
      "loss": 0.0418,
      "step": 11858
    },
    {
      "epoch": 2.9240631163708084,
      "grad_norm": 0.349609375,
      "learning_rate": 8.789507311518528e-05,
      "loss": 0.0435,
      "step": 11860
    },
    {
      "epoch": 2.9245562130177514,
      "grad_norm": 0.314453125,
      "learning_rate": 8.786090233461339e-05,
      "loss": 0.0464,
      "step": 11862
    },
    {
      "epoch": 2.9250493096646943,
      "grad_norm": 0.2734375,
      "learning_rate": 8.78267329925924e-05,
      "loss": 0.0376,
      "step": 11864
    },
    {
      "epoch": 2.925542406311637,
      "grad_norm": 0.365234375,
      "learning_rate": 8.779256509317148e-05,
      "loss": 0.0413,
      "step": 11866
    },
    {
      "epoch": 2.92603550295858,
      "grad_norm": 0.255859375,
      "learning_rate": 8.775839864039978e-05,
      "loss": 0.0399,
      "step": 11868
    },
    {
      "epoch": 2.926528599605523,
      "grad_norm": 0.392578125,
      "learning_rate": 8.772423363832618e-05,
      "loss": 0.0438,
      "step": 11870
    },
    {
      "epoch": 2.9270216962524653,
      "grad_norm": 0.298828125,
      "learning_rate": 8.769007009099946e-05,
      "loss": 0.0426,
      "step": 11872
    },
    {
      "epoch": 2.9275147928994083,
      "grad_norm": 0.27734375,
      "learning_rate": 8.765590800246816e-05,
      "loss": 0.0383,
      "step": 11874
    },
    {
      "epoch": 2.9280078895463513,
      "grad_norm": 0.39453125,
      "learning_rate": 8.762174737678063e-05,
      "loss": 0.0434,
      "step": 11876
    },
    {
      "epoch": 2.928500986193294,
      "grad_norm": 0.365234375,
      "learning_rate": 8.75875882179852e-05,
      "loss": 0.0412,
      "step": 11878
    },
    {
      "epoch": 2.9289940828402368,
      "grad_norm": 0.33984375,
      "learning_rate": 8.755343053012982e-05,
      "loss": 0.0434,
      "step": 11880
    },
    {
      "epoch": 2.9294871794871797,
      "grad_norm": 0.33203125,
      "learning_rate": 8.751927431726245e-05,
      "loss": 0.0383,
      "step": 11882
    },
    {
      "epoch": 2.9299802761341223,
      "grad_norm": 0.3046875,
      "learning_rate": 8.748511958343076e-05,
      "loss": 0.0394,
      "step": 11884
    },
    {
      "epoch": 2.9304733727810652,
      "grad_norm": 0.4453125,
      "learning_rate": 8.745096633268227e-05,
      "loss": 0.0421,
      "step": 11886
    },
    {
      "epoch": 2.9309664694280078,
      "grad_norm": 0.333984375,
      "learning_rate": 8.74168145690643e-05,
      "loss": 0.0397,
      "step": 11888
    },
    {
      "epoch": 2.9314595660749507,
      "grad_norm": 0.44140625,
      "learning_rate": 8.738266429662411e-05,
      "loss": 0.0472,
      "step": 11890
    },
    {
      "epoch": 2.9319526627218933,
      "grad_norm": 0.30859375,
      "learning_rate": 8.734851551940863e-05,
      "loss": 0.0422,
      "step": 11892
    },
    {
      "epoch": 2.9324457593688362,
      "grad_norm": 0.5,
      "learning_rate": 8.731436824146468e-05,
      "loss": 0.04,
      "step": 11894
    },
    {
      "epoch": 2.932938856015779,
      "grad_norm": 0.52734375,
      "learning_rate": 8.728022246683894e-05,
      "loss": 0.0415,
      "step": 11896
    },
    {
      "epoch": 2.9334319526627217,
      "grad_norm": 0.314453125,
      "learning_rate": 8.724607819957785e-05,
      "loss": 0.04,
      "step": 11898
    },
    {
      "epoch": 2.9339250493096647,
      "grad_norm": 0.30859375,
      "learning_rate": 8.721193544372772e-05,
      "loss": 0.0422,
      "step": 11900
    },
    {
      "epoch": 2.9344181459566077,
      "grad_norm": 0.302734375,
      "learning_rate": 8.717779420333459e-05,
      "loss": 0.0417,
      "step": 11902
    },
    {
      "epoch": 2.93491124260355,
      "grad_norm": 0.302734375,
      "learning_rate": 8.714365448244448e-05,
      "loss": 0.045,
      "step": 11904
    },
    {
      "epoch": 2.935404339250493,
      "grad_norm": 0.47265625,
      "learning_rate": 8.71095162851031e-05,
      "loss": 0.0411,
      "step": 11906
    },
    {
      "epoch": 2.935897435897436,
      "grad_norm": 0.306640625,
      "learning_rate": 8.707537961535597e-05,
      "loss": 0.0385,
      "step": 11908
    },
    {
      "epoch": 2.9363905325443787,
      "grad_norm": 0.3828125,
      "learning_rate": 8.704124447724855e-05,
      "loss": 0.0445,
      "step": 11910
    },
    {
      "epoch": 2.9368836291913216,
      "grad_norm": 0.306640625,
      "learning_rate": 8.7007110874826e-05,
      "loss": 0.0397,
      "step": 11912
    },
    {
      "epoch": 2.937376725838264,
      "grad_norm": 0.287109375,
      "learning_rate": 8.697297881213338e-05,
      "loss": 0.0416,
      "step": 11914
    },
    {
      "epoch": 2.937869822485207,
      "grad_norm": 0.267578125,
      "learning_rate": 8.693884829321544e-05,
      "loss": 0.0366,
      "step": 11916
    },
    {
      "epoch": 2.9383629191321496,
      "grad_norm": 0.294921875,
      "learning_rate": 8.690471932211697e-05,
      "loss": 0.0385,
      "step": 11918
    },
    {
      "epoch": 2.9388560157790926,
      "grad_norm": 0.3359375,
      "learning_rate": 8.687059190288231e-05,
      "loss": 0.0442,
      "step": 11920
    },
    {
      "epoch": 2.9393491124260356,
      "grad_norm": 0.29296875,
      "learning_rate": 8.683646603955587e-05,
      "loss": 0.038,
      "step": 11922
    },
    {
      "epoch": 2.939842209072978,
      "grad_norm": 0.30078125,
      "learning_rate": 8.68023417361817e-05,
      "loss": 0.0407,
      "step": 11924
    },
    {
      "epoch": 2.940335305719921,
      "grad_norm": 0.365234375,
      "learning_rate": 8.676821899680369e-05,
      "loss": 0.0426,
      "step": 11926
    },
    {
      "epoch": 2.940828402366864,
      "grad_norm": 0.330078125,
      "learning_rate": 8.673409782546564e-05,
      "loss": 0.0407,
      "step": 11928
    },
    {
      "epoch": 2.9413214990138066,
      "grad_norm": 0.365234375,
      "learning_rate": 8.669997822621102e-05,
      "loss": 0.0408,
      "step": 11930
    },
    {
      "epoch": 2.9418145956607495,
      "grad_norm": 0.3046875,
      "learning_rate": 8.666586020308328e-05,
      "loss": 0.0382,
      "step": 11932
    },
    {
      "epoch": 2.9423076923076925,
      "grad_norm": 0.34375,
      "learning_rate": 8.66317437601255e-05,
      "loss": 0.0424,
      "step": 11934
    },
    {
      "epoch": 2.942800788954635,
      "grad_norm": 0.392578125,
      "learning_rate": 8.659762890138079e-05,
      "loss": 0.0422,
      "step": 11936
    },
    {
      "epoch": 2.943293885601578,
      "grad_norm": 0.345703125,
      "learning_rate": 8.656351563089184e-05,
      "loss": 0.0406,
      "step": 11938
    },
    {
      "epoch": 2.943786982248521,
      "grad_norm": 0.30859375,
      "learning_rate": 8.652940395270134e-05,
      "loss": 0.0431,
      "step": 11940
    },
    {
      "epoch": 2.9442800788954635,
      "grad_norm": 0.34765625,
      "learning_rate": 8.649529387085168e-05,
      "loss": 0.0399,
      "step": 11942
    },
    {
      "epoch": 2.9447731755424065,
      "grad_norm": 0.361328125,
      "learning_rate": 8.646118538938506e-05,
      "loss": 0.044,
      "step": 11944
    },
    {
      "epoch": 2.945266272189349,
      "grad_norm": 0.30078125,
      "learning_rate": 8.642707851234359e-05,
      "loss": 0.0392,
      "step": 11946
    },
    {
      "epoch": 2.945759368836292,
      "grad_norm": 0.341796875,
      "learning_rate": 8.639297324376906e-05,
      "loss": 0.043,
      "step": 11948
    },
    {
      "epoch": 2.9462524654832345,
      "grad_norm": 0.294921875,
      "learning_rate": 8.635886958770323e-05,
      "loss": 0.0394,
      "step": 11950
    },
    {
      "epoch": 2.9467455621301775,
      "grad_norm": 0.294921875,
      "learning_rate": 8.632476754818748e-05,
      "loss": 0.0407,
      "step": 11952
    },
    {
      "epoch": 2.9472386587771204,
      "grad_norm": 0.30078125,
      "learning_rate": 8.629066712926314e-05,
      "loss": 0.0403,
      "step": 11954
    },
    {
      "epoch": 2.947731755424063,
      "grad_norm": 0.302734375,
      "learning_rate": 8.625656833497128e-05,
      "loss": 0.0386,
      "step": 11956
    },
    {
      "epoch": 2.948224852071006,
      "grad_norm": 0.439453125,
      "learning_rate": 8.622247116935282e-05,
      "loss": 0.0478,
      "step": 11958
    },
    {
      "epoch": 2.948717948717949,
      "grad_norm": 0.322265625,
      "learning_rate": 8.618837563644844e-05,
      "loss": 0.0394,
      "step": 11960
    },
    {
      "epoch": 2.9492110453648914,
      "grad_norm": 0.4296875,
      "learning_rate": 8.615428174029864e-05,
      "loss": 0.042,
      "step": 11962
    },
    {
      "epoch": 2.9497041420118344,
      "grad_norm": 0.294921875,
      "learning_rate": 8.61201894849438e-05,
      "loss": 0.0409,
      "step": 11964
    },
    {
      "epoch": 2.9501972386587774,
      "grad_norm": 0.3125,
      "learning_rate": 8.608609887442398e-05,
      "loss": 0.0417,
      "step": 11966
    },
    {
      "epoch": 2.95069033530572,
      "grad_norm": 0.306640625,
      "learning_rate": 8.605200991277916e-05,
      "loss": 0.0441,
      "step": 11968
    },
    {
      "epoch": 2.951183431952663,
      "grad_norm": 0.37109375,
      "learning_rate": 8.601792260404897e-05,
      "loss": 0.0444,
      "step": 11970
    },
    {
      "epoch": 2.9516765285996054,
      "grad_norm": 0.40234375,
      "learning_rate": 8.598383695227308e-05,
      "loss": 0.0435,
      "step": 11972
    },
    {
      "epoch": 2.9521696252465484,
      "grad_norm": 0.29296875,
      "learning_rate": 8.594975296149076e-05,
      "loss": 0.0412,
      "step": 11974
    },
    {
      "epoch": 2.952662721893491,
      "grad_norm": 0.36328125,
      "learning_rate": 8.591567063574112e-05,
      "loss": 0.0455,
      "step": 11976
    },
    {
      "epoch": 2.953155818540434,
      "grad_norm": 0.439453125,
      "learning_rate": 8.588158997906319e-05,
      "loss": 0.0405,
      "step": 11978
    },
    {
      "epoch": 2.953648915187377,
      "grad_norm": 0.43359375,
      "learning_rate": 8.584751099549565e-05,
      "loss": 0.0436,
      "step": 11980
    },
    {
      "epoch": 2.9541420118343193,
      "grad_norm": 0.287109375,
      "learning_rate": 8.581343368907708e-05,
      "loss": 0.0364,
      "step": 11982
    },
    {
      "epoch": 2.9546351084812623,
      "grad_norm": 0.455078125,
      "learning_rate": 8.577935806384583e-05,
      "loss": 0.0397,
      "step": 11984
    },
    {
      "epoch": 2.9551282051282053,
      "grad_norm": 0.30859375,
      "learning_rate": 8.574528412384006e-05,
      "loss": 0.0414,
      "step": 11986
    },
    {
      "epoch": 2.955621301775148,
      "grad_norm": 0.373046875,
      "learning_rate": 8.571121187309766e-05,
      "loss": 0.0399,
      "step": 11988
    },
    {
      "epoch": 2.956114398422091,
      "grad_norm": 0.357421875,
      "learning_rate": 8.567714131565648e-05,
      "loss": 0.035,
      "step": 11990
    },
    {
      "epoch": 2.9566074950690338,
      "grad_norm": 0.36328125,
      "learning_rate": 8.564307245555403e-05,
      "loss": 0.041,
      "step": 11992
    },
    {
      "epoch": 2.9571005917159763,
      "grad_norm": 0.28515625,
      "learning_rate": 8.560900529682763e-05,
      "loss": 0.0431,
      "step": 11994
    },
    {
      "epoch": 2.9575936883629192,
      "grad_norm": 0.404296875,
      "learning_rate": 8.557493984351445e-05,
      "loss": 0.0387,
      "step": 11996
    },
    {
      "epoch": 2.9580867850098618,
      "grad_norm": 0.3671875,
      "learning_rate": 8.554087609965143e-05,
      "loss": 0.0423,
      "step": 11998
    },
    {
      "epoch": 2.9585798816568047,
      "grad_norm": 0.384765625,
      "learning_rate": 8.550681406927535e-05,
      "loss": 0.051,
      "step": 12000
    },
    {
      "epoch": 2.9590729783037473,
      "grad_norm": 0.3046875,
      "learning_rate": 8.547275375642267e-05,
      "loss": 0.0409,
      "step": 12002
    },
    {
      "epoch": 2.9595660749506902,
      "grad_norm": 0.287109375,
      "learning_rate": 8.543869516512984e-05,
      "loss": 0.0429,
      "step": 12004
    },
    {
      "epoch": 2.960059171597633,
      "grad_norm": 0.33984375,
      "learning_rate": 8.540463829943292e-05,
      "loss": 0.0457,
      "step": 12006
    },
    {
      "epoch": 2.9605522682445757,
      "grad_norm": 0.52734375,
      "learning_rate": 8.537058316336783e-05,
      "loss": 0.0438,
      "step": 12008
    },
    {
      "epoch": 2.9610453648915187,
      "grad_norm": 0.330078125,
      "learning_rate": 8.533652976097032e-05,
      "loss": 0.0426,
      "step": 12010
    },
    {
      "epoch": 2.9615384615384617,
      "grad_norm": 0.34765625,
      "learning_rate": 8.53024780962759e-05,
      "loss": 0.0442,
      "step": 12012
    },
    {
      "epoch": 2.962031558185404,
      "grad_norm": 0.3515625,
      "learning_rate": 8.526842817331992e-05,
      "loss": 0.0401,
      "step": 12014
    },
    {
      "epoch": 2.962524654832347,
      "grad_norm": 0.349609375,
      "learning_rate": 8.52343799961374e-05,
      "loss": 0.0436,
      "step": 12016
    },
    {
      "epoch": 2.96301775147929,
      "grad_norm": 0.36328125,
      "learning_rate": 8.520033356876332e-05,
      "loss": 0.0411,
      "step": 12018
    },
    {
      "epoch": 2.9635108481262327,
      "grad_norm": 0.30859375,
      "learning_rate": 8.516628889523234e-05,
      "loss": 0.0405,
      "step": 12020
    },
    {
      "epoch": 2.9640039447731756,
      "grad_norm": 0.4765625,
      "learning_rate": 8.513224597957893e-05,
      "loss": 0.0438,
      "step": 12022
    },
    {
      "epoch": 2.9644970414201186,
      "grad_norm": 0.298828125,
      "learning_rate": 8.509820482583742e-05,
      "loss": 0.0397,
      "step": 12024
    },
    {
      "epoch": 2.964990138067061,
      "grad_norm": 0.322265625,
      "learning_rate": 8.506416543804182e-05,
      "loss": 0.0406,
      "step": 12026
    },
    {
      "epoch": 2.965483234714004,
      "grad_norm": 0.361328125,
      "learning_rate": 8.503012782022602e-05,
      "loss": 0.0398,
      "step": 12028
    },
    {
      "epoch": 2.9659763313609466,
      "grad_norm": 0.30078125,
      "learning_rate": 8.499609197642361e-05,
      "loss": 0.0412,
      "step": 12030
    },
    {
      "epoch": 2.9664694280078896,
      "grad_norm": 0.28515625,
      "learning_rate": 8.496205791066811e-05,
      "loss": 0.0406,
      "step": 12032
    },
    {
      "epoch": 2.966962524654832,
      "grad_norm": 0.34765625,
      "learning_rate": 8.49280256269927e-05,
      "loss": 0.0378,
      "step": 12034
    },
    {
      "epoch": 2.967455621301775,
      "grad_norm": 0.3125,
      "learning_rate": 8.48939951294304e-05,
      "loss": 0.0399,
      "step": 12036
    },
    {
      "epoch": 2.967948717948718,
      "grad_norm": 0.291015625,
      "learning_rate": 8.485996642201404e-05,
      "loss": 0.0395,
      "step": 12038
    },
    {
      "epoch": 2.9684418145956606,
      "grad_norm": 0.330078125,
      "learning_rate": 8.482593950877616e-05,
      "loss": 0.0363,
      "step": 12040
    },
    {
      "epoch": 2.9689349112426036,
      "grad_norm": 0.3515625,
      "learning_rate": 8.47919143937492e-05,
      "loss": 0.0419,
      "step": 12042
    },
    {
      "epoch": 2.9694280078895465,
      "grad_norm": 0.33203125,
      "learning_rate": 8.475789108096526e-05,
      "loss": 0.0489,
      "step": 12044
    },
    {
      "epoch": 2.969921104536489,
      "grad_norm": 0.2890625,
      "learning_rate": 8.472386957445635e-05,
      "loss": 0.0393,
      "step": 12046
    },
    {
      "epoch": 2.970414201183432,
      "grad_norm": 0.294921875,
      "learning_rate": 8.468984987825417e-05,
      "loss": 0.038,
      "step": 12048
    },
    {
      "epoch": 2.970907297830375,
      "grad_norm": 0.28125,
      "learning_rate": 8.465583199639028e-05,
      "loss": 0.0366,
      "step": 12050
    },
    {
      "epoch": 2.9714003944773175,
      "grad_norm": 0.376953125,
      "learning_rate": 8.462181593289596e-05,
      "loss": 0.0417,
      "step": 12052
    },
    {
      "epoch": 2.9718934911242605,
      "grad_norm": 0.29296875,
      "learning_rate": 8.458780169180232e-05,
      "loss": 0.0372,
      "step": 12054
    },
    {
      "epoch": 2.972386587771203,
      "grad_norm": 0.34765625,
      "learning_rate": 8.455378927714024e-05,
      "loss": 0.0407,
      "step": 12056
    },
    {
      "epoch": 2.972879684418146,
      "grad_norm": 0.2890625,
      "learning_rate": 8.451977869294034e-05,
      "loss": 0.0349,
      "step": 12058
    },
    {
      "epoch": 2.9733727810650885,
      "grad_norm": 0.37890625,
      "learning_rate": 8.448576994323311e-05,
      "loss": 0.0419,
      "step": 12060
    },
    {
      "epoch": 2.9738658777120315,
      "grad_norm": 0.302734375,
      "learning_rate": 8.445176303204875e-05,
      "loss": 0.0419,
      "step": 12062
    },
    {
      "epoch": 2.9743589743589745,
      "grad_norm": 0.314453125,
      "learning_rate": 8.44177579634173e-05,
      "loss": 0.0434,
      "step": 12064
    },
    {
      "epoch": 2.974852071005917,
      "grad_norm": 0.294921875,
      "learning_rate": 8.438375474136853e-05,
      "loss": 0.0383,
      "step": 12066
    },
    {
      "epoch": 2.97534516765286,
      "grad_norm": 0.30078125,
      "learning_rate": 8.434975336993202e-05,
      "loss": 0.039,
      "step": 12068
    },
    {
      "epoch": 2.975838264299803,
      "grad_norm": 0.294921875,
      "learning_rate": 8.431575385313708e-05,
      "loss": 0.0419,
      "step": 12070
    },
    {
      "epoch": 2.9763313609467454,
      "grad_norm": 0.291015625,
      "learning_rate": 8.428175619501291e-05,
      "loss": 0.0407,
      "step": 12072
    },
    {
      "epoch": 2.9768244575936884,
      "grad_norm": 0.34765625,
      "learning_rate": 8.424776039958841e-05,
      "loss": 0.0374,
      "step": 12074
    },
    {
      "epoch": 2.9773175542406314,
      "grad_norm": 0.33984375,
      "learning_rate": 8.421376647089218e-05,
      "loss": 0.0394,
      "step": 12076
    },
    {
      "epoch": 2.977810650887574,
      "grad_norm": 0.28125,
      "learning_rate": 8.417977441295283e-05,
      "loss": 0.0372,
      "step": 12078
    },
    {
      "epoch": 2.978303747534517,
      "grad_norm": 0.29296875,
      "learning_rate": 8.41457842297985e-05,
      "loss": 0.0361,
      "step": 12080
    },
    {
      "epoch": 2.9787968441814594,
      "grad_norm": 0.35546875,
      "learning_rate": 8.41117959254573e-05,
      "loss": 0.0427,
      "step": 12082
    },
    {
      "epoch": 2.9792899408284024,
      "grad_norm": 0.330078125,
      "learning_rate": 8.407780950395694e-05,
      "loss": 0.0405,
      "step": 12084
    },
    {
      "epoch": 2.979783037475345,
      "grad_norm": 0.302734375,
      "learning_rate": 8.40438249693251e-05,
      "loss": 0.0414,
      "step": 12086
    },
    {
      "epoch": 2.980276134122288,
      "grad_norm": 0.310546875,
      "learning_rate": 8.400984232558908e-05,
      "loss": 0.0402,
      "step": 12088
    },
    {
      "epoch": 2.980769230769231,
      "grad_norm": 0.298828125,
      "learning_rate": 8.397586157677598e-05,
      "loss": 0.0435,
      "step": 12090
    },
    {
      "epoch": 2.9812623274161734,
      "grad_norm": 0.2890625,
      "learning_rate": 8.39418827269128e-05,
      "loss": 0.0385,
      "step": 12092
    },
    {
      "epoch": 2.9817554240631163,
      "grad_norm": 0.369140625,
      "learning_rate": 8.390790578002615e-05,
      "loss": 0.0387,
      "step": 12094
    },
    {
      "epoch": 2.9822485207100593,
      "grad_norm": 0.427734375,
      "learning_rate": 8.387393074014252e-05,
      "loss": 0.0432,
      "step": 12096
    },
    {
      "epoch": 2.982741617357002,
      "grad_norm": 0.341796875,
      "learning_rate": 8.383995761128811e-05,
      "loss": 0.039,
      "step": 12098
    },
    {
      "epoch": 2.983234714003945,
      "grad_norm": 0.3125,
      "learning_rate": 8.380598639748897e-05,
      "loss": 0.0434,
      "step": 12100
    },
    {
      "epoch": 2.9837278106508878,
      "grad_norm": 0.318359375,
      "learning_rate": 8.377201710277084e-05,
      "loss": 0.0413,
      "step": 12102
    },
    {
      "epoch": 2.9842209072978303,
      "grad_norm": 0.419921875,
      "learning_rate": 8.373804973115931e-05,
      "loss": 0.0463,
      "step": 12104
    },
    {
      "epoch": 2.9847140039447733,
      "grad_norm": 0.37890625,
      "learning_rate": 8.370408428667971e-05,
      "loss": 0.0431,
      "step": 12106
    },
    {
      "epoch": 2.9852071005917162,
      "grad_norm": 0.326171875,
      "learning_rate": 8.367012077335706e-05,
      "loss": 0.0415,
      "step": 12108
    },
    {
      "epoch": 2.9857001972386588,
      "grad_norm": 0.3125,
      "learning_rate": 8.36361591952163e-05,
      "loss": 0.0393,
      "step": 12110
    },
    {
      "epoch": 2.9861932938856017,
      "grad_norm": 0.3125,
      "learning_rate": 8.3602199556282e-05,
      "loss": 0.0371,
      "step": 12112
    },
    {
      "epoch": 2.9866863905325443,
      "grad_norm": 0.30859375,
      "learning_rate": 8.356824186057864e-05,
      "loss": 0.035,
      "step": 12114
    },
    {
      "epoch": 2.9871794871794872,
      "grad_norm": 0.3203125,
      "learning_rate": 8.353428611213033e-05,
      "loss": 0.0418,
      "step": 12116
    },
    {
      "epoch": 2.9876725838264298,
      "grad_norm": 0.357421875,
      "learning_rate": 8.35003323149611e-05,
      "loss": 0.0355,
      "step": 12118
    },
    {
      "epoch": 2.9881656804733727,
      "grad_norm": 0.380859375,
      "learning_rate": 8.346638047309458e-05,
      "loss": 0.0376,
      "step": 12120
    },
    {
      "epoch": 2.9886587771203157,
      "grad_norm": 0.306640625,
      "learning_rate": 8.34324305905543e-05,
      "loss": 0.0422,
      "step": 12122
    },
    {
      "epoch": 2.989151873767258,
      "grad_norm": 0.32421875,
      "learning_rate": 8.339848267136349e-05,
      "loss": 0.0393,
      "step": 12124
    },
    {
      "epoch": 2.989644970414201,
      "grad_norm": 0.38671875,
      "learning_rate": 8.336453671954516e-05,
      "loss": 0.0429,
      "step": 12126
    },
    {
      "epoch": 2.990138067061144,
      "grad_norm": 0.34765625,
      "learning_rate": 8.333059273912214e-05,
      "loss": 0.0395,
      "step": 12128
    },
    {
      "epoch": 2.9906311637080867,
      "grad_norm": 0.40625,
      "learning_rate": 8.32966507341169e-05,
      "loss": 0.0413,
      "step": 12130
    },
    {
      "epoch": 2.9911242603550297,
      "grad_norm": 0.3203125,
      "learning_rate": 8.326271070855185e-05,
      "loss": 0.0423,
      "step": 12132
    },
    {
      "epoch": 2.9916173570019726,
      "grad_norm": 0.392578125,
      "learning_rate": 8.322877266644902e-05,
      "loss": 0.0403,
      "step": 12134
    },
    {
      "epoch": 2.992110453648915,
      "grad_norm": 0.376953125,
      "learning_rate": 8.319483661183029e-05,
      "loss": 0.0385,
      "step": 12136
    },
    {
      "epoch": 2.992603550295858,
      "grad_norm": 0.408203125,
      "learning_rate": 8.316090254871725e-05,
      "loss": 0.0382,
      "step": 12138
    },
    {
      "epoch": 2.9930966469428006,
      "grad_norm": 0.30078125,
      "learning_rate": 8.312697048113121e-05,
      "loss": 0.041,
      "step": 12140
    },
    {
      "epoch": 2.9935897435897436,
      "grad_norm": 0.43359375,
      "learning_rate": 8.309304041309344e-05,
      "loss": 0.0433,
      "step": 12142
    },
    {
      "epoch": 2.994082840236686,
      "grad_norm": 0.4296875,
      "learning_rate": 8.305911234862473e-05,
      "loss": 0.0426,
      "step": 12144
    },
    {
      "epoch": 2.994575936883629,
      "grad_norm": 0.3515625,
      "learning_rate": 8.302518629174582e-05,
      "loss": 0.0365,
      "step": 12146
    },
    {
      "epoch": 2.995069033530572,
      "grad_norm": 0.40625,
      "learning_rate": 8.29912622464771e-05,
      "loss": 0.0449,
      "step": 12148
    },
    {
      "epoch": 2.9955621301775146,
      "grad_norm": 0.4609375,
      "learning_rate": 8.29573402168388e-05,
      "loss": 0.0391,
      "step": 12150
    },
    {
      "epoch": 2.9960552268244576,
      "grad_norm": 0.3046875,
      "learning_rate": 8.292342020685079e-05,
      "loss": 0.0385,
      "step": 12152
    },
    {
      "epoch": 2.9965483234714005,
      "grad_norm": 0.365234375,
      "learning_rate": 8.288950222053286e-05,
      "loss": 0.0405,
      "step": 12154
    },
    {
      "epoch": 2.997041420118343,
      "grad_norm": 0.384765625,
      "learning_rate": 8.285558626190447e-05,
      "loss": 0.0413,
      "step": 12156
    },
    {
      "epoch": 2.997534516765286,
      "grad_norm": 0.36328125,
      "learning_rate": 8.282167233498476e-05,
      "loss": 0.0402,
      "step": 12158
    },
    {
      "epoch": 2.998027613412229,
      "grad_norm": 0.32421875,
      "learning_rate": 8.278776044379286e-05,
      "loss": 0.0427,
      "step": 12160
    },
    {
      "epoch": 2.9985207100591715,
      "grad_norm": 0.431640625,
      "learning_rate": 8.275385059234741e-05,
      "loss": 0.0455,
      "step": 12162
    },
    {
      "epoch": 2.9990138067061145,
      "grad_norm": 0.2734375,
      "learning_rate": 8.271994278466698e-05,
      "loss": 0.0388,
      "step": 12164
    },
    {
      "epoch": 2.9995069033530575,
      "grad_norm": 0.384765625,
      "learning_rate": 8.268603702476977e-05,
      "loss": 0.0402,
      "step": 12166
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.466796875,
      "learning_rate": 8.265213331667388e-05,
      "loss": 0.0356,
      "step": 12168
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.0602399967610836,
      "eval_runtime": 73.8138,
      "eval_samples_per_second": 215.8,
      "eval_steps_per_second": 1.693,
      "step": 12168
    },
    {
      "epoch": 3.000493096646943,
      "grad_norm": 0.267578125,
      "learning_rate": 8.261823166439706e-05,
      "loss": 0.0281,
      "step": 12170
    },
    {
      "epoch": 3.0009861932938855,
      "grad_norm": 0.287109375,
      "learning_rate": 8.258433207195679e-05,
      "loss": 0.024,
      "step": 12172
    },
    {
      "epoch": 3.0014792899408285,
      "grad_norm": 0.255859375,
      "learning_rate": 8.255043454337043e-05,
      "loss": 0.0246,
      "step": 12174
    },
    {
      "epoch": 3.001972386587771,
      "grad_norm": 0.265625,
      "learning_rate": 8.2516539082655e-05,
      "loss": 0.0279,
      "step": 12176
    },
    {
      "epoch": 3.002465483234714,
      "grad_norm": 0.271484375,
      "learning_rate": 8.248264569382732e-05,
      "loss": 0.0258,
      "step": 12178
    },
    {
      "epoch": 3.002958579881657,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.244875438090388e-05,
      "loss": 0.0239,
      "step": 12180
    },
    {
      "epoch": 3.0034516765285995,
      "grad_norm": 0.287109375,
      "learning_rate": 8.241486514790109e-05,
      "loss": 0.0305,
      "step": 12182
    },
    {
      "epoch": 3.0039447731755424,
      "grad_norm": 0.310546875,
      "learning_rate": 8.238097799883492e-05,
      "loss": 0.0325,
      "step": 12184
    },
    {
      "epoch": 3.0044378698224854,
      "grad_norm": 0.2734375,
      "learning_rate": 8.234709293772127e-05,
      "loss": 0.0293,
      "step": 12186
    },
    {
      "epoch": 3.004930966469428,
      "grad_norm": 0.26171875,
      "learning_rate": 8.231320996857567e-05,
      "loss": 0.0285,
      "step": 12188
    },
    {
      "epoch": 3.005424063116371,
      "grad_norm": 0.28125,
      "learning_rate": 8.227932909541341e-05,
      "loss": 0.0267,
      "step": 12190
    },
    {
      "epoch": 3.0059171597633134,
      "grad_norm": 0.283203125,
      "learning_rate": 8.224545032224963e-05,
      "loss": 0.0293,
      "step": 12192
    },
    {
      "epoch": 3.0064102564102564,
      "grad_norm": 0.263671875,
      "learning_rate": 8.221157365309907e-05,
      "loss": 0.0274,
      "step": 12194
    },
    {
      "epoch": 3.0069033530571994,
      "grad_norm": 0.31640625,
      "learning_rate": 8.217769909197636e-05,
      "loss": 0.0261,
      "step": 12196
    },
    {
      "epoch": 3.007396449704142,
      "grad_norm": 0.34765625,
      "learning_rate": 8.214382664289582e-05,
      "loss": 0.0262,
      "step": 12198
    },
    {
      "epoch": 3.007889546351085,
      "grad_norm": 0.26953125,
      "learning_rate": 8.210995630987152e-05,
      "loss": 0.0283,
      "step": 12200
    },
    {
      "epoch": 3.008382642998028,
      "grad_norm": 0.26171875,
      "learning_rate": 8.207608809691729e-05,
      "loss": 0.0278,
      "step": 12202
    },
    {
      "epoch": 3.0088757396449703,
      "grad_norm": 0.29296875,
      "learning_rate": 8.204222200804671e-05,
      "loss": 0.0273,
      "step": 12204
    },
    {
      "epoch": 3.0093688362919133,
      "grad_norm": 0.251953125,
      "learning_rate": 8.200835804727308e-05,
      "loss": 0.0274,
      "step": 12206
    },
    {
      "epoch": 3.009861932938856,
      "grad_norm": 0.3125,
      "learning_rate": 8.197449621860943e-05,
      "loss": 0.0287,
      "step": 12208
    },
    {
      "epoch": 3.010355029585799,
      "grad_norm": 0.30859375,
      "learning_rate": 8.194063652606867e-05,
      "loss": 0.027,
      "step": 12210
    },
    {
      "epoch": 3.010848126232742,
      "grad_norm": 0.283203125,
      "learning_rate": 8.190677897366325e-05,
      "loss": 0.024,
      "step": 12212
    },
    {
      "epoch": 3.0113412228796843,
      "grad_norm": 0.27734375,
      "learning_rate": 8.187292356540561e-05,
      "loss": 0.0302,
      "step": 12214
    },
    {
      "epoch": 3.0118343195266273,
      "grad_norm": 0.33984375,
      "learning_rate": 8.18390703053077e-05,
      "loss": 0.0258,
      "step": 12216
    },
    {
      "epoch": 3.01232741617357,
      "grad_norm": 0.2470703125,
      "learning_rate": 8.180521919738138e-05,
      "loss": 0.0251,
      "step": 12218
    },
    {
      "epoch": 3.0128205128205128,
      "grad_norm": 0.265625,
      "learning_rate": 8.177137024563818e-05,
      "loss": 0.0282,
      "step": 12220
    },
    {
      "epoch": 3.0133136094674557,
      "grad_norm": 0.2578125,
      "learning_rate": 8.173752345408933e-05,
      "loss": 0.0256,
      "step": 12222
    },
    {
      "epoch": 3.0138067061143983,
      "grad_norm": 0.244140625,
      "learning_rate": 8.170367882674595e-05,
      "loss": 0.0259,
      "step": 12224
    },
    {
      "epoch": 3.0142998027613412,
      "grad_norm": 0.29296875,
      "learning_rate": 8.166983636761875e-05,
      "loss": 0.0253,
      "step": 12226
    },
    {
      "epoch": 3.014792899408284,
      "grad_norm": 0.271484375,
      "learning_rate": 8.163599608071833e-05,
      "loss": 0.0267,
      "step": 12228
    },
    {
      "epoch": 3.0152859960552267,
      "grad_norm": 0.251953125,
      "learning_rate": 8.160215797005488e-05,
      "loss": 0.0309,
      "step": 12230
    },
    {
      "epoch": 3.0157790927021697,
      "grad_norm": 0.25,
      "learning_rate": 8.156832203963846e-05,
      "loss": 0.0256,
      "step": 12232
    },
    {
      "epoch": 3.0162721893491122,
      "grad_norm": 0.263671875,
      "learning_rate": 8.153448829347873e-05,
      "loss": 0.0353,
      "step": 12234
    },
    {
      "epoch": 3.016765285996055,
      "grad_norm": 0.2275390625,
      "learning_rate": 8.15006567355853e-05,
      "loss": 0.0211,
      "step": 12236
    },
    {
      "epoch": 3.017258382642998,
      "grad_norm": 0.294921875,
      "learning_rate": 8.14668273699673e-05,
      "loss": 0.0236,
      "step": 12238
    },
    {
      "epoch": 3.0177514792899407,
      "grad_norm": 0.259765625,
      "learning_rate": 8.143300020063372e-05,
      "loss": 0.0285,
      "step": 12240
    },
    {
      "epoch": 3.0182445759368837,
      "grad_norm": 0.2373046875,
      "learning_rate": 8.13991752315933e-05,
      "loss": 0.0272,
      "step": 12242
    },
    {
      "epoch": 3.0187376725838266,
      "grad_norm": 0.251953125,
      "learning_rate": 8.136535246685445e-05,
      "loss": 0.0254,
      "step": 12244
    },
    {
      "epoch": 3.019230769230769,
      "grad_norm": 0.30078125,
      "learning_rate": 8.133153191042541e-05,
      "loss": 0.0259,
      "step": 12246
    },
    {
      "epoch": 3.019723865877712,
      "grad_norm": 0.28515625,
      "learning_rate": 8.129771356631403e-05,
      "loss": 0.0313,
      "step": 12248
    },
    {
      "epoch": 3.0202169625246547,
      "grad_norm": 0.31640625,
      "learning_rate": 8.126389743852806e-05,
      "loss": 0.0265,
      "step": 12250
    },
    {
      "epoch": 3.0207100591715976,
      "grad_norm": 0.28515625,
      "learning_rate": 8.123008353107479e-05,
      "loss": 0.0304,
      "step": 12252
    },
    {
      "epoch": 3.0212031558185406,
      "grad_norm": 0.287109375,
      "learning_rate": 8.119627184796149e-05,
      "loss": 0.0236,
      "step": 12254
    },
    {
      "epoch": 3.021696252465483,
      "grad_norm": 0.32421875,
      "learning_rate": 8.116246239319492e-05,
      "loss": 0.0264,
      "step": 12256
    },
    {
      "epoch": 3.022189349112426,
      "grad_norm": 0.267578125,
      "learning_rate": 8.112865517078174e-05,
      "loss": 0.0271,
      "step": 12258
    },
    {
      "epoch": 3.0226824457593686,
      "grad_norm": 0.22265625,
      "learning_rate": 8.109485018472832e-05,
      "loss": 0.0241,
      "step": 12260
    },
    {
      "epoch": 3.0231755424063116,
      "grad_norm": 0.259765625,
      "learning_rate": 8.106104743904064e-05,
      "loss": 0.0321,
      "step": 12262
    },
    {
      "epoch": 3.0236686390532546,
      "grad_norm": 0.265625,
      "learning_rate": 8.102724693772462e-05,
      "loss": 0.0292,
      "step": 12264
    },
    {
      "epoch": 3.024161735700197,
      "grad_norm": 0.3125,
      "learning_rate": 8.099344868478574e-05,
      "loss": 0.0269,
      "step": 12266
    },
    {
      "epoch": 3.02465483234714,
      "grad_norm": 0.275390625,
      "learning_rate": 8.095965268422933e-05,
      "loss": 0.0276,
      "step": 12268
    },
    {
      "epoch": 3.025147928994083,
      "grad_norm": 0.2490234375,
      "learning_rate": 8.092585894006041e-05,
      "loss": 0.026,
      "step": 12270
    },
    {
      "epoch": 3.0256410256410255,
      "grad_norm": 0.279296875,
      "learning_rate": 8.089206745628367e-05,
      "loss": 0.025,
      "step": 12272
    },
    {
      "epoch": 3.0261341222879685,
      "grad_norm": 0.26953125,
      "learning_rate": 8.085827823690364e-05,
      "loss": 0.0279,
      "step": 12274
    },
    {
      "epoch": 3.026627218934911,
      "grad_norm": 0.279296875,
      "learning_rate": 8.082449128592447e-05,
      "loss": 0.0267,
      "step": 12276
    },
    {
      "epoch": 3.027120315581854,
      "grad_norm": 0.26953125,
      "learning_rate": 8.079070660735019e-05,
      "loss": 0.024,
      "step": 12278
    },
    {
      "epoch": 3.027613412228797,
      "grad_norm": 0.322265625,
      "learning_rate": 8.07569242051844e-05,
      "loss": 0.0282,
      "step": 12280
    },
    {
      "epoch": 3.0281065088757395,
      "grad_norm": 0.2314453125,
      "learning_rate": 8.072314408343056e-05,
      "loss": 0.0262,
      "step": 12282
    },
    {
      "epoch": 3.0285996055226825,
      "grad_norm": 0.251953125,
      "learning_rate": 8.068936624609176e-05,
      "loss": 0.0309,
      "step": 12284
    },
    {
      "epoch": 3.0290927021696255,
      "grad_norm": 0.2353515625,
      "learning_rate": 8.065559069717088e-05,
      "loss": 0.0224,
      "step": 12286
    },
    {
      "epoch": 3.029585798816568,
      "grad_norm": 0.275390625,
      "learning_rate": 8.062181744067052e-05,
      "loss": 0.0259,
      "step": 12288
    },
    {
      "epoch": 3.030078895463511,
      "grad_norm": 0.275390625,
      "learning_rate": 8.058804648059294e-05,
      "loss": 0.0261,
      "step": 12290
    },
    {
      "epoch": 3.0305719921104535,
      "grad_norm": 0.275390625,
      "learning_rate": 8.05542778209403e-05,
      "loss": 0.0289,
      "step": 12292
    },
    {
      "epoch": 3.0310650887573964,
      "grad_norm": 0.23828125,
      "learning_rate": 8.052051146571423e-05,
      "loss": 0.029,
      "step": 12294
    },
    {
      "epoch": 3.0315581854043394,
      "grad_norm": 0.25,
      "learning_rate": 8.048674741891638e-05,
      "loss": 0.024,
      "step": 12296
    },
    {
      "epoch": 3.032051282051282,
      "grad_norm": 0.263671875,
      "learning_rate": 8.045298568454788e-05,
      "loss": 0.0271,
      "step": 12298
    },
    {
      "epoch": 3.032544378698225,
      "grad_norm": 0.263671875,
      "learning_rate": 8.041922626660973e-05,
      "loss": 0.0279,
      "step": 12300
    },
    {
      "epoch": 3.033037475345168,
      "grad_norm": 0.2197265625,
      "learning_rate": 8.038546916910259e-05,
      "loss": 0.0247,
      "step": 12302
    },
    {
      "epoch": 3.0335305719921104,
      "grad_norm": 0.26171875,
      "learning_rate": 8.035171439602684e-05,
      "loss": 0.0269,
      "step": 12304
    },
    {
      "epoch": 3.0340236686390534,
      "grad_norm": 0.296875,
      "learning_rate": 8.031796195138266e-05,
      "loss": 0.0255,
      "step": 12306
    },
    {
      "epoch": 3.034516765285996,
      "grad_norm": 0.248046875,
      "learning_rate": 8.028421183916986e-05,
      "loss": 0.0281,
      "step": 12308
    },
    {
      "epoch": 3.035009861932939,
      "grad_norm": 0.27734375,
      "learning_rate": 8.025046406338806e-05,
      "loss": 0.0261,
      "step": 12310
    },
    {
      "epoch": 3.035502958579882,
      "grad_norm": 0.283203125,
      "learning_rate": 8.02167186280365e-05,
      "loss": 0.0244,
      "step": 12312
    },
    {
      "epoch": 3.0359960552268244,
      "grad_norm": 0.2353515625,
      "learning_rate": 8.018297553711429e-05,
      "loss": 0.0263,
      "step": 12314
    },
    {
      "epoch": 3.0364891518737673,
      "grad_norm": 0.287109375,
      "learning_rate": 8.014923479462006e-05,
      "loss": 0.0259,
      "step": 12316
    },
    {
      "epoch": 3.03698224852071,
      "grad_norm": 0.25,
      "learning_rate": 8.01154964045524e-05,
      "loss": 0.0283,
      "step": 12318
    },
    {
      "epoch": 3.037475345167653,
      "grad_norm": 0.232421875,
      "learning_rate": 8.008176037090942e-05,
      "loss": 0.0244,
      "step": 12320
    },
    {
      "epoch": 3.037968441814596,
      "grad_norm": 0.2412109375,
      "learning_rate": 8.0048026697689e-05,
      "loss": 0.0225,
      "step": 12322
    },
    {
      "epoch": 3.0384615384615383,
      "grad_norm": 0.279296875,
      "learning_rate": 8.001429538888887e-05,
      "loss": 0.0264,
      "step": 12324
    },
    {
      "epoch": 3.0389546351084813,
      "grad_norm": 0.255859375,
      "learning_rate": 7.998056644850631e-05,
      "loss": 0.024,
      "step": 12326
    },
    {
      "epoch": 3.0394477317554243,
      "grad_norm": 0.27734375,
      "learning_rate": 7.99468398805384e-05,
      "loss": 0.0217,
      "step": 12328
    },
    {
      "epoch": 3.039940828402367,
      "grad_norm": 0.291015625,
      "learning_rate": 7.99131156889819e-05,
      "loss": 0.0265,
      "step": 12330
    },
    {
      "epoch": 3.0404339250493098,
      "grad_norm": 0.2236328125,
      "learning_rate": 7.987939387783337e-05,
      "loss": 0.0241,
      "step": 12332
    },
    {
      "epoch": 3.0409270216962523,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.984567445108896e-05,
      "loss": 0.023,
      "step": 12334
    },
    {
      "epoch": 3.0414201183431953,
      "grad_norm": 0.27734375,
      "learning_rate": 7.981195741274472e-05,
      "loss": 0.0295,
      "step": 12336
    },
    {
      "epoch": 3.0419132149901382,
      "grad_norm": 0.244140625,
      "learning_rate": 7.977824276679623e-05,
      "loss": 0.0236,
      "step": 12338
    },
    {
      "epoch": 3.0424063116370808,
      "grad_norm": 0.224609375,
      "learning_rate": 7.974453051723885e-05,
      "loss": 0.025,
      "step": 12340
    },
    {
      "epoch": 3.0428994082840237,
      "grad_norm": 0.2392578125,
      "learning_rate": 7.971082066806772e-05,
      "loss": 0.0236,
      "step": 12342
    },
    {
      "epoch": 3.0433925049309662,
      "grad_norm": 0.287109375,
      "learning_rate": 7.967711322327757e-05,
      "loss": 0.0285,
      "step": 12344
    },
    {
      "epoch": 3.043885601577909,
      "grad_norm": 0.310546875,
      "learning_rate": 7.964340818686304e-05,
      "loss": 0.0263,
      "step": 12346
    },
    {
      "epoch": 3.044378698224852,
      "grad_norm": 0.2353515625,
      "learning_rate": 7.960970556281823e-05,
      "loss": 0.026,
      "step": 12348
    },
    {
      "epoch": 3.0448717948717947,
      "grad_norm": 0.27734375,
      "learning_rate": 7.95760053551372e-05,
      "loss": 0.0257,
      "step": 12350
    },
    {
      "epoch": 3.0453648915187377,
      "grad_norm": 0.251953125,
      "learning_rate": 7.954230756781358e-05,
      "loss": 0.0256,
      "step": 12352
    },
    {
      "epoch": 3.0458579881656807,
      "grad_norm": 0.2373046875,
      "learning_rate": 7.950861220484071e-05,
      "loss": 0.027,
      "step": 12354
    },
    {
      "epoch": 3.046351084812623,
      "grad_norm": 0.255859375,
      "learning_rate": 7.947491927021172e-05,
      "loss": 0.0268,
      "step": 12356
    },
    {
      "epoch": 3.046844181459566,
      "grad_norm": 0.302734375,
      "learning_rate": 7.944122876791934e-05,
      "loss": 0.0262,
      "step": 12358
    },
    {
      "epoch": 3.0473372781065087,
      "grad_norm": 0.28125,
      "learning_rate": 7.94075407019562e-05,
      "loss": 0.0255,
      "step": 12360
    },
    {
      "epoch": 3.0478303747534516,
      "grad_norm": 0.25390625,
      "learning_rate": 7.93738550763144e-05,
      "loss": 0.0238,
      "step": 12362
    },
    {
      "epoch": 3.0483234714003946,
      "grad_norm": 0.27734375,
      "learning_rate": 7.9340171894986e-05,
      "loss": 0.0278,
      "step": 12364
    },
    {
      "epoch": 3.048816568047337,
      "grad_norm": 0.28125,
      "learning_rate": 7.930649116196253e-05,
      "loss": 0.0231,
      "step": 12366
    },
    {
      "epoch": 3.04930966469428,
      "grad_norm": 0.259765625,
      "learning_rate": 7.927281288123544e-05,
      "loss": 0.0253,
      "step": 12368
    },
    {
      "epoch": 3.049802761341223,
      "grad_norm": 0.3125,
      "learning_rate": 7.923913705679573e-05,
      "loss": 0.0271,
      "step": 12370
    },
    {
      "epoch": 3.0502958579881656,
      "grad_norm": 0.32421875,
      "learning_rate": 7.920546369263417e-05,
      "loss": 0.0247,
      "step": 12372
    },
    {
      "epoch": 3.0507889546351086,
      "grad_norm": 0.365234375,
      "learning_rate": 7.917179279274129e-05,
      "loss": 0.0262,
      "step": 12374
    },
    {
      "epoch": 3.051282051282051,
      "grad_norm": 0.291015625,
      "learning_rate": 7.913812436110723e-05,
      "loss": 0.0272,
      "step": 12376
    },
    {
      "epoch": 3.051775147928994,
      "grad_norm": 0.30859375,
      "learning_rate": 7.910445840172194e-05,
      "loss": 0.0254,
      "step": 12378
    },
    {
      "epoch": 3.052268244575937,
      "grad_norm": 0.27734375,
      "learning_rate": 7.907079491857498e-05,
      "loss": 0.026,
      "step": 12380
    },
    {
      "epoch": 3.0527613412228796,
      "grad_norm": 0.341796875,
      "learning_rate": 7.903713391565569e-05,
      "loss": 0.0285,
      "step": 12382
    },
    {
      "epoch": 3.0532544378698225,
      "grad_norm": 0.29296875,
      "learning_rate": 7.900347539695303e-05,
      "loss": 0.026,
      "step": 12384
    },
    {
      "epoch": 3.0537475345167655,
      "grad_norm": 0.27734375,
      "learning_rate": 7.896981936645581e-05,
      "loss": 0.0236,
      "step": 12386
    },
    {
      "epoch": 3.054240631163708,
      "grad_norm": 0.2578125,
      "learning_rate": 7.893616582815242e-05,
      "loss": 0.0249,
      "step": 12388
    },
    {
      "epoch": 3.054733727810651,
      "grad_norm": 0.25,
      "learning_rate": 7.890251478603094e-05,
      "loss": 0.0258,
      "step": 12390
    },
    {
      "epoch": 3.0552268244575935,
      "grad_norm": 0.287109375,
      "learning_rate": 7.886886624407929e-05,
      "loss": 0.0238,
      "step": 12392
    },
    {
      "epoch": 3.0557199211045365,
      "grad_norm": 0.234375,
      "learning_rate": 7.883522020628494e-05,
      "loss": 0.0259,
      "step": 12394
    },
    {
      "epoch": 3.0562130177514795,
      "grad_norm": 0.25,
      "learning_rate": 7.88015766766352e-05,
      "loss": 0.0279,
      "step": 12396
    },
    {
      "epoch": 3.056706114398422,
      "grad_norm": 0.2216796875,
      "learning_rate": 7.876793565911693e-05,
      "loss": 0.0264,
      "step": 12398
    },
    {
      "epoch": 3.057199211045365,
      "grad_norm": 0.2421875,
      "learning_rate": 7.873429715771687e-05,
      "loss": 0.0273,
      "step": 12400
    },
    {
      "epoch": 3.0576923076923075,
      "grad_norm": 0.26171875,
      "learning_rate": 7.870066117642134e-05,
      "loss": 0.0251,
      "step": 12402
    },
    {
      "epoch": 3.0581854043392505,
      "grad_norm": 0.2099609375,
      "learning_rate": 7.866702771921634e-05,
      "loss": 0.0245,
      "step": 12404
    },
    {
      "epoch": 3.0586785009861934,
      "grad_norm": 0.236328125,
      "learning_rate": 7.86333967900877e-05,
      "loss": 0.0245,
      "step": 12406
    },
    {
      "epoch": 3.059171597633136,
      "grad_norm": 0.22265625,
      "learning_rate": 7.859976839302081e-05,
      "loss": 0.0245,
      "step": 12408
    },
    {
      "epoch": 3.059664694280079,
      "grad_norm": 0.25,
      "learning_rate": 7.856614253200087e-05,
      "loss": 0.0258,
      "step": 12410
    },
    {
      "epoch": 3.060157790927022,
      "grad_norm": 0.283203125,
      "learning_rate": 7.853251921101268e-05,
      "loss": 0.0259,
      "step": 12412
    },
    {
      "epoch": 3.0606508875739644,
      "grad_norm": 0.240234375,
      "learning_rate": 7.849889843404086e-05,
      "loss": 0.0228,
      "step": 12414
    },
    {
      "epoch": 3.0611439842209074,
      "grad_norm": 0.271484375,
      "learning_rate": 7.846528020506957e-05,
      "loss": 0.0256,
      "step": 12416
    },
    {
      "epoch": 3.06163708086785,
      "grad_norm": 0.2451171875,
      "learning_rate": 7.843166452808287e-05,
      "loss": 0.0265,
      "step": 12418
    },
    {
      "epoch": 3.062130177514793,
      "grad_norm": 0.2412109375,
      "learning_rate": 7.839805140706433e-05,
      "loss": 0.0278,
      "step": 12420
    },
    {
      "epoch": 3.062623274161736,
      "grad_norm": 0.248046875,
      "learning_rate": 7.836444084599729e-05,
      "loss": 0.0257,
      "step": 12422
    },
    {
      "epoch": 3.0631163708086784,
      "grad_norm": 0.2119140625,
      "learning_rate": 7.833083284886483e-05,
      "loss": 0.0231,
      "step": 12424
    },
    {
      "epoch": 3.0636094674556213,
      "grad_norm": 0.255859375,
      "learning_rate": 7.829722741964963e-05,
      "loss": 0.0254,
      "step": 12426
    },
    {
      "epoch": 3.064102564102564,
      "grad_norm": 0.26953125,
      "learning_rate": 7.826362456233418e-05,
      "loss": 0.0279,
      "step": 12428
    },
    {
      "epoch": 3.064595660749507,
      "grad_norm": 0.279296875,
      "learning_rate": 7.823002428090055e-05,
      "loss": 0.0236,
      "step": 12430
    },
    {
      "epoch": 3.06508875739645,
      "grad_norm": 0.271484375,
      "learning_rate": 7.819642657933062e-05,
      "loss": 0.0278,
      "step": 12432
    },
    {
      "epoch": 3.0655818540433923,
      "grad_norm": 0.25,
      "learning_rate": 7.816283146160589e-05,
      "loss": 0.0267,
      "step": 12434
    },
    {
      "epoch": 3.0660749506903353,
      "grad_norm": 0.248046875,
      "learning_rate": 7.812923893170753e-05,
      "loss": 0.0235,
      "step": 12436
    },
    {
      "epoch": 3.0665680473372783,
      "grad_norm": 0.263671875,
      "learning_rate": 7.809564899361647e-05,
      "loss": 0.0257,
      "step": 12438
    },
    {
      "epoch": 3.067061143984221,
      "grad_norm": 0.28125,
      "learning_rate": 7.806206165131329e-05,
      "loss": 0.0274,
      "step": 12440
    },
    {
      "epoch": 3.0675542406311638,
      "grad_norm": 0.232421875,
      "learning_rate": 7.802847690877832e-05,
      "loss": 0.023,
      "step": 12442
    },
    {
      "epoch": 3.0680473372781063,
      "grad_norm": 0.265625,
      "learning_rate": 7.799489476999147e-05,
      "loss": 0.0251,
      "step": 12444
    },
    {
      "epoch": 3.0685404339250493,
      "grad_norm": 0.2470703125,
      "learning_rate": 7.796131523893249e-05,
      "loss": 0.0277,
      "step": 12446
    },
    {
      "epoch": 3.0690335305719922,
      "grad_norm": 0.232421875,
      "learning_rate": 7.792773831958067e-05,
      "loss": 0.0231,
      "step": 12448
    },
    {
      "epoch": 3.0695266272189348,
      "grad_norm": 0.2421875,
      "learning_rate": 7.789416401591513e-05,
      "loss": 0.0245,
      "step": 12450
    },
    {
      "epoch": 3.0700197238658777,
      "grad_norm": 0.255859375,
      "learning_rate": 7.786059233191457e-05,
      "loss": 0.023,
      "step": 12452
    },
    {
      "epoch": 3.0705128205128207,
      "grad_norm": 0.2431640625,
      "learning_rate": 7.782702327155738e-05,
      "loss": 0.0256,
      "step": 12454
    },
    {
      "epoch": 3.0710059171597632,
      "grad_norm": 0.275390625,
      "learning_rate": 7.77934568388218e-05,
      "loss": 0.0249,
      "step": 12456
    },
    {
      "epoch": 3.071499013806706,
      "grad_norm": 0.228515625,
      "learning_rate": 7.775989303768551e-05,
      "loss": 0.0278,
      "step": 12458
    },
    {
      "epoch": 3.0719921104536487,
      "grad_norm": 0.2314453125,
      "learning_rate": 7.77263318721261e-05,
      "loss": 0.025,
      "step": 12460
    },
    {
      "epoch": 3.0724852071005917,
      "grad_norm": 0.2431640625,
      "learning_rate": 7.769277334612072e-05,
      "loss": 0.0255,
      "step": 12462
    },
    {
      "epoch": 3.0729783037475347,
      "grad_norm": 0.30859375,
      "learning_rate": 7.765921746364626e-05,
      "loss": 0.0287,
      "step": 12464
    },
    {
      "epoch": 3.073471400394477,
      "grad_norm": 0.26171875,
      "learning_rate": 7.762566422867922e-05,
      "loss": 0.0239,
      "step": 12466
    },
    {
      "epoch": 3.07396449704142,
      "grad_norm": 0.2734375,
      "learning_rate": 7.759211364519594e-05,
      "loss": 0.025,
      "step": 12468
    },
    {
      "epoch": 3.074457593688363,
      "grad_norm": 0.234375,
      "learning_rate": 7.75585657171723e-05,
      "loss": 0.0239,
      "step": 12470
    },
    {
      "epoch": 3.0749506903353057,
      "grad_norm": 0.3203125,
      "learning_rate": 7.752502044858388e-05,
      "loss": 0.0261,
      "step": 12472
    },
    {
      "epoch": 3.0754437869822486,
      "grad_norm": 0.34765625,
      "learning_rate": 7.749147784340606e-05,
      "loss": 0.0224,
      "step": 12474
    },
    {
      "epoch": 3.075936883629191,
      "grad_norm": 0.279296875,
      "learning_rate": 7.745793790561377e-05,
      "loss": 0.0258,
      "step": 12476
    },
    {
      "epoch": 3.076429980276134,
      "grad_norm": 0.2333984375,
      "learning_rate": 7.742440063918171e-05,
      "loss": 0.0259,
      "step": 12478
    },
    {
      "epoch": 3.076923076923077,
      "grad_norm": 0.2578125,
      "learning_rate": 7.739086604808417e-05,
      "loss": 0.024,
      "step": 12480
    },
    {
      "epoch": 3.0774161735700196,
      "grad_norm": 0.26953125,
      "learning_rate": 7.73573341362953e-05,
      "loss": 0.0273,
      "step": 12482
    },
    {
      "epoch": 3.0779092702169626,
      "grad_norm": 0.298828125,
      "learning_rate": 7.732380490778874e-05,
      "loss": 0.0242,
      "step": 12484
    },
    {
      "epoch": 3.078402366863905,
      "grad_norm": 0.255859375,
      "learning_rate": 7.729027836653787e-05,
      "loss": 0.0258,
      "step": 12486
    },
    {
      "epoch": 3.078895463510848,
      "grad_norm": 0.30078125,
      "learning_rate": 7.725675451651586e-05,
      "loss": 0.0253,
      "step": 12488
    },
    {
      "epoch": 3.079388560157791,
      "grad_norm": 0.28125,
      "learning_rate": 7.72232333616954e-05,
      "loss": 0.0254,
      "step": 12490
    },
    {
      "epoch": 3.0798816568047336,
      "grad_norm": 0.3125,
      "learning_rate": 7.718971490604896e-05,
      "loss": 0.0243,
      "step": 12492
    },
    {
      "epoch": 3.0803747534516766,
      "grad_norm": 0.3046875,
      "learning_rate": 7.715619915354862e-05,
      "loss": 0.0245,
      "step": 12494
    },
    {
      "epoch": 3.0808678500986195,
      "grad_norm": 0.2275390625,
      "learning_rate": 7.712268610816628e-05,
      "loss": 0.0229,
      "step": 12496
    },
    {
      "epoch": 3.081360946745562,
      "grad_norm": 0.259765625,
      "learning_rate": 7.708917577387331e-05,
      "loss": 0.0235,
      "step": 12498
    },
    {
      "epoch": 3.081854043392505,
      "grad_norm": 0.306640625,
      "learning_rate": 7.705566815464098e-05,
      "loss": 0.0257,
      "step": 12500
    },
    {
      "epoch": 3.0823471400394475,
      "grad_norm": 0.2431640625,
      "learning_rate": 7.702216325444007e-05,
      "loss": 0.0238,
      "step": 12502
    },
    {
      "epoch": 3.0828402366863905,
      "grad_norm": 0.2333984375,
      "learning_rate": 7.69886610772411e-05,
      "loss": 0.0224,
      "step": 12504
    },
    {
      "epoch": 3.0833333333333335,
      "grad_norm": 0.259765625,
      "learning_rate": 7.695516162701429e-05,
      "loss": 0.024,
      "step": 12506
    },
    {
      "epoch": 3.083826429980276,
      "grad_norm": 0.2333984375,
      "learning_rate": 7.692166490772942e-05,
      "loss": 0.0276,
      "step": 12508
    },
    {
      "epoch": 3.084319526627219,
      "grad_norm": 0.2470703125,
      "learning_rate": 7.688817092335618e-05,
      "loss": 0.0242,
      "step": 12510
    },
    {
      "epoch": 3.084812623274162,
      "grad_norm": 0.263671875,
      "learning_rate": 7.68546796778637e-05,
      "loss": 0.0287,
      "step": 12512
    },
    {
      "epoch": 3.0853057199211045,
      "grad_norm": 0.279296875,
      "learning_rate": 7.682119117522092e-05,
      "loss": 0.0263,
      "step": 12514
    },
    {
      "epoch": 3.0857988165680474,
      "grad_norm": 0.2734375,
      "learning_rate": 7.678770541939638e-05,
      "loss": 0.0271,
      "step": 12516
    },
    {
      "epoch": 3.08629191321499,
      "grad_norm": 0.251953125,
      "learning_rate": 7.675422241435839e-05,
      "loss": 0.0265,
      "step": 12518
    },
    {
      "epoch": 3.086785009861933,
      "grad_norm": 0.279296875,
      "learning_rate": 7.672074216407478e-05,
      "loss": 0.0242,
      "step": 12520
    },
    {
      "epoch": 3.087278106508876,
      "grad_norm": 0.224609375,
      "learning_rate": 7.668726467251323e-05,
      "loss": 0.0246,
      "step": 12522
    },
    {
      "epoch": 3.0877712031558184,
      "grad_norm": 0.220703125,
      "learning_rate": 7.665378994364099e-05,
      "loss": 0.0255,
      "step": 12524
    },
    {
      "epoch": 3.0882642998027614,
      "grad_norm": 0.232421875,
      "learning_rate": 7.662031798142493e-05,
      "loss": 0.0215,
      "step": 12526
    },
    {
      "epoch": 3.088757396449704,
      "grad_norm": 0.25,
      "learning_rate": 7.658684878983178e-05,
      "loss": 0.0267,
      "step": 12528
    },
    {
      "epoch": 3.089250493096647,
      "grad_norm": 0.302734375,
      "learning_rate": 7.655338237282775e-05,
      "loss": 0.026,
      "step": 12530
    },
    {
      "epoch": 3.08974358974359,
      "grad_norm": 0.302734375,
      "learning_rate": 7.651991873437883e-05,
      "loss": 0.0227,
      "step": 12532
    },
    {
      "epoch": 3.0902366863905324,
      "grad_norm": 0.2060546875,
      "learning_rate": 7.648645787845064e-05,
      "loss": 0.0231,
      "step": 12534
    },
    {
      "epoch": 3.0907297830374754,
      "grad_norm": 0.2451171875,
      "learning_rate": 7.645299980900846e-05,
      "loss": 0.0252,
      "step": 12536
    },
    {
      "epoch": 3.0912228796844183,
      "grad_norm": 0.2353515625,
      "learning_rate": 7.641954453001728e-05,
      "loss": 0.0251,
      "step": 12538
    },
    {
      "epoch": 3.091715976331361,
      "grad_norm": 0.2265625,
      "learning_rate": 7.638609204544169e-05,
      "loss": 0.0218,
      "step": 12540
    },
    {
      "epoch": 3.092209072978304,
      "grad_norm": 0.302734375,
      "learning_rate": 7.63526423592461e-05,
      "loss": 0.0223,
      "step": 12542
    },
    {
      "epoch": 3.0927021696252464,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.631919547539437e-05,
      "loss": 0.0224,
      "step": 12544
    },
    {
      "epoch": 3.0931952662721893,
      "grad_norm": 0.2734375,
      "learning_rate": 7.628575139785024e-05,
      "loss": 0.0237,
      "step": 12546
    },
    {
      "epoch": 3.0936883629191323,
      "grad_norm": 0.259765625,
      "learning_rate": 7.625231013057691e-05,
      "loss": 0.0252,
      "step": 12548
    },
    {
      "epoch": 3.094181459566075,
      "grad_norm": 0.275390625,
      "learning_rate": 7.621887167753748e-05,
      "loss": 0.0258,
      "step": 12550
    },
    {
      "epoch": 3.094674556213018,
      "grad_norm": 0.29296875,
      "learning_rate": 7.618543604269452e-05,
      "loss": 0.0266,
      "step": 12552
    },
    {
      "epoch": 3.0951676528599608,
      "grad_norm": 0.259765625,
      "learning_rate": 7.615200323001032e-05,
      "loss": 0.0244,
      "step": 12554
    },
    {
      "epoch": 3.0956607495069033,
      "grad_norm": 0.20703125,
      "learning_rate": 7.611857324344692e-05,
      "loss": 0.0235,
      "step": 12556
    },
    {
      "epoch": 3.0961538461538463,
      "grad_norm": 0.27734375,
      "learning_rate": 7.60851460869659e-05,
      "loss": 0.0263,
      "step": 12558
    },
    {
      "epoch": 3.096646942800789,
      "grad_norm": 0.279296875,
      "learning_rate": 7.605172176452862e-05,
      "loss": 0.0309,
      "step": 12560
    },
    {
      "epoch": 3.0971400394477318,
      "grad_norm": 0.296875,
      "learning_rate": 7.6018300280096e-05,
      "loss": 0.0242,
      "step": 12562
    },
    {
      "epoch": 3.0976331360946747,
      "grad_norm": 0.26171875,
      "learning_rate": 7.598488163762872e-05,
      "loss": 0.0252,
      "step": 12564
    },
    {
      "epoch": 3.0981262327416172,
      "grad_norm": 0.333984375,
      "learning_rate": 7.595146584108704e-05,
      "loss": 0.0283,
      "step": 12566
    },
    {
      "epoch": 3.09861932938856,
      "grad_norm": 0.32421875,
      "learning_rate": 7.59180528944309e-05,
      "loss": 0.0294,
      "step": 12568
    },
    {
      "epoch": 3.0991124260355027,
      "grad_norm": 0.19140625,
      "learning_rate": 7.588464280161998e-05,
      "loss": 0.0235,
      "step": 12570
    },
    {
      "epoch": 3.0996055226824457,
      "grad_norm": 0.3046875,
      "learning_rate": 7.585123556661351e-05,
      "loss": 0.0224,
      "step": 12572
    },
    {
      "epoch": 3.1000986193293887,
      "grad_norm": 0.373046875,
      "learning_rate": 7.581783119337048e-05,
      "loss": 0.0281,
      "step": 12574
    },
    {
      "epoch": 3.100591715976331,
      "grad_norm": 0.263671875,
      "learning_rate": 7.578442968584943e-05,
      "loss": 0.0263,
      "step": 12576
    },
    {
      "epoch": 3.101084812623274,
      "grad_norm": 0.2734375,
      "learning_rate": 7.57510310480087e-05,
      "loss": 0.0236,
      "step": 12578
    },
    {
      "epoch": 3.101577909270217,
      "grad_norm": 0.2578125,
      "learning_rate": 7.571763528380615e-05,
      "loss": 0.0235,
      "step": 12580
    },
    {
      "epoch": 3.1020710059171597,
      "grad_norm": 0.279296875,
      "learning_rate": 7.568424239719943e-05,
      "loss": 0.025,
      "step": 12582
    },
    {
      "epoch": 3.1025641025641026,
      "grad_norm": 0.2734375,
      "learning_rate": 7.565085239214574e-05,
      "loss": 0.0286,
      "step": 12584
    },
    {
      "epoch": 3.103057199211045,
      "grad_norm": 0.234375,
      "learning_rate": 7.561746527260197e-05,
      "loss": 0.024,
      "step": 12586
    },
    {
      "epoch": 3.103550295857988,
      "grad_norm": 0.21875,
      "learning_rate": 7.55840810425247e-05,
      "loss": 0.0238,
      "step": 12588
    },
    {
      "epoch": 3.104043392504931,
      "grad_norm": 0.33984375,
      "learning_rate": 7.555069970587017e-05,
      "loss": 0.0329,
      "step": 12590
    },
    {
      "epoch": 3.1045364891518736,
      "grad_norm": 0.314453125,
      "learning_rate": 7.551732126659422e-05,
      "loss": 0.0311,
      "step": 12592
    },
    {
      "epoch": 3.1050295857988166,
      "grad_norm": 0.267578125,
      "learning_rate": 7.548394572865237e-05,
      "loss": 0.0234,
      "step": 12594
    },
    {
      "epoch": 3.1055226824457596,
      "grad_norm": 0.318359375,
      "learning_rate": 7.545057309599986e-05,
      "loss": 0.0212,
      "step": 12596
    },
    {
      "epoch": 3.106015779092702,
      "grad_norm": 0.330078125,
      "learning_rate": 7.54172033725915e-05,
      "loss": 0.0318,
      "step": 12598
    },
    {
      "epoch": 3.106508875739645,
      "grad_norm": 0.341796875,
      "learning_rate": 7.538383656238179e-05,
      "loss": 0.0311,
      "step": 12600
    },
    {
      "epoch": 3.1070019723865876,
      "grad_norm": 0.298828125,
      "learning_rate": 7.53504726693249e-05,
      "loss": 0.0274,
      "step": 12602
    },
    {
      "epoch": 3.1074950690335306,
      "grad_norm": 0.28125,
      "learning_rate": 7.531711169737461e-05,
      "loss": 0.024,
      "step": 12604
    },
    {
      "epoch": 3.1079881656804735,
      "grad_norm": 0.27734375,
      "learning_rate": 7.528375365048442e-05,
      "loss": 0.0329,
      "step": 12606
    },
    {
      "epoch": 3.108481262327416,
      "grad_norm": 0.33984375,
      "learning_rate": 7.525039853260736e-05,
      "loss": 0.0309,
      "step": 12608
    },
    {
      "epoch": 3.108974358974359,
      "grad_norm": 0.25,
      "learning_rate": 7.52170463476963e-05,
      "loss": 0.025,
      "step": 12610
    },
    {
      "epoch": 3.109467455621302,
      "grad_norm": 0.294921875,
      "learning_rate": 7.518369709970362e-05,
      "loss": 0.0261,
      "step": 12612
    },
    {
      "epoch": 3.1099605522682445,
      "grad_norm": 0.31640625,
      "learning_rate": 7.515035079258138e-05,
      "loss": 0.0324,
      "step": 12614
    },
    {
      "epoch": 3.1104536489151875,
      "grad_norm": 0.283203125,
      "learning_rate": 7.511700743028134e-05,
      "loss": 0.0365,
      "step": 12616
    },
    {
      "epoch": 3.11094674556213,
      "grad_norm": 0.259765625,
      "learning_rate": 7.508366701675482e-05,
      "loss": 0.027,
      "step": 12618
    },
    {
      "epoch": 3.111439842209073,
      "grad_norm": 0.306640625,
      "learning_rate": 7.505032955595289e-05,
      "loss": 0.0274,
      "step": 12620
    },
    {
      "epoch": 3.111932938856016,
      "grad_norm": 0.296875,
      "learning_rate": 7.501699505182617e-05,
      "loss": 0.0329,
      "step": 12622
    },
    {
      "epoch": 3.1124260355029585,
      "grad_norm": 0.322265625,
      "learning_rate": 7.498366350832506e-05,
      "loss": 0.0343,
      "step": 12624
    },
    {
      "epoch": 3.1129191321499015,
      "grad_norm": 0.271484375,
      "learning_rate": 7.495033492939949e-05,
      "loss": 0.0247,
      "step": 12626
    },
    {
      "epoch": 3.113412228796844,
      "grad_norm": 0.2734375,
      "learning_rate": 7.491700931899909e-05,
      "loss": 0.022,
      "step": 12628
    },
    {
      "epoch": 3.113905325443787,
      "grad_norm": 0.345703125,
      "learning_rate": 7.488368668107312e-05,
      "loss": 0.0359,
      "step": 12630
    },
    {
      "epoch": 3.11439842209073,
      "grad_norm": 0.3515625,
      "learning_rate": 7.485036701957053e-05,
      "loss": 0.033,
      "step": 12632
    },
    {
      "epoch": 3.1148915187376724,
      "grad_norm": 0.2734375,
      "learning_rate": 7.481705033843989e-05,
      "loss": 0.0249,
      "step": 12634
    },
    {
      "epoch": 3.1153846153846154,
      "grad_norm": 0.265625,
      "learning_rate": 7.478373664162932e-05,
      "loss": 0.0256,
      "step": 12636
    },
    {
      "epoch": 3.1158777120315584,
      "grad_norm": 0.322265625,
      "learning_rate": 7.47504259330868e-05,
      "loss": 0.0328,
      "step": 12638
    },
    {
      "epoch": 3.116370808678501,
      "grad_norm": 0.333984375,
      "learning_rate": 7.471711821675973e-05,
      "loss": 0.0368,
      "step": 12640
    },
    {
      "epoch": 3.116863905325444,
      "grad_norm": 0.287109375,
      "learning_rate": 7.468381349659537e-05,
      "loss": 0.0217,
      "step": 12642
    },
    {
      "epoch": 3.1173570019723864,
      "grad_norm": 0.265625,
      "learning_rate": 7.465051177654043e-05,
      "loss": 0.0306,
      "step": 12644
    },
    {
      "epoch": 3.1178500986193294,
      "grad_norm": 0.306640625,
      "learning_rate": 7.46172130605414e-05,
      "loss": 0.0382,
      "step": 12646
    },
    {
      "epoch": 3.1183431952662723,
      "grad_norm": 0.3203125,
      "learning_rate": 7.458391735254431e-05,
      "loss": 0.0332,
      "step": 12648
    },
    {
      "epoch": 3.118836291913215,
      "grad_norm": 0.3359375,
      "learning_rate": 7.455062465649494e-05,
      "loss": 0.0321,
      "step": 12650
    },
    {
      "epoch": 3.119329388560158,
      "grad_norm": 0.244140625,
      "learning_rate": 7.451733497633865e-05,
      "loss": 0.0269,
      "step": 12652
    },
    {
      "epoch": 3.1198224852071004,
      "grad_norm": 0.302734375,
      "learning_rate": 7.44840483160204e-05,
      "loss": 0.0345,
      "step": 12654
    },
    {
      "epoch": 3.1203155818540433,
      "grad_norm": 0.3203125,
      "learning_rate": 7.445076467948493e-05,
      "loss": 0.0324,
      "step": 12656
    },
    {
      "epoch": 3.1208086785009863,
      "grad_norm": 0.240234375,
      "learning_rate": 7.441748407067648e-05,
      "loss": 0.0267,
      "step": 12658
    },
    {
      "epoch": 3.121301775147929,
      "grad_norm": 0.291015625,
      "learning_rate": 7.438420649353901e-05,
      "loss": 0.029,
      "step": 12660
    },
    {
      "epoch": 3.121794871794872,
      "grad_norm": 0.30859375,
      "learning_rate": 7.435093195201605e-05,
      "loss": 0.0364,
      "step": 12662
    },
    {
      "epoch": 3.1222879684418148,
      "grad_norm": 0.32421875,
      "learning_rate": 7.43176604500509e-05,
      "loss": 0.035,
      "step": 12664
    },
    {
      "epoch": 3.1227810650887573,
      "grad_norm": 0.298828125,
      "learning_rate": 7.428439199158639e-05,
      "loss": 0.0311,
      "step": 12666
    },
    {
      "epoch": 3.1232741617357003,
      "grad_norm": 0.326171875,
      "learning_rate": 7.425112658056496e-05,
      "loss": 0.0287,
      "step": 12668
    },
    {
      "epoch": 3.123767258382643,
      "grad_norm": 0.310546875,
      "learning_rate": 7.421786422092884e-05,
      "loss": 0.0353,
      "step": 12670
    },
    {
      "epoch": 3.1242603550295858,
      "grad_norm": 0.33203125,
      "learning_rate": 7.418460491661973e-05,
      "loss": 0.0354,
      "step": 12672
    },
    {
      "epoch": 3.1247534516765287,
      "grad_norm": 0.275390625,
      "learning_rate": 7.415134867157909e-05,
      "loss": 0.0343,
      "step": 12674
    },
    {
      "epoch": 3.1252465483234713,
      "grad_norm": 0.369140625,
      "learning_rate": 7.411809548974792e-05,
      "loss": 0.0305,
      "step": 12676
    },
    {
      "epoch": 3.1257396449704142,
      "grad_norm": 0.28515625,
      "learning_rate": 7.408484537506698e-05,
      "loss": 0.0323,
      "step": 12678
    },
    {
      "epoch": 3.126232741617357,
      "grad_norm": 0.33203125,
      "learning_rate": 7.405159833147651e-05,
      "loss": 0.0353,
      "step": 12680
    },
    {
      "epoch": 3.1267258382642997,
      "grad_norm": 0.3203125,
      "learning_rate": 7.401835436291656e-05,
      "loss": 0.0334,
      "step": 12682
    },
    {
      "epoch": 3.1272189349112427,
      "grad_norm": 0.30078125,
      "learning_rate": 7.398511347332669e-05,
      "loss": 0.0378,
      "step": 12684
    },
    {
      "epoch": 3.1277120315581852,
      "grad_norm": 0.314453125,
      "learning_rate": 7.395187566664607e-05,
      "loss": 0.038,
      "step": 12686
    },
    {
      "epoch": 3.128205128205128,
      "grad_norm": 0.3203125,
      "learning_rate": 7.391864094681367e-05,
      "loss": 0.0391,
      "step": 12688
    },
    {
      "epoch": 3.128698224852071,
      "grad_norm": 0.302734375,
      "learning_rate": 7.388540931776789e-05,
      "loss": 0.0336,
      "step": 12690
    },
    {
      "epoch": 3.1291913214990137,
      "grad_norm": 0.296875,
      "learning_rate": 7.385218078344694e-05,
      "loss": 0.0325,
      "step": 12692
    },
    {
      "epoch": 3.1296844181459567,
      "grad_norm": 0.33203125,
      "learning_rate": 7.381895534778853e-05,
      "loss": 0.0342,
      "step": 12694
    },
    {
      "epoch": 3.1301775147928996,
      "grad_norm": 0.357421875,
      "learning_rate": 7.378573301473012e-05,
      "loss": 0.0361,
      "step": 12696
    },
    {
      "epoch": 3.130670611439842,
      "grad_norm": 0.31640625,
      "learning_rate": 7.375251378820868e-05,
      "loss": 0.0316,
      "step": 12698
    },
    {
      "epoch": 3.131163708086785,
      "grad_norm": 0.298828125,
      "learning_rate": 7.371929767216094e-05,
      "loss": 0.0289,
      "step": 12700
    },
    {
      "epoch": 3.1316568047337277,
      "grad_norm": 0.330078125,
      "learning_rate": 7.368608467052316e-05,
      "loss": 0.0343,
      "step": 12702
    },
    {
      "epoch": 3.1321499013806706,
      "grad_norm": 0.294921875,
      "learning_rate": 7.365287478723122e-05,
      "loss": 0.0367,
      "step": 12704
    },
    {
      "epoch": 3.1326429980276136,
      "grad_norm": 0.333984375,
      "learning_rate": 7.361966802622075e-05,
      "loss": 0.0327,
      "step": 12706
    },
    {
      "epoch": 3.133136094674556,
      "grad_norm": 0.302734375,
      "learning_rate": 7.358646439142688e-05,
      "loss": 0.0334,
      "step": 12708
    },
    {
      "epoch": 3.133629191321499,
      "grad_norm": 0.30859375,
      "learning_rate": 7.35532638867845e-05,
      "loss": 0.0373,
      "step": 12710
    },
    {
      "epoch": 3.1341222879684416,
      "grad_norm": 0.5390625,
      "learning_rate": 7.352006651622798e-05,
      "loss": 0.0365,
      "step": 12712
    },
    {
      "epoch": 3.1346153846153846,
      "grad_norm": 0.392578125,
      "learning_rate": 7.348687228369144e-05,
      "loss": 0.0423,
      "step": 12714
    },
    {
      "epoch": 3.1351084812623276,
      "grad_norm": 0.337890625,
      "learning_rate": 7.345368119310856e-05,
      "loss": 0.0368,
      "step": 12716
    },
    {
      "epoch": 3.13560157790927,
      "grad_norm": 0.3125,
      "learning_rate": 7.342049324841265e-05,
      "loss": 0.0357,
      "step": 12718
    },
    {
      "epoch": 3.136094674556213,
      "grad_norm": 0.365234375,
      "learning_rate": 7.338730845353671e-05,
      "loss": 0.0345,
      "step": 12720
    },
    {
      "epoch": 3.136587771203156,
      "grad_norm": 0.3125,
      "learning_rate": 7.335412681241329e-05,
      "loss": 0.0342,
      "step": 12722
    },
    {
      "epoch": 3.1370808678500985,
      "grad_norm": 0.34765625,
      "learning_rate": 7.332094832897464e-05,
      "loss": 0.0336,
      "step": 12724
    },
    {
      "epoch": 3.1375739644970415,
      "grad_norm": 0.33203125,
      "learning_rate": 7.328777300715256e-05,
      "loss": 0.0388,
      "step": 12726
    },
    {
      "epoch": 3.138067061143984,
      "grad_norm": 0.37890625,
      "learning_rate": 7.325460085087852e-05,
      "loss": 0.0353,
      "step": 12728
    },
    {
      "epoch": 3.138560157790927,
      "grad_norm": 0.330078125,
      "learning_rate": 7.322143186408358e-05,
      "loss": 0.0318,
      "step": 12730
    },
    {
      "epoch": 3.13905325443787,
      "grad_norm": 0.296875,
      "learning_rate": 7.318826605069851e-05,
      "loss": 0.0335,
      "step": 12732
    },
    {
      "epoch": 3.1395463510848125,
      "grad_norm": 0.34765625,
      "learning_rate": 7.315510341465363e-05,
      "loss": 0.0344,
      "step": 12734
    },
    {
      "epoch": 3.1400394477317555,
      "grad_norm": 0.416015625,
      "learning_rate": 7.312194395987882e-05,
      "loss": 0.0422,
      "step": 12736
    },
    {
      "epoch": 3.140532544378698,
      "grad_norm": 0.32421875,
      "learning_rate": 7.308878769030376e-05,
      "loss": 0.0363,
      "step": 12738
    },
    {
      "epoch": 3.141025641025641,
      "grad_norm": 0.431640625,
      "learning_rate": 7.30556346098576e-05,
      "loss": 0.0387,
      "step": 12740
    },
    {
      "epoch": 3.141518737672584,
      "grad_norm": 0.4609375,
      "learning_rate": 7.302248472246919e-05,
      "loss": 0.0455,
      "step": 12742
    },
    {
      "epoch": 3.1420118343195265,
      "grad_norm": 0.328125,
      "learning_rate": 7.298933803206691e-05,
      "loss": 0.0384,
      "step": 12744
    },
    {
      "epoch": 3.1425049309664694,
      "grad_norm": 0.3046875,
      "learning_rate": 7.295619454257894e-05,
      "loss": 0.0384,
      "step": 12746
    },
    {
      "epoch": 3.1429980276134124,
      "grad_norm": 0.478515625,
      "learning_rate": 7.292305425793291e-05,
      "loss": 0.0388,
      "step": 12748
    },
    {
      "epoch": 3.143491124260355,
      "grad_norm": 0.44921875,
      "learning_rate": 7.288991718205609e-05,
      "loss": 0.0406,
      "step": 12750
    },
    {
      "epoch": 3.143984220907298,
      "grad_norm": 0.32421875,
      "learning_rate": 7.285678331887547e-05,
      "loss": 0.0396,
      "step": 12752
    },
    {
      "epoch": 3.1444773175542404,
      "grad_norm": 0.435546875,
      "learning_rate": 7.282365267231756e-05,
      "loss": 0.0353,
      "step": 12754
    },
    {
      "epoch": 3.1449704142011834,
      "grad_norm": 0.3203125,
      "learning_rate": 7.279052524630858e-05,
      "loss": 0.0346,
      "step": 12756
    },
    {
      "epoch": 3.1454635108481264,
      "grad_norm": 0.34765625,
      "learning_rate": 7.275740104477422e-05,
      "loss": 0.04,
      "step": 12758
    },
    {
      "epoch": 3.145956607495069,
      "grad_norm": 0.47265625,
      "learning_rate": 7.272428007163998e-05,
      "loss": 0.0377,
      "step": 12760
    },
    {
      "epoch": 3.146449704142012,
      "grad_norm": 0.365234375,
      "learning_rate": 7.269116233083082e-05,
      "loss": 0.0375,
      "step": 12762
    },
    {
      "epoch": 3.146942800788955,
      "grad_norm": 0.2890625,
      "learning_rate": 7.265804782627141e-05,
      "loss": 0.0374,
      "step": 12764
    },
    {
      "epoch": 3.1474358974358974,
      "grad_norm": 0.37890625,
      "learning_rate": 7.2624936561886e-05,
      "loss": 0.0417,
      "step": 12766
    },
    {
      "epoch": 3.1479289940828403,
      "grad_norm": 0.291015625,
      "learning_rate": 7.259182854159845e-05,
      "loss": 0.0403,
      "step": 12768
    },
    {
      "epoch": 3.148422090729783,
      "grad_norm": 0.357421875,
      "learning_rate": 7.255872376933223e-05,
      "loss": 0.0426,
      "step": 12770
    },
    {
      "epoch": 3.148915187376726,
      "grad_norm": 0.32421875,
      "learning_rate": 7.252562224901044e-05,
      "loss": 0.0378,
      "step": 12772
    },
    {
      "epoch": 3.149408284023669,
      "grad_norm": 0.359375,
      "learning_rate": 7.249252398455584e-05,
      "loss": 0.0399,
      "step": 12774
    },
    {
      "epoch": 3.1499013806706113,
      "grad_norm": 0.380859375,
      "learning_rate": 7.245942897989069e-05,
      "loss": 0.0424,
      "step": 12776
    },
    {
      "epoch": 3.1503944773175543,
      "grad_norm": 0.390625,
      "learning_rate": 7.242633723893701e-05,
      "loss": 0.0391,
      "step": 12778
    },
    {
      "epoch": 3.1508875739644973,
      "grad_norm": 0.294921875,
      "learning_rate": 7.239324876561631e-05,
      "loss": 0.0361,
      "step": 12780
    },
    {
      "epoch": 3.15138067061144,
      "grad_norm": 0.341796875,
      "learning_rate": 7.236016356384978e-05,
      "loss": 0.0402,
      "step": 12782
    },
    {
      "epoch": 3.1518737672583828,
      "grad_norm": 0.3046875,
      "learning_rate": 7.23270816375582e-05,
      "loss": 0.0358,
      "step": 12784
    },
    {
      "epoch": 3.1523668639053253,
      "grad_norm": 0.30859375,
      "learning_rate": 7.22940029906619e-05,
      "loss": 0.038,
      "step": 12786
    },
    {
      "epoch": 3.1528599605522682,
      "grad_norm": 0.396484375,
      "learning_rate": 7.226092762708099e-05,
      "loss": 0.0382,
      "step": 12788
    },
    {
      "epoch": 3.153353057199211,
      "grad_norm": 0.384765625,
      "learning_rate": 7.222785555073496e-05,
      "loss": 0.0404,
      "step": 12790
    },
    {
      "epoch": 3.1538461538461537,
      "grad_norm": 0.328125,
      "learning_rate": 7.219478676554318e-05,
      "loss": 0.041,
      "step": 12792
    },
    {
      "epoch": 3.1543392504930967,
      "grad_norm": 0.29296875,
      "learning_rate": 7.216172127542437e-05,
      "loss": 0.0356,
      "step": 12794
    },
    {
      "epoch": 3.1548323471400392,
      "grad_norm": 0.357421875,
      "learning_rate": 7.212865908429705e-05,
      "loss": 0.0391,
      "step": 12796
    },
    {
      "epoch": 3.155325443786982,
      "grad_norm": 0.306640625,
      "learning_rate": 7.209560019607924e-05,
      "loss": 0.0374,
      "step": 12798
    },
    {
      "epoch": 3.155818540433925,
      "grad_norm": 0.359375,
      "learning_rate": 7.206254461468858e-05,
      "loss": 0.0426,
      "step": 12800
    },
    {
      "epoch": 3.1563116370808677,
      "grad_norm": 0.349609375,
      "learning_rate": 7.202949234404239e-05,
      "loss": 0.0349,
      "step": 12802
    },
    {
      "epoch": 3.1568047337278107,
      "grad_norm": 0.462890625,
      "learning_rate": 7.199644338805749e-05,
      "loss": 0.0379,
      "step": 12804
    },
    {
      "epoch": 3.1572978303747536,
      "grad_norm": 0.345703125,
      "learning_rate": 7.196339775065042e-05,
      "loss": 0.0371,
      "step": 12806
    },
    {
      "epoch": 3.157790927021696,
      "grad_norm": 0.359375,
      "learning_rate": 7.193035543573725e-05,
      "loss": 0.0404,
      "step": 12808
    },
    {
      "epoch": 3.158284023668639,
      "grad_norm": 0.322265625,
      "learning_rate": 7.189731644723371e-05,
      "loss": 0.0391,
      "step": 12810
    },
    {
      "epoch": 3.1587771203155817,
      "grad_norm": 0.359375,
      "learning_rate": 7.186428078905503e-05,
      "loss": 0.0406,
      "step": 12812
    },
    {
      "epoch": 3.1592702169625246,
      "grad_norm": 0.53515625,
      "learning_rate": 7.18312484651162e-05,
      "loss": 0.0458,
      "step": 12814
    },
    {
      "epoch": 3.1597633136094676,
      "grad_norm": 0.412109375,
      "learning_rate": 7.17982194793317e-05,
      "loss": 0.0441,
      "step": 12816
    },
    {
      "epoch": 3.16025641025641,
      "grad_norm": 0.345703125,
      "learning_rate": 7.176519383561562e-05,
      "loss": 0.0365,
      "step": 12818
    },
    {
      "epoch": 3.160749506903353,
      "grad_norm": 0.42578125,
      "learning_rate": 7.173217153788174e-05,
      "loss": 0.0408,
      "step": 12820
    },
    {
      "epoch": 3.1612426035502956,
      "grad_norm": 0.36328125,
      "learning_rate": 7.169915259004336e-05,
      "loss": 0.0429,
      "step": 12822
    },
    {
      "epoch": 3.1617357001972386,
      "grad_norm": 0.412109375,
      "learning_rate": 7.166613699601341e-05,
      "loss": 0.0503,
      "step": 12824
    },
    {
      "epoch": 3.1622287968441816,
      "grad_norm": 0.359375,
      "learning_rate": 7.163312475970438e-05,
      "loss": 0.0383,
      "step": 12826
    },
    {
      "epoch": 3.162721893491124,
      "grad_norm": 0.41015625,
      "learning_rate": 7.160011588502851e-05,
      "loss": 0.0365,
      "step": 12828
    },
    {
      "epoch": 3.163214990138067,
      "grad_norm": 0.361328125,
      "learning_rate": 7.156711037589741e-05,
      "loss": 0.0412,
      "step": 12830
    },
    {
      "epoch": 3.16370808678501,
      "grad_norm": 0.373046875,
      "learning_rate": 7.153410823622253e-05,
      "loss": 0.0433,
      "step": 12832
    },
    {
      "epoch": 3.1642011834319526,
      "grad_norm": 0.302734375,
      "learning_rate": 7.150110946991476e-05,
      "loss": 0.0388,
      "step": 12834
    },
    {
      "epoch": 3.1646942800788955,
      "grad_norm": 0.369140625,
      "learning_rate": 7.14681140808846e-05,
      "loss": 0.0354,
      "step": 12836
    },
    {
      "epoch": 3.1651873767258385,
      "grad_norm": 0.34765625,
      "learning_rate": 7.143512207304226e-05,
      "loss": 0.0389,
      "step": 12838
    },
    {
      "epoch": 3.165680473372781,
      "grad_norm": 0.31640625,
      "learning_rate": 7.140213345029741e-05,
      "loss": 0.0403,
      "step": 12840
    },
    {
      "epoch": 3.166173570019724,
      "grad_norm": 0.35546875,
      "learning_rate": 7.136914821655943e-05,
      "loss": 0.0369,
      "step": 12842
    },
    {
      "epoch": 3.1666666666666665,
      "grad_norm": 0.345703125,
      "learning_rate": 7.133616637573721e-05,
      "loss": 0.0404,
      "step": 12844
    },
    {
      "epoch": 3.1671597633136095,
      "grad_norm": 0.306640625,
      "learning_rate": 7.130318793173936e-05,
      "loss": 0.0423,
      "step": 12846
    },
    {
      "epoch": 3.1676528599605525,
      "grad_norm": 0.353515625,
      "learning_rate": 7.127021288847395e-05,
      "loss": 0.0424,
      "step": 12848
    },
    {
      "epoch": 3.168145956607495,
      "grad_norm": 0.322265625,
      "learning_rate": 7.12372412498487e-05,
      "loss": 0.0347,
      "step": 12850
    },
    {
      "epoch": 3.168639053254438,
      "grad_norm": 0.333984375,
      "learning_rate": 7.120427301977097e-05,
      "loss": 0.039,
      "step": 12852
    },
    {
      "epoch": 3.1691321499013805,
      "grad_norm": 0.388671875,
      "learning_rate": 7.117130820214764e-05,
      "loss": 0.0422,
      "step": 12854
    },
    {
      "epoch": 3.1696252465483234,
      "grad_norm": 0.361328125,
      "learning_rate": 7.113834680088527e-05,
      "loss": 0.0429,
      "step": 12856
    },
    {
      "epoch": 3.1701183431952664,
      "grad_norm": 0.357421875,
      "learning_rate": 7.110538881988992e-05,
      "loss": 0.041,
      "step": 12858
    },
    {
      "epoch": 3.170611439842209,
      "grad_norm": 0.390625,
      "learning_rate": 7.107243426306736e-05,
      "loss": 0.0414,
      "step": 12860
    },
    {
      "epoch": 3.171104536489152,
      "grad_norm": 0.349609375,
      "learning_rate": 7.103948313432283e-05,
      "loss": 0.0404,
      "step": 12862
    },
    {
      "epoch": 3.171597633136095,
      "grad_norm": 0.41015625,
      "learning_rate": 7.100653543756125e-05,
      "loss": 0.0425,
      "step": 12864
    },
    {
      "epoch": 3.1720907297830374,
      "grad_norm": 0.326171875,
      "learning_rate": 7.097359117668711e-05,
      "loss": 0.0432,
      "step": 12866
    },
    {
      "epoch": 3.1725838264299804,
      "grad_norm": 0.294921875,
      "learning_rate": 7.094065035560443e-05,
      "loss": 0.0362,
      "step": 12868
    },
    {
      "epoch": 3.173076923076923,
      "grad_norm": 0.349609375,
      "learning_rate": 7.090771297821698e-05,
      "loss": 0.045,
      "step": 12870
    },
    {
      "epoch": 3.173570019723866,
      "grad_norm": 0.388671875,
      "learning_rate": 7.087477904842793e-05,
      "loss": 0.0472,
      "step": 12872
    },
    {
      "epoch": 3.174063116370809,
      "grad_norm": 0.322265625,
      "learning_rate": 7.084184857014023e-05,
      "loss": 0.0398,
      "step": 12874
    },
    {
      "epoch": 3.1745562130177514,
      "grad_norm": 0.310546875,
      "learning_rate": 7.080892154725622e-05,
      "loss": 0.037,
      "step": 12876
    },
    {
      "epoch": 3.1750493096646943,
      "grad_norm": 0.35546875,
      "learning_rate": 7.077599798367804e-05,
      "loss": 0.0481,
      "step": 12878
    },
    {
      "epoch": 3.175542406311637,
      "grad_norm": 0.353515625,
      "learning_rate": 7.074307788330725e-05,
      "loss": 0.0427,
      "step": 12880
    },
    {
      "epoch": 3.17603550295858,
      "grad_norm": 0.369140625,
      "learning_rate": 7.071016125004503e-05,
      "loss": 0.0442,
      "step": 12882
    },
    {
      "epoch": 3.176528599605523,
      "grad_norm": 0.38671875,
      "learning_rate": 7.067724808779231e-05,
      "loss": 0.0429,
      "step": 12884
    },
    {
      "epoch": 3.1770216962524653,
      "grad_norm": 0.345703125,
      "learning_rate": 7.064433840044933e-05,
      "loss": 0.0456,
      "step": 12886
    },
    {
      "epoch": 3.1775147928994083,
      "grad_norm": 0.357421875,
      "learning_rate": 7.061143219191621e-05,
      "loss": 0.0447,
      "step": 12888
    },
    {
      "epoch": 3.1780078895463513,
      "grad_norm": 0.357421875,
      "learning_rate": 7.057852946609244e-05,
      "loss": 0.0401,
      "step": 12890
    },
    {
      "epoch": 3.178500986193294,
      "grad_norm": 0.3359375,
      "learning_rate": 7.054563022687722e-05,
      "loss": 0.0375,
      "step": 12892
    },
    {
      "epoch": 3.1789940828402368,
      "grad_norm": 0.326171875,
      "learning_rate": 7.051273447816922e-05,
      "loss": 0.0393,
      "step": 12894
    },
    {
      "epoch": 3.1794871794871793,
      "grad_norm": 0.359375,
      "learning_rate": 7.047984222386686e-05,
      "loss": 0.0431,
      "step": 12896
    },
    {
      "epoch": 3.1799802761341223,
      "grad_norm": 0.32421875,
      "learning_rate": 7.044695346786802e-05,
      "loss": 0.0395,
      "step": 12898
    },
    {
      "epoch": 3.1804733727810652,
      "grad_norm": 0.337890625,
      "learning_rate": 7.041406821407016e-05,
      "loss": 0.0409,
      "step": 12900
    },
    {
      "epoch": 3.1809664694280078,
      "grad_norm": 0.353515625,
      "learning_rate": 7.038118646637044e-05,
      "loss": 0.0402,
      "step": 12902
    },
    {
      "epoch": 3.1814595660749507,
      "grad_norm": 0.3515625,
      "learning_rate": 7.034830822866549e-05,
      "loss": 0.0411,
      "step": 12904
    },
    {
      "epoch": 3.1819526627218937,
      "grad_norm": 0.32421875,
      "learning_rate": 7.031543350485157e-05,
      "loss": 0.0375,
      "step": 12906
    },
    {
      "epoch": 3.1824457593688362,
      "grad_norm": 0.357421875,
      "learning_rate": 7.028256229882448e-05,
      "loss": 0.0378,
      "step": 12908
    },
    {
      "epoch": 3.182938856015779,
      "grad_norm": 0.330078125,
      "learning_rate": 7.024969461447972e-05,
      "loss": 0.0413,
      "step": 12910
    },
    {
      "epoch": 3.1834319526627217,
      "grad_norm": 0.310546875,
      "learning_rate": 7.02168304557122e-05,
      "loss": 0.0403,
      "step": 12912
    },
    {
      "epoch": 3.1839250493096647,
      "grad_norm": 0.3359375,
      "learning_rate": 7.018396982641662e-05,
      "loss": 0.0381,
      "step": 12914
    },
    {
      "epoch": 3.1844181459566077,
      "grad_norm": 0.373046875,
      "learning_rate": 7.015111273048708e-05,
      "loss": 0.0403,
      "step": 12916
    },
    {
      "epoch": 3.18491124260355,
      "grad_norm": 0.392578125,
      "learning_rate": 7.011825917181732e-05,
      "loss": 0.0457,
      "step": 12918
    },
    {
      "epoch": 3.185404339250493,
      "grad_norm": 0.423828125,
      "learning_rate": 7.00854091543007e-05,
      "loss": 0.0406,
      "step": 12920
    },
    {
      "epoch": 3.185897435897436,
      "grad_norm": 0.359375,
      "learning_rate": 7.005256268183007e-05,
      "loss": 0.0412,
      "step": 12922
    },
    {
      "epoch": 3.1863905325443787,
      "grad_norm": 0.326171875,
      "learning_rate": 7.001971975829801e-05,
      "loss": 0.0392,
      "step": 12924
    },
    {
      "epoch": 3.1868836291913216,
      "grad_norm": 0.3203125,
      "learning_rate": 6.998688038759651e-05,
      "loss": 0.0426,
      "step": 12926
    },
    {
      "epoch": 3.187376725838264,
      "grad_norm": 0.3671875,
      "learning_rate": 6.99540445736173e-05,
      "loss": 0.0457,
      "step": 12928
    },
    {
      "epoch": 3.187869822485207,
      "grad_norm": 0.30859375,
      "learning_rate": 6.992121232025157e-05,
      "loss": 0.0408,
      "step": 12930
    },
    {
      "epoch": 3.18836291913215,
      "grad_norm": 0.3125,
      "learning_rate": 6.988838363139009e-05,
      "loss": 0.0396,
      "step": 12932
    },
    {
      "epoch": 3.1888560157790926,
      "grad_norm": 0.376953125,
      "learning_rate": 6.985555851092329e-05,
      "loss": 0.0425,
      "step": 12934
    },
    {
      "epoch": 3.1893491124260356,
      "grad_norm": 0.37890625,
      "learning_rate": 6.982273696274106e-05,
      "loss": 0.0455,
      "step": 12936
    },
    {
      "epoch": 3.189842209072978,
      "grad_norm": 0.3671875,
      "learning_rate": 6.978991899073306e-05,
      "loss": 0.0478,
      "step": 12938
    },
    {
      "epoch": 3.190335305719921,
      "grad_norm": 0.373046875,
      "learning_rate": 6.975710459878825e-05,
      "loss": 0.0403,
      "step": 12940
    },
    {
      "epoch": 3.190828402366864,
      "grad_norm": 0.3671875,
      "learning_rate": 6.972429379079548e-05,
      "loss": 0.0432,
      "step": 12942
    },
    {
      "epoch": 3.1913214990138066,
      "grad_norm": 0.375,
      "learning_rate": 6.969148657064289e-05,
      "loss": 0.0433,
      "step": 12944
    },
    {
      "epoch": 3.1918145956607495,
      "grad_norm": 0.349609375,
      "learning_rate": 6.965868294221839e-05,
      "loss": 0.042,
      "step": 12946
    },
    {
      "epoch": 3.1923076923076925,
      "grad_norm": 0.322265625,
      "learning_rate": 6.962588290940937e-05,
      "loss": 0.0452,
      "step": 12948
    },
    {
      "epoch": 3.192800788954635,
      "grad_norm": 0.357421875,
      "learning_rate": 6.959308647610276e-05,
      "loss": 0.0414,
      "step": 12950
    },
    {
      "epoch": 3.193293885601578,
      "grad_norm": 0.345703125,
      "learning_rate": 6.956029364618521e-05,
      "loss": 0.0416,
      "step": 12952
    },
    {
      "epoch": 3.1937869822485205,
      "grad_norm": 0.369140625,
      "learning_rate": 6.952750442354278e-05,
      "loss": 0.0409,
      "step": 12954
    },
    {
      "epoch": 3.1942800788954635,
      "grad_norm": 0.345703125,
      "learning_rate": 6.949471881206126e-05,
      "loss": 0.0416,
      "step": 12956
    },
    {
      "epoch": 3.1947731755424065,
      "grad_norm": 0.3125,
      "learning_rate": 6.946193681562588e-05,
      "loss": 0.0452,
      "step": 12958
    },
    {
      "epoch": 3.195266272189349,
      "grad_norm": 0.39453125,
      "learning_rate": 6.942915843812148e-05,
      "loss": 0.0447,
      "step": 12960
    },
    {
      "epoch": 3.195759368836292,
      "grad_norm": 0.349609375,
      "learning_rate": 6.939638368343246e-05,
      "loss": 0.0429,
      "step": 12962
    },
    {
      "epoch": 3.1962524654832345,
      "grad_norm": 0.353515625,
      "learning_rate": 6.936361255544289e-05,
      "loss": 0.047,
      "step": 12964
    },
    {
      "epoch": 3.1967455621301775,
      "grad_norm": 0.34375,
      "learning_rate": 6.933084505803626e-05,
      "loss": 0.0494,
      "step": 12966
    },
    {
      "epoch": 3.1972386587771204,
      "grad_norm": 0.390625,
      "learning_rate": 6.92980811950957e-05,
      "loss": 0.0499,
      "step": 12968
    },
    {
      "epoch": 3.197731755424063,
      "grad_norm": 0.337890625,
      "learning_rate": 6.926532097050398e-05,
      "loss": 0.0396,
      "step": 12970
    },
    {
      "epoch": 3.198224852071006,
      "grad_norm": 0.396484375,
      "learning_rate": 6.923256438814331e-05,
      "loss": 0.0363,
      "step": 12972
    },
    {
      "epoch": 3.198717948717949,
      "grad_norm": 0.359375,
      "learning_rate": 6.919981145189556e-05,
      "loss": 0.0423,
      "step": 12974
    },
    {
      "epoch": 3.1992110453648914,
      "grad_norm": 0.361328125,
      "learning_rate": 6.916706216564206e-05,
      "loss": 0.0431,
      "step": 12976
    },
    {
      "epoch": 3.1997041420118344,
      "grad_norm": 0.353515625,
      "learning_rate": 6.91343165332639e-05,
      "loss": 0.0409,
      "step": 12978
    },
    {
      "epoch": 3.200197238658777,
      "grad_norm": 0.34375,
      "learning_rate": 6.910157455864156e-05,
      "loss": 0.0433,
      "step": 12980
    },
    {
      "epoch": 3.20069033530572,
      "grad_norm": 0.345703125,
      "learning_rate": 6.906883624565512e-05,
      "loss": 0.0462,
      "step": 12982
    },
    {
      "epoch": 3.201183431952663,
      "grad_norm": 0.330078125,
      "learning_rate": 6.90361015981843e-05,
      "loss": 0.0394,
      "step": 12984
    },
    {
      "epoch": 3.2016765285996054,
      "grad_norm": 0.326171875,
      "learning_rate": 6.900337062010831e-05,
      "loss": 0.0409,
      "step": 12986
    },
    {
      "epoch": 3.2021696252465484,
      "grad_norm": 0.380859375,
      "learning_rate": 6.8970643315306e-05,
      "loss": 0.0457,
      "step": 12988
    },
    {
      "epoch": 3.2026627218934913,
      "grad_norm": 0.33984375,
      "learning_rate": 6.893791968765564e-05,
      "loss": 0.0485,
      "step": 12990
    },
    {
      "epoch": 3.203155818540434,
      "grad_norm": 0.412109375,
      "learning_rate": 6.890519974103528e-05,
      "loss": 0.0426,
      "step": 12992
    },
    {
      "epoch": 3.203648915187377,
      "grad_norm": 0.328125,
      "learning_rate": 6.887248347932232e-05,
      "loss": 0.0383,
      "step": 12994
    },
    {
      "epoch": 3.2041420118343193,
      "grad_norm": 0.416015625,
      "learning_rate": 6.883977090639389e-05,
      "loss": 0.0373,
      "step": 12996
    },
    {
      "epoch": 3.2046351084812623,
      "grad_norm": 0.32421875,
      "learning_rate": 6.880706202612659e-05,
      "loss": 0.0429,
      "step": 12998
    },
    {
      "epoch": 3.2051282051282053,
      "grad_norm": 0.34765625,
      "learning_rate": 6.877435684239659e-05,
      "loss": 0.0418,
      "step": 13000
    },
    {
      "epoch": 3.205621301775148,
      "grad_norm": 0.44921875,
      "learning_rate": 6.874165535907967e-05,
      "loss": 0.0408,
      "step": 13002
    },
    {
      "epoch": 3.206114398422091,
      "grad_norm": 0.361328125,
      "learning_rate": 6.870895758005105e-05,
      "loss": 0.0433,
      "step": 13004
    },
    {
      "epoch": 3.2066074950690338,
      "grad_norm": 0.3515625,
      "learning_rate": 6.867626350918571e-05,
      "loss": 0.0439,
      "step": 13006
    },
    {
      "epoch": 3.2071005917159763,
      "grad_norm": 0.3515625,
      "learning_rate": 6.864357315035801e-05,
      "loss": 0.0463,
      "step": 13008
    },
    {
      "epoch": 3.2075936883629192,
      "grad_norm": 0.5703125,
      "learning_rate": 6.861088650744198e-05,
      "loss": 0.0428,
      "step": 13010
    },
    {
      "epoch": 3.2080867850098618,
      "grad_norm": 0.330078125,
      "learning_rate": 6.857820358431117e-05,
      "loss": 0.0419,
      "step": 13012
    },
    {
      "epoch": 3.2085798816568047,
      "grad_norm": 0.376953125,
      "learning_rate": 6.854552438483865e-05,
      "loss": 0.0438,
      "step": 13014
    },
    {
      "epoch": 3.2090729783037477,
      "grad_norm": 0.51953125,
      "learning_rate": 6.851284891289713e-05,
      "loss": 0.0469,
      "step": 13016
    },
    {
      "epoch": 3.2095660749506902,
      "grad_norm": 0.5546875,
      "learning_rate": 6.848017717235877e-05,
      "loss": 0.0456,
      "step": 13018
    },
    {
      "epoch": 3.210059171597633,
      "grad_norm": 0.337890625,
      "learning_rate": 6.844750916709544e-05,
      "loss": 0.0455,
      "step": 13020
    },
    {
      "epoch": 3.2105522682445757,
      "grad_norm": 0.380859375,
      "learning_rate": 6.84148449009784e-05,
      "loss": 0.0481,
      "step": 13022
    },
    {
      "epoch": 3.2110453648915187,
      "grad_norm": 0.333984375,
      "learning_rate": 6.838218437787863e-05,
      "loss": 0.048,
      "step": 13024
    },
    {
      "epoch": 3.2115384615384617,
      "grad_norm": 0.328125,
      "learning_rate": 6.834952760166653e-05,
      "loss": 0.0412,
      "step": 13026
    },
    {
      "epoch": 3.212031558185404,
      "grad_norm": 0.375,
      "learning_rate": 6.831687457621214e-05,
      "loss": 0.0482,
      "step": 13028
    },
    {
      "epoch": 3.212524654832347,
      "grad_norm": 0.31640625,
      "learning_rate": 6.828422530538501e-05,
      "loss": 0.0442,
      "step": 13030
    },
    {
      "epoch": 3.21301775147929,
      "grad_norm": 0.3359375,
      "learning_rate": 6.82515797930542e-05,
      "loss": 0.0486,
      "step": 13032
    },
    {
      "epoch": 3.2135108481262327,
      "grad_norm": 0.384765625,
      "learning_rate": 6.82189380430885e-05,
      "loss": 0.042,
      "step": 13034
    },
    {
      "epoch": 3.2140039447731756,
      "grad_norm": 0.3359375,
      "learning_rate": 6.818630005935605e-05,
      "loss": 0.0427,
      "step": 13036
    },
    {
      "epoch": 3.214497041420118,
      "grad_norm": 0.390625,
      "learning_rate": 6.81536658457247e-05,
      "loss": 0.0419,
      "step": 13038
    },
    {
      "epoch": 3.214990138067061,
      "grad_norm": 0.3515625,
      "learning_rate": 6.812103540606174e-05,
      "loss": 0.0453,
      "step": 13040
    },
    {
      "epoch": 3.215483234714004,
      "grad_norm": 0.345703125,
      "learning_rate": 6.808840874423407e-05,
      "loss": 0.0414,
      "step": 13042
    },
    {
      "epoch": 3.2159763313609466,
      "grad_norm": 0.36328125,
      "learning_rate": 6.805578586410812e-05,
      "loss": 0.0468,
      "step": 13044
    },
    {
      "epoch": 3.2164694280078896,
      "grad_norm": 0.390625,
      "learning_rate": 6.802316676954993e-05,
      "loss": 0.0453,
      "step": 13046
    },
    {
      "epoch": 3.216962524654832,
      "grad_norm": 0.3515625,
      "learning_rate": 6.799055146442503e-05,
      "loss": 0.0432,
      "step": 13048
    },
    {
      "epoch": 3.217455621301775,
      "grad_norm": 0.384765625,
      "learning_rate": 6.795793995259845e-05,
      "loss": 0.0429,
      "step": 13050
    },
    {
      "epoch": 3.217948717948718,
      "grad_norm": 0.36328125,
      "learning_rate": 6.792533223793492e-05,
      "loss": 0.0441,
      "step": 13052
    },
    {
      "epoch": 3.2184418145956606,
      "grad_norm": 0.310546875,
      "learning_rate": 6.789272832429858e-05,
      "loss": 0.0438,
      "step": 13054
    },
    {
      "epoch": 3.2189349112426036,
      "grad_norm": 0.314453125,
      "learning_rate": 6.786012821555324e-05,
      "loss": 0.0447,
      "step": 13056
    },
    {
      "epoch": 3.2194280078895465,
      "grad_norm": 0.4375,
      "learning_rate": 6.782753191556208e-05,
      "loss": 0.0452,
      "step": 13058
    },
    {
      "epoch": 3.219921104536489,
      "grad_norm": 0.380859375,
      "learning_rate": 6.779493942818807e-05,
      "loss": 0.0423,
      "step": 13060
    },
    {
      "epoch": 3.220414201183432,
      "grad_norm": 0.41796875,
      "learning_rate": 6.776235075729356e-05,
      "loss": 0.0411,
      "step": 13062
    },
    {
      "epoch": 3.2209072978303745,
      "grad_norm": 0.41796875,
      "learning_rate": 6.772976590674042e-05,
      "loss": 0.0429,
      "step": 13064
    },
    {
      "epoch": 3.2214003944773175,
      "grad_norm": 0.34375,
      "learning_rate": 6.769718488039023e-05,
      "loss": 0.0446,
      "step": 13066
    },
    {
      "epoch": 3.2218934911242605,
      "grad_norm": 0.314453125,
      "learning_rate": 6.766460768210399e-05,
      "loss": 0.0424,
      "step": 13068
    },
    {
      "epoch": 3.222386587771203,
      "grad_norm": 0.314453125,
      "learning_rate": 6.763203431574227e-05,
      "loss": 0.0428,
      "step": 13070
    },
    {
      "epoch": 3.222879684418146,
      "grad_norm": 0.3359375,
      "learning_rate": 6.759946478516519e-05,
      "loss": 0.0439,
      "step": 13072
    },
    {
      "epoch": 3.223372781065089,
      "grad_norm": 0.33984375,
      "learning_rate": 6.756689909423247e-05,
      "loss": 0.0436,
      "step": 13074
    },
    {
      "epoch": 3.2238658777120315,
      "grad_norm": 0.318359375,
      "learning_rate": 6.753433724680323e-05,
      "loss": 0.042,
      "step": 13076
    },
    {
      "epoch": 3.2243589743589745,
      "grad_norm": 0.31640625,
      "learning_rate": 6.750177924673634e-05,
      "loss": 0.0454,
      "step": 13078
    },
    {
      "epoch": 3.224852071005917,
      "grad_norm": 0.31640625,
      "learning_rate": 6.746922509789006e-05,
      "loss": 0.0391,
      "step": 13080
    },
    {
      "epoch": 3.22534516765286,
      "grad_norm": 0.341796875,
      "learning_rate": 6.743667480412222e-05,
      "loss": 0.0433,
      "step": 13082
    },
    {
      "epoch": 3.225838264299803,
      "grad_norm": 0.39453125,
      "learning_rate": 6.740412836929025e-05,
      "loss": 0.0431,
      "step": 13084
    },
    {
      "epoch": 3.2263313609467454,
      "grad_norm": 0.345703125,
      "learning_rate": 6.7371585797251e-05,
      "loss": 0.0449,
      "step": 13086
    },
    {
      "epoch": 3.2268244575936884,
      "grad_norm": 0.318359375,
      "learning_rate": 6.733904709186105e-05,
      "loss": 0.0421,
      "step": 13088
    },
    {
      "epoch": 3.2273175542406314,
      "grad_norm": 0.3515625,
      "learning_rate": 6.730651225697635e-05,
      "loss": 0.0472,
      "step": 13090
    },
    {
      "epoch": 3.227810650887574,
      "grad_norm": 0.359375,
      "learning_rate": 6.727398129645252e-05,
      "loss": 0.0457,
      "step": 13092
    },
    {
      "epoch": 3.228303747534517,
      "grad_norm": 0.345703125,
      "learning_rate": 6.724145421414461e-05,
      "loss": 0.047,
      "step": 13094
    },
    {
      "epoch": 3.2287968441814594,
      "grad_norm": 0.328125,
      "learning_rate": 6.720893101390728e-05,
      "loss": 0.044,
      "step": 13096
    },
    {
      "epoch": 3.2292899408284024,
      "grad_norm": 0.37109375,
      "learning_rate": 6.71764116995947e-05,
      "loss": 0.0444,
      "step": 13098
    },
    {
      "epoch": 3.2297830374753453,
      "grad_norm": 0.330078125,
      "learning_rate": 6.714389627506059e-05,
      "loss": 0.0428,
      "step": 13100
    },
    {
      "epoch": 3.230276134122288,
      "grad_norm": 0.37890625,
      "learning_rate": 6.711138474415823e-05,
      "loss": 0.0486,
      "step": 13102
    },
    {
      "epoch": 3.230769230769231,
      "grad_norm": 0.376953125,
      "learning_rate": 6.707887711074037e-05,
      "loss": 0.0457,
      "step": 13104
    },
    {
      "epoch": 3.2312623274161734,
      "grad_norm": 0.326171875,
      "learning_rate": 6.704637337865942e-05,
      "loss": 0.0449,
      "step": 13106
    },
    {
      "epoch": 3.2317554240631163,
      "grad_norm": 0.369140625,
      "learning_rate": 6.701387355176721e-05,
      "loss": 0.0448,
      "step": 13108
    },
    {
      "epoch": 3.2322485207100593,
      "grad_norm": 0.3515625,
      "learning_rate": 6.698137763391516e-05,
      "loss": 0.0414,
      "step": 13110
    },
    {
      "epoch": 3.232741617357002,
      "grad_norm": 0.341796875,
      "learning_rate": 6.694888562895419e-05,
      "loss": 0.0451,
      "step": 13112
    },
    {
      "epoch": 3.233234714003945,
      "grad_norm": 0.365234375,
      "learning_rate": 6.691639754073482e-05,
      "loss": 0.0432,
      "step": 13114
    },
    {
      "epoch": 3.2337278106508878,
      "grad_norm": 0.345703125,
      "learning_rate": 6.688391337310706e-05,
      "loss": 0.0402,
      "step": 13116
    },
    {
      "epoch": 3.2342209072978303,
      "grad_norm": 0.33984375,
      "learning_rate": 6.685143312992044e-05,
      "loss": 0.0412,
      "step": 13118
    },
    {
      "epoch": 3.2347140039447733,
      "grad_norm": 0.314453125,
      "learning_rate": 6.68189568150241e-05,
      "loss": 0.0432,
      "step": 13120
    },
    {
      "epoch": 3.235207100591716,
      "grad_norm": 0.302734375,
      "learning_rate": 6.678648443226662e-05,
      "loss": 0.0418,
      "step": 13122
    },
    {
      "epoch": 3.2357001972386588,
      "grad_norm": 0.34375,
      "learning_rate": 6.675401598549619e-05,
      "loss": 0.0436,
      "step": 13124
    },
    {
      "epoch": 3.2361932938856017,
      "grad_norm": 0.333984375,
      "learning_rate": 6.672155147856045e-05,
      "loss": 0.0417,
      "step": 13126
    },
    {
      "epoch": 3.2366863905325443,
      "grad_norm": 0.353515625,
      "learning_rate": 6.668909091530671e-05,
      "loss": 0.0467,
      "step": 13128
    },
    {
      "epoch": 3.2371794871794872,
      "grad_norm": 0.400390625,
      "learning_rate": 6.665663429958167e-05,
      "loss": 0.0444,
      "step": 13130
    },
    {
      "epoch": 3.2376725838264298,
      "grad_norm": 0.3125,
      "learning_rate": 6.66241816352316e-05,
      "loss": 0.0453,
      "step": 13132
    },
    {
      "epoch": 3.2381656804733727,
      "grad_norm": 0.3984375,
      "learning_rate": 6.659173292610239e-05,
      "loss": 0.0425,
      "step": 13134
    },
    {
      "epoch": 3.2386587771203157,
      "grad_norm": 0.390625,
      "learning_rate": 6.655928817603933e-05,
      "loss": 0.0464,
      "step": 13136
    },
    {
      "epoch": 3.239151873767258,
      "grad_norm": 0.27734375,
      "learning_rate": 6.652684738888734e-05,
      "loss": 0.0409,
      "step": 13138
    },
    {
      "epoch": 3.239644970414201,
      "grad_norm": 0.345703125,
      "learning_rate": 6.649441056849082e-05,
      "loss": 0.043,
      "step": 13140
    },
    {
      "epoch": 3.240138067061144,
      "grad_norm": 0.36328125,
      "learning_rate": 6.646197771869374e-05,
      "loss": 0.0428,
      "step": 13142
    },
    {
      "epoch": 3.2406311637080867,
      "grad_norm": 0.341796875,
      "learning_rate": 6.642954884333955e-05,
      "loss": 0.0435,
      "step": 13144
    },
    {
      "epoch": 3.2411242603550297,
      "grad_norm": 0.359375,
      "learning_rate": 6.639712394627122e-05,
      "loss": 0.0452,
      "step": 13146
    },
    {
      "epoch": 3.2416173570019726,
      "grad_norm": 0.369140625,
      "learning_rate": 6.636470303133136e-05,
      "loss": 0.0426,
      "step": 13148
    },
    {
      "epoch": 3.242110453648915,
      "grad_norm": 0.341796875,
      "learning_rate": 6.633228610236195e-05,
      "loss": 0.0426,
      "step": 13150
    },
    {
      "epoch": 3.242603550295858,
      "grad_norm": 0.35546875,
      "learning_rate": 6.629987316320463e-05,
      "loss": 0.0445,
      "step": 13152
    },
    {
      "epoch": 3.2430966469428006,
      "grad_norm": 0.3984375,
      "learning_rate": 6.626746421770047e-05,
      "loss": 0.0417,
      "step": 13154
    },
    {
      "epoch": 3.2435897435897436,
      "grad_norm": 0.306640625,
      "learning_rate": 6.62350592696902e-05,
      "loss": 0.0424,
      "step": 13156
    },
    {
      "epoch": 3.2440828402366866,
      "grad_norm": 0.40625,
      "learning_rate": 6.620265832301386e-05,
      "loss": 0.044,
      "step": 13158
    },
    {
      "epoch": 3.244575936883629,
      "grad_norm": 0.35546875,
      "learning_rate": 6.617026138151125e-05,
      "loss": 0.0419,
      "step": 13160
    },
    {
      "epoch": 3.245069033530572,
      "grad_norm": 0.34375,
      "learning_rate": 6.613786844902155e-05,
      "loss": 0.0417,
      "step": 13162
    },
    {
      "epoch": 3.2455621301775146,
      "grad_norm": 0.376953125,
      "learning_rate": 6.610547952938349e-05,
      "loss": 0.0405,
      "step": 13164
    },
    {
      "epoch": 3.2460552268244576,
      "grad_norm": 0.35546875,
      "learning_rate": 6.607309462643534e-05,
      "loss": 0.0449,
      "step": 13166
    },
    {
      "epoch": 3.2465483234714005,
      "grad_norm": 0.380859375,
      "learning_rate": 6.60407137440149e-05,
      "loss": 0.0387,
      "step": 13168
    },
    {
      "epoch": 3.247041420118343,
      "grad_norm": 0.37109375,
      "learning_rate": 6.600833688595951e-05,
      "loss": 0.0437,
      "step": 13170
    },
    {
      "epoch": 3.247534516765286,
      "grad_norm": 0.4296875,
      "learning_rate": 6.597596405610593e-05,
      "loss": 0.048,
      "step": 13172
    },
    {
      "epoch": 3.248027613412229,
      "grad_norm": 0.396484375,
      "learning_rate": 6.594359525829063e-05,
      "loss": 0.0434,
      "step": 13174
    },
    {
      "epoch": 3.2485207100591715,
      "grad_norm": 0.408203125,
      "learning_rate": 6.591123049634942e-05,
      "loss": 0.0471,
      "step": 13176
    },
    {
      "epoch": 3.2490138067061145,
      "grad_norm": 0.375,
      "learning_rate": 6.587886977411773e-05,
      "loss": 0.0427,
      "step": 13178
    },
    {
      "epoch": 3.249506903353057,
      "grad_norm": 0.349609375,
      "learning_rate": 6.584651309543049e-05,
      "loss": 0.0492,
      "step": 13180
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.32421875,
      "learning_rate": 6.58141604641221e-05,
      "loss": 0.046,
      "step": 13182
    },
    {
      "epoch": 3.250493096646943,
      "grad_norm": 0.353515625,
      "learning_rate": 6.578181188402661e-05,
      "loss": 0.0442,
      "step": 13184
    },
    {
      "epoch": 3.2509861932938855,
      "grad_norm": 0.41796875,
      "learning_rate": 6.57494673589774e-05,
      "loss": 0.0417,
      "step": 13186
    },
    {
      "epoch": 3.2514792899408285,
      "grad_norm": 0.361328125,
      "learning_rate": 6.57171268928076e-05,
      "loss": 0.0435,
      "step": 13188
    },
    {
      "epoch": 3.251972386587771,
      "grad_norm": 0.388671875,
      "learning_rate": 6.568479048934964e-05,
      "loss": 0.0467,
      "step": 13190
    },
    {
      "epoch": 3.252465483234714,
      "grad_norm": 0.349609375,
      "learning_rate": 6.565245815243559e-05,
      "loss": 0.0444,
      "step": 13192
    },
    {
      "epoch": 3.252958579881657,
      "grad_norm": 0.359375,
      "learning_rate": 6.562012988589707e-05,
      "loss": 0.0427,
      "step": 13194
    },
    {
      "epoch": 3.2534516765285995,
      "grad_norm": 0.431640625,
      "learning_rate": 6.558780569356507e-05,
      "loss": 0.0411,
      "step": 13196
    },
    {
      "epoch": 3.2539447731755424,
      "grad_norm": 0.337890625,
      "learning_rate": 6.555548557927025e-05,
      "loss": 0.0448,
      "step": 13198
    },
    {
      "epoch": 3.2544378698224854,
      "grad_norm": 0.33984375,
      "learning_rate": 6.552316954684267e-05,
      "loss": 0.0442,
      "step": 13200
    },
    {
      "epoch": 3.254930966469428,
      "grad_norm": 0.330078125,
      "learning_rate": 6.549085760011202e-05,
      "loss": 0.0453,
      "step": 13202
    },
    {
      "epoch": 3.255424063116371,
      "grad_norm": 0.37109375,
      "learning_rate": 6.545854974290741e-05,
      "loss": 0.0463,
      "step": 13204
    },
    {
      "epoch": 3.2559171597633134,
      "grad_norm": 0.376953125,
      "learning_rate": 6.542624597905751e-05,
      "loss": 0.0444,
      "step": 13206
    },
    {
      "epoch": 3.2564102564102564,
      "grad_norm": 0.3515625,
      "learning_rate": 6.539394631239048e-05,
      "loss": 0.0499,
      "step": 13208
    },
    {
      "epoch": 3.2569033530571994,
      "grad_norm": 0.373046875,
      "learning_rate": 6.536165074673409e-05,
      "loss": 0.0439,
      "step": 13210
    },
    {
      "epoch": 3.257396449704142,
      "grad_norm": 0.357421875,
      "learning_rate": 6.532935928591544e-05,
      "loss": 0.0484,
      "step": 13212
    },
    {
      "epoch": 3.257889546351085,
      "grad_norm": 0.33203125,
      "learning_rate": 6.529707193376127e-05,
      "loss": 0.0487,
      "step": 13214
    },
    {
      "epoch": 3.2583826429980274,
      "grad_norm": 0.408203125,
      "learning_rate": 6.526478869409785e-05,
      "loss": 0.0479,
      "step": 13216
    },
    {
      "epoch": 3.2588757396449703,
      "grad_norm": 0.423828125,
      "learning_rate": 6.523250957075089e-05,
      "loss": 0.042,
      "step": 13218
    },
    {
      "epoch": 3.2593688362919133,
      "grad_norm": 0.390625,
      "learning_rate": 6.52002345675457e-05,
      "loss": 0.0494,
      "step": 13220
    },
    {
      "epoch": 3.259861932938856,
      "grad_norm": 0.36328125,
      "learning_rate": 6.516796368830698e-05,
      "loss": 0.0417,
      "step": 13222
    },
    {
      "epoch": 3.260355029585799,
      "grad_norm": 0.390625,
      "learning_rate": 6.513569693685907e-05,
      "loss": 0.0472,
      "step": 13224
    },
    {
      "epoch": 3.260848126232742,
      "grad_norm": 0.423828125,
      "learning_rate": 6.510343431702568e-05,
      "loss": 0.0485,
      "step": 13226
    },
    {
      "epoch": 3.2613412228796843,
      "grad_norm": 0.396484375,
      "learning_rate": 6.507117583263021e-05,
      "loss": 0.043,
      "step": 13228
    },
    {
      "epoch": 3.2618343195266273,
      "grad_norm": 0.376953125,
      "learning_rate": 6.503892148749542e-05,
      "loss": 0.0475,
      "step": 13230
    },
    {
      "epoch": 3.2623274161735702,
      "grad_norm": 0.330078125,
      "learning_rate": 6.50066712854436e-05,
      "loss": 0.0462,
      "step": 13232
    },
    {
      "epoch": 3.2628205128205128,
      "grad_norm": 0.43359375,
      "learning_rate": 6.497442523029663e-05,
      "loss": 0.0449,
      "step": 13234
    },
    {
      "epoch": 3.2633136094674557,
      "grad_norm": 0.328125,
      "learning_rate": 6.494218332587583e-05,
      "loss": 0.0411,
      "step": 13236
    },
    {
      "epoch": 3.2638067061143983,
      "grad_norm": 0.357421875,
      "learning_rate": 6.490994557600204e-05,
      "loss": 0.0477,
      "step": 13238
    },
    {
      "epoch": 3.2642998027613412,
      "grad_norm": 0.33984375,
      "learning_rate": 6.48777119844956e-05,
      "loss": 0.0452,
      "step": 13240
    },
    {
      "epoch": 3.264792899408284,
      "grad_norm": 0.380859375,
      "learning_rate": 6.484548255517642e-05,
      "loss": 0.046,
      "step": 13242
    },
    {
      "epoch": 3.2652859960552267,
      "grad_norm": 0.34765625,
      "learning_rate": 6.481325729186383e-05,
      "loss": 0.0421,
      "step": 13244
    },
    {
      "epoch": 3.2657790927021697,
      "grad_norm": 0.462890625,
      "learning_rate": 6.478103619837665e-05,
      "loss": 0.0468,
      "step": 13246
    },
    {
      "epoch": 3.2662721893491122,
      "grad_norm": 0.443359375,
      "learning_rate": 6.474881927853339e-05,
      "loss": 0.044,
      "step": 13248
    },
    {
      "epoch": 3.266765285996055,
      "grad_norm": 0.443359375,
      "learning_rate": 6.471660653615184e-05,
      "loss": 0.0443,
      "step": 13250
    },
    {
      "epoch": 3.267258382642998,
      "grad_norm": 0.3515625,
      "learning_rate": 6.468439797504942e-05,
      "loss": 0.048,
      "step": 13252
    },
    {
      "epoch": 3.2677514792899407,
      "grad_norm": 0.376953125,
      "learning_rate": 6.465219359904298e-05,
      "loss": 0.046,
      "step": 13254
    },
    {
      "epoch": 3.2682445759368837,
      "grad_norm": 0.326171875,
      "learning_rate": 6.4619993411949e-05,
      "loss": 0.0531,
      "step": 13256
    },
    {
      "epoch": 3.2687376725838266,
      "grad_norm": 0.373046875,
      "learning_rate": 6.458779741758328e-05,
      "loss": 0.0445,
      "step": 13258
    },
    {
      "epoch": 3.269230769230769,
      "grad_norm": 0.380859375,
      "learning_rate": 6.455560561976135e-05,
      "loss": 0.0448,
      "step": 13260
    },
    {
      "epoch": 3.269723865877712,
      "grad_norm": 0.4453125,
      "learning_rate": 6.452341802229803e-05,
      "loss": 0.0486,
      "step": 13262
    },
    {
      "epoch": 3.2702169625246547,
      "grad_norm": 0.365234375,
      "learning_rate": 6.449123462900774e-05,
      "loss": 0.0405,
      "step": 13264
    },
    {
      "epoch": 3.2707100591715976,
      "grad_norm": 0.37890625,
      "learning_rate": 6.445905544370441e-05,
      "loss": 0.046,
      "step": 13266
    },
    {
      "epoch": 3.2712031558185406,
      "grad_norm": 0.43359375,
      "learning_rate": 6.442688047020142e-05,
      "loss": 0.0442,
      "step": 13268
    },
    {
      "epoch": 3.271696252465483,
      "grad_norm": 0.380859375,
      "learning_rate": 6.439470971231174e-05,
      "loss": 0.0396,
      "step": 13270
    },
    {
      "epoch": 3.272189349112426,
      "grad_norm": 0.3359375,
      "learning_rate": 6.436254317384772e-05,
      "loss": 0.0471,
      "step": 13272
    },
    {
      "epoch": 3.2726824457593686,
      "grad_norm": 0.345703125,
      "learning_rate": 6.433038085862134e-05,
      "loss": 0.0437,
      "step": 13274
    },
    {
      "epoch": 3.2731755424063116,
      "grad_norm": 0.41796875,
      "learning_rate": 6.429822277044399e-05,
      "loss": 0.0462,
      "step": 13276
    },
    {
      "epoch": 3.2736686390532546,
      "grad_norm": 0.302734375,
      "learning_rate": 6.426606891312655e-05,
      "loss": 0.0453,
      "step": 13278
    },
    {
      "epoch": 3.274161735700197,
      "grad_norm": 0.373046875,
      "learning_rate": 6.423391929047948e-05,
      "loss": 0.0465,
      "step": 13280
    },
    {
      "epoch": 3.27465483234714,
      "grad_norm": 0.357421875,
      "learning_rate": 6.420177390631263e-05,
      "loss": 0.0405,
      "step": 13282
    },
    {
      "epoch": 3.275147928994083,
      "grad_norm": 0.326171875,
      "learning_rate": 6.416963276443547e-05,
      "loss": 0.0428,
      "step": 13284
    },
    {
      "epoch": 3.2756410256410255,
      "grad_norm": 0.349609375,
      "learning_rate": 6.413749586865684e-05,
      "loss": 0.0446,
      "step": 13286
    },
    {
      "epoch": 3.2761341222879685,
      "grad_norm": 0.375,
      "learning_rate": 6.410536322278523e-05,
      "loss": 0.0426,
      "step": 13288
    },
    {
      "epoch": 3.2766272189349115,
      "grad_norm": 0.37109375,
      "learning_rate": 6.407323483062848e-05,
      "loss": 0.0444,
      "step": 13290
    },
    {
      "epoch": 3.277120315581854,
      "grad_norm": 0.3515625,
      "learning_rate": 6.404111069599399e-05,
      "loss": 0.042,
      "step": 13292
    },
    {
      "epoch": 3.277613412228797,
      "grad_norm": 0.341796875,
      "learning_rate": 6.400899082268865e-05,
      "loss": 0.0443,
      "step": 13294
    },
    {
      "epoch": 3.2781065088757395,
      "grad_norm": 0.404296875,
      "learning_rate": 6.397687521451882e-05,
      "loss": 0.0466,
      "step": 13296
    },
    {
      "epoch": 3.2785996055226825,
      "grad_norm": 0.32421875,
      "learning_rate": 6.394476387529044e-05,
      "loss": 0.0384,
      "step": 13298
    },
    {
      "epoch": 3.279092702169625,
      "grad_norm": 0.30859375,
      "learning_rate": 6.39126568088088e-05,
      "loss": 0.0387,
      "step": 13300
    },
    {
      "epoch": 3.279585798816568,
      "grad_norm": 0.3359375,
      "learning_rate": 6.388055401887886e-05,
      "loss": 0.0396,
      "step": 13302
    },
    {
      "epoch": 3.280078895463511,
      "grad_norm": 0.35546875,
      "learning_rate": 6.384845550930492e-05,
      "loss": 0.044,
      "step": 13304
    },
    {
      "epoch": 3.2805719921104535,
      "grad_norm": 0.376953125,
      "learning_rate": 6.381636128389085e-05,
      "loss": 0.0418,
      "step": 13306
    },
    {
      "epoch": 3.2810650887573964,
      "grad_norm": 0.365234375,
      "learning_rate": 6.378427134643996e-05,
      "loss": 0.0434,
      "step": 13308
    },
    {
      "epoch": 3.2815581854043394,
      "grad_norm": 0.5234375,
      "learning_rate": 6.375218570075516e-05,
      "loss": 0.0418,
      "step": 13310
    },
    {
      "epoch": 3.282051282051282,
      "grad_norm": 0.400390625,
      "learning_rate": 6.372010435063875e-05,
      "loss": 0.0435,
      "step": 13312
    },
    {
      "epoch": 3.282544378698225,
      "grad_norm": 0.392578125,
      "learning_rate": 6.368802729989247e-05,
      "loss": 0.0523,
      "step": 13314
    },
    {
      "epoch": 3.283037475345168,
      "grad_norm": 0.359375,
      "learning_rate": 6.365595455231774e-05,
      "loss": 0.0449,
      "step": 13316
    },
    {
      "epoch": 3.2835305719921104,
      "grad_norm": 0.36328125,
      "learning_rate": 6.36238861117153e-05,
      "loss": 0.043,
      "step": 13318
    },
    {
      "epoch": 3.2840236686390534,
      "grad_norm": 0.421875,
      "learning_rate": 6.359182198188549e-05,
      "loss": 0.0445,
      "step": 13320
    },
    {
      "epoch": 3.284516765285996,
      "grad_norm": 0.4453125,
      "learning_rate": 6.355976216662799e-05,
      "loss": 0.0462,
      "step": 13322
    },
    {
      "epoch": 3.285009861932939,
      "grad_norm": 0.3203125,
      "learning_rate": 6.352770666974218e-05,
      "loss": 0.0433,
      "step": 13324
    },
    {
      "epoch": 3.285502958579882,
      "grad_norm": 0.44921875,
      "learning_rate": 6.349565549502676e-05,
      "loss": 0.0439,
      "step": 13326
    },
    {
      "epoch": 3.2859960552268244,
      "grad_norm": 0.392578125,
      "learning_rate": 6.346360864627994e-05,
      "loss": 0.0415,
      "step": 13328
    },
    {
      "epoch": 3.2864891518737673,
      "grad_norm": 0.396484375,
      "learning_rate": 6.343156612729952e-05,
      "loss": 0.0439,
      "step": 13330
    },
    {
      "epoch": 3.28698224852071,
      "grad_norm": 0.4453125,
      "learning_rate": 6.339952794188268e-05,
      "loss": 0.0483,
      "step": 13332
    },
    {
      "epoch": 3.287475345167653,
      "grad_norm": 0.46875,
      "learning_rate": 6.336749409382615e-05,
      "loss": 0.0447,
      "step": 13334
    },
    {
      "epoch": 3.287968441814596,
      "grad_norm": 0.419921875,
      "learning_rate": 6.333546458692605e-05,
      "loss": 0.0471,
      "step": 13336
    },
    {
      "epoch": 3.2884615384615383,
      "grad_norm": 0.376953125,
      "learning_rate": 6.330343942497815e-05,
      "loss": 0.0442,
      "step": 13338
    },
    {
      "epoch": 3.2889546351084813,
      "grad_norm": 0.447265625,
      "learning_rate": 6.327141861177752e-05,
      "loss": 0.0428,
      "step": 13340
    },
    {
      "epoch": 3.2894477317554243,
      "grad_norm": 0.58203125,
      "learning_rate": 6.323940215111889e-05,
      "loss": 0.0414,
      "step": 13342
    },
    {
      "epoch": 3.289940828402367,
      "grad_norm": 0.390625,
      "learning_rate": 6.320739004679637e-05,
      "loss": 0.0405,
      "step": 13344
    },
    {
      "epoch": 3.2904339250493098,
      "grad_norm": 0.310546875,
      "learning_rate": 6.317538230260352e-05,
      "loss": 0.0409,
      "step": 13346
    },
    {
      "epoch": 3.2909270216962523,
      "grad_norm": 0.345703125,
      "learning_rate": 6.314337892233348e-05,
      "loss": 0.0456,
      "step": 13348
    },
    {
      "epoch": 3.2914201183431953,
      "grad_norm": 0.439453125,
      "learning_rate": 6.311137990977878e-05,
      "loss": 0.0425,
      "step": 13350
    },
    {
      "epoch": 3.2919132149901382,
      "grad_norm": 0.349609375,
      "learning_rate": 6.307938526873157e-05,
      "loss": 0.0422,
      "step": 13352
    },
    {
      "epoch": 3.2924063116370808,
      "grad_norm": 0.345703125,
      "learning_rate": 6.30473950029833e-05,
      "loss": 0.0413,
      "step": 13354
    },
    {
      "epoch": 3.2928994082840237,
      "grad_norm": 0.31640625,
      "learning_rate": 6.301540911632509e-05,
      "loss": 0.0415,
      "step": 13356
    },
    {
      "epoch": 3.2933925049309662,
      "grad_norm": 0.376953125,
      "learning_rate": 6.298342761254736e-05,
      "loss": 0.0455,
      "step": 13358
    },
    {
      "epoch": 3.293885601577909,
      "grad_norm": 0.373046875,
      "learning_rate": 6.295145049544016e-05,
      "loss": 0.0416,
      "step": 13360
    },
    {
      "epoch": 3.294378698224852,
      "grad_norm": 0.345703125,
      "learning_rate": 6.291947776879292e-05,
      "loss": 0.045,
      "step": 13362
    },
    {
      "epoch": 3.2948717948717947,
      "grad_norm": 0.318359375,
      "learning_rate": 6.288750943639457e-05,
      "loss": 0.0421,
      "step": 13364
    },
    {
      "epoch": 3.2953648915187377,
      "grad_norm": 0.37890625,
      "learning_rate": 6.285554550203358e-05,
      "loss": 0.0442,
      "step": 13366
    },
    {
      "epoch": 3.2958579881656807,
      "grad_norm": 0.353515625,
      "learning_rate": 6.282358596949781e-05,
      "loss": 0.0431,
      "step": 13368
    },
    {
      "epoch": 3.296351084812623,
      "grad_norm": 0.390625,
      "learning_rate": 6.279163084257471e-05,
      "loss": 0.0455,
      "step": 13370
    },
    {
      "epoch": 3.296844181459566,
      "grad_norm": 0.33203125,
      "learning_rate": 6.275968012505107e-05,
      "loss": 0.0448,
      "step": 13372
    },
    {
      "epoch": 3.297337278106509,
      "grad_norm": 0.3359375,
      "learning_rate": 6.272773382071329e-05,
      "loss": 0.0412,
      "step": 13374
    },
    {
      "epoch": 3.2978303747534516,
      "grad_norm": 0.37109375,
      "learning_rate": 6.269579193334716e-05,
      "loss": 0.0463,
      "step": 13376
    },
    {
      "epoch": 3.2983234714003946,
      "grad_norm": 0.349609375,
      "learning_rate": 6.26638544667379e-05,
      "loss": 0.0448,
      "step": 13378
    },
    {
      "epoch": 3.298816568047337,
      "grad_norm": 0.38671875,
      "learning_rate": 6.26319214246704e-05,
      "loss": 0.0435,
      "step": 13380
    },
    {
      "epoch": 3.29930966469428,
      "grad_norm": 0.357421875,
      "learning_rate": 6.25999928109288e-05,
      "loss": 0.0427,
      "step": 13382
    },
    {
      "epoch": 3.2998027613412226,
      "grad_norm": 0.37890625,
      "learning_rate": 6.256806862929694e-05,
      "loss": 0.0446,
      "step": 13384
    },
    {
      "epoch": 3.3002958579881656,
      "grad_norm": 0.330078125,
      "learning_rate": 6.253614888355789e-05,
      "loss": 0.0409,
      "step": 13386
    },
    {
      "epoch": 3.3007889546351086,
      "grad_norm": 0.330078125,
      "learning_rate": 6.250423357749439e-05,
      "loss": 0.0445,
      "step": 13388
    },
    {
      "epoch": 3.301282051282051,
      "grad_norm": 0.361328125,
      "learning_rate": 6.247232271488852e-05,
      "loss": 0.0459,
      "step": 13390
    },
    {
      "epoch": 3.301775147928994,
      "grad_norm": 0.34375,
      "learning_rate": 6.244041629952199e-05,
      "loss": 0.043,
      "step": 13392
    },
    {
      "epoch": 3.302268244575937,
      "grad_norm": 0.341796875,
      "learning_rate": 6.240851433517583e-05,
      "loss": 0.0408,
      "step": 13394
    },
    {
      "epoch": 3.3027613412228796,
      "grad_norm": 0.30859375,
      "learning_rate": 6.237661682563056e-05,
      "loss": 0.0464,
      "step": 13396
    },
    {
      "epoch": 3.3032544378698225,
      "grad_norm": 0.32421875,
      "learning_rate": 6.23447237746663e-05,
      "loss": 0.0443,
      "step": 13398
    },
    {
      "epoch": 3.3037475345167655,
      "grad_norm": 0.388671875,
      "learning_rate": 6.231283518606248e-05,
      "loss": 0.0501,
      "step": 13400
    },
    {
      "epoch": 3.304240631163708,
      "grad_norm": 0.349609375,
      "learning_rate": 6.228095106359815e-05,
      "loss": 0.0456,
      "step": 13402
    },
    {
      "epoch": 3.304733727810651,
      "grad_norm": 0.376953125,
      "learning_rate": 6.224907141105165e-05,
      "loss": 0.0445,
      "step": 13404
    },
    {
      "epoch": 3.3052268244575935,
      "grad_norm": 0.369140625,
      "learning_rate": 6.221719623220102e-05,
      "loss": 0.0443,
      "step": 13406
    },
    {
      "epoch": 3.3057199211045365,
      "grad_norm": 0.349609375,
      "learning_rate": 6.218532553082359e-05,
      "loss": 0.0471,
      "step": 13408
    },
    {
      "epoch": 3.3062130177514795,
      "grad_norm": 0.37890625,
      "learning_rate": 6.215345931069617e-05,
      "loss": 0.043,
      "step": 13410
    },
    {
      "epoch": 3.306706114398422,
      "grad_norm": 0.376953125,
      "learning_rate": 6.212159757559516e-05,
      "loss": 0.0403,
      "step": 13412
    },
    {
      "epoch": 3.307199211045365,
      "grad_norm": 0.375,
      "learning_rate": 6.208974032929633e-05,
      "loss": 0.0436,
      "step": 13414
    },
    {
      "epoch": 3.3076923076923075,
      "grad_norm": 0.318359375,
      "learning_rate": 6.205788757557493e-05,
      "loss": 0.0427,
      "step": 13416
    },
    {
      "epoch": 3.3081854043392505,
      "grad_norm": 0.30078125,
      "learning_rate": 6.202603931820567e-05,
      "loss": 0.0467,
      "step": 13418
    },
    {
      "epoch": 3.3086785009861934,
      "grad_norm": 0.314453125,
      "learning_rate": 6.19941955609628e-05,
      "loss": 0.041,
      "step": 13420
    },
    {
      "epoch": 3.309171597633136,
      "grad_norm": 0.353515625,
      "learning_rate": 6.196235630761991e-05,
      "loss": 0.0393,
      "step": 13422
    },
    {
      "epoch": 3.309664694280079,
      "grad_norm": 0.421875,
      "learning_rate": 6.193052156195023e-05,
      "loss": 0.0459,
      "step": 13424
    },
    {
      "epoch": 3.310157790927022,
      "grad_norm": 0.41015625,
      "learning_rate": 6.189869132772628e-05,
      "loss": 0.0477,
      "step": 13426
    },
    {
      "epoch": 3.3106508875739644,
      "grad_norm": 0.345703125,
      "learning_rate": 6.186686560872013e-05,
      "loss": 0.0439,
      "step": 13428
    },
    {
      "epoch": 3.3111439842209074,
      "grad_norm": 0.373046875,
      "learning_rate": 6.183504440870331e-05,
      "loss": 0.0447,
      "step": 13430
    },
    {
      "epoch": 3.31163708086785,
      "grad_norm": 0.3984375,
      "learning_rate": 6.18032277314468e-05,
      "loss": 0.0447,
      "step": 13432
    },
    {
      "epoch": 3.312130177514793,
      "grad_norm": 0.326171875,
      "learning_rate": 6.177141558072111e-05,
      "loss": 0.0491,
      "step": 13434
    },
    {
      "epoch": 3.312623274161736,
      "grad_norm": 0.375,
      "learning_rate": 6.173960796029605e-05,
      "loss": 0.0404,
      "step": 13436
    },
    {
      "epoch": 3.3131163708086784,
      "grad_norm": 0.314453125,
      "learning_rate": 6.170780487394112e-05,
      "loss": 0.0434,
      "step": 13438
    },
    {
      "epoch": 3.3136094674556213,
      "grad_norm": 0.328125,
      "learning_rate": 6.167600632542508e-05,
      "loss": 0.0395,
      "step": 13440
    },
    {
      "epoch": 3.314102564102564,
      "grad_norm": 0.412109375,
      "learning_rate": 6.164421231851629e-05,
      "loss": 0.041,
      "step": 13442
    },
    {
      "epoch": 3.314595660749507,
      "grad_norm": 0.322265625,
      "learning_rate": 6.16124228569825e-05,
      "loss": 0.0462,
      "step": 13444
    },
    {
      "epoch": 3.31508875739645,
      "grad_norm": 0.3515625,
      "learning_rate": 6.158063794459087e-05,
      "loss": 0.0451,
      "step": 13446
    },
    {
      "epoch": 3.3155818540433923,
      "grad_norm": 0.421875,
      "learning_rate": 6.154885758510822e-05,
      "loss": 0.0425,
      "step": 13448
    },
    {
      "epoch": 3.3160749506903353,
      "grad_norm": 0.3359375,
      "learning_rate": 6.151708178230056e-05,
      "loss": 0.0416,
      "step": 13450
    },
    {
      "epoch": 3.3165680473372783,
      "grad_norm": 0.3515625,
      "learning_rate": 6.148531053993362e-05,
      "loss": 0.0444,
      "step": 13452
    },
    {
      "epoch": 3.317061143984221,
      "grad_norm": 0.37109375,
      "learning_rate": 6.145354386177241e-05,
      "loss": 0.0411,
      "step": 13454
    },
    {
      "epoch": 3.3175542406311638,
      "grad_norm": 0.34375,
      "learning_rate": 6.142178175158149e-05,
      "loss": 0.0409,
      "step": 13456
    },
    {
      "epoch": 3.3180473372781067,
      "grad_norm": 0.486328125,
      "learning_rate": 6.139002421312481e-05,
      "loss": 0.0437,
      "step": 13458
    },
    {
      "epoch": 3.3185404339250493,
      "grad_norm": 0.328125,
      "learning_rate": 6.13582712501658e-05,
      "loss": 0.0428,
      "step": 13460
    },
    {
      "epoch": 3.3190335305719922,
      "grad_norm": 0.33984375,
      "learning_rate": 6.132652286646742e-05,
      "loss": 0.0437,
      "step": 13462
    },
    {
      "epoch": 3.3195266272189348,
      "grad_norm": 0.35546875,
      "learning_rate": 6.129477906579198e-05,
      "loss": 0.0437,
      "step": 13464
    },
    {
      "epoch": 3.3200197238658777,
      "grad_norm": 0.3984375,
      "learning_rate": 6.126303985190136e-05,
      "loss": 0.0434,
      "step": 13466
    },
    {
      "epoch": 3.3205128205128207,
      "grad_norm": 0.328125,
      "learning_rate": 6.123130522855678e-05,
      "loss": 0.0434,
      "step": 13468
    },
    {
      "epoch": 3.3210059171597632,
      "grad_norm": 0.3359375,
      "learning_rate": 6.1199575199519e-05,
      "loss": 0.0412,
      "step": 13470
    },
    {
      "epoch": 3.321499013806706,
      "grad_norm": 0.3671875,
      "learning_rate": 6.116784976854816e-05,
      "loss": 0.0424,
      "step": 13472
    },
    {
      "epoch": 3.3219921104536487,
      "grad_norm": 0.35546875,
      "learning_rate": 6.113612893940397e-05,
      "loss": 0.0416,
      "step": 13474
    },
    {
      "epoch": 3.3224852071005917,
      "grad_norm": 0.3828125,
      "learning_rate": 6.11044127158455e-05,
      "loss": 0.044,
      "step": 13476
    },
    {
      "epoch": 3.3229783037475347,
      "grad_norm": 0.384765625,
      "learning_rate": 6.107270110163123e-05,
      "loss": 0.0461,
      "step": 13478
    },
    {
      "epoch": 3.323471400394477,
      "grad_norm": 0.349609375,
      "learning_rate": 6.104099410051927e-05,
      "loss": 0.0427,
      "step": 13480
    },
    {
      "epoch": 3.32396449704142,
      "grad_norm": 0.4296875,
      "learning_rate": 6.100929171626701e-05,
      "loss": 0.0407,
      "step": 13482
    },
    {
      "epoch": 3.324457593688363,
      "grad_norm": 0.310546875,
      "learning_rate": 6.09775939526314e-05,
      "loss": 0.0394,
      "step": 13484
    },
    {
      "epoch": 3.3249506903353057,
      "grad_norm": 0.365234375,
      "learning_rate": 6.094590081336873e-05,
      "loss": 0.0452,
      "step": 13486
    },
    {
      "epoch": 3.3254437869822486,
      "grad_norm": 0.38671875,
      "learning_rate": 6.091421230223491e-05,
      "loss": 0.0422,
      "step": 13488
    },
    {
      "epoch": 3.325936883629191,
      "grad_norm": 0.337890625,
      "learning_rate": 6.088252842298512e-05,
      "loss": 0.0447,
      "step": 13490
    },
    {
      "epoch": 3.326429980276134,
      "grad_norm": 0.609375,
      "learning_rate": 6.085084917937416e-05,
      "loss": 0.0468,
      "step": 13492
    },
    {
      "epoch": 3.326923076923077,
      "grad_norm": 0.310546875,
      "learning_rate": 6.081917457515614e-05,
      "loss": 0.0407,
      "step": 13494
    },
    {
      "epoch": 3.3274161735700196,
      "grad_norm": 0.3828125,
      "learning_rate": 6.078750461408468e-05,
      "loss": 0.0404,
      "step": 13496
    },
    {
      "epoch": 3.3279092702169626,
      "grad_norm": 0.404296875,
      "learning_rate": 6.0755839299912864e-05,
      "loss": 0.0437,
      "step": 13498
    },
    {
      "epoch": 3.328402366863905,
      "grad_norm": 0.33984375,
      "learning_rate": 6.0724178636393184e-05,
      "loss": 0.0456,
      "step": 13500
    },
    {
      "epoch": 3.328895463510848,
      "grad_norm": 0.32421875,
      "learning_rate": 6.069252262727765e-05,
      "loss": 0.0429,
      "step": 13502
    },
    {
      "epoch": 3.329388560157791,
      "grad_norm": 0.396484375,
      "learning_rate": 6.066087127631761e-05,
      "loss": 0.0392,
      "step": 13504
    },
    {
      "epoch": 3.3298816568047336,
      "grad_norm": 0.3984375,
      "learning_rate": 6.0629224587264013e-05,
      "loss": 0.045,
      "step": 13506
    },
    {
      "epoch": 3.3303747534516766,
      "grad_norm": 0.326171875,
      "learning_rate": 6.059758256386712e-05,
      "loss": 0.0413,
      "step": 13508
    },
    {
      "epoch": 3.3308678500986195,
      "grad_norm": 0.337890625,
      "learning_rate": 6.0565945209876664e-05,
      "loss": 0.0444,
      "step": 13510
    },
    {
      "epoch": 3.331360946745562,
      "grad_norm": 0.392578125,
      "learning_rate": 6.053431252904189e-05,
      "loss": 0.0393,
      "step": 13512
    },
    {
      "epoch": 3.331854043392505,
      "grad_norm": 0.33203125,
      "learning_rate": 6.050268452511138e-05,
      "loss": 0.0482,
      "step": 13514
    },
    {
      "epoch": 3.3323471400394475,
      "grad_norm": 0.3828125,
      "learning_rate": 6.0471061201833346e-05,
      "loss": 0.0441,
      "step": 13516
    },
    {
      "epoch": 3.3328402366863905,
      "grad_norm": 0.3203125,
      "learning_rate": 6.04394425629552e-05,
      "loss": 0.0422,
      "step": 13518
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 0.3828125,
      "learning_rate": 6.040782861222403e-05,
      "loss": 0.0456,
      "step": 13520
    },
    {
      "epoch": 3.333826429980276,
      "grad_norm": 0.380859375,
      "learning_rate": 6.037621935338621e-05,
      "loss": 0.0369,
      "step": 13522
    },
    {
      "epoch": 3.334319526627219,
      "grad_norm": 0.322265625,
      "learning_rate": 6.034461479018765e-05,
      "loss": 0.0374,
      "step": 13524
    },
    {
      "epoch": 3.3348126232741615,
      "grad_norm": 0.314453125,
      "learning_rate": 6.031301492637364e-05,
      "loss": 0.0401,
      "step": 13526
    },
    {
      "epoch": 3.3353057199211045,
      "grad_norm": 0.40234375,
      "learning_rate": 6.0281419765688904e-05,
      "loss": 0.0416,
      "step": 13528
    },
    {
      "epoch": 3.3357988165680474,
      "grad_norm": 0.310546875,
      "learning_rate": 6.024982931187773e-05,
      "loss": 0.0387,
      "step": 13530
    },
    {
      "epoch": 3.33629191321499,
      "grad_norm": 0.333984375,
      "learning_rate": 6.021824356868366e-05,
      "loss": 0.0431,
      "step": 13532
    },
    {
      "epoch": 3.336785009861933,
      "grad_norm": 0.318359375,
      "learning_rate": 6.01866625398499e-05,
      "loss": 0.04,
      "step": 13534
    },
    {
      "epoch": 3.337278106508876,
      "grad_norm": 0.37890625,
      "learning_rate": 6.015508622911889e-05,
      "loss": 0.0422,
      "step": 13536
    },
    {
      "epoch": 3.3377712031558184,
      "grad_norm": 0.353515625,
      "learning_rate": 6.012351464023266e-05,
      "loss": 0.0408,
      "step": 13538
    },
    {
      "epoch": 3.3382642998027614,
      "grad_norm": 0.302734375,
      "learning_rate": 6.009194777693258e-05,
      "loss": 0.0456,
      "step": 13540
    },
    {
      "epoch": 3.3387573964497044,
      "grad_norm": 0.349609375,
      "learning_rate": 6.006038564295946e-05,
      "loss": 0.0419,
      "step": 13542
    },
    {
      "epoch": 3.339250493096647,
      "grad_norm": 0.34765625,
      "learning_rate": 6.002882824205368e-05,
      "loss": 0.04,
      "step": 13544
    },
    {
      "epoch": 3.33974358974359,
      "grad_norm": 0.341796875,
      "learning_rate": 5.9997275577954894e-05,
      "loss": 0.0436,
      "step": 13546
    },
    {
      "epoch": 3.3402366863905324,
      "grad_norm": 0.314453125,
      "learning_rate": 5.9965727654402315e-05,
      "loss": 0.0414,
      "step": 13548
    },
    {
      "epoch": 3.3407297830374754,
      "grad_norm": 0.333984375,
      "learning_rate": 5.993418447513452e-05,
      "loss": 0.0402,
      "step": 13550
    },
    {
      "epoch": 3.3412228796844183,
      "grad_norm": 0.427734375,
      "learning_rate": 5.99026460438896e-05,
      "loss": 0.0453,
      "step": 13552
    },
    {
      "epoch": 3.341715976331361,
      "grad_norm": 0.34375,
      "learning_rate": 5.987111236440495e-05,
      "loss": 0.0395,
      "step": 13554
    },
    {
      "epoch": 3.342209072978304,
      "grad_norm": 0.453125,
      "learning_rate": 5.983958344041757e-05,
      "loss": 0.0436,
      "step": 13556
    },
    {
      "epoch": 3.3427021696252464,
      "grad_norm": 0.33203125,
      "learning_rate": 5.980805927566378e-05,
      "loss": 0.0374,
      "step": 13558
    },
    {
      "epoch": 3.3431952662721893,
      "grad_norm": 0.400390625,
      "learning_rate": 5.977653987387933e-05,
      "loss": 0.0406,
      "step": 13560
    },
    {
      "epoch": 3.3436883629191323,
      "grad_norm": 0.37890625,
      "learning_rate": 5.9745025238799526e-05,
      "loss": 0.0476,
      "step": 13562
    },
    {
      "epoch": 3.344181459566075,
      "grad_norm": 0.345703125,
      "learning_rate": 5.971351537415897e-05,
      "loss": 0.044,
      "step": 13564
    },
    {
      "epoch": 3.344674556213018,
      "grad_norm": 0.36328125,
      "learning_rate": 5.9682010283691794e-05,
      "loss": 0.0445,
      "step": 13566
    },
    {
      "epoch": 3.3451676528599608,
      "grad_norm": 0.3515625,
      "learning_rate": 5.9650509971131465e-05,
      "loss": 0.0444,
      "step": 13568
    },
    {
      "epoch": 3.3456607495069033,
      "grad_norm": 0.443359375,
      "learning_rate": 5.961901444021101e-05,
      "loss": 0.0451,
      "step": 13570
    },
    {
      "epoch": 3.3461538461538463,
      "grad_norm": 0.380859375,
      "learning_rate": 5.958752369466278e-05,
      "loss": 0.0475,
      "step": 13572
    },
    {
      "epoch": 3.346646942800789,
      "grad_norm": 0.349609375,
      "learning_rate": 5.955603773821866e-05,
      "loss": 0.0452,
      "step": 13574
    },
    {
      "epoch": 3.3471400394477318,
      "grad_norm": 0.3359375,
      "learning_rate": 5.952455657460988e-05,
      "loss": 0.0418,
      "step": 13576
    },
    {
      "epoch": 3.3476331360946747,
      "grad_norm": 0.328125,
      "learning_rate": 5.949308020756712e-05,
      "loss": 0.0428,
      "step": 13578
    },
    {
      "epoch": 3.3481262327416172,
      "grad_norm": 0.375,
      "learning_rate": 5.9461608640820535e-05,
      "loss": 0.0387,
      "step": 13580
    },
    {
      "epoch": 3.34861932938856,
      "grad_norm": 0.396484375,
      "learning_rate": 5.943014187809961e-05,
      "loss": 0.0417,
      "step": 13582
    },
    {
      "epoch": 3.3491124260355027,
      "grad_norm": 0.365234375,
      "learning_rate": 5.939867992313344e-05,
      "loss": 0.0402,
      "step": 13584
    },
    {
      "epoch": 3.3496055226824457,
      "grad_norm": 0.2890625,
      "learning_rate": 5.9367222779650334e-05,
      "loss": 0.0388,
      "step": 13586
    },
    {
      "epoch": 3.3500986193293887,
      "grad_norm": 0.318359375,
      "learning_rate": 5.933577045137823e-05,
      "loss": 0.0394,
      "step": 13588
    },
    {
      "epoch": 3.350591715976331,
      "grad_norm": 0.3671875,
      "learning_rate": 5.9304322942044376e-05,
      "loss": 0.0413,
      "step": 13590
    },
    {
      "epoch": 3.351084812623274,
      "grad_norm": 0.3515625,
      "learning_rate": 5.927288025537545e-05,
      "loss": 0.0429,
      "step": 13592
    },
    {
      "epoch": 3.351577909270217,
      "grad_norm": 0.357421875,
      "learning_rate": 5.924144239509764e-05,
      "loss": 0.0426,
      "step": 13594
    },
    {
      "epoch": 3.3520710059171597,
      "grad_norm": 0.3203125,
      "learning_rate": 5.9210009364936416e-05,
      "loss": 0.0405,
      "step": 13596
    },
    {
      "epoch": 3.3525641025641026,
      "grad_norm": 0.345703125,
      "learning_rate": 5.917858116861687e-05,
      "loss": 0.0414,
      "step": 13598
    },
    {
      "epoch": 3.3530571992110456,
      "grad_norm": 0.41015625,
      "learning_rate": 5.914715780986333e-05,
      "loss": 0.0422,
      "step": 13600
    },
    {
      "epoch": 3.353550295857988,
      "grad_norm": 0.31640625,
      "learning_rate": 5.911573929239971e-05,
      "loss": 0.0416,
      "step": 13602
    },
    {
      "epoch": 3.354043392504931,
      "grad_norm": 0.28125,
      "learning_rate": 5.9084325619949256e-05,
      "loss": 0.0395,
      "step": 13604
    },
    {
      "epoch": 3.3545364891518736,
      "grad_norm": 0.376953125,
      "learning_rate": 5.9052916796234656e-05,
      "loss": 0.0395,
      "step": 13606
    },
    {
      "epoch": 3.3550295857988166,
      "grad_norm": 0.32421875,
      "learning_rate": 5.9021512824978056e-05,
      "loss": 0.0398,
      "step": 13608
    },
    {
      "epoch": 3.355522682445759,
      "grad_norm": 0.37109375,
      "learning_rate": 5.899011370990093e-05,
      "loss": 0.039,
      "step": 13610
    },
    {
      "epoch": 3.356015779092702,
      "grad_norm": 0.380859375,
      "learning_rate": 5.8958719454724346e-05,
      "loss": 0.0417,
      "step": 13612
    },
    {
      "epoch": 3.356508875739645,
      "grad_norm": 0.30078125,
      "learning_rate": 5.892733006316861e-05,
      "loss": 0.0405,
      "step": 13614
    },
    {
      "epoch": 3.3570019723865876,
      "grad_norm": 0.3515625,
      "learning_rate": 5.889594553895361e-05,
      "loss": 0.0419,
      "step": 13616
    },
    {
      "epoch": 3.3574950690335306,
      "grad_norm": 0.46484375,
      "learning_rate": 5.886456588579855e-05,
      "loss": 0.0459,
      "step": 13618
    },
    {
      "epoch": 3.3579881656804735,
      "grad_norm": 0.31640625,
      "learning_rate": 5.883319110742213e-05,
      "loss": 0.0372,
      "step": 13620
    },
    {
      "epoch": 3.358481262327416,
      "grad_norm": 0.314453125,
      "learning_rate": 5.8801821207542365e-05,
      "loss": 0.0427,
      "step": 13622
    },
    {
      "epoch": 3.358974358974359,
      "grad_norm": 0.365234375,
      "learning_rate": 5.8770456189876844e-05,
      "loss": 0.042,
      "step": 13624
    },
    {
      "epoch": 3.359467455621302,
      "grad_norm": 0.3515625,
      "learning_rate": 5.873909605814245e-05,
      "loss": 0.0355,
      "step": 13626
    },
    {
      "epoch": 3.3599605522682445,
      "grad_norm": 0.30078125,
      "learning_rate": 5.8707740816055526e-05,
      "loss": 0.0358,
      "step": 13628
    },
    {
      "epoch": 3.3604536489151875,
      "grad_norm": 0.333984375,
      "learning_rate": 5.867639046733189e-05,
      "loss": 0.0445,
      "step": 13630
    },
    {
      "epoch": 3.36094674556213,
      "grad_norm": 0.3359375,
      "learning_rate": 5.8645045015686686e-05,
      "loss": 0.0445,
      "step": 13632
    },
    {
      "epoch": 3.361439842209073,
      "grad_norm": 0.4375,
      "learning_rate": 5.861370446483456e-05,
      "loss": 0.0419,
      "step": 13634
    },
    {
      "epoch": 3.361932938856016,
      "grad_norm": 0.2890625,
      "learning_rate": 5.8582368818489484e-05,
      "loss": 0.0351,
      "step": 13636
    },
    {
      "epoch": 3.3624260355029585,
      "grad_norm": 0.431640625,
      "learning_rate": 5.855103808036501e-05,
      "loss": 0.0421,
      "step": 13638
    },
    {
      "epoch": 3.3629191321499015,
      "grad_norm": 0.294921875,
      "learning_rate": 5.851971225417393e-05,
      "loss": 0.0387,
      "step": 13640
    },
    {
      "epoch": 3.363412228796844,
      "grad_norm": 0.353515625,
      "learning_rate": 5.84883913436285e-05,
      "loss": 0.042,
      "step": 13642
    },
    {
      "epoch": 3.363905325443787,
      "grad_norm": 0.34375,
      "learning_rate": 5.8457075352440514e-05,
      "loss": 0.0418,
      "step": 13644
    },
    {
      "epoch": 3.36439842209073,
      "grad_norm": 0.357421875,
      "learning_rate": 5.842576428432103e-05,
      "loss": 0.0417,
      "step": 13646
    },
    {
      "epoch": 3.3648915187376724,
      "grad_norm": 0.3359375,
      "learning_rate": 5.8394458142980615e-05,
      "loss": 0.0415,
      "step": 13648
    },
    {
      "epoch": 3.3653846153846154,
      "grad_norm": 0.396484375,
      "learning_rate": 5.836315693212916e-05,
      "loss": 0.0451,
      "step": 13650
    },
    {
      "epoch": 3.3658777120315584,
      "grad_norm": 0.337890625,
      "learning_rate": 5.83318606554761e-05,
      "loss": 0.044,
      "step": 13652
    },
    {
      "epoch": 3.366370808678501,
      "grad_norm": 0.31640625,
      "learning_rate": 5.83005693167302e-05,
      "loss": 0.0452,
      "step": 13654
    },
    {
      "epoch": 3.366863905325444,
      "grad_norm": 0.32421875,
      "learning_rate": 5.82692829195997e-05,
      "loss": 0.042,
      "step": 13656
    },
    {
      "epoch": 3.3673570019723864,
      "grad_norm": 0.341796875,
      "learning_rate": 5.823800146779212e-05,
      "loss": 0.0407,
      "step": 13658
    },
    {
      "epoch": 3.3678500986193294,
      "grad_norm": 0.302734375,
      "learning_rate": 5.820672496501454e-05,
      "loss": 0.0372,
      "step": 13660
    },
    {
      "epoch": 3.3683431952662723,
      "grad_norm": 0.359375,
      "learning_rate": 5.817545341497339e-05,
      "loss": 0.0389,
      "step": 13662
    },
    {
      "epoch": 3.368836291913215,
      "grad_norm": 0.310546875,
      "learning_rate": 5.814418682137454e-05,
      "loss": 0.0389,
      "step": 13664
    },
    {
      "epoch": 3.369329388560158,
      "grad_norm": 0.41796875,
      "learning_rate": 5.811292518792324e-05,
      "loss": 0.039,
      "step": 13666
    },
    {
      "epoch": 3.3698224852071004,
      "grad_norm": 0.423828125,
      "learning_rate": 5.8081668518324176e-05,
      "loss": 0.0394,
      "step": 13668
    },
    {
      "epoch": 3.3703155818540433,
      "grad_norm": 0.330078125,
      "learning_rate": 5.805041681628146e-05,
      "loss": 0.0384,
      "step": 13670
    },
    {
      "epoch": 3.3708086785009863,
      "grad_norm": 0.330078125,
      "learning_rate": 5.801917008549853e-05,
      "loss": 0.0416,
      "step": 13672
    },
    {
      "epoch": 3.371301775147929,
      "grad_norm": 0.35546875,
      "learning_rate": 5.798792832967834e-05,
      "loss": 0.044,
      "step": 13674
    },
    {
      "epoch": 3.371794871794872,
      "grad_norm": 0.31640625,
      "learning_rate": 5.795669155252321e-05,
      "loss": 0.0419,
      "step": 13676
    },
    {
      "epoch": 3.3722879684418148,
      "grad_norm": 0.302734375,
      "learning_rate": 5.792545975773489e-05,
      "loss": 0.0374,
      "step": 13678
    },
    {
      "epoch": 3.3727810650887573,
      "grad_norm": 0.4765625,
      "learning_rate": 5.789423294901448e-05,
      "loss": 0.0422,
      "step": 13680
    },
    {
      "epoch": 3.3732741617357003,
      "grad_norm": 0.380859375,
      "learning_rate": 5.786301113006256e-05,
      "loss": 0.0449,
      "step": 13682
    },
    {
      "epoch": 3.3737672583826432,
      "grad_norm": 0.357421875,
      "learning_rate": 5.7831794304579124e-05,
      "loss": 0.0426,
      "step": 13684
    },
    {
      "epoch": 3.3742603550295858,
      "grad_norm": 0.373046875,
      "learning_rate": 5.780058247626341e-05,
      "loss": 0.0419,
      "step": 13686
    },
    {
      "epoch": 3.3747534516765287,
      "grad_norm": 0.310546875,
      "learning_rate": 5.776937564881437e-05,
      "loss": 0.0369,
      "step": 13688
    },
    {
      "epoch": 3.3752465483234713,
      "grad_norm": 0.310546875,
      "learning_rate": 5.773817382593008e-05,
      "loss": 0.0429,
      "step": 13690
    },
    {
      "epoch": 3.3757396449704142,
      "grad_norm": 0.318359375,
      "learning_rate": 5.7706977011308126e-05,
      "loss": 0.0424,
      "step": 13692
    },
    {
      "epoch": 3.3762327416173568,
      "grad_norm": 0.3125,
      "learning_rate": 5.767578520864554e-05,
      "loss": 0.0435,
      "step": 13694
    },
    {
      "epoch": 3.3767258382642997,
      "grad_norm": 0.291015625,
      "learning_rate": 5.764459842163872e-05,
      "loss": 0.0379,
      "step": 13696
    },
    {
      "epoch": 3.3772189349112427,
      "grad_norm": 0.337890625,
      "learning_rate": 5.7613416653983497e-05,
      "loss": 0.0372,
      "step": 13698
    },
    {
      "epoch": 3.3777120315581852,
      "grad_norm": 0.283203125,
      "learning_rate": 5.758223990937499e-05,
      "loss": 0.0354,
      "step": 13700
    },
    {
      "epoch": 3.378205128205128,
      "grad_norm": 0.298828125,
      "learning_rate": 5.7551068191507954e-05,
      "loss": 0.0374,
      "step": 13702
    },
    {
      "epoch": 3.378698224852071,
      "grad_norm": 0.337890625,
      "learning_rate": 5.751990150407627e-05,
      "loss": 0.0401,
      "step": 13704
    },
    {
      "epoch": 3.3791913214990137,
      "grad_norm": 0.349609375,
      "learning_rate": 5.7488739850773496e-05,
      "loss": 0.0443,
      "step": 13706
    },
    {
      "epoch": 3.3796844181459567,
      "grad_norm": 0.349609375,
      "learning_rate": 5.745758323529239e-05,
      "loss": 0.0395,
      "step": 13708
    },
    {
      "epoch": 3.3801775147928996,
      "grad_norm": 0.443359375,
      "learning_rate": 5.742643166132519e-05,
      "loss": 0.0421,
      "step": 13710
    },
    {
      "epoch": 3.380670611439842,
      "grad_norm": 0.3203125,
      "learning_rate": 5.739528513256357e-05,
      "loss": 0.0433,
      "step": 13712
    },
    {
      "epoch": 3.381163708086785,
      "grad_norm": 0.34375,
      "learning_rate": 5.736414365269846e-05,
      "loss": 0.0401,
      "step": 13714
    },
    {
      "epoch": 3.3816568047337277,
      "grad_norm": 0.318359375,
      "learning_rate": 5.733300722542045e-05,
      "loss": 0.0383,
      "step": 13716
    },
    {
      "epoch": 3.3821499013806706,
      "grad_norm": 0.314453125,
      "learning_rate": 5.730187585441924e-05,
      "loss": 0.0392,
      "step": 13718
    },
    {
      "epoch": 3.3826429980276136,
      "grad_norm": 0.341796875,
      "learning_rate": 5.727074954338421e-05,
      "loss": 0.0428,
      "step": 13720
    },
    {
      "epoch": 3.383136094674556,
      "grad_norm": 0.33984375,
      "learning_rate": 5.723962829600388e-05,
      "loss": 0.0375,
      "step": 13722
    },
    {
      "epoch": 3.383629191321499,
      "grad_norm": 0.349609375,
      "learning_rate": 5.720851211596634e-05,
      "loss": 0.0379,
      "step": 13724
    },
    {
      "epoch": 3.3841222879684416,
      "grad_norm": 0.37109375,
      "learning_rate": 5.7177401006959075e-05,
      "loss": 0.041,
      "step": 13726
    },
    {
      "epoch": 3.3846153846153846,
      "grad_norm": 0.400390625,
      "learning_rate": 5.714629497266881e-05,
      "loss": 0.0447,
      "step": 13728
    },
    {
      "epoch": 3.3851084812623276,
      "grad_norm": 0.416015625,
      "learning_rate": 5.7115194016781934e-05,
      "loss": 0.0403,
      "step": 13730
    },
    {
      "epoch": 3.38560157790927,
      "grad_norm": 0.416015625,
      "learning_rate": 5.7084098142983924e-05,
      "loss": 0.039,
      "step": 13732
    },
    {
      "epoch": 3.386094674556213,
      "grad_norm": 0.291015625,
      "learning_rate": 5.7053007354959975e-05,
      "loss": 0.0339,
      "step": 13734
    },
    {
      "epoch": 3.386587771203156,
      "grad_norm": 0.326171875,
      "learning_rate": 5.7021921656394394e-05,
      "loss": 0.0355,
      "step": 13736
    },
    {
      "epoch": 3.3870808678500985,
      "grad_norm": 0.333984375,
      "learning_rate": 5.699084105097106e-05,
      "loss": 0.041,
      "step": 13738
    },
    {
      "epoch": 3.3875739644970415,
      "grad_norm": 0.349609375,
      "learning_rate": 5.6959765542373235e-05,
      "loss": 0.0368,
      "step": 13740
    },
    {
      "epoch": 3.388067061143984,
      "grad_norm": 0.375,
      "learning_rate": 5.692869513428341e-05,
      "loss": 0.0406,
      "step": 13742
    },
    {
      "epoch": 3.388560157790927,
      "grad_norm": 0.361328125,
      "learning_rate": 5.689762983038378e-05,
      "loss": 0.0442,
      "step": 13744
    },
    {
      "epoch": 3.38905325443787,
      "grad_norm": 0.353515625,
      "learning_rate": 5.686656963435558e-05,
      "loss": 0.0404,
      "step": 13746
    },
    {
      "epoch": 3.3895463510848125,
      "grad_norm": 0.3125,
      "learning_rate": 5.683551454987979e-05,
      "loss": 0.0403,
      "step": 13748
    },
    {
      "epoch": 3.3900394477317555,
      "grad_norm": 0.306640625,
      "learning_rate": 5.680446458063649e-05,
      "loss": 0.038,
      "step": 13750
    },
    {
      "epoch": 3.390532544378698,
      "grad_norm": 0.349609375,
      "learning_rate": 5.67734197303053e-05,
      "loss": 0.043,
      "step": 13752
    },
    {
      "epoch": 3.391025641025641,
      "grad_norm": 0.306640625,
      "learning_rate": 5.6742380002565235e-05,
      "loss": 0.0397,
      "step": 13754
    },
    {
      "epoch": 3.391518737672584,
      "grad_norm": 0.333984375,
      "learning_rate": 5.671134540109464e-05,
      "loss": 0.04,
      "step": 13756
    },
    {
      "epoch": 3.3920118343195265,
      "grad_norm": 0.326171875,
      "learning_rate": 5.668031592957136e-05,
      "loss": 0.04,
      "step": 13758
    },
    {
      "epoch": 3.3925049309664694,
      "grad_norm": 0.326171875,
      "learning_rate": 5.6649291591672424e-05,
      "loss": 0.0379,
      "step": 13760
    },
    {
      "epoch": 3.3929980276134124,
      "grad_norm": 0.3671875,
      "learning_rate": 5.661827239107456e-05,
      "loss": 0.0394,
      "step": 13762
    },
    {
      "epoch": 3.393491124260355,
      "grad_norm": 0.337890625,
      "learning_rate": 5.658725833145357e-05,
      "loss": 0.0412,
      "step": 13764
    },
    {
      "epoch": 3.393984220907298,
      "grad_norm": 0.2890625,
      "learning_rate": 5.6556249416484873e-05,
      "loss": 0.0424,
      "step": 13766
    },
    {
      "epoch": 3.394477317554241,
      "grad_norm": 0.3046875,
      "learning_rate": 5.6525245649843184e-05,
      "loss": 0.0412,
      "step": 13768
    },
    {
      "epoch": 3.3949704142011834,
      "grad_norm": 0.328125,
      "learning_rate": 5.649424703520261e-05,
      "loss": 0.0375,
      "step": 13770
    },
    {
      "epoch": 3.3954635108481264,
      "grad_norm": 0.29296875,
      "learning_rate": 5.6463253576236694e-05,
      "loss": 0.0365,
      "step": 13772
    },
    {
      "epoch": 3.395956607495069,
      "grad_norm": 0.322265625,
      "learning_rate": 5.643226527661824e-05,
      "loss": 0.0405,
      "step": 13774
    },
    {
      "epoch": 3.396449704142012,
      "grad_norm": 0.294921875,
      "learning_rate": 5.6401282140019685e-05,
      "loss": 0.0404,
      "step": 13776
    },
    {
      "epoch": 3.3969428007889544,
      "grad_norm": 0.322265625,
      "learning_rate": 5.637030417011258e-05,
      "loss": 0.0409,
      "step": 13778
    },
    {
      "epoch": 3.3974358974358974,
      "grad_norm": 0.291015625,
      "learning_rate": 5.633933137056801e-05,
      "loss": 0.0401,
      "step": 13780
    },
    {
      "epoch": 3.3979289940828403,
      "grad_norm": 0.35546875,
      "learning_rate": 5.630836374505647e-05,
      "loss": 0.0348,
      "step": 13782
    },
    {
      "epoch": 3.398422090729783,
      "grad_norm": 0.39453125,
      "learning_rate": 5.627740129724775e-05,
      "loss": 0.0449,
      "step": 13784
    },
    {
      "epoch": 3.398915187376726,
      "grad_norm": 0.28515625,
      "learning_rate": 5.62464440308111e-05,
      "loss": 0.0401,
      "step": 13786
    },
    {
      "epoch": 3.399408284023669,
      "grad_norm": 0.34765625,
      "learning_rate": 5.621549194941511e-05,
      "loss": 0.0387,
      "step": 13788
    },
    {
      "epoch": 3.3999013806706113,
      "grad_norm": 0.306640625,
      "learning_rate": 5.6184545056727836e-05,
      "loss": 0.0385,
      "step": 13790
    },
    {
      "epoch": 3.4003944773175543,
      "grad_norm": 0.349609375,
      "learning_rate": 5.6153603356416575e-05,
      "loss": 0.037,
      "step": 13792
    },
    {
      "epoch": 3.4008875739644973,
      "grad_norm": 0.3203125,
      "learning_rate": 5.612266685214811e-05,
      "loss": 0.0413,
      "step": 13794
    },
    {
      "epoch": 3.40138067061144,
      "grad_norm": 0.341796875,
      "learning_rate": 5.60917355475886e-05,
      "loss": 0.0437,
      "step": 13796
    },
    {
      "epoch": 3.4018737672583828,
      "grad_norm": 0.3203125,
      "learning_rate": 5.60608094464036e-05,
      "loss": 0.0433,
      "step": 13798
    },
    {
      "epoch": 3.4023668639053253,
      "grad_norm": 0.375,
      "learning_rate": 5.602988855225798e-05,
      "loss": 0.0438,
      "step": 13800
    },
    {
      "epoch": 3.4028599605522682,
      "grad_norm": 0.3046875,
      "learning_rate": 5.599897286881608e-05,
      "loss": 0.0396,
      "step": 13802
    },
    {
      "epoch": 3.403353057199211,
      "grad_norm": 0.37890625,
      "learning_rate": 5.5968062399741593e-05,
      "loss": 0.0395,
      "step": 13804
    },
    {
      "epoch": 3.4038461538461537,
      "grad_norm": 0.298828125,
      "learning_rate": 5.5937157148697514e-05,
      "loss": 0.0397,
      "step": 13806
    },
    {
      "epoch": 3.4043392504930967,
      "grad_norm": 0.337890625,
      "learning_rate": 5.590625711934634e-05,
      "loss": 0.0359,
      "step": 13808
    },
    {
      "epoch": 3.4048323471400392,
      "grad_norm": 0.30078125,
      "learning_rate": 5.587536231534989e-05,
      "loss": 0.0376,
      "step": 13810
    },
    {
      "epoch": 3.405325443786982,
      "grad_norm": 0.30859375,
      "learning_rate": 5.5844472740369355e-05,
      "loss": 0.0363,
      "step": 13812
    },
    {
      "epoch": 3.405818540433925,
      "grad_norm": 0.2890625,
      "learning_rate": 5.581358839806534e-05,
      "loss": 0.0372,
      "step": 13814
    },
    {
      "epoch": 3.4063116370808677,
      "grad_norm": 0.345703125,
      "learning_rate": 5.57827092920978e-05,
      "loss": 0.0405,
      "step": 13816
    },
    {
      "epoch": 3.4068047337278107,
      "grad_norm": 0.3515625,
      "learning_rate": 5.575183542612609e-05,
      "loss": 0.0392,
      "step": 13818
    },
    {
      "epoch": 3.4072978303747536,
      "grad_norm": 0.314453125,
      "learning_rate": 5.572096680380896e-05,
      "loss": 0.0366,
      "step": 13820
    },
    {
      "epoch": 3.407790927021696,
      "grad_norm": 0.3359375,
      "learning_rate": 5.5690103428804465e-05,
      "loss": 0.0379,
      "step": 13822
    },
    {
      "epoch": 3.408284023668639,
      "grad_norm": 0.326171875,
      "learning_rate": 5.565924530477009e-05,
      "loss": 0.0368,
      "step": 13824
    },
    {
      "epoch": 3.4087771203155817,
      "grad_norm": 0.333984375,
      "learning_rate": 5.562839243536273e-05,
      "loss": 0.0391,
      "step": 13826
    },
    {
      "epoch": 3.4092702169625246,
      "grad_norm": 0.30859375,
      "learning_rate": 5.5597544824238604e-05,
      "loss": 0.0405,
      "step": 13828
    },
    {
      "epoch": 3.4097633136094676,
      "grad_norm": 0.330078125,
      "learning_rate": 5.556670247505331e-05,
      "loss": 0.0361,
      "step": 13830
    },
    {
      "epoch": 3.41025641025641,
      "grad_norm": 0.322265625,
      "learning_rate": 5.553586539146187e-05,
      "loss": 0.0405,
      "step": 13832
    },
    {
      "epoch": 3.410749506903353,
      "grad_norm": 0.330078125,
      "learning_rate": 5.5505033577118684e-05,
      "loss": 0.0392,
      "step": 13834
    },
    {
      "epoch": 3.4112426035502956,
      "grad_norm": 0.3125,
      "learning_rate": 5.5474207035677364e-05,
      "loss": 0.0326,
      "step": 13836
    },
    {
      "epoch": 3.4117357001972386,
      "grad_norm": 0.3125,
      "learning_rate": 5.544338577079118e-05,
      "loss": 0.0349,
      "step": 13838
    },
    {
      "epoch": 3.4122287968441816,
      "grad_norm": 0.291015625,
      "learning_rate": 5.541256978611251e-05,
      "loss": 0.0384,
      "step": 13840
    },
    {
      "epoch": 3.412721893491124,
      "grad_norm": 0.3671875,
      "learning_rate": 5.5381759085293284e-05,
      "loss": 0.0388,
      "step": 13842
    },
    {
      "epoch": 3.413214990138067,
      "grad_norm": 0.384765625,
      "learning_rate": 5.535095367198471e-05,
      "loss": 0.0414,
      "step": 13844
    },
    {
      "epoch": 3.41370808678501,
      "grad_norm": 0.310546875,
      "learning_rate": 5.5320153549837415e-05,
      "loss": 0.0385,
      "step": 13846
    },
    {
      "epoch": 3.4142011834319526,
      "grad_norm": 0.376953125,
      "learning_rate": 5.528935872250143e-05,
      "loss": 0.0384,
      "step": 13848
    },
    {
      "epoch": 3.4146942800788955,
      "grad_norm": 0.37890625,
      "learning_rate": 5.5258569193626e-05,
      "loss": 0.0455,
      "step": 13850
    },
    {
      "epoch": 3.4151873767258385,
      "grad_norm": 0.380859375,
      "learning_rate": 5.5227784966859984e-05,
      "loss": 0.043,
      "step": 13852
    },
    {
      "epoch": 3.415680473372781,
      "grad_norm": 0.310546875,
      "learning_rate": 5.519700604585139e-05,
      "loss": 0.0379,
      "step": 13854
    },
    {
      "epoch": 3.416173570019724,
      "grad_norm": 0.28125,
      "learning_rate": 5.5166232434247725e-05,
      "loss": 0.0423,
      "step": 13856
    },
    {
      "epoch": 3.4166666666666665,
      "grad_norm": 0.3515625,
      "learning_rate": 5.5135464135695846e-05,
      "loss": 0.0397,
      "step": 13858
    },
    {
      "epoch": 3.4171597633136095,
      "grad_norm": 0.310546875,
      "learning_rate": 5.5104701153841956e-05,
      "loss": 0.0392,
      "step": 13860
    },
    {
      "epoch": 3.4176528599605525,
      "grad_norm": 0.388671875,
      "learning_rate": 5.507394349233168e-05,
      "loss": 0.0362,
      "step": 13862
    },
    {
      "epoch": 3.418145956607495,
      "grad_norm": 0.345703125,
      "learning_rate": 5.504319115480985e-05,
      "loss": 0.0403,
      "step": 13864
    },
    {
      "epoch": 3.418639053254438,
      "grad_norm": 0.330078125,
      "learning_rate": 5.501244414492096e-05,
      "loss": 0.0353,
      "step": 13866
    },
    {
      "epoch": 3.4191321499013805,
      "grad_norm": 0.306640625,
      "learning_rate": 5.498170246630854e-05,
      "loss": 0.0341,
      "step": 13868
    },
    {
      "epoch": 3.4196252465483234,
      "grad_norm": 0.314453125,
      "learning_rate": 5.495096612261581e-05,
      "loss": 0.0361,
      "step": 13870
    },
    {
      "epoch": 3.4201183431952664,
      "grad_norm": 0.341796875,
      "learning_rate": 5.492023511748506e-05,
      "loss": 0.0331,
      "step": 13872
    },
    {
      "epoch": 3.420611439842209,
      "grad_norm": 0.291015625,
      "learning_rate": 5.488950945455815e-05,
      "loss": 0.0386,
      "step": 13874
    },
    {
      "epoch": 3.421104536489152,
      "grad_norm": 0.35546875,
      "learning_rate": 5.485878913747629e-05,
      "loss": 0.0362,
      "step": 13876
    },
    {
      "epoch": 3.421597633136095,
      "grad_norm": 0.328125,
      "learning_rate": 5.482807416987985e-05,
      "loss": 0.0394,
      "step": 13878
    },
    {
      "epoch": 3.4220907297830374,
      "grad_norm": 0.3203125,
      "learning_rate": 5.479736455540892e-05,
      "loss": 0.0404,
      "step": 13880
    },
    {
      "epoch": 3.4225838264299804,
      "grad_norm": 0.271484375,
      "learning_rate": 5.4766660297702586e-05,
      "loss": 0.0371,
      "step": 13882
    },
    {
      "epoch": 3.423076923076923,
      "grad_norm": 0.31640625,
      "learning_rate": 5.473596140039965e-05,
      "loss": 0.0402,
      "step": 13884
    },
    {
      "epoch": 3.423570019723866,
      "grad_norm": 0.39453125,
      "learning_rate": 5.4705267867137955e-05,
      "loss": 0.0394,
      "step": 13886
    },
    {
      "epoch": 3.424063116370809,
      "grad_norm": 0.337890625,
      "learning_rate": 5.467457970155492e-05,
      "loss": 0.0401,
      "step": 13888
    },
    {
      "epoch": 3.4245562130177514,
      "grad_norm": 0.427734375,
      "learning_rate": 5.464389690728728e-05,
      "loss": 0.0378,
      "step": 13890
    },
    {
      "epoch": 3.4250493096646943,
      "grad_norm": 0.251953125,
      "learning_rate": 5.461321948797103e-05,
      "loss": 0.0366,
      "step": 13892
    },
    {
      "epoch": 3.425542406311637,
      "grad_norm": 0.33984375,
      "learning_rate": 5.458254744724176e-05,
      "loss": 0.0406,
      "step": 13894
    },
    {
      "epoch": 3.42603550295858,
      "grad_norm": 0.341796875,
      "learning_rate": 5.45518807887341e-05,
      "loss": 0.0425,
      "step": 13896
    },
    {
      "epoch": 3.426528599605523,
      "grad_norm": 0.298828125,
      "learning_rate": 5.4521219516082404e-05,
      "loss": 0.0376,
      "step": 13898
    },
    {
      "epoch": 3.4270216962524653,
      "grad_norm": 0.296875,
      "learning_rate": 5.4490563632920076e-05,
      "loss": 0.0394,
      "step": 13900
    },
    {
      "epoch": 3.4275147928994083,
      "grad_norm": 0.318359375,
      "learning_rate": 5.445991314288006e-05,
      "loss": 0.0365,
      "step": 13902
    },
    {
      "epoch": 3.4280078895463513,
      "grad_norm": 0.30078125,
      "learning_rate": 5.442926804959463e-05,
      "loss": 0.0389,
      "step": 13904
    },
    {
      "epoch": 3.428500986193294,
      "grad_norm": 0.29296875,
      "learning_rate": 5.439862835669529e-05,
      "loss": 0.0362,
      "step": 13906
    },
    {
      "epoch": 3.4289940828402368,
      "grad_norm": 0.337890625,
      "learning_rate": 5.436799406781317e-05,
      "loss": 0.0366,
      "step": 13908
    },
    {
      "epoch": 3.4294871794871793,
      "grad_norm": 0.29296875,
      "learning_rate": 5.433736518657846e-05,
      "loss": 0.0387,
      "step": 13910
    },
    {
      "epoch": 3.4299802761341223,
      "grad_norm": 0.314453125,
      "learning_rate": 5.4306741716621e-05,
      "loss": 0.0369,
      "step": 13912
    },
    {
      "epoch": 3.4304733727810652,
      "grad_norm": 0.30859375,
      "learning_rate": 5.427612366156971e-05,
      "loss": 0.0349,
      "step": 13914
    },
    {
      "epoch": 3.4309664694280078,
      "grad_norm": 0.330078125,
      "learning_rate": 5.424551102505308e-05,
      "loss": 0.0373,
      "step": 13916
    },
    {
      "epoch": 3.4314595660749507,
      "grad_norm": 0.384765625,
      "learning_rate": 5.4214903810698835e-05,
      "loss": 0.0392,
      "step": 13918
    },
    {
      "epoch": 3.4319526627218933,
      "grad_norm": 0.279296875,
      "learning_rate": 5.418430202213414e-05,
      "loss": 0.0337,
      "step": 13920
    },
    {
      "epoch": 3.4324457593688362,
      "grad_norm": 0.37109375,
      "learning_rate": 5.415370566298548e-05,
      "loss": 0.0402,
      "step": 13922
    },
    {
      "epoch": 3.432938856015779,
      "grad_norm": 0.296875,
      "learning_rate": 5.412311473687859e-05,
      "loss": 0.0362,
      "step": 13924
    },
    {
      "epoch": 3.4334319526627217,
      "grad_norm": 0.328125,
      "learning_rate": 5.409252924743884e-05,
      "loss": 0.0394,
      "step": 13926
    },
    {
      "epoch": 3.4339250493096647,
      "grad_norm": 0.36328125,
      "learning_rate": 5.406194919829064e-05,
      "loss": 0.0365,
      "step": 13928
    },
    {
      "epoch": 3.4344181459566077,
      "grad_norm": 0.388671875,
      "learning_rate": 5.4031374593057946e-05,
      "loss": 0.0412,
      "step": 13930
    },
    {
      "epoch": 3.43491124260355,
      "grad_norm": 0.328125,
      "learning_rate": 5.400080543536402e-05,
      "loss": 0.0417,
      "step": 13932
    },
    {
      "epoch": 3.435404339250493,
      "grad_norm": 0.310546875,
      "learning_rate": 5.397024172883147e-05,
      "loss": 0.0375,
      "step": 13934
    },
    {
      "epoch": 3.435897435897436,
      "grad_norm": 0.341796875,
      "learning_rate": 5.3939683477082324e-05,
      "loss": 0.0404,
      "step": 13936
    },
    {
      "epoch": 3.4363905325443787,
      "grad_norm": 0.41015625,
      "learning_rate": 5.390913068373777e-05,
      "loss": 0.0414,
      "step": 13938
    },
    {
      "epoch": 3.4368836291913216,
      "grad_norm": 0.490234375,
      "learning_rate": 5.3878583352418645e-05,
      "loss": 0.0395,
      "step": 13940
    },
    {
      "epoch": 3.437376725838264,
      "grad_norm": 0.30078125,
      "learning_rate": 5.384804148674487e-05,
      "loss": 0.0353,
      "step": 13942
    },
    {
      "epoch": 3.437869822485207,
      "grad_norm": 0.390625,
      "learning_rate": 5.381750509033585e-05,
      "loss": 0.0339,
      "step": 13944
    },
    {
      "epoch": 3.43836291913215,
      "grad_norm": 0.3046875,
      "learning_rate": 5.378697416681033e-05,
      "loss": 0.0382,
      "step": 13946
    },
    {
      "epoch": 3.4388560157790926,
      "grad_norm": 0.314453125,
      "learning_rate": 5.3756448719786403e-05,
      "loss": 0.04,
      "step": 13948
    },
    {
      "epoch": 3.4393491124260356,
      "grad_norm": 0.333984375,
      "learning_rate": 5.372592875288147e-05,
      "loss": 0.0371,
      "step": 13950
    },
    {
      "epoch": 3.439842209072978,
      "grad_norm": 0.326171875,
      "learning_rate": 5.3695414269712366e-05,
      "loss": 0.037,
      "step": 13952
    },
    {
      "epoch": 3.440335305719921,
      "grad_norm": 0.357421875,
      "learning_rate": 5.366490527389523e-05,
      "loss": 0.0373,
      "step": 13954
    },
    {
      "epoch": 3.440828402366864,
      "grad_norm": 0.345703125,
      "learning_rate": 5.363440176904547e-05,
      "loss": 0.0377,
      "step": 13956
    },
    {
      "epoch": 3.4413214990138066,
      "grad_norm": 0.369140625,
      "learning_rate": 5.360390375877799e-05,
      "loss": 0.0349,
      "step": 13958
    },
    {
      "epoch": 3.4418145956607495,
      "grad_norm": 0.328125,
      "learning_rate": 5.357341124670694e-05,
      "loss": 0.0366,
      "step": 13960
    },
    {
      "epoch": 3.4423076923076925,
      "grad_norm": 0.349609375,
      "learning_rate": 5.354292423644589e-05,
      "loss": 0.039,
      "step": 13962
    },
    {
      "epoch": 3.442800788954635,
      "grad_norm": 0.34765625,
      "learning_rate": 5.3512442731607694e-05,
      "loss": 0.035,
      "step": 13964
    },
    {
      "epoch": 3.443293885601578,
      "grad_norm": 0.3046875,
      "learning_rate": 5.3481966735804586e-05,
      "loss": 0.0364,
      "step": 13966
    },
    {
      "epoch": 3.4437869822485205,
      "grad_norm": 0.30078125,
      "learning_rate": 5.345149625264814e-05,
      "loss": 0.0354,
      "step": 13968
    },
    {
      "epoch": 3.4442800788954635,
      "grad_norm": 0.353515625,
      "learning_rate": 5.342103128574932e-05,
      "loss": 0.039,
      "step": 13970
    },
    {
      "epoch": 3.4447731755424065,
      "grad_norm": 0.31640625,
      "learning_rate": 5.339057183871832e-05,
      "loss": 0.0356,
      "step": 13972
    },
    {
      "epoch": 3.445266272189349,
      "grad_norm": 0.349609375,
      "learning_rate": 5.33601179151648e-05,
      "loss": 0.0395,
      "step": 13974
    },
    {
      "epoch": 3.445759368836292,
      "grad_norm": 0.3125,
      "learning_rate": 5.33296695186977e-05,
      "loss": 0.0382,
      "step": 13976
    },
    {
      "epoch": 3.4462524654832345,
      "grad_norm": 0.30078125,
      "learning_rate": 5.3299226652925336e-05,
      "loss": 0.037,
      "step": 13978
    },
    {
      "epoch": 3.4467455621301775,
      "grad_norm": 0.396484375,
      "learning_rate": 5.326878932145536e-05,
      "loss": 0.0344,
      "step": 13980
    },
    {
      "epoch": 3.4472386587771204,
      "grad_norm": 0.34375,
      "learning_rate": 5.323835752789477e-05,
      "loss": 0.0428,
      "step": 13982
    },
    {
      "epoch": 3.447731755424063,
      "grad_norm": 0.2734375,
      "learning_rate": 5.3207931275849934e-05,
      "loss": 0.0361,
      "step": 13984
    },
    {
      "epoch": 3.448224852071006,
      "grad_norm": 0.298828125,
      "learning_rate": 5.3177510568926434e-05,
      "loss": 0.0351,
      "step": 13986
    },
    {
      "epoch": 3.448717948717949,
      "grad_norm": 0.28125,
      "learning_rate": 5.314709541072938e-05,
      "loss": 0.0342,
      "step": 13988
    },
    {
      "epoch": 3.4492110453648914,
      "grad_norm": 0.373046875,
      "learning_rate": 5.311668580486311e-05,
      "loss": 0.036,
      "step": 13990
    },
    {
      "epoch": 3.4497041420118344,
      "grad_norm": 0.35546875,
      "learning_rate": 5.308628175493132e-05,
      "loss": 0.0383,
      "step": 13992
    },
    {
      "epoch": 3.4501972386587774,
      "grad_norm": 0.337890625,
      "learning_rate": 5.3055883264537085e-05,
      "loss": 0.0371,
      "step": 13994
    },
    {
      "epoch": 3.45069033530572,
      "grad_norm": 0.34375,
      "learning_rate": 5.302549033728279e-05,
      "loss": 0.0405,
      "step": 13996
    },
    {
      "epoch": 3.451183431952663,
      "grad_norm": 0.3046875,
      "learning_rate": 5.299510297677015e-05,
      "loss": 0.0329,
      "step": 13998
    },
    {
      "epoch": 3.4516765285996054,
      "grad_norm": 0.412109375,
      "learning_rate": 5.296472118660025e-05,
      "loss": 0.0372,
      "step": 14000
    },
    {
      "epoch": 3.4521696252465484,
      "grad_norm": 0.318359375,
      "learning_rate": 5.293434497037355e-05,
      "loss": 0.0406,
      "step": 14002
    },
    {
      "epoch": 3.452662721893491,
      "grad_norm": 0.30859375,
      "learning_rate": 5.290397433168972e-05,
      "loss": 0.0396,
      "step": 14004
    },
    {
      "epoch": 3.453155818540434,
      "grad_norm": 0.361328125,
      "learning_rate": 5.2873609274147864e-05,
      "loss": 0.0344,
      "step": 14006
    },
    {
      "epoch": 3.453648915187377,
      "grad_norm": 0.310546875,
      "learning_rate": 5.284324980134644e-05,
      "loss": 0.038,
      "step": 14008
    },
    {
      "epoch": 3.4541420118343193,
      "grad_norm": 0.2890625,
      "learning_rate": 5.28128959168832e-05,
      "loss": 0.0383,
      "step": 14010
    },
    {
      "epoch": 3.4546351084812623,
      "grad_norm": 0.443359375,
      "learning_rate": 5.278254762435525e-05,
      "loss": 0.0425,
      "step": 14012
    },
    {
      "epoch": 3.4551282051282053,
      "grad_norm": 0.408203125,
      "learning_rate": 5.275220492735904e-05,
      "loss": 0.0382,
      "step": 14014
    },
    {
      "epoch": 3.455621301775148,
      "grad_norm": 0.32421875,
      "learning_rate": 5.272186782949038e-05,
      "loss": 0.0368,
      "step": 14016
    },
    {
      "epoch": 3.456114398422091,
      "grad_norm": 0.322265625,
      "learning_rate": 5.2691536334344274e-05,
      "loss": 0.0343,
      "step": 14018
    },
    {
      "epoch": 3.4566074950690338,
      "grad_norm": 0.41796875,
      "learning_rate": 5.266121044551533e-05,
      "loss": 0.0375,
      "step": 14020
    },
    {
      "epoch": 3.4571005917159763,
      "grad_norm": 0.4296875,
      "learning_rate": 5.2630890166597215e-05,
      "loss": 0.0387,
      "step": 14022
    },
    {
      "epoch": 3.4575936883629192,
      "grad_norm": 0.3046875,
      "learning_rate": 5.260057550118307e-05,
      "loss": 0.039,
      "step": 14024
    },
    {
      "epoch": 3.4580867850098618,
      "grad_norm": 0.279296875,
      "learning_rate": 5.25702664528654e-05,
      "loss": 0.0391,
      "step": 14026
    },
    {
      "epoch": 3.4585798816568047,
      "grad_norm": 0.298828125,
      "learning_rate": 5.253996302523596e-05,
      "loss": 0.0365,
      "step": 14028
    },
    {
      "epoch": 3.4590729783037477,
      "grad_norm": 0.27734375,
      "learning_rate": 5.250966522188592e-05,
      "loss": 0.0328,
      "step": 14030
    },
    {
      "epoch": 3.4595660749506902,
      "grad_norm": 0.3828125,
      "learning_rate": 5.2479373046405625e-05,
      "loss": 0.0364,
      "step": 14032
    },
    {
      "epoch": 3.460059171597633,
      "grad_norm": 0.314453125,
      "learning_rate": 5.244908650238503e-05,
      "loss": 0.0348,
      "step": 14034
    },
    {
      "epoch": 3.4605522682445757,
      "grad_norm": 0.302734375,
      "learning_rate": 5.241880559341312e-05,
      "loss": 0.0379,
      "step": 14036
    },
    {
      "epoch": 3.4610453648915187,
      "grad_norm": 0.28515625,
      "learning_rate": 5.238853032307842e-05,
      "loss": 0.0354,
      "step": 14038
    },
    {
      "epoch": 3.4615384615384617,
      "grad_norm": 0.3203125,
      "learning_rate": 5.23582606949687e-05,
      "loss": 0.0348,
      "step": 14040
    },
    {
      "epoch": 3.462031558185404,
      "grad_norm": 0.3046875,
      "learning_rate": 5.232799671267109e-05,
      "loss": 0.037,
      "step": 14042
    },
    {
      "epoch": 3.462524654832347,
      "grad_norm": 0.31640625,
      "learning_rate": 5.229773837977208e-05,
      "loss": 0.0393,
      "step": 14044
    },
    {
      "epoch": 3.46301775147929,
      "grad_norm": 0.361328125,
      "learning_rate": 5.226748569985731e-05,
      "loss": 0.0353,
      "step": 14046
    },
    {
      "epoch": 3.4635108481262327,
      "grad_norm": 0.3125,
      "learning_rate": 5.223723867651208e-05,
      "loss": 0.0437,
      "step": 14048
    },
    {
      "epoch": 3.4640039447731756,
      "grad_norm": 0.302734375,
      "learning_rate": 5.220699731332071e-05,
      "loss": 0.0375,
      "step": 14050
    },
    {
      "epoch": 3.464497041420118,
      "grad_norm": 0.3515625,
      "learning_rate": 5.2176761613866974e-05,
      "loss": 0.0391,
      "step": 14052
    },
    {
      "epoch": 3.464990138067061,
      "grad_norm": 0.341796875,
      "learning_rate": 5.214653158173401e-05,
      "loss": 0.0371,
      "step": 14054
    },
    {
      "epoch": 3.465483234714004,
      "grad_norm": 0.306640625,
      "learning_rate": 5.2116307220504245e-05,
      "loss": 0.0392,
      "step": 14056
    },
    {
      "epoch": 3.4659763313609466,
      "grad_norm": 0.29296875,
      "learning_rate": 5.208608853375945e-05,
      "loss": 0.0367,
      "step": 14058
    },
    {
      "epoch": 3.4664694280078896,
      "grad_norm": 0.337890625,
      "learning_rate": 5.205587552508061e-05,
      "loss": 0.0377,
      "step": 14060
    },
    {
      "epoch": 3.466962524654832,
      "grad_norm": 0.322265625,
      "learning_rate": 5.202566819804828e-05,
      "loss": 0.038,
      "step": 14062
    },
    {
      "epoch": 3.467455621301775,
      "grad_norm": 0.318359375,
      "learning_rate": 5.199546655624208e-05,
      "loss": 0.0393,
      "step": 14064
    },
    {
      "epoch": 3.467948717948718,
      "grad_norm": 0.349609375,
      "learning_rate": 5.1965270603241124e-05,
      "loss": 0.0404,
      "step": 14066
    },
    {
      "epoch": 3.4684418145956606,
      "grad_norm": 0.345703125,
      "learning_rate": 5.1935080342623786e-05,
      "loss": 0.0411,
      "step": 14068
    },
    {
      "epoch": 3.4689349112426036,
      "grad_norm": 0.294921875,
      "learning_rate": 5.1904895777967796e-05,
      "loss": 0.0355,
      "step": 14070
    },
    {
      "epoch": 3.4694280078895465,
      "grad_norm": 0.31640625,
      "learning_rate": 5.1874716912850216e-05,
      "loss": 0.0342,
      "step": 14072
    },
    {
      "epoch": 3.469921104536489,
      "grad_norm": 0.337890625,
      "learning_rate": 5.184454375084731e-05,
      "loss": 0.0393,
      "step": 14074
    },
    {
      "epoch": 3.470414201183432,
      "grad_norm": 0.349609375,
      "learning_rate": 5.1814376295534895e-05,
      "loss": 0.0385,
      "step": 14076
    },
    {
      "epoch": 3.470907297830375,
      "grad_norm": 0.29296875,
      "learning_rate": 5.178421455048786e-05,
      "loss": 0.0369,
      "step": 14078
    },
    {
      "epoch": 3.4714003944773175,
      "grad_norm": 0.291015625,
      "learning_rate": 5.175405851928068e-05,
      "loss": 0.0366,
      "step": 14080
    },
    {
      "epoch": 3.4718934911242605,
      "grad_norm": 0.2734375,
      "learning_rate": 5.1723908205486896e-05,
      "loss": 0.0347,
      "step": 14082
    },
    {
      "epoch": 3.472386587771203,
      "grad_norm": 0.35546875,
      "learning_rate": 5.1693763612679526e-05,
      "loss": 0.0357,
      "step": 14084
    },
    {
      "epoch": 3.472879684418146,
      "grad_norm": 0.2890625,
      "learning_rate": 5.1663624744430915e-05,
      "loss": 0.0394,
      "step": 14086
    },
    {
      "epoch": 3.4733727810650885,
      "grad_norm": 0.302734375,
      "learning_rate": 5.163349160431257e-05,
      "loss": 0.0347,
      "step": 14088
    },
    {
      "epoch": 3.4738658777120315,
      "grad_norm": 0.33203125,
      "learning_rate": 5.1603364195895596e-05,
      "loss": 0.0361,
      "step": 14090
    },
    {
      "epoch": 3.4743589743589745,
      "grad_norm": 0.287109375,
      "learning_rate": 5.15732425227501e-05,
      "loss": 0.0352,
      "step": 14092
    },
    {
      "epoch": 3.474852071005917,
      "grad_norm": 0.31640625,
      "learning_rate": 5.1543126588445825e-05,
      "loss": 0.0399,
      "step": 14094
    },
    {
      "epoch": 3.47534516765286,
      "grad_norm": 0.392578125,
      "learning_rate": 5.151301639655155e-05,
      "loss": 0.038,
      "step": 14096
    },
    {
      "epoch": 3.475838264299803,
      "grad_norm": 0.384765625,
      "learning_rate": 5.148291195063555e-05,
      "loss": 0.0349,
      "step": 14098
    },
    {
      "epoch": 3.4763313609467454,
      "grad_norm": 0.33984375,
      "learning_rate": 5.1452813254265364e-05,
      "loss": 0.0365,
      "step": 14100
    },
    {
      "epoch": 3.4768244575936884,
      "grad_norm": 0.3671875,
      "learning_rate": 5.142272031100788e-05,
      "loss": 0.0377,
      "step": 14102
    },
    {
      "epoch": 3.4773175542406314,
      "grad_norm": 0.322265625,
      "learning_rate": 5.139263312442928e-05,
      "loss": 0.0357,
      "step": 14104
    },
    {
      "epoch": 3.477810650887574,
      "grad_norm": 0.271484375,
      "learning_rate": 5.1362551698094964e-05,
      "loss": 0.0357,
      "step": 14106
    },
    {
      "epoch": 3.478303747534517,
      "grad_norm": 0.318359375,
      "learning_rate": 5.133247603556992e-05,
      "loss": 0.0394,
      "step": 14108
    },
    {
      "epoch": 3.4787968441814594,
      "grad_norm": 0.341796875,
      "learning_rate": 5.130240614041815e-05,
      "loss": 0.0358,
      "step": 14110
    },
    {
      "epoch": 3.4792899408284024,
      "grad_norm": 0.33203125,
      "learning_rate": 5.1272342016203126e-05,
      "loss": 0.0343,
      "step": 14112
    },
    {
      "epoch": 3.4797830374753453,
      "grad_norm": 0.376953125,
      "learning_rate": 5.124228366648765e-05,
      "loss": 0.0382,
      "step": 14114
    },
    {
      "epoch": 3.480276134122288,
      "grad_norm": 0.361328125,
      "learning_rate": 5.1212231094833784e-05,
      "loss": 0.039,
      "step": 14116
    },
    {
      "epoch": 3.480769230769231,
      "grad_norm": 0.30859375,
      "learning_rate": 5.118218430480297e-05,
      "loss": 0.0391,
      "step": 14118
    },
    {
      "epoch": 3.4812623274161734,
      "grad_norm": 0.36328125,
      "learning_rate": 5.11521432999558e-05,
      "loss": 0.0396,
      "step": 14120
    },
    {
      "epoch": 3.4817554240631163,
      "grad_norm": 0.291015625,
      "learning_rate": 5.112210808385246e-05,
      "loss": 0.0371,
      "step": 14122
    },
    {
      "epoch": 3.4822485207100593,
      "grad_norm": 0.2734375,
      "learning_rate": 5.109207866005217e-05,
      "loss": 0.0371,
      "step": 14124
    },
    {
      "epoch": 3.482741617357002,
      "grad_norm": 0.2890625,
      "learning_rate": 5.106205503211363e-05,
      "loss": 0.0344,
      "step": 14126
    },
    {
      "epoch": 3.483234714003945,
      "grad_norm": 0.33984375,
      "learning_rate": 5.1032037203594794e-05,
      "loss": 0.0385,
      "step": 14128
    },
    {
      "epoch": 3.4837278106508878,
      "grad_norm": 0.29296875,
      "learning_rate": 5.100202517805297e-05,
      "loss": 0.0355,
      "step": 14130
    },
    {
      "epoch": 3.4842209072978303,
      "grad_norm": 0.31640625,
      "learning_rate": 5.097201895904471e-05,
      "loss": 0.036,
      "step": 14132
    },
    {
      "epoch": 3.4847140039447733,
      "grad_norm": 0.341796875,
      "learning_rate": 5.094201855012596e-05,
      "loss": 0.0376,
      "step": 14134
    },
    {
      "epoch": 3.485207100591716,
      "grad_norm": 0.279296875,
      "learning_rate": 5.0912023954851953e-05,
      "loss": 0.0366,
      "step": 14136
    },
    {
      "epoch": 3.4857001972386588,
      "grad_norm": 0.296875,
      "learning_rate": 5.088203517677712e-05,
      "loss": 0.0373,
      "step": 14138
    },
    {
      "epoch": 3.4861932938856017,
      "grad_norm": 0.294921875,
      "learning_rate": 5.085205221945537e-05,
      "loss": 0.0318,
      "step": 14140
    },
    {
      "epoch": 3.4866863905325443,
      "grad_norm": 0.33984375,
      "learning_rate": 5.082207508643985e-05,
      "loss": 0.0374,
      "step": 14142
    },
    {
      "epoch": 3.4871794871794872,
      "grad_norm": 0.287109375,
      "learning_rate": 5.0792103781282997e-05,
      "loss": 0.0366,
      "step": 14144
    },
    {
      "epoch": 3.4876725838264298,
      "grad_norm": 0.36328125,
      "learning_rate": 5.0762138307536586e-05,
      "loss": 0.038,
      "step": 14146
    },
    {
      "epoch": 3.4881656804733727,
      "grad_norm": 0.28125,
      "learning_rate": 5.07321786687517e-05,
      "loss": 0.0351,
      "step": 14148
    },
    {
      "epoch": 3.4886587771203157,
      "grad_norm": 0.326171875,
      "learning_rate": 5.0702224868478706e-05,
      "loss": 0.0378,
      "step": 14150
    },
    {
      "epoch": 3.489151873767258,
      "grad_norm": 0.345703125,
      "learning_rate": 5.0672276910267366e-05,
      "loss": 0.0366,
      "step": 14152
    },
    {
      "epoch": 3.489644970414201,
      "grad_norm": 0.322265625,
      "learning_rate": 5.064233479766657e-05,
      "loss": 0.042,
      "step": 14154
    },
    {
      "epoch": 3.490138067061144,
      "grad_norm": 0.302734375,
      "learning_rate": 5.061239853422468e-05,
      "loss": 0.0338,
      "step": 14156
    },
    {
      "epoch": 3.4906311637080867,
      "grad_norm": 0.30078125,
      "learning_rate": 5.058246812348931e-05,
      "loss": 0.0358,
      "step": 14158
    },
    {
      "epoch": 3.4911242603550297,
      "grad_norm": 0.283203125,
      "learning_rate": 5.0552543569007384e-05,
      "loss": 0.0329,
      "step": 14160
    },
    {
      "epoch": 3.4916173570019726,
      "grad_norm": 0.296875,
      "learning_rate": 5.052262487432513e-05,
      "loss": 0.0351,
      "step": 14162
    },
    {
      "epoch": 3.492110453648915,
      "grad_norm": 0.3203125,
      "learning_rate": 5.0492712042988065e-05,
      "loss": 0.0348,
      "step": 14164
    },
    {
      "epoch": 3.492603550295858,
      "grad_norm": 0.33984375,
      "learning_rate": 5.046280507854106e-05,
      "loss": 0.0359,
      "step": 14166
    },
    {
      "epoch": 3.4930966469428006,
      "grad_norm": 0.302734375,
      "learning_rate": 5.0432903984528225e-05,
      "loss": 0.0349,
      "step": 14168
    },
    {
      "epoch": 3.4935897435897436,
      "grad_norm": 0.310546875,
      "learning_rate": 5.0403008764492995e-05,
      "loss": 0.0372,
      "step": 14170
    },
    {
      "epoch": 3.4940828402366866,
      "grad_norm": 0.29296875,
      "learning_rate": 5.037311942197815e-05,
      "loss": 0.0294,
      "step": 14172
    },
    {
      "epoch": 3.494575936883629,
      "grad_norm": 0.294921875,
      "learning_rate": 5.0343235960525725e-05,
      "loss": 0.0352,
      "step": 14174
    },
    {
      "epoch": 3.495069033530572,
      "grad_norm": 0.287109375,
      "learning_rate": 5.03133583836771e-05,
      "loss": 0.0356,
      "step": 14176
    },
    {
      "epoch": 3.4955621301775146,
      "grad_norm": 0.31640625,
      "learning_rate": 5.028348669497292e-05,
      "loss": 0.0419,
      "step": 14178
    },
    {
      "epoch": 3.4960552268244576,
      "grad_norm": 0.306640625,
      "learning_rate": 5.025362089795317e-05,
      "loss": 0.0373,
      "step": 14180
    },
    {
      "epoch": 3.4965483234714005,
      "grad_norm": 0.3359375,
      "learning_rate": 5.0223760996157024e-05,
      "loss": 0.0397,
      "step": 14182
    },
    {
      "epoch": 3.497041420118343,
      "grad_norm": 0.400390625,
      "learning_rate": 5.01939069931232e-05,
      "loss": 0.0364,
      "step": 14184
    },
    {
      "epoch": 3.497534516765286,
      "grad_norm": 0.3046875,
      "learning_rate": 5.016405889238942e-05,
      "loss": 0.0375,
      "step": 14186
    },
    {
      "epoch": 3.498027613412229,
      "grad_norm": 0.333984375,
      "learning_rate": 5.013421669749293e-05,
      "loss": 0.0376,
      "step": 14188
    },
    {
      "epoch": 3.4985207100591715,
      "grad_norm": 0.330078125,
      "learning_rate": 5.010438041197017e-05,
      "loss": 0.0369,
      "step": 14190
    },
    {
      "epoch": 3.4990138067061145,
      "grad_norm": 0.34375,
      "learning_rate": 5.0074550039356905e-05,
      "loss": 0.0383,
      "step": 14192
    },
    {
      "epoch": 3.499506903353057,
      "grad_norm": 0.376953125,
      "learning_rate": 5.004472558318827e-05,
      "loss": 0.0449,
      "step": 14194
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.33203125,
      "learning_rate": 5.001490704699847e-05,
      "loss": 0.0441,
      "step": 14196
    },
    {
      "epoch": 3.5004930966469425,
      "grad_norm": 0.29296875,
      "learning_rate": 4.998509443432134e-05,
      "loss": 0.037,
      "step": 14198
    },
    {
      "epoch": 3.5009861932938855,
      "grad_norm": 0.265625,
      "learning_rate": 4.9955287748689736e-05,
      "loss": 0.036,
      "step": 14200
    },
    {
      "epoch": 3.5014792899408285,
      "grad_norm": 0.36328125,
      "learning_rate": 4.992548699363595e-05,
      "loss": 0.0381,
      "step": 14202
    },
    {
      "epoch": 3.501972386587771,
      "grad_norm": 0.310546875,
      "learning_rate": 4.9895692172691534e-05,
      "loss": 0.0412,
      "step": 14204
    },
    {
      "epoch": 3.502465483234714,
      "grad_norm": 0.302734375,
      "learning_rate": 4.986590328938734e-05,
      "loss": 0.0361,
      "step": 14206
    },
    {
      "epoch": 3.502958579881657,
      "grad_norm": 0.29296875,
      "learning_rate": 4.9836120347253554e-05,
      "loss": 0.0367,
      "step": 14208
    },
    {
      "epoch": 3.5034516765285995,
      "grad_norm": 0.322265625,
      "learning_rate": 4.980634334981952e-05,
      "loss": 0.0412,
      "step": 14210
    },
    {
      "epoch": 3.5039447731755424,
      "grad_norm": 0.322265625,
      "learning_rate": 4.9776572300614125e-05,
      "loss": 0.0351,
      "step": 14212
    },
    {
      "epoch": 3.5044378698224854,
      "grad_norm": 0.376953125,
      "learning_rate": 4.9746807203165225e-05,
      "loss": 0.0364,
      "step": 14214
    },
    {
      "epoch": 3.504930966469428,
      "grad_norm": 0.27734375,
      "learning_rate": 4.971704806100035e-05,
      "loss": 0.036,
      "step": 14216
    },
    {
      "epoch": 3.505424063116371,
      "grad_norm": 0.345703125,
      "learning_rate": 4.968729487764596e-05,
      "loss": 0.0355,
      "step": 14218
    },
    {
      "epoch": 3.505917159763314,
      "grad_norm": 0.279296875,
      "learning_rate": 4.965754765662806e-05,
      "loss": 0.038,
      "step": 14220
    },
    {
      "epoch": 3.5064102564102564,
      "grad_norm": 0.380859375,
      "learning_rate": 4.962780640147187e-05,
      "loss": 0.0372,
      "step": 14222
    },
    {
      "epoch": 3.5069033530571994,
      "grad_norm": 0.3828125,
      "learning_rate": 4.959807111570178e-05,
      "loss": 0.0377,
      "step": 14224
    },
    {
      "epoch": 3.507396449704142,
      "grad_norm": 0.263671875,
      "learning_rate": 4.956834180284176e-05,
      "loss": 0.0383,
      "step": 14226
    },
    {
      "epoch": 3.507889546351085,
      "grad_norm": 0.310546875,
      "learning_rate": 4.9538618466414724e-05,
      "loss": 0.0346,
      "step": 14228
    },
    {
      "epoch": 3.5083826429980274,
      "grad_norm": 0.3359375,
      "learning_rate": 4.9508901109943215e-05,
      "loss": 0.0361,
      "step": 14230
    },
    {
      "epoch": 3.5088757396449703,
      "grad_norm": 0.345703125,
      "learning_rate": 4.947918973694881e-05,
      "loss": 0.0367,
      "step": 14232
    },
    {
      "epoch": 3.5093688362919133,
      "grad_norm": 0.3046875,
      "learning_rate": 4.944948435095248e-05,
      "loss": 0.0359,
      "step": 14234
    },
    {
      "epoch": 3.509861932938856,
      "grad_norm": 0.33984375,
      "learning_rate": 4.9419784955474524e-05,
      "loss": 0.0314,
      "step": 14236
    },
    {
      "epoch": 3.510355029585799,
      "grad_norm": 0.296875,
      "learning_rate": 4.9390091554034393e-05,
      "loss": 0.0352,
      "step": 14238
    },
    {
      "epoch": 3.510848126232742,
      "grad_norm": 0.388671875,
      "learning_rate": 4.936040415015105e-05,
      "loss": 0.0385,
      "step": 14240
    },
    {
      "epoch": 3.5113412228796843,
      "grad_norm": 0.3203125,
      "learning_rate": 4.9330722747342474e-05,
      "loss": 0.0362,
      "step": 14242
    },
    {
      "epoch": 3.5118343195266273,
      "grad_norm": 0.353515625,
      "learning_rate": 4.930104734912623e-05,
      "loss": 0.0381,
      "step": 14244
    },
    {
      "epoch": 3.5123274161735702,
      "grad_norm": 0.314453125,
      "learning_rate": 4.92713779590189e-05,
      "loss": 0.0361,
      "step": 14246
    },
    {
      "epoch": 3.5128205128205128,
      "grad_norm": 0.318359375,
      "learning_rate": 4.9241714580536526e-05,
      "loss": 0.037,
      "step": 14248
    },
    {
      "epoch": 3.5133136094674557,
      "grad_norm": 0.478515625,
      "learning_rate": 4.9212057217194405e-05,
      "loss": 0.0399,
      "step": 14250
    },
    {
      "epoch": 3.5138067061143983,
      "grad_norm": 0.3828125,
      "learning_rate": 4.918240587250699e-05,
      "loss": 0.0357,
      "step": 14252
    },
    {
      "epoch": 3.5142998027613412,
      "grad_norm": 0.275390625,
      "learning_rate": 4.915276054998828e-05,
      "loss": 0.0346,
      "step": 14254
    },
    {
      "epoch": 3.5147928994082838,
      "grad_norm": 0.337890625,
      "learning_rate": 4.9123121253151275e-05,
      "loss": 0.0364,
      "step": 14256
    },
    {
      "epoch": 3.5152859960552267,
      "grad_norm": 0.279296875,
      "learning_rate": 4.909348798550852e-05,
      "loss": 0.0346,
      "step": 14258
    },
    {
      "epoch": 3.5157790927021697,
      "grad_norm": 0.296875,
      "learning_rate": 4.906386075057164e-05,
      "loss": 0.0349,
      "step": 14260
    },
    {
      "epoch": 3.5162721893491122,
      "grad_norm": 0.30859375,
      "learning_rate": 4.903423955185165e-05,
      "loss": 0.0358,
      "step": 14262
    },
    {
      "epoch": 3.516765285996055,
      "grad_norm": 0.3046875,
      "learning_rate": 4.9004624392858835e-05,
      "loss": 0.0352,
      "step": 14264
    },
    {
      "epoch": 3.517258382642998,
      "grad_norm": 0.37109375,
      "learning_rate": 4.897501527710274e-05,
      "loss": 0.0351,
      "step": 14266
    },
    {
      "epoch": 3.5177514792899407,
      "grad_norm": 0.326171875,
      "learning_rate": 4.8945412208092265e-05,
      "loss": 0.0351,
      "step": 14268
    },
    {
      "epoch": 3.5182445759368837,
      "grad_norm": 0.29296875,
      "learning_rate": 4.891581518933541e-05,
      "loss": 0.0389,
      "step": 14270
    },
    {
      "epoch": 3.5187376725838266,
      "grad_norm": 0.291015625,
      "learning_rate": 4.888622422433976e-05,
      "loss": 0.0347,
      "step": 14272
    },
    {
      "epoch": 3.519230769230769,
      "grad_norm": 0.341796875,
      "learning_rate": 4.8856639316611885e-05,
      "loss": 0.0376,
      "step": 14274
    },
    {
      "epoch": 3.519723865877712,
      "grad_norm": 0.322265625,
      "learning_rate": 4.882706046965779e-05,
      "loss": 0.0384,
      "step": 14276
    },
    {
      "epoch": 3.520216962524655,
      "grad_norm": 0.29296875,
      "learning_rate": 4.879748768698276e-05,
      "loss": 0.0373,
      "step": 14278
    },
    {
      "epoch": 3.5207100591715976,
      "grad_norm": 0.30078125,
      "learning_rate": 4.8767920972091296e-05,
      "loss": 0.0355,
      "step": 14280
    },
    {
      "epoch": 3.5212031558185406,
      "grad_norm": 0.31640625,
      "learning_rate": 4.8738360328487256e-05,
      "loss": 0.0355,
      "step": 14282
    },
    {
      "epoch": 3.521696252465483,
      "grad_norm": 0.33203125,
      "learning_rate": 4.8708805759673714e-05,
      "loss": 0.0344,
      "step": 14284
    },
    {
      "epoch": 3.522189349112426,
      "grad_norm": 0.310546875,
      "learning_rate": 4.86792572691531e-05,
      "loss": 0.0346,
      "step": 14286
    },
    {
      "epoch": 3.5226824457593686,
      "grad_norm": 0.33203125,
      "learning_rate": 4.864971486042701e-05,
      "loss": 0.0374,
      "step": 14288
    },
    {
      "epoch": 3.5231755424063116,
      "grad_norm": 0.29296875,
      "learning_rate": 4.862017853699641e-05,
      "loss": 0.0345,
      "step": 14290
    },
    {
      "epoch": 3.5236686390532546,
      "grad_norm": 0.359375,
      "learning_rate": 4.8590648302361516e-05,
      "loss": 0.0384,
      "step": 14292
    },
    {
      "epoch": 3.524161735700197,
      "grad_norm": 0.349609375,
      "learning_rate": 4.8561124160021834e-05,
      "loss": 0.0397,
      "step": 14294
    },
    {
      "epoch": 3.52465483234714,
      "grad_norm": 0.365234375,
      "learning_rate": 4.8531606113476134e-05,
      "loss": 0.0363,
      "step": 14296
    },
    {
      "epoch": 3.525147928994083,
      "grad_norm": 0.279296875,
      "learning_rate": 4.850209416622248e-05,
      "loss": 0.0349,
      "step": 14298
    },
    {
      "epoch": 3.5256410256410255,
      "grad_norm": 0.328125,
      "learning_rate": 4.8472588321758226e-05,
      "loss": 0.037,
      "step": 14300
    },
    {
      "epoch": 3.5261341222879685,
      "grad_norm": 0.3984375,
      "learning_rate": 4.844308858357992e-05,
      "loss": 0.0366,
      "step": 14302
    },
    {
      "epoch": 3.5266272189349115,
      "grad_norm": 0.333984375,
      "learning_rate": 4.841359495518346e-05,
      "loss": 0.0376,
      "step": 14304
    },
    {
      "epoch": 3.527120315581854,
      "grad_norm": 0.3359375,
      "learning_rate": 4.838410744006403e-05,
      "loss": 0.039,
      "step": 14306
    },
    {
      "epoch": 3.527613412228797,
      "grad_norm": 0.26171875,
      "learning_rate": 4.8354626041716045e-05,
      "loss": 0.0333,
      "step": 14308
    },
    {
      "epoch": 3.5281065088757395,
      "grad_norm": 0.34375,
      "learning_rate": 4.832515076363323e-05,
      "loss": 0.034,
      "step": 14310
    },
    {
      "epoch": 3.5285996055226825,
      "grad_norm": 0.3359375,
      "learning_rate": 4.829568160930858e-05,
      "loss": 0.0369,
      "step": 14312
    },
    {
      "epoch": 3.529092702169625,
      "grad_norm": 0.294921875,
      "learning_rate": 4.826621858223431e-05,
      "loss": 0.0366,
      "step": 14314
    },
    {
      "epoch": 3.529585798816568,
      "grad_norm": 0.291015625,
      "learning_rate": 4.8236761685902045e-05,
      "loss": 0.0333,
      "step": 14316
    },
    {
      "epoch": 3.530078895463511,
      "grad_norm": 0.3515625,
      "learning_rate": 4.820731092380247e-05,
      "loss": 0.0413,
      "step": 14318
    },
    {
      "epoch": 3.5305719921104535,
      "grad_norm": 0.3203125,
      "learning_rate": 4.8177866299425746e-05,
      "loss": 0.0385,
      "step": 14320
    },
    {
      "epoch": 3.5310650887573964,
      "grad_norm": 0.275390625,
      "learning_rate": 4.814842781626119e-05,
      "loss": 0.037,
      "step": 14322
    },
    {
      "epoch": 3.5315581854043394,
      "grad_norm": 0.34765625,
      "learning_rate": 4.811899547779743e-05,
      "loss": 0.0402,
      "step": 14324
    },
    {
      "epoch": 3.532051282051282,
      "grad_norm": 0.30859375,
      "learning_rate": 4.8089569287522396e-05,
      "loss": 0.0377,
      "step": 14326
    },
    {
      "epoch": 3.532544378698225,
      "grad_norm": 0.294921875,
      "learning_rate": 4.8060149248923215e-05,
      "loss": 0.0372,
      "step": 14328
    },
    {
      "epoch": 3.533037475345168,
      "grad_norm": 0.29296875,
      "learning_rate": 4.803073536548639e-05,
      "loss": 0.0354,
      "step": 14330
    },
    {
      "epoch": 3.5335305719921104,
      "grad_norm": 0.3515625,
      "learning_rate": 4.800132764069755e-05,
      "loss": 0.0344,
      "step": 14332
    },
    {
      "epoch": 3.5340236686390534,
      "grad_norm": 0.314453125,
      "learning_rate": 4.79719260780417e-05,
      "loss": 0.0358,
      "step": 14334
    },
    {
      "epoch": 3.534516765285996,
      "grad_norm": 0.328125,
      "learning_rate": 4.7942530681003084e-05,
      "loss": 0.0337,
      "step": 14336
    },
    {
      "epoch": 3.535009861932939,
      "grad_norm": 0.2734375,
      "learning_rate": 4.791314145306526e-05,
      "loss": 0.0309,
      "step": 14338
    },
    {
      "epoch": 3.5355029585798814,
      "grad_norm": 0.3046875,
      "learning_rate": 4.788375839771097e-05,
      "loss": 0.0393,
      "step": 14340
    },
    {
      "epoch": 3.5359960552268244,
      "grad_norm": 0.291015625,
      "learning_rate": 4.7854381518422296e-05,
      "loss": 0.0375,
      "step": 14342
    },
    {
      "epoch": 3.5364891518737673,
      "grad_norm": 0.376953125,
      "learning_rate": 4.782501081868059e-05,
      "loss": 0.0364,
      "step": 14344
    },
    {
      "epoch": 3.53698224852071,
      "grad_norm": 0.3359375,
      "learning_rate": 4.7795646301966344e-05,
      "loss": 0.034,
      "step": 14346
    },
    {
      "epoch": 3.537475345167653,
      "grad_norm": 0.30859375,
      "learning_rate": 4.7766287971759546e-05,
      "loss": 0.0346,
      "step": 14348
    },
    {
      "epoch": 3.537968441814596,
      "grad_norm": 0.279296875,
      "learning_rate": 4.7736935831539233e-05,
      "loss": 0.0353,
      "step": 14350
    },
    {
      "epoch": 3.5384615384615383,
      "grad_norm": 0.2890625,
      "learning_rate": 4.7707589884783824e-05,
      "loss": 0.0345,
      "step": 14352
    },
    {
      "epoch": 3.5389546351084813,
      "grad_norm": 0.41796875,
      "learning_rate": 4.767825013497097e-05,
      "loss": 0.0366,
      "step": 14354
    },
    {
      "epoch": 3.5394477317554243,
      "grad_norm": 0.33984375,
      "learning_rate": 4.764891658557761e-05,
      "loss": 0.0391,
      "step": 14356
    },
    {
      "epoch": 3.539940828402367,
      "grad_norm": 0.3515625,
      "learning_rate": 4.761958924007995e-05,
      "loss": 0.0365,
      "step": 14358
    },
    {
      "epoch": 3.5404339250493098,
      "grad_norm": 0.322265625,
      "learning_rate": 4.759026810195335e-05,
      "loss": 0.035,
      "step": 14360
    },
    {
      "epoch": 3.5409270216962527,
      "grad_norm": 0.30859375,
      "learning_rate": 4.756095317467269e-05,
      "loss": 0.0385,
      "step": 14362
    },
    {
      "epoch": 3.5414201183431953,
      "grad_norm": 0.322265625,
      "learning_rate": 4.7531644461711755e-05,
      "loss": 0.0353,
      "step": 14364
    },
    {
      "epoch": 3.5419132149901382,
      "grad_norm": 0.30078125,
      "learning_rate": 4.7502341966544e-05,
      "loss": 0.0367,
      "step": 14366
    },
    {
      "epoch": 3.5424063116370808,
      "grad_norm": 0.337890625,
      "learning_rate": 4.747304569264178e-05,
      "loss": 0.0368,
      "step": 14368
    },
    {
      "epoch": 3.5428994082840237,
      "grad_norm": 0.328125,
      "learning_rate": 4.7443755643476915e-05,
      "loss": 0.0394,
      "step": 14370
    },
    {
      "epoch": 3.5433925049309662,
      "grad_norm": 0.32421875,
      "learning_rate": 4.741447182252049e-05,
      "loss": 0.0369,
      "step": 14372
    },
    {
      "epoch": 3.543885601577909,
      "grad_norm": 0.275390625,
      "learning_rate": 4.738519423324267e-05,
      "loss": 0.0338,
      "step": 14374
    },
    {
      "epoch": 3.544378698224852,
      "grad_norm": 0.3359375,
      "learning_rate": 4.735592287911318e-05,
      "loss": 0.0344,
      "step": 14376
    },
    {
      "epoch": 3.5448717948717947,
      "grad_norm": 0.271484375,
      "learning_rate": 4.732665776360068e-05,
      "loss": 0.0339,
      "step": 14378
    },
    {
      "epoch": 3.5453648915187377,
      "grad_norm": 0.302734375,
      "learning_rate": 4.729739889017339e-05,
      "loss": 0.0378,
      "step": 14380
    },
    {
      "epoch": 3.5458579881656807,
      "grad_norm": 0.2890625,
      "learning_rate": 4.726814626229856e-05,
      "loss": 0.0324,
      "step": 14382
    },
    {
      "epoch": 3.546351084812623,
      "grad_norm": 0.345703125,
      "learning_rate": 4.723889988344281e-05,
      "loss": 0.0371,
      "step": 14384
    },
    {
      "epoch": 3.546844181459566,
      "grad_norm": 0.29296875,
      "learning_rate": 4.720965975707204e-05,
      "loss": 0.0368,
      "step": 14386
    },
    {
      "epoch": 3.547337278106509,
      "grad_norm": 0.2890625,
      "learning_rate": 4.718042588665127e-05,
      "loss": 0.0403,
      "step": 14388
    },
    {
      "epoch": 3.5478303747534516,
      "grad_norm": 0.3359375,
      "learning_rate": 4.7151198275645004e-05,
      "loss": 0.0347,
      "step": 14390
    },
    {
      "epoch": 3.5483234714003946,
      "grad_norm": 0.359375,
      "learning_rate": 4.712197692751673e-05,
      "loss": 0.0395,
      "step": 14392
    },
    {
      "epoch": 3.548816568047337,
      "grad_norm": 0.3125,
      "learning_rate": 4.70927618457295e-05,
      "loss": 0.0359,
      "step": 14394
    },
    {
      "epoch": 3.54930966469428,
      "grad_norm": 0.271484375,
      "learning_rate": 4.7063553033745365e-05,
      "loss": 0.0349,
      "step": 14396
    },
    {
      "epoch": 3.5498027613412226,
      "grad_norm": 0.306640625,
      "learning_rate": 4.703435049502575e-05,
      "loss": 0.0358,
      "step": 14398
    },
    {
      "epoch": 3.5502958579881656,
      "grad_norm": 0.357421875,
      "learning_rate": 4.700515423303134e-05,
      "loss": 0.0393,
      "step": 14400
    },
    {
      "epoch": 3.5507889546351086,
      "grad_norm": 0.328125,
      "learning_rate": 4.697596425122199e-05,
      "loss": 0.0349,
      "step": 14402
    },
    {
      "epoch": 3.551282051282051,
      "grad_norm": 0.294921875,
      "learning_rate": 4.694678055305699e-05,
      "loss": 0.0344,
      "step": 14404
    },
    {
      "epoch": 3.551775147928994,
      "grad_norm": 0.3203125,
      "learning_rate": 4.6917603141994626e-05,
      "loss": 0.0358,
      "step": 14406
    },
    {
      "epoch": 3.552268244575937,
      "grad_norm": 0.408203125,
      "learning_rate": 4.688843202149275e-05,
      "loss": 0.0363,
      "step": 14408
    },
    {
      "epoch": 3.5527613412228796,
      "grad_norm": 0.2734375,
      "learning_rate": 4.685926719500817e-05,
      "loss": 0.0356,
      "step": 14410
    },
    {
      "epoch": 3.5532544378698225,
      "grad_norm": 0.2890625,
      "learning_rate": 4.683010866599713e-05,
      "loss": 0.0402,
      "step": 14412
    },
    {
      "epoch": 3.5537475345167655,
      "grad_norm": 0.3515625,
      "learning_rate": 4.6800956437915066e-05,
      "loss": 0.0337,
      "step": 14414
    },
    {
      "epoch": 3.554240631163708,
      "grad_norm": 0.26953125,
      "learning_rate": 4.6771810514216706e-05,
      "loss": 0.034,
      "step": 14416
    },
    {
      "epoch": 3.554733727810651,
      "grad_norm": 0.3125,
      "learning_rate": 4.674267089835601e-05,
      "loss": 0.0345,
      "step": 14418
    },
    {
      "epoch": 3.5552268244575935,
      "grad_norm": 0.353515625,
      "learning_rate": 4.67135375937861e-05,
      "loss": 0.0364,
      "step": 14420
    },
    {
      "epoch": 3.5557199211045365,
      "grad_norm": 0.28515625,
      "learning_rate": 4.6684410603959574e-05,
      "loss": 0.0362,
      "step": 14422
    },
    {
      "epoch": 3.556213017751479,
      "grad_norm": 0.279296875,
      "learning_rate": 4.665528993232803e-05,
      "loss": 0.0356,
      "step": 14424
    },
    {
      "epoch": 3.556706114398422,
      "grad_norm": 0.35546875,
      "learning_rate": 4.662617558234246e-05,
      "loss": 0.0365,
      "step": 14426
    },
    {
      "epoch": 3.557199211045365,
      "grad_norm": 0.29296875,
      "learning_rate": 4.659706755745309e-05,
      "loss": 0.0349,
      "step": 14428
    },
    {
      "epoch": 3.5576923076923075,
      "grad_norm": 0.306640625,
      "learning_rate": 4.656796586110938e-05,
      "loss": 0.0374,
      "step": 14430
    },
    {
      "epoch": 3.5581854043392505,
      "grad_norm": 0.294921875,
      "learning_rate": 4.653887049676007e-05,
      "loss": 0.0376,
      "step": 14432
    },
    {
      "epoch": 3.5586785009861934,
      "grad_norm": 0.296875,
      "learning_rate": 4.650978146785302e-05,
      "loss": 0.0346,
      "step": 14434
    },
    {
      "epoch": 3.559171597633136,
      "grad_norm": 0.3359375,
      "learning_rate": 4.6480698777835605e-05,
      "loss": 0.0379,
      "step": 14436
    },
    {
      "epoch": 3.559664694280079,
      "grad_norm": 0.361328125,
      "learning_rate": 4.645162243015415e-05,
      "loss": 0.0335,
      "step": 14438
    },
    {
      "epoch": 3.560157790927022,
      "grad_norm": 0.35546875,
      "learning_rate": 4.642255242825442e-05,
      "loss": 0.0347,
      "step": 14440
    },
    {
      "epoch": 3.5606508875739644,
      "grad_norm": 0.326171875,
      "learning_rate": 4.6393488775581364e-05,
      "loss": 0.0375,
      "step": 14442
    },
    {
      "epoch": 3.5611439842209074,
      "grad_norm": 0.328125,
      "learning_rate": 4.636443147557918e-05,
      "loss": 0.0369,
      "step": 14444
    },
    {
      "epoch": 3.5616370808678504,
      "grad_norm": 0.302734375,
      "learning_rate": 4.633538053169134e-05,
      "loss": 0.0339,
      "step": 14446
    },
    {
      "epoch": 3.562130177514793,
      "grad_norm": 0.30078125,
      "learning_rate": 4.630633594736053e-05,
      "loss": 0.0338,
      "step": 14448
    },
    {
      "epoch": 3.562623274161736,
      "grad_norm": 0.33203125,
      "learning_rate": 4.627729772602874e-05,
      "loss": 0.037,
      "step": 14450
    },
    {
      "epoch": 3.5631163708086784,
      "grad_norm": 0.361328125,
      "learning_rate": 4.6248265871137065e-05,
      "loss": 0.0374,
      "step": 14452
    },
    {
      "epoch": 3.5636094674556213,
      "grad_norm": 0.32421875,
      "learning_rate": 4.621924038612601e-05,
      "loss": 0.0305,
      "step": 14454
    },
    {
      "epoch": 3.564102564102564,
      "grad_norm": 0.330078125,
      "learning_rate": 4.619022127443523e-05,
      "loss": 0.0351,
      "step": 14456
    },
    {
      "epoch": 3.564595660749507,
      "grad_norm": 0.333984375,
      "learning_rate": 4.616120853950367e-05,
      "loss": 0.0356,
      "step": 14458
    },
    {
      "epoch": 3.56508875739645,
      "grad_norm": 0.314453125,
      "learning_rate": 4.613220218476949e-05,
      "loss": 0.0386,
      "step": 14460
    },
    {
      "epoch": 3.5655818540433923,
      "grad_norm": 0.369140625,
      "learning_rate": 4.610320221367013e-05,
      "loss": 0.0348,
      "step": 14462
    },
    {
      "epoch": 3.5660749506903353,
      "grad_norm": 0.33203125,
      "learning_rate": 4.607420862964225e-05,
      "loss": 0.0355,
      "step": 14464
    },
    {
      "epoch": 3.5665680473372783,
      "grad_norm": 0.27734375,
      "learning_rate": 4.604522143612169e-05,
      "loss": 0.0358,
      "step": 14466
    },
    {
      "epoch": 3.567061143984221,
      "grad_norm": 0.3046875,
      "learning_rate": 4.601624063654364e-05,
      "loss": 0.0357,
      "step": 14468
    },
    {
      "epoch": 3.5675542406311638,
      "grad_norm": 0.3359375,
      "learning_rate": 4.5987266234342486e-05,
      "loss": 0.0355,
      "step": 14470
    },
    {
      "epoch": 3.5680473372781067,
      "grad_norm": 0.3125,
      "learning_rate": 4.5958298232951846e-05,
      "loss": 0.0367,
      "step": 14472
    },
    {
      "epoch": 3.5685404339250493,
      "grad_norm": 0.28125,
      "learning_rate": 4.59293366358046e-05,
      "loss": 0.0369,
      "step": 14474
    },
    {
      "epoch": 3.5690335305719922,
      "grad_norm": 0.35546875,
      "learning_rate": 4.590038144633285e-05,
      "loss": 0.034,
      "step": 14476
    },
    {
      "epoch": 3.5695266272189348,
      "grad_norm": 0.32421875,
      "learning_rate": 4.587143266796795e-05,
      "loss": 0.0337,
      "step": 14478
    },
    {
      "epoch": 3.5700197238658777,
      "grad_norm": 0.359375,
      "learning_rate": 4.584249030414054e-05,
      "loss": 0.0377,
      "step": 14480
    },
    {
      "epoch": 3.5705128205128203,
      "grad_norm": 0.28515625,
      "learning_rate": 4.5813554358280366e-05,
      "loss": 0.0368,
      "step": 14482
    },
    {
      "epoch": 3.5710059171597632,
      "grad_norm": 0.42578125,
      "learning_rate": 4.578462483381654e-05,
      "loss": 0.0419,
      "step": 14484
    },
    {
      "epoch": 3.571499013806706,
      "grad_norm": 0.318359375,
      "learning_rate": 4.5755701734177356e-05,
      "loss": 0.0407,
      "step": 14486
    },
    {
      "epoch": 3.5719921104536487,
      "grad_norm": 0.314453125,
      "learning_rate": 4.57267850627904e-05,
      "loss": 0.0338,
      "step": 14488
    },
    {
      "epoch": 3.5724852071005917,
      "grad_norm": 0.3984375,
      "learning_rate": 4.5697874823082424e-05,
      "loss": 0.0357,
      "step": 14490
    },
    {
      "epoch": 3.5729783037475347,
      "grad_norm": 0.33984375,
      "learning_rate": 4.566897101847947e-05,
      "loss": 0.0368,
      "step": 14492
    },
    {
      "epoch": 3.573471400394477,
      "grad_norm": 0.349609375,
      "learning_rate": 4.5640073652406836e-05,
      "loss": 0.0378,
      "step": 14494
    },
    {
      "epoch": 3.57396449704142,
      "grad_norm": 0.330078125,
      "learning_rate": 4.5611182728288895e-05,
      "loss": 0.0341,
      "step": 14496
    },
    {
      "epoch": 3.574457593688363,
      "grad_norm": 0.28125,
      "learning_rate": 4.558229824954955e-05,
      "loss": 0.0372,
      "step": 14498
    },
    {
      "epoch": 3.5749506903353057,
      "grad_norm": 0.33203125,
      "learning_rate": 4.555342021961165e-05,
      "loss": 0.0371,
      "step": 14500
    },
    {
      "epoch": 3.5754437869822486,
      "grad_norm": 0.310546875,
      "learning_rate": 4.552454864189746e-05,
      "loss": 0.0344,
      "step": 14502
    },
    {
      "epoch": 3.5759368836291916,
      "grad_norm": 0.2890625,
      "learning_rate": 4.5495683519828395e-05,
      "loss": 0.0362,
      "step": 14504
    },
    {
      "epoch": 3.576429980276134,
      "grad_norm": 0.328125,
      "learning_rate": 4.546682485682515e-05,
      "loss": 0.0367,
      "step": 14506
    },
    {
      "epoch": 3.5769230769230766,
      "grad_norm": 0.291015625,
      "learning_rate": 4.543797265630767e-05,
      "loss": 0.0393,
      "step": 14508
    },
    {
      "epoch": 3.5774161735700196,
      "grad_norm": 0.330078125,
      "learning_rate": 4.540912692169499e-05,
      "loss": 0.0336,
      "step": 14510
    },
    {
      "epoch": 3.5779092702169626,
      "grad_norm": 0.369140625,
      "learning_rate": 4.538028765640564e-05,
      "loss": 0.0389,
      "step": 14512
    },
    {
      "epoch": 3.578402366863905,
      "grad_norm": 0.3984375,
      "learning_rate": 4.5351454863857124e-05,
      "loss": 0.0397,
      "step": 14514
    },
    {
      "epoch": 3.578895463510848,
      "grad_norm": 0.32421875,
      "learning_rate": 4.532262854746633e-05,
      "loss": 0.0385,
      "step": 14516
    },
    {
      "epoch": 3.579388560157791,
      "grad_norm": 0.361328125,
      "learning_rate": 4.529380871064933e-05,
      "loss": 0.0403,
      "step": 14518
    },
    {
      "epoch": 3.5798816568047336,
      "grad_norm": 0.326171875,
      "learning_rate": 4.526499535682144e-05,
      "loss": 0.0354,
      "step": 14520
    },
    {
      "epoch": 3.5803747534516766,
      "grad_norm": 0.310546875,
      "learning_rate": 4.5236188489397225e-05,
      "loss": 0.0391,
      "step": 14522
    },
    {
      "epoch": 3.5808678500986195,
      "grad_norm": 0.287109375,
      "learning_rate": 4.520738811179037e-05,
      "loss": 0.0361,
      "step": 14524
    },
    {
      "epoch": 3.581360946745562,
      "grad_norm": 0.3125,
      "learning_rate": 4.5178594227414026e-05,
      "loss": 0.0389,
      "step": 14526
    },
    {
      "epoch": 3.581854043392505,
      "grad_norm": 0.341796875,
      "learning_rate": 4.514980683968028e-05,
      "loss": 0.0348,
      "step": 14528
    },
    {
      "epoch": 3.582347140039448,
      "grad_norm": 0.287109375,
      "learning_rate": 4.512102595200073e-05,
      "loss": 0.0346,
      "step": 14530
    },
    {
      "epoch": 3.5828402366863905,
      "grad_norm": 0.27734375,
      "learning_rate": 4.509225156778597e-05,
      "loss": 0.0357,
      "step": 14532
    },
    {
      "epoch": 3.5833333333333335,
      "grad_norm": 0.302734375,
      "learning_rate": 4.506348369044597e-05,
      "loss": 0.0329,
      "step": 14534
    },
    {
      "epoch": 3.583826429980276,
      "grad_norm": 0.380859375,
      "learning_rate": 4.5034722323389886e-05,
      "loss": 0.0348,
      "step": 14536
    },
    {
      "epoch": 3.584319526627219,
      "grad_norm": 0.396484375,
      "learning_rate": 4.500596747002609e-05,
      "loss": 0.0379,
      "step": 14538
    },
    {
      "epoch": 3.5848126232741615,
      "grad_norm": 0.3203125,
      "learning_rate": 4.497721913376223e-05,
      "loss": 0.0401,
      "step": 14540
    },
    {
      "epoch": 3.5853057199211045,
      "grad_norm": 0.275390625,
      "learning_rate": 4.494847731800503e-05,
      "loss": 0.0332,
      "step": 14542
    },
    {
      "epoch": 3.5857988165680474,
      "grad_norm": 0.37890625,
      "learning_rate": 4.491974202616072e-05,
      "loss": 0.0378,
      "step": 14544
    },
    {
      "epoch": 3.58629191321499,
      "grad_norm": 0.33203125,
      "learning_rate": 4.4891013261634474e-05,
      "loss": 0.0403,
      "step": 14546
    },
    {
      "epoch": 3.586785009861933,
      "grad_norm": 0.326171875,
      "learning_rate": 4.486229102783084e-05,
      "loss": 0.04,
      "step": 14548
    },
    {
      "epoch": 3.587278106508876,
      "grad_norm": 0.30859375,
      "learning_rate": 4.483357532815356e-05,
      "loss": 0.0327,
      "step": 14550
    },
    {
      "epoch": 3.5877712031558184,
      "grad_norm": 0.294921875,
      "learning_rate": 4.480486616600563e-05,
      "loss": 0.0357,
      "step": 14552
    },
    {
      "epoch": 3.5882642998027614,
      "grad_norm": 0.291015625,
      "learning_rate": 4.477616354478926e-05,
      "loss": 0.0361,
      "step": 14554
    },
    {
      "epoch": 3.5887573964497044,
      "grad_norm": 0.341796875,
      "learning_rate": 4.4747467467905765e-05,
      "loss": 0.033,
      "step": 14556
    },
    {
      "epoch": 3.589250493096647,
      "grad_norm": 0.30859375,
      "learning_rate": 4.471877793875594e-05,
      "loss": 0.0392,
      "step": 14558
    },
    {
      "epoch": 3.58974358974359,
      "grad_norm": 0.28125,
      "learning_rate": 4.469009496073955e-05,
      "loss": 0.0377,
      "step": 14560
    },
    {
      "epoch": 3.5902366863905324,
      "grad_norm": 0.2890625,
      "learning_rate": 4.466141853725571e-05,
      "loss": 0.038,
      "step": 14562
    },
    {
      "epoch": 3.5907297830374754,
      "grad_norm": 0.328125,
      "learning_rate": 4.463274867170274e-05,
      "loss": 0.0344,
      "step": 14564
    },
    {
      "epoch": 3.591222879684418,
      "grad_norm": 0.251953125,
      "learning_rate": 4.460408536747819e-05,
      "loss": 0.0313,
      "step": 14566
    },
    {
      "epoch": 3.591715976331361,
      "grad_norm": 0.3046875,
      "learning_rate": 4.457542862797885e-05,
      "loss": 0.0348,
      "step": 14568
    },
    {
      "epoch": 3.592209072978304,
      "grad_norm": 0.3359375,
      "learning_rate": 4.45467784566006e-05,
      "loss": 0.0378,
      "step": 14570
    },
    {
      "epoch": 3.5927021696252464,
      "grad_norm": 0.29296875,
      "learning_rate": 4.451813485673878e-05,
      "loss": 0.0352,
      "step": 14572
    },
    {
      "epoch": 3.5931952662721893,
      "grad_norm": 0.30859375,
      "learning_rate": 4.4489497831787716e-05,
      "loss": 0.0362,
      "step": 14574
    },
    {
      "epoch": 3.5936883629191323,
      "grad_norm": 0.390625,
      "learning_rate": 4.44608673851411e-05,
      "loss": 0.0387,
      "step": 14576
    },
    {
      "epoch": 3.594181459566075,
      "grad_norm": 0.3125,
      "learning_rate": 4.443224352019179e-05,
      "loss": 0.0362,
      "step": 14578
    },
    {
      "epoch": 3.594674556213018,
      "grad_norm": 0.3359375,
      "learning_rate": 4.440362624033185e-05,
      "loss": 0.0344,
      "step": 14580
    },
    {
      "epoch": 3.5951676528599608,
      "grad_norm": 0.32421875,
      "learning_rate": 4.437501554895267e-05,
      "loss": 0.0383,
      "step": 14582
    },
    {
      "epoch": 3.5956607495069033,
      "grad_norm": 0.365234375,
      "learning_rate": 4.4346411449444636e-05,
      "loss": 0.0382,
      "step": 14584
    },
    {
      "epoch": 3.5961538461538463,
      "grad_norm": 0.265625,
      "learning_rate": 4.431781394519765e-05,
      "loss": 0.0355,
      "step": 14586
    },
    {
      "epoch": 3.5966469428007892,
      "grad_norm": 0.328125,
      "learning_rate": 4.428922303960056e-05,
      "loss": 0.038,
      "step": 14588
    },
    {
      "epoch": 3.5971400394477318,
      "grad_norm": 0.287109375,
      "learning_rate": 4.426063873604159e-05,
      "loss": 0.0363,
      "step": 14590
    },
    {
      "epoch": 3.5976331360946747,
      "grad_norm": 0.345703125,
      "learning_rate": 4.4232061037908146e-05,
      "loss": 0.0363,
      "step": 14592
    },
    {
      "epoch": 3.5981262327416172,
      "grad_norm": 0.306640625,
      "learning_rate": 4.420348994858683e-05,
      "loss": 0.0327,
      "step": 14594
    },
    {
      "epoch": 3.59861932938856,
      "grad_norm": 0.326171875,
      "learning_rate": 4.4174925471463524e-05,
      "loss": 0.0343,
      "step": 14596
    },
    {
      "epoch": 3.5991124260355027,
      "grad_norm": 0.3203125,
      "learning_rate": 4.414636760992317e-05,
      "loss": 0.0346,
      "step": 14598
    },
    {
      "epoch": 3.5996055226824457,
      "grad_norm": 0.37109375,
      "learning_rate": 4.411781636735016e-05,
      "loss": 0.032,
      "step": 14600
    },
    {
      "epoch": 3.6000986193293887,
      "grad_norm": 0.330078125,
      "learning_rate": 4.408927174712789e-05,
      "loss": 0.0361,
      "step": 14602
    },
    {
      "epoch": 3.600591715976331,
      "grad_norm": 0.32421875,
      "learning_rate": 4.4060733752639074e-05,
      "loss": 0.036,
      "step": 14604
    },
    {
      "epoch": 3.601084812623274,
      "grad_norm": 0.28515625,
      "learning_rate": 4.403220238726564e-05,
      "loss": 0.0358,
      "step": 14606
    },
    {
      "epoch": 3.601577909270217,
      "grad_norm": 0.384765625,
      "learning_rate": 4.400367765438871e-05,
      "loss": 0.0349,
      "step": 14608
    },
    {
      "epoch": 3.6020710059171597,
      "grad_norm": 0.314453125,
      "learning_rate": 4.3975159557388615e-05,
      "loss": 0.0377,
      "step": 14610
    },
    {
      "epoch": 3.6025641025641026,
      "grad_norm": 0.302734375,
      "learning_rate": 4.394664809964492e-05,
      "loss": 0.0365,
      "step": 14612
    },
    {
      "epoch": 3.6030571992110456,
      "grad_norm": 0.314453125,
      "learning_rate": 4.391814328453642e-05,
      "loss": 0.0396,
      "step": 14614
    },
    {
      "epoch": 3.603550295857988,
      "grad_norm": 0.294921875,
      "learning_rate": 4.388964511544099e-05,
      "loss": 0.0361,
      "step": 14616
    },
    {
      "epoch": 3.604043392504931,
      "grad_norm": 0.302734375,
      "learning_rate": 4.386115359573596e-05,
      "loss": 0.0352,
      "step": 14618
    },
    {
      "epoch": 3.6045364891518736,
      "grad_norm": 0.30078125,
      "learning_rate": 4.383266872879764e-05,
      "loss": 0.035,
      "step": 14620
    },
    {
      "epoch": 3.6050295857988166,
      "grad_norm": 0.291015625,
      "learning_rate": 4.380419051800167e-05,
      "loss": 0.037,
      "step": 14622
    },
    {
      "epoch": 3.605522682445759,
      "grad_norm": 0.373046875,
      "learning_rate": 4.377571896672289e-05,
      "loss": 0.0363,
      "step": 14624
    },
    {
      "epoch": 3.606015779092702,
      "grad_norm": 0.29296875,
      "learning_rate": 4.374725407833532e-05,
      "loss": 0.0383,
      "step": 14626
    },
    {
      "epoch": 3.606508875739645,
      "grad_norm": 0.3203125,
      "learning_rate": 4.371879585621222e-05,
      "loss": 0.0348,
      "step": 14628
    },
    {
      "epoch": 3.6070019723865876,
      "grad_norm": 0.3359375,
      "learning_rate": 4.3690344303726036e-05,
      "loss": 0.0341,
      "step": 14630
    },
    {
      "epoch": 3.6074950690335306,
      "grad_norm": 0.359375,
      "learning_rate": 4.36618994242485e-05,
      "loss": 0.0354,
      "step": 14632
    },
    {
      "epoch": 3.6079881656804735,
      "grad_norm": 0.30078125,
      "learning_rate": 4.3633461221150384e-05,
      "loss": 0.0357,
      "step": 14634
    },
    {
      "epoch": 3.608481262327416,
      "grad_norm": 0.33203125,
      "learning_rate": 4.360502969780182e-05,
      "loss": 0.0369,
      "step": 14636
    },
    {
      "epoch": 3.608974358974359,
      "grad_norm": 0.310546875,
      "learning_rate": 4.3576604857572114e-05,
      "loss": 0.0339,
      "step": 14638
    },
    {
      "epoch": 3.609467455621302,
      "grad_norm": 0.306640625,
      "learning_rate": 4.354818670382976e-05,
      "loss": 0.0368,
      "step": 14640
    },
    {
      "epoch": 3.6099605522682445,
      "grad_norm": 0.32421875,
      "learning_rate": 4.351977523994247e-05,
      "loss": 0.0373,
      "step": 14642
    },
    {
      "epoch": 3.6104536489151875,
      "grad_norm": 0.3203125,
      "learning_rate": 4.349137046927715e-05,
      "loss": 0.0386,
      "step": 14644
    },
    {
      "epoch": 3.61094674556213,
      "grad_norm": 0.2890625,
      "learning_rate": 4.3462972395199976e-05,
      "loss": 0.0316,
      "step": 14646
    },
    {
      "epoch": 3.611439842209073,
      "grad_norm": 0.330078125,
      "learning_rate": 4.3434581021076206e-05,
      "loss": 0.0343,
      "step": 14648
    },
    {
      "epoch": 3.6119329388560155,
      "grad_norm": 0.2734375,
      "learning_rate": 4.340619635027039e-05,
      "loss": 0.0359,
      "step": 14650
    },
    {
      "epoch": 3.6124260355029585,
      "grad_norm": 0.314453125,
      "learning_rate": 4.3377818386146294e-05,
      "loss": 0.0377,
      "step": 14652
    },
    {
      "epoch": 3.6129191321499015,
      "grad_norm": 0.373046875,
      "learning_rate": 4.334944713206687e-05,
      "loss": 0.0357,
      "step": 14654
    },
    {
      "epoch": 3.613412228796844,
      "grad_norm": 0.318359375,
      "learning_rate": 4.3321082591394235e-05,
      "loss": 0.0358,
      "step": 14656
    },
    {
      "epoch": 3.613905325443787,
      "grad_norm": 0.294921875,
      "learning_rate": 4.3292724767489776e-05,
      "loss": 0.034,
      "step": 14658
    },
    {
      "epoch": 3.61439842209073,
      "grad_norm": 0.416015625,
      "learning_rate": 4.326437366371405e-05,
      "loss": 0.0377,
      "step": 14660
    },
    {
      "epoch": 3.6148915187376724,
      "grad_norm": 0.2578125,
      "learning_rate": 4.323602928342685e-05,
      "loss": 0.0349,
      "step": 14662
    },
    {
      "epoch": 3.6153846153846154,
      "grad_norm": 0.318359375,
      "learning_rate": 4.320769162998707e-05,
      "loss": 0.037,
      "step": 14664
    },
    {
      "epoch": 3.6158777120315584,
      "grad_norm": 0.2890625,
      "learning_rate": 4.317936070675291e-05,
      "loss": 0.0337,
      "step": 14666
    },
    {
      "epoch": 3.616370808678501,
      "grad_norm": 0.345703125,
      "learning_rate": 4.3151036517081765e-05,
      "loss": 0.0371,
      "step": 14668
    },
    {
      "epoch": 3.616863905325444,
      "grad_norm": 0.30859375,
      "learning_rate": 4.31227190643302e-05,
      "loss": 0.0338,
      "step": 14670
    },
    {
      "epoch": 3.617357001972387,
      "grad_norm": 0.330078125,
      "learning_rate": 4.309440835185399e-05,
      "loss": 0.0339,
      "step": 14672
    },
    {
      "epoch": 3.6178500986193294,
      "grad_norm": 0.29296875,
      "learning_rate": 4.30661043830081e-05,
      "loss": 0.034,
      "step": 14674
    },
    {
      "epoch": 3.6183431952662723,
      "grad_norm": 0.33203125,
      "learning_rate": 4.303780716114675e-05,
      "loss": 0.0381,
      "step": 14676
    },
    {
      "epoch": 3.618836291913215,
      "grad_norm": 0.30078125,
      "learning_rate": 4.300951668962323e-05,
      "loss": 0.0353,
      "step": 14678
    },
    {
      "epoch": 3.619329388560158,
      "grad_norm": 0.3046875,
      "learning_rate": 4.298123297179024e-05,
      "loss": 0.0313,
      "step": 14680
    },
    {
      "epoch": 3.6198224852071004,
      "grad_norm": 0.36328125,
      "learning_rate": 4.2952956010999455e-05,
      "loss": 0.0334,
      "step": 14682
    },
    {
      "epoch": 3.6203155818540433,
      "grad_norm": 0.365234375,
      "learning_rate": 4.29246858106019e-05,
      "loss": 0.0344,
      "step": 14684
    },
    {
      "epoch": 3.6208086785009863,
      "grad_norm": 0.333984375,
      "learning_rate": 4.2896422373947745e-05,
      "loss": 0.0406,
      "step": 14686
    },
    {
      "epoch": 3.621301775147929,
      "grad_norm": 0.373046875,
      "learning_rate": 4.2868165704386366e-05,
      "loss": 0.0338,
      "step": 14688
    },
    {
      "epoch": 3.621794871794872,
      "grad_norm": 0.333984375,
      "learning_rate": 4.283991580526636e-05,
      "loss": 0.0405,
      "step": 14690
    },
    {
      "epoch": 3.6222879684418148,
      "grad_norm": 0.3046875,
      "learning_rate": 4.281167267993541e-05,
      "loss": 0.0358,
      "step": 14692
    },
    {
      "epoch": 3.6227810650887573,
      "grad_norm": 0.33203125,
      "learning_rate": 4.278343633174061e-05,
      "loss": 0.0367,
      "step": 14694
    },
    {
      "epoch": 3.6232741617357003,
      "grad_norm": 0.361328125,
      "learning_rate": 4.275520676402803e-05,
      "loss": 0.0305,
      "step": 14696
    },
    {
      "epoch": 3.6237672583826432,
      "grad_norm": 0.337890625,
      "learning_rate": 4.272698398014306e-05,
      "loss": 0.0346,
      "step": 14698
    },
    {
      "epoch": 3.6242603550295858,
      "grad_norm": 0.322265625,
      "learning_rate": 4.2698767983430265e-05,
      "loss": 0.0361,
      "step": 14700
    },
    {
      "epoch": 3.6247534516765287,
      "grad_norm": 0.328125,
      "learning_rate": 4.267055877723338e-05,
      "loss": 0.034,
      "step": 14702
    },
    {
      "epoch": 3.6252465483234713,
      "grad_norm": 0.298828125,
      "learning_rate": 4.264235636489542e-05,
      "loss": 0.031,
      "step": 14704
    },
    {
      "epoch": 3.6257396449704142,
      "grad_norm": 0.296875,
      "learning_rate": 4.261416074975839e-05,
      "loss": 0.0364,
      "step": 14706
    },
    {
      "epoch": 3.6262327416173568,
      "grad_norm": 0.353515625,
      "learning_rate": 4.2585971935163794e-05,
      "loss": 0.034,
      "step": 14708
    },
    {
      "epoch": 3.6267258382642997,
      "grad_norm": 0.275390625,
      "learning_rate": 4.255778992445201e-05,
      "loss": 0.0367,
      "step": 14710
    },
    {
      "epoch": 3.6272189349112427,
      "grad_norm": 0.337890625,
      "learning_rate": 4.252961472096289e-05,
      "loss": 0.0355,
      "step": 14712
    },
    {
      "epoch": 3.6277120315581852,
      "grad_norm": 0.2890625,
      "learning_rate": 4.2501446328035286e-05,
      "loss": 0.035,
      "step": 14714
    },
    {
      "epoch": 3.628205128205128,
      "grad_norm": 0.318359375,
      "learning_rate": 4.247328474900732e-05,
      "loss": 0.0361,
      "step": 14716
    },
    {
      "epoch": 3.628698224852071,
      "grad_norm": 0.3125,
      "learning_rate": 4.244512998721633e-05,
      "loss": 0.0396,
      "step": 14718
    },
    {
      "epoch": 3.6291913214990137,
      "grad_norm": 0.30859375,
      "learning_rate": 4.241698204599871e-05,
      "loss": 0.0387,
      "step": 14720
    },
    {
      "epoch": 3.6296844181459567,
      "grad_norm": 0.306640625,
      "learning_rate": 4.238884092869029e-05,
      "loss": 0.0364,
      "step": 14722
    },
    {
      "epoch": 3.6301775147928996,
      "grad_norm": 0.291015625,
      "learning_rate": 4.23607066386258e-05,
      "loss": 0.0347,
      "step": 14724
    },
    {
      "epoch": 3.630670611439842,
      "grad_norm": 0.283203125,
      "learning_rate": 4.233257917913948e-05,
      "loss": 0.0332,
      "step": 14726
    },
    {
      "epoch": 3.631163708086785,
      "grad_norm": 0.328125,
      "learning_rate": 4.230445855356445e-05,
      "loss": 0.0348,
      "step": 14728
    },
    {
      "epoch": 3.6316568047337277,
      "grad_norm": 0.3046875,
      "learning_rate": 4.227634476523322e-05,
      "loss": 0.0366,
      "step": 14730
    },
    {
      "epoch": 3.6321499013806706,
      "grad_norm": 0.279296875,
      "learning_rate": 4.2248237817477455e-05,
      "loss": 0.0357,
      "step": 14732
    },
    {
      "epoch": 3.632642998027613,
      "grad_norm": 0.337890625,
      "learning_rate": 4.222013771362788e-05,
      "loss": 0.0362,
      "step": 14734
    },
    {
      "epoch": 3.633136094674556,
      "grad_norm": 0.291015625,
      "learning_rate": 4.219204445701467e-05,
      "loss": 0.0368,
      "step": 14736
    },
    {
      "epoch": 3.633629191321499,
      "grad_norm": 0.29296875,
      "learning_rate": 4.216395805096687e-05,
      "loss": 0.0316,
      "step": 14738
    },
    {
      "epoch": 3.6341222879684416,
      "grad_norm": 0.310546875,
      "learning_rate": 4.213587849881303e-05,
      "loss": 0.0355,
      "step": 14740
    },
    {
      "epoch": 3.6346153846153846,
      "grad_norm": 0.314453125,
      "learning_rate": 4.210780580388062e-05,
      "loss": 0.0372,
      "step": 14742
    },
    {
      "epoch": 3.6351084812623276,
      "grad_norm": 0.345703125,
      "learning_rate": 4.207973996949646e-05,
      "loss": 0.0335,
      "step": 14744
    },
    {
      "epoch": 3.63560157790927,
      "grad_norm": 0.306640625,
      "learning_rate": 4.205168099898652e-05,
      "loss": 0.0362,
      "step": 14746
    },
    {
      "epoch": 3.636094674556213,
      "grad_norm": 0.31640625,
      "learning_rate": 4.202362889567586e-05,
      "loss": 0.0374,
      "step": 14748
    },
    {
      "epoch": 3.636587771203156,
      "grad_norm": 0.2734375,
      "learning_rate": 4.199558366288895e-05,
      "loss": 0.0322,
      "step": 14750
    },
    {
      "epoch": 3.6370808678500985,
      "grad_norm": 0.279296875,
      "learning_rate": 4.196754530394915e-05,
      "loss": 0.0373,
      "step": 14752
    },
    {
      "epoch": 3.6375739644970415,
      "grad_norm": 0.302734375,
      "learning_rate": 4.193951382217932e-05,
      "loss": 0.0383,
      "step": 14754
    },
    {
      "epoch": 3.6380670611439845,
      "grad_norm": 0.40234375,
      "learning_rate": 4.1911489220901236e-05,
      "loss": 0.0369,
      "step": 14756
    },
    {
      "epoch": 3.638560157790927,
      "grad_norm": 0.30078125,
      "learning_rate": 4.1883471503436e-05,
      "loss": 0.0355,
      "step": 14758
    },
    {
      "epoch": 3.63905325443787,
      "grad_norm": 0.46484375,
      "learning_rate": 4.185546067310388e-05,
      "loss": 0.0383,
      "step": 14760
    },
    {
      "epoch": 3.6395463510848125,
      "grad_norm": 0.384765625,
      "learning_rate": 4.1827456733224305e-05,
      "loss": 0.0359,
      "step": 14762
    },
    {
      "epoch": 3.6400394477317555,
      "grad_norm": 0.296875,
      "learning_rate": 4.179945968711594e-05,
      "loss": 0.0325,
      "step": 14764
    },
    {
      "epoch": 3.640532544378698,
      "grad_norm": 0.296875,
      "learning_rate": 4.1771469538096464e-05,
      "loss": 0.0362,
      "step": 14766
    },
    {
      "epoch": 3.641025641025641,
      "grad_norm": 0.30078125,
      "learning_rate": 4.1743486289483037e-05,
      "loss": 0.0322,
      "step": 14768
    },
    {
      "epoch": 3.641518737672584,
      "grad_norm": 0.298828125,
      "learning_rate": 4.171550994459171e-05,
      "loss": 0.0374,
      "step": 14770
    },
    {
      "epoch": 3.6420118343195265,
      "grad_norm": 0.2890625,
      "learning_rate": 4.1687540506737874e-05,
      "loss": 0.0374,
      "step": 14772
    },
    {
      "epoch": 3.6425049309664694,
      "grad_norm": 0.322265625,
      "learning_rate": 4.1659577979236075e-05,
      "loss": 0.0349,
      "step": 14774
    },
    {
      "epoch": 3.6429980276134124,
      "grad_norm": 0.279296875,
      "learning_rate": 4.16316223654e-05,
      "loss": 0.0365,
      "step": 14776
    },
    {
      "epoch": 3.643491124260355,
      "grad_norm": 0.302734375,
      "learning_rate": 4.160367366854261e-05,
      "loss": 0.0334,
      "step": 14778
    },
    {
      "epoch": 3.643984220907298,
      "grad_norm": 0.28125,
      "learning_rate": 4.157573189197586e-05,
      "loss": 0.0351,
      "step": 14780
    },
    {
      "epoch": 3.644477317554241,
      "grad_norm": 0.31640625,
      "learning_rate": 4.154779703901114e-05,
      "loss": 0.0345,
      "step": 14782
    },
    {
      "epoch": 3.6449704142011834,
      "grad_norm": 0.27734375,
      "learning_rate": 4.15198691129588e-05,
      "loss": 0.0349,
      "step": 14784
    },
    {
      "epoch": 3.6454635108481264,
      "grad_norm": 0.29296875,
      "learning_rate": 4.1491948117128474e-05,
      "loss": 0.0315,
      "step": 14786
    },
    {
      "epoch": 3.645956607495069,
      "grad_norm": 0.3515625,
      "learning_rate": 4.146403405482897e-05,
      "loss": 0.0328,
      "step": 14788
    },
    {
      "epoch": 3.646449704142012,
      "grad_norm": 0.271484375,
      "learning_rate": 4.1436126929368255e-05,
      "loss": 0.0337,
      "step": 14790
    },
    {
      "epoch": 3.6469428007889544,
      "grad_norm": 0.359375,
      "learning_rate": 4.140822674405347e-05,
      "loss": 0.0384,
      "step": 14792
    },
    {
      "epoch": 3.6474358974358974,
      "grad_norm": 0.32421875,
      "learning_rate": 4.138033350219096e-05,
      "loss": 0.0338,
      "step": 14794
    },
    {
      "epoch": 3.6479289940828403,
      "grad_norm": 0.296875,
      "learning_rate": 4.1352447207086256e-05,
      "loss": 0.0353,
      "step": 14796
    },
    {
      "epoch": 3.648422090729783,
      "grad_norm": 0.376953125,
      "learning_rate": 4.132456786204395e-05,
      "loss": 0.0352,
      "step": 14798
    },
    {
      "epoch": 3.648915187376726,
      "grad_norm": 0.330078125,
      "learning_rate": 4.1296695470367974e-05,
      "loss": 0.0308,
      "step": 14800
    },
    {
      "epoch": 3.649408284023669,
      "grad_norm": 0.3046875,
      "learning_rate": 4.126883003536134e-05,
      "loss": 0.0369,
      "step": 14802
    },
    {
      "epoch": 3.6499013806706113,
      "grad_norm": 0.34765625,
      "learning_rate": 4.124097156032625e-05,
      "loss": 0.038,
      "step": 14804
    },
    {
      "epoch": 3.6503944773175543,
      "grad_norm": 0.27734375,
      "learning_rate": 4.12131200485641e-05,
      "loss": 0.0336,
      "step": 14806
    },
    {
      "epoch": 3.6508875739644973,
      "grad_norm": 0.298828125,
      "learning_rate": 4.118527550337546e-05,
      "loss": 0.0338,
      "step": 14808
    },
    {
      "epoch": 3.65138067061144,
      "grad_norm": 0.279296875,
      "learning_rate": 4.115743792806004e-05,
      "loss": 0.0377,
      "step": 14810
    },
    {
      "epoch": 3.6518737672583828,
      "grad_norm": 0.29296875,
      "learning_rate": 4.1129607325916797e-05,
      "loss": 0.035,
      "step": 14812
    },
    {
      "epoch": 3.6523668639053253,
      "grad_norm": 0.3125,
      "learning_rate": 4.1101783700243744e-05,
      "loss": 0.0372,
      "step": 14814
    },
    {
      "epoch": 3.6528599605522682,
      "grad_norm": 0.306640625,
      "learning_rate": 4.107396705433817e-05,
      "loss": 0.0408,
      "step": 14816
    },
    {
      "epoch": 3.6533530571992108,
      "grad_norm": 0.3125,
      "learning_rate": 4.1046157391496496e-05,
      "loss": 0.0358,
      "step": 14818
    },
    {
      "epoch": 3.6538461538461537,
      "grad_norm": 0.359375,
      "learning_rate": 4.101835471501433e-05,
      "loss": 0.037,
      "step": 14820
    },
    {
      "epoch": 3.6543392504930967,
      "grad_norm": 0.3046875,
      "learning_rate": 4.099055902818644e-05,
      "loss": 0.0345,
      "step": 14822
    },
    {
      "epoch": 3.6548323471400392,
      "grad_norm": 0.294921875,
      "learning_rate": 4.096277033430678e-05,
      "loss": 0.0381,
      "step": 14824
    },
    {
      "epoch": 3.655325443786982,
      "grad_norm": 0.3046875,
      "learning_rate": 4.093498863666848e-05,
      "loss": 0.036,
      "step": 14826
    },
    {
      "epoch": 3.655818540433925,
      "grad_norm": 0.357421875,
      "learning_rate": 4.090721393856377e-05,
      "loss": 0.0369,
      "step": 14828
    },
    {
      "epoch": 3.6563116370808677,
      "grad_norm": 0.30859375,
      "learning_rate": 4.0879446243284145e-05,
      "loss": 0.0349,
      "step": 14830
    },
    {
      "epoch": 3.6568047337278107,
      "grad_norm": 0.353515625,
      "learning_rate": 4.0851685554120226e-05,
      "loss": 0.0387,
      "step": 14832
    },
    {
      "epoch": 3.6572978303747536,
      "grad_norm": 0.2734375,
      "learning_rate": 4.082393187436181e-05,
      "loss": 0.0353,
      "step": 14834
    },
    {
      "epoch": 3.657790927021696,
      "grad_norm": 0.31640625,
      "learning_rate": 4.079618520729786e-05,
      "loss": 0.0344,
      "step": 14836
    },
    {
      "epoch": 3.658284023668639,
      "grad_norm": 0.341796875,
      "learning_rate": 4.076844555621653e-05,
      "loss": 0.0347,
      "step": 14838
    },
    {
      "epoch": 3.658777120315582,
      "grad_norm": 0.298828125,
      "learning_rate": 4.0740712924405125e-05,
      "loss": 0.0329,
      "step": 14840
    },
    {
      "epoch": 3.6592702169625246,
      "grad_norm": 0.328125,
      "learning_rate": 4.071298731515004e-05,
      "loss": 0.0354,
      "step": 14842
    },
    {
      "epoch": 3.6597633136094676,
      "grad_norm": 0.30859375,
      "learning_rate": 4.068526873173704e-05,
      "loss": 0.0343,
      "step": 14844
    },
    {
      "epoch": 3.66025641025641,
      "grad_norm": 0.27734375,
      "learning_rate": 4.065755717745083e-05,
      "loss": 0.0316,
      "step": 14846
    },
    {
      "epoch": 3.660749506903353,
      "grad_norm": 0.306640625,
      "learning_rate": 4.062985265557543e-05,
      "loss": 0.035,
      "step": 14848
    },
    {
      "epoch": 3.6612426035502956,
      "grad_norm": 0.296875,
      "learning_rate": 4.0602155169393964e-05,
      "loss": 0.0349,
      "step": 14850
    },
    {
      "epoch": 3.6617357001972386,
      "grad_norm": 0.296875,
      "learning_rate": 4.0574464722188745e-05,
      "loss": 0.0321,
      "step": 14852
    },
    {
      "epoch": 3.6622287968441816,
      "grad_norm": 0.29296875,
      "learning_rate": 4.054678131724128e-05,
      "loss": 0.0348,
      "step": 14854
    },
    {
      "epoch": 3.662721893491124,
      "grad_norm": 0.296875,
      "learning_rate": 4.05191049578321e-05,
      "loss": 0.0348,
      "step": 14856
    },
    {
      "epoch": 3.663214990138067,
      "grad_norm": 0.30078125,
      "learning_rate": 4.049143564724116e-05,
      "loss": 0.0363,
      "step": 14858
    },
    {
      "epoch": 3.66370808678501,
      "grad_norm": 0.359375,
      "learning_rate": 4.0463773388747306e-05,
      "loss": 0.039,
      "step": 14860
    },
    {
      "epoch": 3.6642011834319526,
      "grad_norm": 0.37109375,
      "learning_rate": 4.0436118185628714e-05,
      "loss": 0.0378,
      "step": 14862
    },
    {
      "epoch": 3.6646942800788955,
      "grad_norm": 0.357421875,
      "learning_rate": 4.040847004116268e-05,
      "loss": 0.0365,
      "step": 14864
    },
    {
      "epoch": 3.6651873767258385,
      "grad_norm": 0.33984375,
      "learning_rate": 4.038082895862567e-05,
      "loss": 0.0379,
      "step": 14866
    },
    {
      "epoch": 3.665680473372781,
      "grad_norm": 0.30078125,
      "learning_rate": 4.0353194941293324e-05,
      "loss": 0.0386,
      "step": 14868
    },
    {
      "epoch": 3.666173570019724,
      "grad_norm": 0.419921875,
      "learning_rate": 4.032556799244034e-05,
      "loss": 0.0434,
      "step": 14870
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 0.3046875,
      "learning_rate": 4.02979481153408e-05,
      "loss": 0.0331,
      "step": 14872
    },
    {
      "epoch": 3.6671597633136095,
      "grad_norm": 0.291015625,
      "learning_rate": 4.027033531326767e-05,
      "loss": 0.0372,
      "step": 14874
    },
    {
      "epoch": 3.667652859960552,
      "grad_norm": 0.3671875,
      "learning_rate": 4.024272958949336e-05,
      "loss": 0.0332,
      "step": 14876
    },
    {
      "epoch": 3.668145956607495,
      "grad_norm": 0.359375,
      "learning_rate": 4.0215130947289206e-05,
      "loss": 0.0384,
      "step": 14878
    },
    {
      "epoch": 3.668639053254438,
      "grad_norm": 0.28515625,
      "learning_rate": 4.018753938992582e-05,
      "loss": 0.0324,
      "step": 14880
    },
    {
      "epoch": 3.6691321499013805,
      "grad_norm": 0.298828125,
      "learning_rate": 4.0159954920673005e-05,
      "loss": 0.034,
      "step": 14882
    },
    {
      "epoch": 3.6696252465483234,
      "grad_norm": 0.271484375,
      "learning_rate": 4.013237754279957e-05,
      "loss": 0.034,
      "step": 14884
    },
    {
      "epoch": 3.6701183431952664,
      "grad_norm": 0.392578125,
      "learning_rate": 4.0104807259573716e-05,
      "loss": 0.0352,
      "step": 14886
    },
    {
      "epoch": 3.670611439842209,
      "grad_norm": 0.33984375,
      "learning_rate": 4.007724407426255e-05,
      "loss": 0.032,
      "step": 14888
    },
    {
      "epoch": 3.671104536489152,
      "grad_norm": 0.373046875,
      "learning_rate": 4.00496879901326e-05,
      "loss": 0.0386,
      "step": 14890
    },
    {
      "epoch": 3.671597633136095,
      "grad_norm": 0.294921875,
      "learning_rate": 4.002213901044931e-05,
      "loss": 0.0386,
      "step": 14892
    },
    {
      "epoch": 3.6720907297830374,
      "grad_norm": 0.34375,
      "learning_rate": 3.999459713847742e-05,
      "loss": 0.0337,
      "step": 14894
    },
    {
      "epoch": 3.6725838264299804,
      "grad_norm": 0.341796875,
      "learning_rate": 3.9967062377480824e-05,
      "loss": 0.0368,
      "step": 14896
    },
    {
      "epoch": 3.6730769230769234,
      "grad_norm": 0.3125,
      "learning_rate": 3.993953473072245e-05,
      "loss": 0.0354,
      "step": 14898
    },
    {
      "epoch": 3.673570019723866,
      "grad_norm": 0.310546875,
      "learning_rate": 3.991201420146461e-05,
      "loss": 0.0329,
      "step": 14900
    },
    {
      "epoch": 3.6740631163708084,
      "grad_norm": 0.291015625,
      "learning_rate": 3.9884500792968515e-05,
      "loss": 0.0372,
      "step": 14902
    },
    {
      "epoch": 3.6745562130177514,
      "grad_norm": 0.322265625,
      "learning_rate": 3.985699450849478e-05,
      "loss": 0.0368,
      "step": 14904
    },
    {
      "epoch": 3.6750493096646943,
      "grad_norm": 0.31640625,
      "learning_rate": 3.982949535130297e-05,
      "loss": 0.0364,
      "step": 14906
    },
    {
      "epoch": 3.675542406311637,
      "grad_norm": 0.287109375,
      "learning_rate": 3.980200332465189e-05,
      "loss": 0.0327,
      "step": 14908
    },
    {
      "epoch": 3.67603550295858,
      "grad_norm": 0.3046875,
      "learning_rate": 3.977451843179957e-05,
      "loss": 0.0354,
      "step": 14910
    },
    {
      "epoch": 3.676528599605523,
      "grad_norm": 0.28125,
      "learning_rate": 3.9747040676003e-05,
      "loss": 0.0313,
      "step": 14912
    },
    {
      "epoch": 3.6770216962524653,
      "grad_norm": 0.31640625,
      "learning_rate": 3.971957006051858e-05,
      "loss": 0.0356,
      "step": 14914
    },
    {
      "epoch": 3.6775147928994083,
      "grad_norm": 0.294921875,
      "learning_rate": 3.9692106588601605e-05,
      "loss": 0.0345,
      "step": 14916
    },
    {
      "epoch": 3.6780078895463513,
      "grad_norm": 0.330078125,
      "learning_rate": 3.966465026350679e-05,
      "loss": 0.0369,
      "step": 14918
    },
    {
      "epoch": 3.678500986193294,
      "grad_norm": 0.263671875,
      "learning_rate": 3.963720108848775e-05,
      "loss": 0.0362,
      "step": 14920
    },
    {
      "epoch": 3.6789940828402368,
      "grad_norm": 0.353515625,
      "learning_rate": 3.96097590667974e-05,
      "loss": 0.0359,
      "step": 14922
    },
    {
      "epoch": 3.6794871794871797,
      "grad_norm": 0.33984375,
      "learning_rate": 3.958232420168778e-05,
      "loss": 0.0335,
      "step": 14924
    },
    {
      "epoch": 3.6799802761341223,
      "grad_norm": 0.31640625,
      "learning_rate": 3.955489649641007e-05,
      "loss": 0.0376,
      "step": 14926
    },
    {
      "epoch": 3.6804733727810652,
      "grad_norm": 0.30078125,
      "learning_rate": 3.952747595421464e-05,
      "loss": 0.0408,
      "step": 14928
    },
    {
      "epoch": 3.6809664694280078,
      "grad_norm": 0.2734375,
      "learning_rate": 3.9500062578350885e-05,
      "loss": 0.0334,
      "step": 14930
    },
    {
      "epoch": 3.6814595660749507,
      "grad_norm": 0.369140625,
      "learning_rate": 3.947265637206756e-05,
      "loss": 0.0368,
      "step": 14932
    },
    {
      "epoch": 3.6819526627218933,
      "grad_norm": 0.322265625,
      "learning_rate": 3.9445257338612373e-05,
      "loss": 0.0337,
      "step": 14934
    },
    {
      "epoch": 3.6824457593688362,
      "grad_norm": 0.322265625,
      "learning_rate": 3.9417865481232274e-05,
      "loss": 0.0359,
      "step": 14936
    },
    {
      "epoch": 3.682938856015779,
      "grad_norm": 0.302734375,
      "learning_rate": 3.939048080317337e-05,
      "loss": 0.0342,
      "step": 14938
    },
    {
      "epoch": 3.6834319526627217,
      "grad_norm": 0.33984375,
      "learning_rate": 3.9363103307680894e-05,
      "loss": 0.0344,
      "step": 14940
    },
    {
      "epoch": 3.6839250493096647,
      "grad_norm": 0.310546875,
      "learning_rate": 3.933573299799922e-05,
      "loss": 0.0317,
      "step": 14942
    },
    {
      "epoch": 3.6844181459566077,
      "grad_norm": 0.296875,
      "learning_rate": 3.93083698773719e-05,
      "loss": 0.0366,
      "step": 14944
    },
    {
      "epoch": 3.68491124260355,
      "grad_norm": 0.3203125,
      "learning_rate": 3.928101394904165e-05,
      "loss": 0.0319,
      "step": 14946
    },
    {
      "epoch": 3.685404339250493,
      "grad_norm": 0.306640625,
      "learning_rate": 3.92536652162502e-05,
      "loss": 0.0344,
      "step": 14948
    },
    {
      "epoch": 3.685897435897436,
      "grad_norm": 0.3203125,
      "learning_rate": 3.9226323682238605e-05,
      "loss": 0.0342,
      "step": 14950
    },
    {
      "epoch": 3.6863905325443787,
      "grad_norm": 0.298828125,
      "learning_rate": 3.9198989350246963e-05,
      "loss": 0.0304,
      "step": 14952
    },
    {
      "epoch": 3.6868836291913216,
      "grad_norm": 0.27734375,
      "learning_rate": 3.917166222351455e-05,
      "loss": 0.0325,
      "step": 14954
    },
    {
      "epoch": 3.687376725838264,
      "grad_norm": 0.294921875,
      "learning_rate": 3.9144342305279794e-05,
      "loss": 0.0348,
      "step": 14956
    },
    {
      "epoch": 3.687869822485207,
      "grad_norm": 0.3203125,
      "learning_rate": 3.911702959878024e-05,
      "loss": 0.0341,
      "step": 14958
    },
    {
      "epoch": 3.6883629191321496,
      "grad_norm": 0.3203125,
      "learning_rate": 3.9089724107252645e-05,
      "loss": 0.0338,
      "step": 14960
    },
    {
      "epoch": 3.6888560157790926,
      "grad_norm": 0.314453125,
      "learning_rate": 3.906242583393279e-05,
      "loss": 0.0355,
      "step": 14962
    },
    {
      "epoch": 3.6893491124260356,
      "grad_norm": 0.298828125,
      "learning_rate": 3.90351347820557e-05,
      "loss": 0.0315,
      "step": 14964
    },
    {
      "epoch": 3.689842209072978,
      "grad_norm": 0.314453125,
      "learning_rate": 3.900785095485553e-05,
      "loss": 0.0374,
      "step": 14966
    },
    {
      "epoch": 3.690335305719921,
      "grad_norm": 0.3125,
      "learning_rate": 3.898057435556556e-05,
      "loss": 0.029,
      "step": 14968
    },
    {
      "epoch": 3.690828402366864,
      "grad_norm": 0.294921875,
      "learning_rate": 3.8953304987418205e-05,
      "loss": 0.0317,
      "step": 14970
    },
    {
      "epoch": 3.6913214990138066,
      "grad_norm": 0.26953125,
      "learning_rate": 3.8926042853645064e-05,
      "loss": 0.0322,
      "step": 14972
    },
    {
      "epoch": 3.6918145956607495,
      "grad_norm": 0.27734375,
      "learning_rate": 3.889878795747683e-05,
      "loss": 0.0326,
      "step": 14974
    },
    {
      "epoch": 3.6923076923076925,
      "grad_norm": 0.349609375,
      "learning_rate": 3.8871540302143406e-05,
      "loss": 0.0338,
      "step": 14976
    },
    {
      "epoch": 3.692800788954635,
      "grad_norm": 0.30859375,
      "learning_rate": 3.884429989087371e-05,
      "loss": 0.0316,
      "step": 14978
    },
    {
      "epoch": 3.693293885601578,
      "grad_norm": 0.380859375,
      "learning_rate": 3.8817066726895945e-05,
      "loss": 0.0379,
      "step": 14980
    },
    {
      "epoch": 3.693786982248521,
      "grad_norm": 0.27734375,
      "learning_rate": 3.878984081343736e-05,
      "loss": 0.0335,
      "step": 14982
    },
    {
      "epoch": 3.6942800788954635,
      "grad_norm": 0.291015625,
      "learning_rate": 3.876262215372439e-05,
      "loss": 0.0344,
      "step": 14984
    },
    {
      "epoch": 3.6947731755424065,
      "grad_norm": 0.29296875,
      "learning_rate": 3.873541075098261e-05,
      "loss": 0.037,
      "step": 14986
    },
    {
      "epoch": 3.695266272189349,
      "grad_norm": 0.302734375,
      "learning_rate": 3.8708206608436715e-05,
      "loss": 0.0355,
      "step": 14988
    },
    {
      "epoch": 3.695759368836292,
      "grad_norm": 0.357421875,
      "learning_rate": 3.868100972931058e-05,
      "loss": 0.0364,
      "step": 14990
    },
    {
      "epoch": 3.6962524654832345,
      "grad_norm": 0.330078125,
      "learning_rate": 3.865382011682711e-05,
      "loss": 0.0357,
      "step": 14992
    },
    {
      "epoch": 3.6967455621301775,
      "grad_norm": 0.3203125,
      "learning_rate": 3.862663777420847e-05,
      "loss": 0.0376,
      "step": 14994
    },
    {
      "epoch": 3.6972386587771204,
      "grad_norm": 0.34765625,
      "learning_rate": 3.8599462704675926e-05,
      "loss": 0.0348,
      "step": 14996
    },
    {
      "epoch": 3.697731755424063,
      "grad_norm": 0.27734375,
      "learning_rate": 3.857229491144987e-05,
      "loss": 0.034,
      "step": 14998
    },
    {
      "epoch": 3.698224852071006,
      "grad_norm": 0.31640625,
      "learning_rate": 3.854513439774984e-05,
      "loss": 0.035,
      "step": 15000
    },
    {
      "epoch": 3.698717948717949,
      "grad_norm": 0.302734375,
      "learning_rate": 3.851798116679449e-05,
      "loss": 0.0351,
      "step": 15002
    },
    {
      "epoch": 3.6992110453648914,
      "grad_norm": 0.29296875,
      "learning_rate": 3.849083522180168e-05,
      "loss": 0.0336,
      "step": 15004
    },
    {
      "epoch": 3.6997041420118344,
      "grad_norm": 0.4140625,
      "learning_rate": 3.846369656598825e-05,
      "loss": 0.0355,
      "step": 15006
    },
    {
      "epoch": 3.7001972386587774,
      "grad_norm": 0.30859375,
      "learning_rate": 3.843656520257043e-05,
      "loss": 0.0358,
      "step": 15008
    },
    {
      "epoch": 3.70069033530572,
      "grad_norm": 0.326171875,
      "learning_rate": 3.840944113476331e-05,
      "loss": 0.0393,
      "step": 15010
    },
    {
      "epoch": 3.701183431952663,
      "grad_norm": 0.330078125,
      "learning_rate": 3.838232436578131e-05,
      "loss": 0.0373,
      "step": 15012
    },
    {
      "epoch": 3.7016765285996054,
      "grad_norm": 0.369140625,
      "learning_rate": 3.8355214898837896e-05,
      "loss": 0.0347,
      "step": 15014
    },
    {
      "epoch": 3.7021696252465484,
      "grad_norm": 0.318359375,
      "learning_rate": 3.832811273714569e-05,
      "loss": 0.0333,
      "step": 15016
    },
    {
      "epoch": 3.702662721893491,
      "grad_norm": 0.37890625,
      "learning_rate": 3.83010178839165e-05,
      "loss": 0.0354,
      "step": 15018
    },
    {
      "epoch": 3.703155818540434,
      "grad_norm": 0.359375,
      "learning_rate": 3.82739303423611e-05,
      "loss": 0.0344,
      "step": 15020
    },
    {
      "epoch": 3.703648915187377,
      "grad_norm": 0.271484375,
      "learning_rate": 3.824685011568966e-05,
      "loss": 0.0346,
      "step": 15022
    },
    {
      "epoch": 3.7041420118343193,
      "grad_norm": 0.34375,
      "learning_rate": 3.82197772071112e-05,
      "loss": 0.0387,
      "step": 15024
    },
    {
      "epoch": 3.7046351084812623,
      "grad_norm": 0.29296875,
      "learning_rate": 3.819271161983414e-05,
      "loss": 0.0355,
      "step": 15026
    },
    {
      "epoch": 3.7051282051282053,
      "grad_norm": 0.296875,
      "learning_rate": 3.816565335706581e-05,
      "loss": 0.0323,
      "step": 15028
    },
    {
      "epoch": 3.705621301775148,
      "grad_norm": 0.314453125,
      "learning_rate": 3.8138602422012793e-05,
      "loss": 0.03,
      "step": 15030
    },
    {
      "epoch": 3.706114398422091,
      "grad_norm": 0.3671875,
      "learning_rate": 3.811155881788082e-05,
      "loss": 0.0319,
      "step": 15032
    },
    {
      "epoch": 3.7066074950690338,
      "grad_norm": 0.32421875,
      "learning_rate": 3.808452254787459e-05,
      "loss": 0.035,
      "step": 15034
    },
    {
      "epoch": 3.7071005917159763,
      "grad_norm": 0.26171875,
      "learning_rate": 3.80574936151982e-05,
      "loss": 0.0332,
      "step": 15036
    },
    {
      "epoch": 3.7075936883629192,
      "grad_norm": 0.265625,
      "learning_rate": 3.8030472023054595e-05,
      "loss": 0.0302,
      "step": 15038
    },
    {
      "epoch": 3.7080867850098618,
      "grad_norm": 0.236328125,
      "learning_rate": 3.800345777464612e-05,
      "loss": 0.0309,
      "step": 15040
    },
    {
      "epoch": 3.7085798816568047,
      "grad_norm": 0.30078125,
      "learning_rate": 3.7976450873174005e-05,
      "loss": 0.0352,
      "step": 15042
    },
    {
      "epoch": 3.7090729783037473,
      "grad_norm": 0.357421875,
      "learning_rate": 3.794945132183876e-05,
      "loss": 0.0388,
      "step": 15044
    },
    {
      "epoch": 3.7095660749506902,
      "grad_norm": 0.28125,
      "learning_rate": 3.792245912384e-05,
      "loss": 0.0344,
      "step": 15046
    },
    {
      "epoch": 3.710059171597633,
      "grad_norm": 0.33203125,
      "learning_rate": 3.7895474282376375e-05,
      "loss": 0.0346,
      "step": 15048
    },
    {
      "epoch": 3.7105522682445757,
      "grad_norm": 0.30859375,
      "learning_rate": 3.786849680064586e-05,
      "loss": 0.0346,
      "step": 15050
    },
    {
      "epoch": 3.7110453648915187,
      "grad_norm": 0.306640625,
      "learning_rate": 3.7841526681845305e-05,
      "loss": 0.0375,
      "step": 15052
    },
    {
      "epoch": 3.7115384615384617,
      "grad_norm": 0.296875,
      "learning_rate": 3.781456392917095e-05,
      "loss": 0.0335,
      "step": 15054
    },
    {
      "epoch": 3.712031558185404,
      "grad_norm": 0.322265625,
      "learning_rate": 3.778760854581794e-05,
      "loss": 0.0329,
      "step": 15056
    },
    {
      "epoch": 3.712524654832347,
      "grad_norm": 0.279296875,
      "learning_rate": 3.7760660534980666e-05,
      "loss": 0.031,
      "step": 15058
    },
    {
      "epoch": 3.71301775147929,
      "grad_norm": 0.3125,
      "learning_rate": 3.773371989985265e-05,
      "loss": 0.0349,
      "step": 15060
    },
    {
      "epoch": 3.7135108481262327,
      "grad_norm": 0.3359375,
      "learning_rate": 3.770678664362641e-05,
      "loss": 0.0353,
      "step": 15062
    },
    {
      "epoch": 3.7140039447731756,
      "grad_norm": 0.3203125,
      "learning_rate": 3.7679860769493825e-05,
      "loss": 0.0372,
      "step": 15064
    },
    {
      "epoch": 3.7144970414201186,
      "grad_norm": 0.341796875,
      "learning_rate": 3.765294228064562e-05,
      "loss": 0.0365,
      "step": 15066
    },
    {
      "epoch": 3.714990138067061,
      "grad_norm": 0.43359375,
      "learning_rate": 3.762603118027193e-05,
      "loss": 0.0372,
      "step": 15068
    },
    {
      "epoch": 3.715483234714004,
      "grad_norm": 0.287109375,
      "learning_rate": 3.7599127471561745e-05,
      "loss": 0.0348,
      "step": 15070
    },
    {
      "epoch": 3.7159763313609466,
      "grad_norm": 0.294921875,
      "learning_rate": 3.757223115770336e-05,
      "loss": 0.0349,
      "step": 15072
    },
    {
      "epoch": 3.7164694280078896,
      "grad_norm": 0.341796875,
      "learning_rate": 3.754534224188415e-05,
      "loss": 0.0336,
      "step": 15074
    },
    {
      "epoch": 3.716962524654832,
      "grad_norm": 0.2734375,
      "learning_rate": 3.751846072729057e-05,
      "loss": 0.0281,
      "step": 15076
    },
    {
      "epoch": 3.717455621301775,
      "grad_norm": 0.314453125,
      "learning_rate": 3.74915866171083e-05,
      "loss": 0.0357,
      "step": 15078
    },
    {
      "epoch": 3.717948717948718,
      "grad_norm": 0.298828125,
      "learning_rate": 3.746471991452193e-05,
      "loss": 0.0374,
      "step": 15080
    },
    {
      "epoch": 3.7184418145956606,
      "grad_norm": 0.337890625,
      "learning_rate": 3.743786062271548e-05,
      "loss": 0.0364,
      "step": 15082
    },
    {
      "epoch": 3.7189349112426036,
      "grad_norm": 0.318359375,
      "learning_rate": 3.741100874487181e-05,
      "loss": 0.0356,
      "step": 15084
    },
    {
      "epoch": 3.7194280078895465,
      "grad_norm": 0.298828125,
      "learning_rate": 3.7384164284173064e-05,
      "loss": 0.034,
      "step": 15086
    },
    {
      "epoch": 3.719921104536489,
      "grad_norm": 0.298828125,
      "learning_rate": 3.7357327243800446e-05,
      "loss": 0.0349,
      "step": 15088
    },
    {
      "epoch": 3.720414201183432,
      "grad_norm": 0.32421875,
      "learning_rate": 3.7330497626934305e-05,
      "loss": 0.0398,
      "step": 15090
    },
    {
      "epoch": 3.720907297830375,
      "grad_norm": 0.349609375,
      "learning_rate": 3.7303675436754126e-05,
      "loss": 0.0353,
      "step": 15092
    },
    {
      "epoch": 3.7214003944773175,
      "grad_norm": 0.326171875,
      "learning_rate": 3.72768606764384e-05,
      "loss": 0.0346,
      "step": 15094
    },
    {
      "epoch": 3.7218934911242605,
      "grad_norm": 0.291015625,
      "learning_rate": 3.725005334916495e-05,
      "loss": 0.0342,
      "step": 15096
    },
    {
      "epoch": 3.722386587771203,
      "grad_norm": 0.294921875,
      "learning_rate": 3.722325345811048e-05,
      "loss": 0.0333,
      "step": 15098
    },
    {
      "epoch": 3.722879684418146,
      "grad_norm": 0.326171875,
      "learning_rate": 3.719646100645099e-05,
      "loss": 0.0331,
      "step": 15100
    },
    {
      "epoch": 3.7233727810650885,
      "grad_norm": 0.337890625,
      "learning_rate": 3.716967599736152e-05,
      "loss": 0.0348,
      "step": 15102
    },
    {
      "epoch": 3.7238658777120315,
      "grad_norm": 0.318359375,
      "learning_rate": 3.714289843401625e-05,
      "loss": 0.0327,
      "step": 15104
    },
    {
      "epoch": 3.7243589743589745,
      "grad_norm": 0.287109375,
      "learning_rate": 3.711612831958845e-05,
      "loss": 0.0327,
      "step": 15106
    },
    {
      "epoch": 3.724852071005917,
      "grad_norm": 0.330078125,
      "learning_rate": 3.7089365657250555e-05,
      "loss": 0.0349,
      "step": 15108
    },
    {
      "epoch": 3.72534516765286,
      "grad_norm": 0.279296875,
      "learning_rate": 3.7062610450174105e-05,
      "loss": 0.0357,
      "step": 15110
    },
    {
      "epoch": 3.725838264299803,
      "grad_norm": 0.296875,
      "learning_rate": 3.703586270152968e-05,
      "loss": 0.037,
      "step": 15112
    },
    {
      "epoch": 3.7263313609467454,
      "grad_norm": 0.263671875,
      "learning_rate": 3.700912241448707e-05,
      "loss": 0.0368,
      "step": 15114
    },
    {
      "epoch": 3.7268244575936884,
      "grad_norm": 0.34765625,
      "learning_rate": 3.698238959221515e-05,
      "loss": 0.0349,
      "step": 15116
    },
    {
      "epoch": 3.7273175542406314,
      "grad_norm": 0.283203125,
      "learning_rate": 3.6955664237881916e-05,
      "loss": 0.0304,
      "step": 15118
    },
    {
      "epoch": 3.727810650887574,
      "grad_norm": 0.28125,
      "learning_rate": 3.692894635465446e-05,
      "loss": 0.0339,
      "step": 15120
    },
    {
      "epoch": 3.728303747534517,
      "grad_norm": 0.298828125,
      "learning_rate": 3.690223594569901e-05,
      "loss": 0.0354,
      "step": 15122
    },
    {
      "epoch": 3.7287968441814594,
      "grad_norm": 0.349609375,
      "learning_rate": 3.687553301418092e-05,
      "loss": 0.0354,
      "step": 15124
    },
    {
      "epoch": 3.7292899408284024,
      "grad_norm": 0.294921875,
      "learning_rate": 3.684883756326457e-05,
      "loss": 0.0335,
      "step": 15126
    },
    {
      "epoch": 3.729783037475345,
      "grad_norm": 0.34375,
      "learning_rate": 3.6822149596113574e-05,
      "loss": 0.0344,
      "step": 15128
    },
    {
      "epoch": 3.730276134122288,
      "grad_norm": 0.29296875,
      "learning_rate": 3.679546911589058e-05,
      "loss": 0.036,
      "step": 15130
    },
    {
      "epoch": 3.730769230769231,
      "grad_norm": 0.330078125,
      "learning_rate": 3.6768796125757384e-05,
      "loss": 0.035,
      "step": 15132
    },
    {
      "epoch": 3.7312623274161734,
      "grad_norm": 0.29296875,
      "learning_rate": 3.674213062887489e-05,
      "loss": 0.0333,
      "step": 15134
    },
    {
      "epoch": 3.7317554240631163,
      "grad_norm": 0.296875,
      "learning_rate": 3.67154726284031e-05,
      "loss": 0.0319,
      "step": 15136
    },
    {
      "epoch": 3.7322485207100593,
      "grad_norm": 0.275390625,
      "learning_rate": 3.668882212750115e-05,
      "loss": 0.0338,
      "step": 15138
    },
    {
      "epoch": 3.732741617357002,
      "grad_norm": 0.294921875,
      "learning_rate": 3.666217912932729e-05,
      "loss": 0.0353,
      "step": 15140
    },
    {
      "epoch": 3.733234714003945,
      "grad_norm": 0.259765625,
      "learning_rate": 3.663554363703879e-05,
      "loss": 0.0332,
      "step": 15142
    },
    {
      "epoch": 3.7337278106508878,
      "grad_norm": 0.3203125,
      "learning_rate": 3.6608915653792174e-05,
      "loss": 0.0355,
      "step": 15144
    },
    {
      "epoch": 3.7342209072978303,
      "grad_norm": 0.32421875,
      "learning_rate": 3.6582295182742964e-05,
      "loss": 0.0334,
      "step": 15146
    },
    {
      "epoch": 3.7347140039447733,
      "grad_norm": 0.322265625,
      "learning_rate": 3.6555682227045864e-05,
      "loss": 0.0344,
      "step": 15148
    },
    {
      "epoch": 3.7352071005917162,
      "grad_norm": 0.33203125,
      "learning_rate": 3.652907678985464e-05,
      "loss": 0.0379,
      "step": 15150
    },
    {
      "epoch": 3.7357001972386588,
      "grad_norm": 0.3203125,
      "learning_rate": 3.65024788743222e-05,
      "loss": 0.0342,
      "step": 15152
    },
    {
      "epoch": 3.7361932938856017,
      "grad_norm": 0.283203125,
      "learning_rate": 3.647588848360054e-05,
      "loss": 0.0334,
      "step": 15154
    },
    {
      "epoch": 3.7366863905325443,
      "grad_norm": 0.32421875,
      "learning_rate": 3.644930562084076e-05,
      "loss": 0.0366,
      "step": 15156
    },
    {
      "epoch": 3.7371794871794872,
      "grad_norm": 0.26953125,
      "learning_rate": 3.6422730289193116e-05,
      "loss": 0.0313,
      "step": 15158
    },
    {
      "epoch": 3.7376725838264298,
      "grad_norm": 0.275390625,
      "learning_rate": 3.6396162491806884e-05,
      "loss": 0.0313,
      "step": 15160
    },
    {
      "epoch": 3.7381656804733727,
      "grad_norm": 0.298828125,
      "learning_rate": 3.63696022318305e-05,
      "loss": 0.0349,
      "step": 15162
    },
    {
      "epoch": 3.7386587771203157,
      "grad_norm": 0.2421875,
      "learning_rate": 3.634304951241153e-05,
      "loss": 0.0304,
      "step": 15164
    },
    {
      "epoch": 3.739151873767258,
      "grad_norm": 0.296875,
      "learning_rate": 3.63165043366966e-05,
      "loss": 0.0343,
      "step": 15166
    },
    {
      "epoch": 3.739644970414201,
      "grad_norm": 0.2890625,
      "learning_rate": 3.628996670783148e-05,
      "loss": 0.0375,
      "step": 15168
    },
    {
      "epoch": 3.740138067061144,
      "grad_norm": 0.322265625,
      "learning_rate": 3.6263436628961e-05,
      "loss": 0.035,
      "step": 15170
    },
    {
      "epoch": 3.7406311637080867,
      "grad_norm": 0.287109375,
      "learning_rate": 3.623691410322918e-05,
      "loss": 0.0339,
      "step": 15172
    },
    {
      "epoch": 3.7411242603550297,
      "grad_norm": 0.3828125,
      "learning_rate": 3.6210399133779014e-05,
      "loss": 0.0361,
      "step": 15174
    },
    {
      "epoch": 3.7416173570019726,
      "grad_norm": 0.291015625,
      "learning_rate": 3.61838917237527e-05,
      "loss": 0.0355,
      "step": 15176
    },
    {
      "epoch": 3.742110453648915,
      "grad_norm": 0.337890625,
      "learning_rate": 3.6157391876291535e-05,
      "loss": 0.0379,
      "step": 15178
    },
    {
      "epoch": 3.742603550295858,
      "grad_norm": 0.32421875,
      "learning_rate": 3.613089959453587e-05,
      "loss": 0.0314,
      "step": 15180
    },
    {
      "epoch": 3.7430966469428006,
      "grad_norm": 0.26953125,
      "learning_rate": 3.6104414881625215e-05,
      "loss": 0.0301,
      "step": 15182
    },
    {
      "epoch": 3.7435897435897436,
      "grad_norm": 0.279296875,
      "learning_rate": 3.607793774069813e-05,
      "loss": 0.0345,
      "step": 15184
    },
    {
      "epoch": 3.744082840236686,
      "grad_norm": 0.31640625,
      "learning_rate": 3.605146817489236e-05,
      "loss": 0.0335,
      "step": 15186
    },
    {
      "epoch": 3.744575936883629,
      "grad_norm": 0.318359375,
      "learning_rate": 3.602500618734459e-05,
      "loss": 0.036,
      "step": 15188
    },
    {
      "epoch": 3.745069033530572,
      "grad_norm": 0.302734375,
      "learning_rate": 3.599855178119084e-05,
      "loss": 0.0307,
      "step": 15190
    },
    {
      "epoch": 3.7455621301775146,
      "grad_norm": 0.279296875,
      "learning_rate": 3.5972104959566e-05,
      "loss": 0.0324,
      "step": 15192
    },
    {
      "epoch": 3.7460552268244576,
      "grad_norm": 0.31640625,
      "learning_rate": 3.5945665725604216e-05,
      "loss": 0.0368,
      "step": 15194
    },
    {
      "epoch": 3.7465483234714005,
      "grad_norm": 0.255859375,
      "learning_rate": 3.591923408243868e-05,
      "loss": 0.0298,
      "step": 15196
    },
    {
      "epoch": 3.747041420118343,
      "grad_norm": 0.302734375,
      "learning_rate": 3.589281003320168e-05,
      "loss": 0.0325,
      "step": 15198
    },
    {
      "epoch": 3.747534516765286,
      "grad_norm": 0.3515625,
      "learning_rate": 3.586639358102466e-05,
      "loss": 0.0342,
      "step": 15200
    },
    {
      "epoch": 3.748027613412229,
      "grad_norm": 0.275390625,
      "learning_rate": 3.583998472903801e-05,
      "loss": 0.0367,
      "step": 15202
    },
    {
      "epoch": 3.7485207100591715,
      "grad_norm": 0.32421875,
      "learning_rate": 3.581358348037146e-05,
      "loss": 0.0363,
      "step": 15204
    },
    {
      "epoch": 3.7490138067061145,
      "grad_norm": 0.33984375,
      "learning_rate": 3.57871898381536e-05,
      "loss": 0.0332,
      "step": 15206
    },
    {
      "epoch": 3.7495069033530575,
      "grad_norm": 0.30078125,
      "learning_rate": 3.5760803805512266e-05,
      "loss": 0.0388,
      "step": 15208
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.29296875,
      "learning_rate": 3.573442538557434e-05,
      "loss": 0.0346,
      "step": 15210
    },
    {
      "epoch": 3.7504930966469425,
      "grad_norm": 0.322265625,
      "learning_rate": 3.570805458146582e-05,
      "loss": 0.034,
      "step": 15212
    },
    {
      "epoch": 3.7509861932938855,
      "grad_norm": 0.275390625,
      "learning_rate": 3.568169139631183e-05,
      "loss": 0.037,
      "step": 15214
    },
    {
      "epoch": 3.7514792899408285,
      "grad_norm": 0.29296875,
      "learning_rate": 3.565533583323646e-05,
      "loss": 0.0336,
      "step": 15216
    },
    {
      "epoch": 3.751972386587771,
      "grad_norm": 0.27734375,
      "learning_rate": 3.56289878953631e-05,
      "loss": 0.0371,
      "step": 15218
    },
    {
      "epoch": 3.752465483234714,
      "grad_norm": 0.32421875,
      "learning_rate": 3.560264758581402e-05,
      "loss": 0.0352,
      "step": 15220
    },
    {
      "epoch": 3.752958579881657,
      "grad_norm": 0.29296875,
      "learning_rate": 3.557631490771081e-05,
      "loss": 0.0362,
      "step": 15222
    },
    {
      "epoch": 3.7534516765285995,
      "grad_norm": 0.31640625,
      "learning_rate": 3.5549989864173946e-05,
      "loss": 0.033,
      "step": 15224
    },
    {
      "epoch": 3.7539447731755424,
      "grad_norm": 0.38671875,
      "learning_rate": 3.5523672458323134e-05,
      "loss": 0.0348,
      "step": 15226
    },
    {
      "epoch": 3.7544378698224854,
      "grad_norm": 0.2578125,
      "learning_rate": 3.549736269327715e-05,
      "loss": 0.0332,
      "step": 15228
    },
    {
      "epoch": 3.754930966469428,
      "grad_norm": 0.310546875,
      "learning_rate": 3.547106057215376e-05,
      "loss": 0.0336,
      "step": 15230
    },
    {
      "epoch": 3.755424063116371,
      "grad_norm": 0.28515625,
      "learning_rate": 3.5444766098070034e-05,
      "loss": 0.0361,
      "step": 15232
    },
    {
      "epoch": 3.755917159763314,
      "grad_norm": 0.279296875,
      "learning_rate": 3.541847927414189e-05,
      "loss": 0.0351,
      "step": 15234
    },
    {
      "epoch": 3.7564102564102564,
      "grad_norm": 0.353515625,
      "learning_rate": 3.539220010348461e-05,
      "loss": 0.0349,
      "step": 15236
    },
    {
      "epoch": 3.7569033530571994,
      "grad_norm": 0.291015625,
      "learning_rate": 3.5365928589212296e-05,
      "loss": 0.0287,
      "step": 15238
    },
    {
      "epoch": 3.757396449704142,
      "grad_norm": 0.298828125,
      "learning_rate": 3.533966473443832e-05,
      "loss": 0.034,
      "step": 15240
    },
    {
      "epoch": 3.757889546351085,
      "grad_norm": 0.369140625,
      "learning_rate": 3.531340854227513e-05,
      "loss": 0.0338,
      "step": 15242
    },
    {
      "epoch": 3.7583826429980274,
      "grad_norm": 0.32421875,
      "learning_rate": 3.528716001583411e-05,
      "loss": 0.0343,
      "step": 15244
    },
    {
      "epoch": 3.7588757396449703,
      "grad_norm": 0.263671875,
      "learning_rate": 3.526091915822601e-05,
      "loss": 0.0313,
      "step": 15246
    },
    {
      "epoch": 3.7593688362919133,
      "grad_norm": 0.3671875,
      "learning_rate": 3.5234685972560386e-05,
      "loss": 0.0343,
      "step": 15248
    },
    {
      "epoch": 3.759861932938856,
      "grad_norm": 0.361328125,
      "learning_rate": 3.520846046194614e-05,
      "loss": 0.035,
      "step": 15250
    },
    {
      "epoch": 3.760355029585799,
      "grad_norm": 0.326171875,
      "learning_rate": 3.5182242629491046e-05,
      "loss": 0.032,
      "step": 15252
    },
    {
      "epoch": 3.760848126232742,
      "grad_norm": 0.32421875,
      "learning_rate": 3.515603247830209e-05,
      "loss": 0.033,
      "step": 15254
    },
    {
      "epoch": 3.7613412228796843,
      "grad_norm": 0.34765625,
      "learning_rate": 3.5129830011485366e-05,
      "loss": 0.0335,
      "step": 15256
    },
    {
      "epoch": 3.7618343195266273,
      "grad_norm": 0.287109375,
      "learning_rate": 3.5103635232145916e-05,
      "loss": 0.0321,
      "step": 15258
    },
    {
      "epoch": 3.7623274161735702,
      "grad_norm": 0.3125,
      "learning_rate": 3.5077448143388084e-05,
      "loss": 0.0334,
      "step": 15260
    },
    {
      "epoch": 3.7628205128205128,
      "grad_norm": 0.287109375,
      "learning_rate": 3.505126874831506e-05,
      "loss": 0.0329,
      "step": 15262
    },
    {
      "epoch": 3.7633136094674557,
      "grad_norm": 0.29296875,
      "learning_rate": 3.502509705002939e-05,
      "loss": 0.0342,
      "step": 15264
    },
    {
      "epoch": 3.7638067061143983,
      "grad_norm": 0.296875,
      "learning_rate": 3.499893305163246e-05,
      "loss": 0.0323,
      "step": 15266
    },
    {
      "epoch": 3.7642998027613412,
      "grad_norm": 0.376953125,
      "learning_rate": 3.497277675622487e-05,
      "loss": 0.036,
      "step": 15268
    },
    {
      "epoch": 3.7647928994082838,
      "grad_norm": 0.353515625,
      "learning_rate": 3.49466281669063e-05,
      "loss": 0.0357,
      "step": 15270
    },
    {
      "epoch": 3.7652859960552267,
      "grad_norm": 0.3046875,
      "learning_rate": 3.4920487286775517e-05,
      "loss": 0.0292,
      "step": 15272
    },
    {
      "epoch": 3.7657790927021697,
      "grad_norm": 0.3359375,
      "learning_rate": 3.4894354118930375e-05,
      "loss": 0.0366,
      "step": 15274
    },
    {
      "epoch": 3.7662721893491122,
      "grad_norm": 0.302734375,
      "learning_rate": 3.4868228666467704e-05,
      "loss": 0.0368,
      "step": 15276
    },
    {
      "epoch": 3.766765285996055,
      "grad_norm": 0.251953125,
      "learning_rate": 3.484211093248366e-05,
      "loss": 0.0312,
      "step": 15278
    },
    {
      "epoch": 3.767258382642998,
      "grad_norm": 0.328125,
      "learning_rate": 3.4816000920073235e-05,
      "loss": 0.0363,
      "step": 15280
    },
    {
      "epoch": 3.7677514792899407,
      "grad_norm": 0.328125,
      "learning_rate": 3.478989863233063e-05,
      "loss": 0.0351,
      "step": 15282
    },
    {
      "epoch": 3.7682445759368837,
      "grad_norm": 0.267578125,
      "learning_rate": 3.476380407234913e-05,
      "loss": 0.0336,
      "step": 15284
    },
    {
      "epoch": 3.7687376725838266,
      "grad_norm": 0.271484375,
      "learning_rate": 3.473771724322108e-05,
      "loss": 0.0345,
      "step": 15286
    },
    {
      "epoch": 3.769230769230769,
      "grad_norm": 0.2890625,
      "learning_rate": 3.4711638148037905e-05,
      "loss": 0.0298,
      "step": 15288
    },
    {
      "epoch": 3.769723865877712,
      "grad_norm": 0.306640625,
      "learning_rate": 3.4685566789890145e-05,
      "loss": 0.0364,
      "step": 15290
    },
    {
      "epoch": 3.770216962524655,
      "grad_norm": 0.330078125,
      "learning_rate": 3.4659503171867414e-05,
      "loss": 0.0324,
      "step": 15292
    },
    {
      "epoch": 3.7707100591715976,
      "grad_norm": 0.306640625,
      "learning_rate": 3.463344729705835e-05,
      "loss": 0.0324,
      "step": 15294
    },
    {
      "epoch": 3.7712031558185406,
      "grad_norm": 0.29296875,
      "learning_rate": 3.4607399168550734e-05,
      "loss": 0.0336,
      "step": 15296
    },
    {
      "epoch": 3.771696252465483,
      "grad_norm": 0.357421875,
      "learning_rate": 3.458135878943142e-05,
      "loss": 0.0352,
      "step": 15298
    },
    {
      "epoch": 3.772189349112426,
      "grad_norm": 0.2890625,
      "learning_rate": 3.455532616278634e-05,
      "loss": 0.0364,
      "step": 15300
    },
    {
      "epoch": 3.7726824457593686,
      "grad_norm": 0.275390625,
      "learning_rate": 3.452930129170051e-05,
      "loss": 0.0335,
      "step": 15302
    },
    {
      "epoch": 3.7731755424063116,
      "grad_norm": 0.28515625,
      "learning_rate": 3.4503284179258025e-05,
      "loss": 0.0334,
      "step": 15304
    },
    {
      "epoch": 3.7736686390532546,
      "grad_norm": 0.35546875,
      "learning_rate": 3.4477274828542074e-05,
      "loss": 0.0322,
      "step": 15306
    },
    {
      "epoch": 3.774161735700197,
      "grad_norm": 0.36328125,
      "learning_rate": 3.445127324263485e-05,
      "loss": 0.0391,
      "step": 15308
    },
    {
      "epoch": 3.77465483234714,
      "grad_norm": 0.361328125,
      "learning_rate": 3.442527942461774e-05,
      "loss": 0.0386,
      "step": 15310
    },
    {
      "epoch": 3.775147928994083,
      "grad_norm": 0.28515625,
      "learning_rate": 3.439929337757113e-05,
      "loss": 0.035,
      "step": 15312
    },
    {
      "epoch": 3.7756410256410255,
      "grad_norm": 0.306640625,
      "learning_rate": 3.4373315104574524e-05,
      "loss": 0.0357,
      "step": 15314
    },
    {
      "epoch": 3.7761341222879685,
      "grad_norm": 0.337890625,
      "learning_rate": 3.434734460870649e-05,
      "loss": 0.0318,
      "step": 15316
    },
    {
      "epoch": 3.7766272189349115,
      "grad_norm": 0.318359375,
      "learning_rate": 3.432138189304468e-05,
      "loss": 0.033,
      "step": 15318
    },
    {
      "epoch": 3.777120315581854,
      "grad_norm": 0.396484375,
      "learning_rate": 3.4295426960665814e-05,
      "loss": 0.0335,
      "step": 15320
    },
    {
      "epoch": 3.777613412228797,
      "grad_norm": 0.33984375,
      "learning_rate": 3.4269479814645745e-05,
      "loss": 0.0332,
      "step": 15322
    },
    {
      "epoch": 3.7781065088757395,
      "grad_norm": 0.3046875,
      "learning_rate": 3.4243540458059264e-05,
      "loss": 0.0364,
      "step": 15324
    },
    {
      "epoch": 3.7785996055226825,
      "grad_norm": 0.302734375,
      "learning_rate": 3.421760889398038e-05,
      "loss": 0.0347,
      "step": 15326
    },
    {
      "epoch": 3.779092702169625,
      "grad_norm": 0.294921875,
      "learning_rate": 3.419168512548212e-05,
      "loss": 0.0326,
      "step": 15328
    },
    {
      "epoch": 3.779585798816568,
      "grad_norm": 0.3203125,
      "learning_rate": 3.4165769155636604e-05,
      "loss": 0.036,
      "step": 15330
    },
    {
      "epoch": 3.780078895463511,
      "grad_norm": 0.318359375,
      "learning_rate": 3.4139860987515014e-05,
      "loss": 0.0328,
      "step": 15332
    },
    {
      "epoch": 3.7805719921104535,
      "grad_norm": 0.2734375,
      "learning_rate": 3.4113960624187615e-05,
      "loss": 0.0326,
      "step": 15334
    },
    {
      "epoch": 3.7810650887573964,
      "grad_norm": 0.31640625,
      "learning_rate": 3.408806806872376e-05,
      "loss": 0.0334,
      "step": 15336
    },
    {
      "epoch": 3.7815581854043394,
      "grad_norm": 0.2890625,
      "learning_rate": 3.40621833241918e-05,
      "loss": 0.0334,
      "step": 15338
    },
    {
      "epoch": 3.782051282051282,
      "grad_norm": 0.296875,
      "learning_rate": 3.403630639365932e-05,
      "loss": 0.0365,
      "step": 15340
    },
    {
      "epoch": 3.782544378698225,
      "grad_norm": 0.443359375,
      "learning_rate": 3.40104372801928e-05,
      "loss": 0.0364,
      "step": 15342
    },
    {
      "epoch": 3.783037475345168,
      "grad_norm": 0.3203125,
      "learning_rate": 3.3984575986857905e-05,
      "loss": 0.0317,
      "step": 15344
    },
    {
      "epoch": 3.7835305719921104,
      "grad_norm": 0.314453125,
      "learning_rate": 3.395872251671932e-05,
      "loss": 0.0372,
      "step": 15346
    },
    {
      "epoch": 3.7840236686390534,
      "grad_norm": 0.337890625,
      "learning_rate": 3.393287687284086e-05,
      "loss": 0.0311,
      "step": 15348
    },
    {
      "epoch": 3.784516765285996,
      "grad_norm": 0.3203125,
      "learning_rate": 3.3907039058285387e-05,
      "loss": 0.0333,
      "step": 15350
    },
    {
      "epoch": 3.785009861932939,
      "grad_norm": 0.330078125,
      "learning_rate": 3.388120907611474e-05,
      "loss": 0.0387,
      "step": 15352
    },
    {
      "epoch": 3.7855029585798814,
      "grad_norm": 0.2734375,
      "learning_rate": 3.3855386929390045e-05,
      "loss": 0.0304,
      "step": 15354
    },
    {
      "epoch": 3.7859960552268244,
      "grad_norm": 0.28125,
      "learning_rate": 3.382957262117127e-05,
      "loss": 0.0319,
      "step": 15356
    },
    {
      "epoch": 3.7864891518737673,
      "grad_norm": 0.28125,
      "learning_rate": 3.3803766154517577e-05,
      "loss": 0.0322,
      "step": 15358
    },
    {
      "epoch": 3.78698224852071,
      "grad_norm": 0.283203125,
      "learning_rate": 3.37779675324872e-05,
      "loss": 0.0345,
      "step": 15360
    },
    {
      "epoch": 3.787475345167653,
      "grad_norm": 0.287109375,
      "learning_rate": 3.37521767581374e-05,
      "loss": 0.0346,
      "step": 15362
    },
    {
      "epoch": 3.787968441814596,
      "grad_norm": 0.32421875,
      "learning_rate": 3.372639383452456e-05,
      "loss": 0.0351,
      "step": 15364
    },
    {
      "epoch": 3.7884615384615383,
      "grad_norm": 0.30078125,
      "learning_rate": 3.370061876470403e-05,
      "loss": 0.0371,
      "step": 15366
    },
    {
      "epoch": 3.7889546351084813,
      "grad_norm": 0.3359375,
      "learning_rate": 3.367485155173039e-05,
      "loss": 0.0326,
      "step": 15368
    },
    {
      "epoch": 3.7894477317554243,
      "grad_norm": 0.27734375,
      "learning_rate": 3.36490921986571e-05,
      "loss": 0.0368,
      "step": 15370
    },
    {
      "epoch": 3.789940828402367,
      "grad_norm": 0.298828125,
      "learning_rate": 3.36233407085369e-05,
      "loss": 0.0352,
      "step": 15372
    },
    {
      "epoch": 3.7904339250493098,
      "grad_norm": 0.32421875,
      "learning_rate": 3.359759708442139e-05,
      "loss": 0.032,
      "step": 15374
    },
    {
      "epoch": 3.7909270216962527,
      "grad_norm": 0.34375,
      "learning_rate": 3.3571861329361385e-05,
      "loss": 0.0365,
      "step": 15376
    },
    {
      "epoch": 3.7914201183431953,
      "grad_norm": 0.283203125,
      "learning_rate": 3.354613344640672e-05,
      "loss": 0.0326,
      "step": 15378
    },
    {
      "epoch": 3.7919132149901382,
      "grad_norm": 0.306640625,
      "learning_rate": 3.352041343860621e-05,
      "loss": 0.0345,
      "step": 15380
    },
    {
      "epoch": 3.7924063116370808,
      "grad_norm": 0.3046875,
      "learning_rate": 3.3494701309007945e-05,
      "loss": 0.0346,
      "step": 15382
    },
    {
      "epoch": 3.7928994082840237,
      "grad_norm": 0.30078125,
      "learning_rate": 3.3468997060658834e-05,
      "loss": 0.0315,
      "step": 15384
    },
    {
      "epoch": 3.7933925049309662,
      "grad_norm": 0.296875,
      "learning_rate": 3.34433006966051e-05,
      "loss": 0.0327,
      "step": 15386
    },
    {
      "epoch": 3.793885601577909,
      "grad_norm": 0.36328125,
      "learning_rate": 3.341761221989179e-05,
      "loss": 0.0339,
      "step": 15388
    },
    {
      "epoch": 3.794378698224852,
      "grad_norm": 0.333984375,
      "learning_rate": 3.339193163356319e-05,
      "loss": 0.0324,
      "step": 15390
    },
    {
      "epoch": 3.7948717948717947,
      "grad_norm": 0.29296875,
      "learning_rate": 3.336625894066262e-05,
      "loss": 0.0336,
      "step": 15392
    },
    {
      "epoch": 3.7953648915187377,
      "grad_norm": 0.306640625,
      "learning_rate": 3.334059414423233e-05,
      "loss": 0.0345,
      "step": 15394
    },
    {
      "epoch": 3.7958579881656807,
      "grad_norm": 0.30859375,
      "learning_rate": 3.331493724731387e-05,
      "loss": 0.0335,
      "step": 15396
    },
    {
      "epoch": 3.796351084812623,
      "grad_norm": 0.3359375,
      "learning_rate": 3.328928825294761e-05,
      "loss": 0.0337,
      "step": 15398
    },
    {
      "epoch": 3.796844181459566,
      "grad_norm": 0.30078125,
      "learning_rate": 3.3263647164173227e-05,
      "loss": 0.0353,
      "step": 15400
    },
    {
      "epoch": 3.797337278106509,
      "grad_norm": 0.30078125,
      "learning_rate": 3.323801398402923e-05,
      "loss": 0.0323,
      "step": 15402
    },
    {
      "epoch": 3.7978303747534516,
      "grad_norm": 0.314453125,
      "learning_rate": 3.321238871555332e-05,
      "loss": 0.0327,
      "step": 15404
    },
    {
      "epoch": 3.7983234714003946,
      "grad_norm": 0.298828125,
      "learning_rate": 3.318677136178228e-05,
      "loss": 0.0339,
      "step": 15406
    },
    {
      "epoch": 3.798816568047337,
      "grad_norm": 0.265625,
      "learning_rate": 3.31611619257518e-05,
      "loss": 0.0295,
      "step": 15408
    },
    {
      "epoch": 3.79930966469428,
      "grad_norm": 0.365234375,
      "learning_rate": 3.313556041049689e-05,
      "loss": 0.0339,
      "step": 15410
    },
    {
      "epoch": 3.7998027613412226,
      "grad_norm": 0.267578125,
      "learning_rate": 3.310996681905132e-05,
      "loss": 0.031,
      "step": 15412
    },
    {
      "epoch": 3.8002958579881656,
      "grad_norm": 0.32421875,
      "learning_rate": 3.308438115444821e-05,
      "loss": 0.0351,
      "step": 15414
    },
    {
      "epoch": 3.8007889546351086,
      "grad_norm": 0.39453125,
      "learning_rate": 3.305880341971951e-05,
      "loss": 0.0358,
      "step": 15416
    },
    {
      "epoch": 3.801282051282051,
      "grad_norm": 0.337890625,
      "learning_rate": 3.3033233617896364e-05,
      "loss": 0.0361,
      "step": 15418
    },
    {
      "epoch": 3.801775147928994,
      "grad_norm": 0.296875,
      "learning_rate": 3.300767175200893e-05,
      "loss": 0.0334,
      "step": 15420
    },
    {
      "epoch": 3.802268244575937,
      "grad_norm": 0.296875,
      "learning_rate": 3.298211782508642e-05,
      "loss": 0.0327,
      "step": 15422
    },
    {
      "epoch": 3.8027613412228796,
      "grad_norm": 0.294921875,
      "learning_rate": 3.295657184015717e-05,
      "loss": 0.0332,
      "step": 15424
    },
    {
      "epoch": 3.8032544378698225,
      "grad_norm": 0.28125,
      "learning_rate": 3.29310338002484e-05,
      "loss": 0.0376,
      "step": 15426
    },
    {
      "epoch": 3.8037475345167655,
      "grad_norm": 0.31640625,
      "learning_rate": 3.290550370838668e-05,
      "loss": 0.0326,
      "step": 15428
    },
    {
      "epoch": 3.804240631163708,
      "grad_norm": 0.2890625,
      "learning_rate": 3.287998156759732e-05,
      "loss": 0.0361,
      "step": 15430
    },
    {
      "epoch": 3.804733727810651,
      "grad_norm": 0.291015625,
      "learning_rate": 3.285446738090491e-05,
      "loss": 0.032,
      "step": 15432
    },
    {
      "epoch": 3.8052268244575935,
      "grad_norm": 0.28125,
      "learning_rate": 3.282896115133301e-05,
      "loss": 0.0338,
      "step": 15434
    },
    {
      "epoch": 3.8057199211045365,
      "grad_norm": 0.345703125,
      "learning_rate": 3.280346288190423e-05,
      "loss": 0.0386,
      "step": 15436
    },
    {
      "epoch": 3.806213017751479,
      "grad_norm": 0.361328125,
      "learning_rate": 3.2777972575640325e-05,
      "loss": 0.0375,
      "step": 15438
    },
    {
      "epoch": 3.806706114398422,
      "grad_norm": 0.310546875,
      "learning_rate": 3.275249023556192e-05,
      "loss": 0.0369,
      "step": 15440
    },
    {
      "epoch": 3.807199211045365,
      "grad_norm": 0.314453125,
      "learning_rate": 3.272701586468895e-05,
      "loss": 0.0337,
      "step": 15442
    },
    {
      "epoch": 3.8076923076923075,
      "grad_norm": 0.34765625,
      "learning_rate": 3.270154946604016e-05,
      "loss": 0.0355,
      "step": 15444
    },
    {
      "epoch": 3.8081854043392505,
      "grad_norm": 0.375,
      "learning_rate": 3.267609104263352e-05,
      "loss": 0.0312,
      "step": 15446
    },
    {
      "epoch": 3.8086785009861934,
      "grad_norm": 0.271484375,
      "learning_rate": 3.2650640597485973e-05,
      "loss": 0.0324,
      "step": 15448
    },
    {
      "epoch": 3.809171597633136,
      "grad_norm": 0.322265625,
      "learning_rate": 3.2625198133613554e-05,
      "loss": 0.0346,
      "step": 15450
    },
    {
      "epoch": 3.809664694280079,
      "grad_norm": 0.349609375,
      "learning_rate": 3.2599763654031325e-05,
      "loss": 0.0363,
      "step": 15452
    },
    {
      "epoch": 3.810157790927022,
      "grad_norm": 0.29296875,
      "learning_rate": 3.257433716175341e-05,
      "loss": 0.0339,
      "step": 15454
    },
    {
      "epoch": 3.8106508875739644,
      "grad_norm": 0.3046875,
      "learning_rate": 3.254891865979304e-05,
      "loss": 0.0346,
      "step": 15456
    },
    {
      "epoch": 3.8111439842209074,
      "grad_norm": 0.3203125,
      "learning_rate": 3.252350815116238e-05,
      "loss": 0.0325,
      "step": 15458
    },
    {
      "epoch": 3.8116370808678504,
      "grad_norm": 0.390625,
      "learning_rate": 3.2498105638872724e-05,
      "loss": 0.0337,
      "step": 15460
    },
    {
      "epoch": 3.812130177514793,
      "grad_norm": 0.326171875,
      "learning_rate": 3.247271112593444e-05,
      "loss": 0.0327,
      "step": 15462
    },
    {
      "epoch": 3.812623274161736,
      "grad_norm": 0.2890625,
      "learning_rate": 3.244732461535692e-05,
      "loss": 0.0346,
      "step": 15464
    },
    {
      "epoch": 3.8131163708086784,
      "grad_norm": 0.328125,
      "learning_rate": 3.242194611014859e-05,
      "loss": 0.033,
      "step": 15466
    },
    {
      "epoch": 3.8136094674556213,
      "grad_norm": 0.29296875,
      "learning_rate": 3.239657561331696e-05,
      "loss": 0.0328,
      "step": 15468
    },
    {
      "epoch": 3.814102564102564,
      "grad_norm": 0.34375,
      "learning_rate": 3.237121312786857e-05,
      "loss": 0.0348,
      "step": 15470
    },
    {
      "epoch": 3.814595660749507,
      "grad_norm": 0.435546875,
      "learning_rate": 3.234585865680905e-05,
      "loss": 0.0352,
      "step": 15472
    },
    {
      "epoch": 3.81508875739645,
      "grad_norm": 0.3359375,
      "learning_rate": 3.232051220314297e-05,
      "loss": 0.0357,
      "step": 15474
    },
    {
      "epoch": 3.8155818540433923,
      "grad_norm": 0.267578125,
      "learning_rate": 3.229517376987407e-05,
      "loss": 0.0334,
      "step": 15476
    },
    {
      "epoch": 3.8160749506903353,
      "grad_norm": 0.291015625,
      "learning_rate": 3.2269843360005094e-05,
      "loss": 0.0368,
      "step": 15478
    },
    {
      "epoch": 3.8165680473372783,
      "grad_norm": 0.298828125,
      "learning_rate": 3.224452097653784e-05,
      "loss": 0.0338,
      "step": 15480
    },
    {
      "epoch": 3.817061143984221,
      "grad_norm": 0.271484375,
      "learning_rate": 3.2219206622473144e-05,
      "loss": 0.0339,
      "step": 15482
    },
    {
      "epoch": 3.8175542406311638,
      "grad_norm": 0.3203125,
      "learning_rate": 3.219390030081091e-05,
      "loss": 0.035,
      "step": 15484
    },
    {
      "epoch": 3.8180473372781067,
      "grad_norm": 0.330078125,
      "learning_rate": 3.2168602014550096e-05,
      "loss": 0.0356,
      "step": 15486
    },
    {
      "epoch": 3.8185404339250493,
      "grad_norm": 0.439453125,
      "learning_rate": 3.214331176668863e-05,
      "loss": 0.0342,
      "step": 15488
    },
    {
      "epoch": 3.8190335305719922,
      "grad_norm": 0.32421875,
      "learning_rate": 3.211802956022359e-05,
      "loss": 0.0365,
      "step": 15490
    },
    {
      "epoch": 3.8195266272189348,
      "grad_norm": 0.2890625,
      "learning_rate": 3.209275539815105e-05,
      "loss": 0.032,
      "step": 15492
    },
    {
      "epoch": 3.8200197238658777,
      "grad_norm": 0.3828125,
      "learning_rate": 3.206748928346614e-05,
      "loss": 0.0329,
      "step": 15494
    },
    {
      "epoch": 3.8205128205128203,
      "grad_norm": 0.30078125,
      "learning_rate": 3.204223121916304e-05,
      "loss": 0.0325,
      "step": 15496
    },
    {
      "epoch": 3.8210059171597632,
      "grad_norm": 0.296875,
      "learning_rate": 3.201698120823496e-05,
      "loss": 0.0362,
      "step": 15498
    },
    {
      "epoch": 3.821499013806706,
      "grad_norm": 0.29296875,
      "learning_rate": 3.199173925367421e-05,
      "loss": 0.0297,
      "step": 15500
    },
    {
      "epoch": 3.8219921104536487,
      "grad_norm": 0.291015625,
      "learning_rate": 3.196650535847201e-05,
      "loss": 0.036,
      "step": 15502
    },
    {
      "epoch": 3.8224852071005917,
      "grad_norm": 0.35546875,
      "learning_rate": 3.1941279525618836e-05,
      "loss": 0.0358,
      "step": 15504
    },
    {
      "epoch": 3.8229783037475347,
      "grad_norm": 0.2890625,
      "learning_rate": 3.1916061758104e-05,
      "loss": 0.0334,
      "step": 15506
    },
    {
      "epoch": 3.823471400394477,
      "grad_norm": 0.2890625,
      "learning_rate": 3.189085205891598e-05,
      "loss": 0.0349,
      "step": 15508
    },
    {
      "epoch": 3.82396449704142,
      "grad_norm": 0.306640625,
      "learning_rate": 3.186565043104226e-05,
      "loss": 0.0356,
      "step": 15510
    },
    {
      "epoch": 3.824457593688363,
      "grad_norm": 0.30859375,
      "learning_rate": 3.184045687746937e-05,
      "loss": 0.0378,
      "step": 15512
    },
    {
      "epoch": 3.8249506903353057,
      "grad_norm": 0.32421875,
      "learning_rate": 3.181527140118293e-05,
      "loss": 0.037,
      "step": 15514
    },
    {
      "epoch": 3.8254437869822486,
      "grad_norm": 0.37109375,
      "learning_rate": 3.179009400516747e-05,
      "loss": 0.0339,
      "step": 15516
    },
    {
      "epoch": 3.8259368836291916,
      "grad_norm": 0.298828125,
      "learning_rate": 3.176492469240676e-05,
      "loss": 0.0345,
      "step": 15518
    },
    {
      "epoch": 3.826429980276134,
      "grad_norm": 0.2890625,
      "learning_rate": 3.173976346588342e-05,
      "loss": 0.0347,
      "step": 15520
    },
    {
      "epoch": 3.8269230769230766,
      "grad_norm": 0.314453125,
      "learning_rate": 3.171461032857923e-05,
      "loss": 0.0329,
      "step": 15522
    },
    {
      "epoch": 3.8274161735700196,
      "grad_norm": 0.310546875,
      "learning_rate": 3.1689465283474964e-05,
      "loss": 0.0326,
      "step": 15524
    },
    {
      "epoch": 3.8279092702169626,
      "grad_norm": 0.31640625,
      "learning_rate": 3.1664328333550466e-05,
      "loss": 0.032,
      "step": 15526
    },
    {
      "epoch": 3.828402366863905,
      "grad_norm": 0.2734375,
      "learning_rate": 3.163919948178462e-05,
      "loss": 0.0324,
      "step": 15528
    },
    {
      "epoch": 3.828895463510848,
      "grad_norm": 0.361328125,
      "learning_rate": 3.161407873115526e-05,
      "loss": 0.0334,
      "step": 15530
    },
    {
      "epoch": 3.829388560157791,
      "grad_norm": 0.279296875,
      "learning_rate": 3.158896608463945e-05,
      "loss": 0.0332,
      "step": 15532
    },
    {
      "epoch": 3.8298816568047336,
      "grad_norm": 0.302734375,
      "learning_rate": 3.1563861545213056e-05,
      "loss": 0.0323,
      "step": 15534
    },
    {
      "epoch": 3.8303747534516766,
      "grad_norm": 0.294921875,
      "learning_rate": 3.153876511585122e-05,
      "loss": 0.0328,
      "step": 15536
    },
    {
      "epoch": 3.8308678500986195,
      "grad_norm": 0.32421875,
      "learning_rate": 3.1513676799527935e-05,
      "loss": 0.0329,
      "step": 15538
    },
    {
      "epoch": 3.831360946745562,
      "grad_norm": 0.27734375,
      "learning_rate": 3.1488596599216326e-05,
      "loss": 0.0358,
      "step": 15540
    },
    {
      "epoch": 3.831854043392505,
      "grad_norm": 0.296875,
      "learning_rate": 3.1463524517888575e-05,
      "loss": 0.0325,
      "step": 15542
    },
    {
      "epoch": 3.832347140039448,
      "grad_norm": 0.3515625,
      "learning_rate": 3.1438460558515756e-05,
      "loss": 0.0361,
      "step": 15544
    },
    {
      "epoch": 3.8328402366863905,
      "grad_norm": 0.306640625,
      "learning_rate": 3.141340472406823e-05,
      "loss": 0.0309,
      "step": 15546
    },
    {
      "epoch": 3.8333333333333335,
      "grad_norm": 0.310546875,
      "learning_rate": 3.138835701751511e-05,
      "loss": 0.0364,
      "step": 15548
    },
    {
      "epoch": 3.833826429980276,
      "grad_norm": 0.302734375,
      "learning_rate": 3.1363317441824825e-05,
      "loss": 0.037,
      "step": 15550
    },
    {
      "epoch": 3.834319526627219,
      "grad_norm": 0.318359375,
      "learning_rate": 3.133828599996462e-05,
      "loss": 0.0338,
      "step": 15552
    },
    {
      "epoch": 3.8348126232741615,
      "grad_norm": 0.318359375,
      "learning_rate": 3.131326269490088e-05,
      "loss": 0.0346,
      "step": 15554
    },
    {
      "epoch": 3.8353057199211045,
      "grad_norm": 0.2890625,
      "learning_rate": 3.1288247529599034e-05,
      "loss": 0.0365,
      "step": 15556
    },
    {
      "epoch": 3.8357988165680474,
      "grad_norm": 0.328125,
      "learning_rate": 3.126324050702343e-05,
      "loss": 0.0334,
      "step": 15558
    },
    {
      "epoch": 3.83629191321499,
      "grad_norm": 0.3203125,
      "learning_rate": 3.1238241630137666e-05,
      "loss": 0.0368,
      "step": 15560
    },
    {
      "epoch": 3.836785009861933,
      "grad_norm": 0.3359375,
      "learning_rate": 3.1213250901904126e-05,
      "loss": 0.0359,
      "step": 15562
    },
    {
      "epoch": 3.837278106508876,
      "grad_norm": 0.271484375,
      "learning_rate": 3.118826832528448e-05,
      "loss": 0.0352,
      "step": 15564
    },
    {
      "epoch": 3.8377712031558184,
      "grad_norm": 0.361328125,
      "learning_rate": 3.116329390323919e-05,
      "loss": 0.0343,
      "step": 15566
    },
    {
      "epoch": 3.8382642998027614,
      "grad_norm": 0.328125,
      "learning_rate": 3.11383276387279e-05,
      "loss": 0.039,
      "step": 15568
    },
    {
      "epoch": 3.8387573964497044,
      "grad_norm": 0.330078125,
      "learning_rate": 3.11133695347093e-05,
      "loss": 0.0349,
      "step": 15570
    },
    {
      "epoch": 3.839250493096647,
      "grad_norm": 0.31640625,
      "learning_rate": 3.108841959414097e-05,
      "loss": 0.0347,
      "step": 15572
    },
    {
      "epoch": 3.83974358974359,
      "grad_norm": 0.361328125,
      "learning_rate": 3.106347781997973e-05,
      "loss": 0.0343,
      "step": 15574
    },
    {
      "epoch": 3.8402366863905324,
      "grad_norm": 0.3359375,
      "learning_rate": 3.103854421518119e-05,
      "loss": 0.0338,
      "step": 15576
    },
    {
      "epoch": 3.8407297830374754,
      "grad_norm": 0.31640625,
      "learning_rate": 3.101361878270026e-05,
      "loss": 0.0333,
      "step": 15578
    },
    {
      "epoch": 3.841222879684418,
      "grad_norm": 0.32421875,
      "learning_rate": 3.0988701525490634e-05,
      "loss": 0.0383,
      "step": 15580
    },
    {
      "epoch": 3.841715976331361,
      "grad_norm": 0.2890625,
      "learning_rate": 3.09637924465052e-05,
      "loss": 0.0335,
      "step": 15582
    },
    {
      "epoch": 3.842209072978304,
      "grad_norm": 0.328125,
      "learning_rate": 3.0938891548695805e-05,
      "loss": 0.0351,
      "step": 15584
    },
    {
      "epoch": 3.8427021696252464,
      "grad_norm": 0.33203125,
      "learning_rate": 3.091399883501335e-05,
      "loss": 0.0372,
      "step": 15586
    },
    {
      "epoch": 3.8431952662721893,
      "grad_norm": 0.369140625,
      "learning_rate": 3.0889114308407786e-05,
      "loss": 0.0351,
      "step": 15588
    },
    {
      "epoch": 3.8436883629191323,
      "grad_norm": 0.298828125,
      "learning_rate": 3.0864237971827995e-05,
      "loss": 0.0353,
      "step": 15590
    },
    {
      "epoch": 3.844181459566075,
      "grad_norm": 0.298828125,
      "learning_rate": 3.0839369828222057e-05,
      "loss": 0.0318,
      "step": 15592
    },
    {
      "epoch": 3.844674556213018,
      "grad_norm": 0.27734375,
      "learning_rate": 3.08145098805369e-05,
      "loss": 0.0348,
      "step": 15594
    },
    {
      "epoch": 3.8451676528599608,
      "grad_norm": 0.275390625,
      "learning_rate": 3.0789658131718624e-05,
      "loss": 0.0337,
      "step": 15596
    },
    {
      "epoch": 3.8456607495069033,
      "grad_norm": 0.30078125,
      "learning_rate": 3.0764814584712263e-05,
      "loss": 0.0344,
      "step": 15598
    },
    {
      "epoch": 3.8461538461538463,
      "grad_norm": 0.28125,
      "learning_rate": 3.0739979242461944e-05,
      "loss": 0.0311,
      "step": 15600
    },
    {
      "epoch": 3.8466469428007892,
      "grad_norm": 0.34375,
      "learning_rate": 3.071515210791078e-05,
      "loss": 0.0364,
      "step": 15602
    },
    {
      "epoch": 3.8471400394477318,
      "grad_norm": 0.291015625,
      "learning_rate": 3.069033318400093e-05,
      "loss": 0.0333,
      "step": 15604
    },
    {
      "epoch": 3.8476331360946747,
      "grad_norm": 0.302734375,
      "learning_rate": 3.0665522473673604e-05,
      "loss": 0.0323,
      "step": 15606
    },
    {
      "epoch": 3.8481262327416172,
      "grad_norm": 0.34765625,
      "learning_rate": 3.064071997986895e-05,
      "loss": 0.0347,
      "step": 15608
    },
    {
      "epoch": 3.84861932938856,
      "grad_norm": 0.353515625,
      "learning_rate": 3.061592570552623e-05,
      "loss": 0.0309,
      "step": 15610
    },
    {
      "epoch": 3.8491124260355027,
      "grad_norm": 0.3203125,
      "learning_rate": 3.05911396535837e-05,
      "loss": 0.0336,
      "step": 15612
    },
    {
      "epoch": 3.8496055226824457,
      "grad_norm": 0.2890625,
      "learning_rate": 3.056636182697865e-05,
      "loss": 0.0318,
      "step": 15614
    },
    {
      "epoch": 3.8500986193293887,
      "grad_norm": 0.291015625,
      "learning_rate": 3.05415922286474e-05,
      "loss": 0.0316,
      "step": 15616
    },
    {
      "epoch": 3.850591715976331,
      "grad_norm": 0.25390625,
      "learning_rate": 3.051683086152527e-05,
      "loss": 0.0313,
      "step": 15618
    },
    {
      "epoch": 3.851084812623274,
      "grad_norm": 0.322265625,
      "learning_rate": 3.0492077728546664e-05,
      "loss": 0.034,
      "step": 15620
    },
    {
      "epoch": 3.851577909270217,
      "grad_norm": 0.326171875,
      "learning_rate": 3.046733283264489e-05,
      "loss": 0.0356,
      "step": 15622
    },
    {
      "epoch": 3.8520710059171597,
      "grad_norm": 0.31640625,
      "learning_rate": 3.0442596176752403e-05,
      "loss": 0.0344,
      "step": 15624
    },
    {
      "epoch": 3.8525641025641026,
      "grad_norm": 0.32421875,
      "learning_rate": 3.0417867763800632e-05,
      "loss": 0.0355,
      "step": 15626
    },
    {
      "epoch": 3.8530571992110456,
      "grad_norm": 0.265625,
      "learning_rate": 3.0393147596720018e-05,
      "loss": 0.0305,
      "step": 15628
    },
    {
      "epoch": 3.853550295857988,
      "grad_norm": 0.275390625,
      "learning_rate": 3.0368435678440056e-05,
      "loss": 0.0337,
      "step": 15630
    },
    {
      "epoch": 3.854043392504931,
      "grad_norm": 0.34375,
      "learning_rate": 3.0343732011889235e-05,
      "loss": 0.0357,
      "step": 15632
    },
    {
      "epoch": 3.8545364891518736,
      "grad_norm": 0.380859375,
      "learning_rate": 3.0319036599995077e-05,
      "loss": 0.0369,
      "step": 15634
    },
    {
      "epoch": 3.8550295857988166,
      "grad_norm": 0.306640625,
      "learning_rate": 3.0294349445684166e-05,
      "loss": 0.0355,
      "step": 15636
    },
    {
      "epoch": 3.855522682445759,
      "grad_norm": 0.314453125,
      "learning_rate": 3.0269670551882013e-05,
      "loss": 0.0371,
      "step": 15638
    },
    {
      "epoch": 3.856015779092702,
      "grad_norm": 0.271484375,
      "learning_rate": 3.0244999921513205e-05,
      "loss": 0.0383,
      "step": 15640
    },
    {
      "epoch": 3.856508875739645,
      "grad_norm": 0.361328125,
      "learning_rate": 3.0220337557501376e-05,
      "loss": 0.0342,
      "step": 15642
    },
    {
      "epoch": 3.8570019723865876,
      "grad_norm": 0.287109375,
      "learning_rate": 3.0195683462769152e-05,
      "loss": 0.0342,
      "step": 15644
    },
    {
      "epoch": 3.8574950690335306,
      "grad_norm": 0.35546875,
      "learning_rate": 3.0171037640238165e-05,
      "loss": 0.0368,
      "step": 15646
    },
    {
      "epoch": 3.8579881656804735,
      "grad_norm": 0.27734375,
      "learning_rate": 3.0146400092829095e-05,
      "loss": 0.0355,
      "step": 15648
    },
    {
      "epoch": 3.858481262327416,
      "grad_norm": 0.384765625,
      "learning_rate": 3.0121770823461658e-05,
      "loss": 0.0314,
      "step": 15650
    },
    {
      "epoch": 3.858974358974359,
      "grad_norm": 0.294921875,
      "learning_rate": 3.0097149835054506e-05,
      "loss": 0.0313,
      "step": 15652
    },
    {
      "epoch": 3.859467455621302,
      "grad_norm": 0.2578125,
      "learning_rate": 3.007253713052538e-05,
      "loss": 0.0314,
      "step": 15654
    },
    {
      "epoch": 3.8599605522682445,
      "grad_norm": 0.296875,
      "learning_rate": 3.0047932712791027e-05,
      "loss": 0.0326,
      "step": 15656
    },
    {
      "epoch": 3.8604536489151875,
      "grad_norm": 0.341796875,
      "learning_rate": 3.0023336584767203e-05,
      "loss": 0.0344,
      "step": 15658
    },
    {
      "epoch": 3.86094674556213,
      "grad_norm": 0.279296875,
      "learning_rate": 2.999874874936871e-05,
      "loss": 0.0336,
      "step": 15660
    },
    {
      "epoch": 3.861439842209073,
      "grad_norm": 0.3671875,
      "learning_rate": 2.9974169209509317e-05,
      "loss": 0.0319,
      "step": 15662
    },
    {
      "epoch": 3.8619329388560155,
      "grad_norm": 0.326171875,
      "learning_rate": 2.9949597968101884e-05,
      "loss": 0.0331,
      "step": 15664
    },
    {
      "epoch": 3.8624260355029585,
      "grad_norm": 0.33203125,
      "learning_rate": 2.9925035028058134e-05,
      "loss": 0.0317,
      "step": 15666
    },
    {
      "epoch": 3.8629191321499015,
      "grad_norm": 0.330078125,
      "learning_rate": 2.9900480392289044e-05,
      "loss": 0.0327,
      "step": 15668
    },
    {
      "epoch": 3.863412228796844,
      "grad_norm": 0.30078125,
      "learning_rate": 2.9875934063704382e-05,
      "loss": 0.0308,
      "step": 15670
    },
    {
      "epoch": 3.863905325443787,
      "grad_norm": 0.2578125,
      "learning_rate": 2.9851396045213066e-05,
      "loss": 0.0322,
      "step": 15672
    },
    {
      "epoch": 3.86439842209073,
      "grad_norm": 0.41796875,
      "learning_rate": 2.9826866339722958e-05,
      "loss": 0.0372,
      "step": 15674
    },
    {
      "epoch": 3.8648915187376724,
      "grad_norm": 0.302734375,
      "learning_rate": 2.9802344950141002e-05,
      "loss": 0.0325,
      "step": 15676
    },
    {
      "epoch": 3.8653846153846154,
      "grad_norm": 0.267578125,
      "learning_rate": 2.977783187937313e-05,
      "loss": 0.0341,
      "step": 15678
    },
    {
      "epoch": 3.8658777120315584,
      "grad_norm": 0.296875,
      "learning_rate": 2.9753327130324194e-05,
      "loss": 0.0353,
      "step": 15680
    },
    {
      "epoch": 3.866370808678501,
      "grad_norm": 0.3203125,
      "learning_rate": 2.972883070589826e-05,
      "loss": 0.0318,
      "step": 15682
    },
    {
      "epoch": 3.866863905325444,
      "grad_norm": 0.33984375,
      "learning_rate": 2.970434260899816e-05,
      "loss": 0.0346,
      "step": 15684
    },
    {
      "epoch": 3.867357001972387,
      "grad_norm": 0.30078125,
      "learning_rate": 2.9679862842526007e-05,
      "loss": 0.0319,
      "step": 15686
    },
    {
      "epoch": 3.8678500986193294,
      "grad_norm": 0.361328125,
      "learning_rate": 2.9655391409382705e-05,
      "loss": 0.0331,
      "step": 15688
    },
    {
      "epoch": 3.8683431952662723,
      "grad_norm": 0.306640625,
      "learning_rate": 2.9630928312468263e-05,
      "loss": 0.0348,
      "step": 15690
    },
    {
      "epoch": 3.868836291913215,
      "grad_norm": 0.283203125,
      "learning_rate": 2.9606473554681712e-05,
      "loss": 0.0274,
      "step": 15692
    },
    {
      "epoch": 3.869329388560158,
      "grad_norm": 0.345703125,
      "learning_rate": 2.9582027138921063e-05,
      "loss": 0.0379,
      "step": 15694
    },
    {
      "epoch": 3.8698224852071004,
      "grad_norm": 0.32421875,
      "learning_rate": 2.9557589068083403e-05,
      "loss": 0.0348,
      "step": 15696
    },
    {
      "epoch": 3.8703155818540433,
      "grad_norm": 0.265625,
      "learning_rate": 2.9533159345064677e-05,
      "loss": 0.029,
      "step": 15698
    },
    {
      "epoch": 3.8708086785009863,
      "grad_norm": 0.291015625,
      "learning_rate": 2.950873797276006e-05,
      "loss": 0.0338,
      "step": 15700
    },
    {
      "epoch": 3.871301775147929,
      "grad_norm": 0.39453125,
      "learning_rate": 2.9484324954063536e-05,
      "loss": 0.0347,
      "step": 15702
    },
    {
      "epoch": 3.871794871794872,
      "grad_norm": 0.306640625,
      "learning_rate": 2.9459920291868215e-05,
      "loss": 0.0297,
      "step": 15704
    },
    {
      "epoch": 3.8722879684418148,
      "grad_norm": 0.333984375,
      "learning_rate": 2.9435523989066182e-05,
      "loss": 0.0305,
      "step": 15706
    },
    {
      "epoch": 3.8727810650887573,
      "grad_norm": 0.271484375,
      "learning_rate": 2.941113604854855e-05,
      "loss": 0.0316,
      "step": 15708
    },
    {
      "epoch": 3.8732741617357003,
      "grad_norm": 0.298828125,
      "learning_rate": 2.9386756473205433e-05,
      "loss": 0.0335,
      "step": 15710
    },
    {
      "epoch": 3.8737672583826432,
      "grad_norm": 0.33984375,
      "learning_rate": 2.9362385265925873e-05,
      "loss": 0.0349,
      "step": 15712
    },
    {
      "epoch": 3.8742603550295858,
      "grad_norm": 0.337890625,
      "learning_rate": 2.93380224295981e-05,
      "loss": 0.0338,
      "step": 15714
    },
    {
      "epoch": 3.8747534516765287,
      "grad_norm": 0.267578125,
      "learning_rate": 2.931366796710917e-05,
      "loss": 0.0322,
      "step": 15716
    },
    {
      "epoch": 3.8752465483234713,
      "grad_norm": 0.30859375,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 0.0355,
      "step": 15718
    },
    {
      "epoch": 3.8757396449704142,
      "grad_norm": 0.33203125,
      "learning_rate": 2.926498417519149e-05,
      "loss": 0.0369,
      "step": 15720
    },
    {
      "epoch": 3.8762327416173568,
      "grad_norm": 0.341796875,
      "learning_rate": 2.9240654851532034e-05,
      "loss": 0.0323,
      "step": 15722
    },
    {
      "epoch": 3.8767258382642997,
      "grad_norm": 0.34375,
      "learning_rate": 2.9216333913250084e-05,
      "loss": 0.0339,
      "step": 15724
    },
    {
      "epoch": 3.8772189349112427,
      "grad_norm": 0.32421875,
      "learning_rate": 2.9192021363227696e-05,
      "loss": 0.0349,
      "step": 15726
    },
    {
      "epoch": 3.8777120315581852,
      "grad_norm": 0.3125,
      "learning_rate": 2.916771720434619e-05,
      "loss": 0.0326,
      "step": 15728
    },
    {
      "epoch": 3.878205128205128,
      "grad_norm": 0.29296875,
      "learning_rate": 2.914342143948563e-05,
      "loss": 0.0361,
      "step": 15730
    },
    {
      "epoch": 3.878698224852071,
      "grad_norm": 0.400390625,
      "learning_rate": 2.9119134071525254e-05,
      "loss": 0.0361,
      "step": 15732
    },
    {
      "epoch": 3.8791913214990137,
      "grad_norm": 0.30859375,
      "learning_rate": 2.9094855103343232e-05,
      "loss": 0.0324,
      "step": 15734
    },
    {
      "epoch": 3.8796844181459567,
      "grad_norm": 0.3125,
      "learning_rate": 2.907058453781677e-05,
      "loss": 0.0341,
      "step": 15736
    },
    {
      "epoch": 3.8801775147928996,
      "grad_norm": 0.318359375,
      "learning_rate": 2.904632237782209e-05,
      "loss": 0.0335,
      "step": 15738
    },
    {
      "epoch": 3.880670611439842,
      "grad_norm": 0.388671875,
      "learning_rate": 2.9022068626234287e-05,
      "loss": 0.0325,
      "step": 15740
    },
    {
      "epoch": 3.881163708086785,
      "grad_norm": 0.296875,
      "learning_rate": 2.8997823285927716e-05,
      "loss": 0.0327,
      "step": 15742
    },
    {
      "epoch": 3.8816568047337277,
      "grad_norm": 0.31640625,
      "learning_rate": 2.8973586359775474e-05,
      "loss": 0.03,
      "step": 15744
    },
    {
      "epoch": 3.8821499013806706,
      "grad_norm": 0.283203125,
      "learning_rate": 2.8949357850649815e-05,
      "loss": 0.0322,
      "step": 15746
    },
    {
      "epoch": 3.882642998027613,
      "grad_norm": 0.36328125,
      "learning_rate": 2.892513776142195e-05,
      "loss": 0.0313,
      "step": 15748
    },
    {
      "epoch": 3.883136094674556,
      "grad_norm": 0.296875,
      "learning_rate": 2.8900926094962088e-05,
      "loss": 0.0369,
      "step": 15750
    },
    {
      "epoch": 3.883629191321499,
      "grad_norm": 0.306640625,
      "learning_rate": 2.887672285413947e-05,
      "loss": 0.0331,
      "step": 15752
    },
    {
      "epoch": 3.8841222879684416,
      "grad_norm": 0.3125,
      "learning_rate": 2.8852528041822247e-05,
      "loss": 0.0298,
      "step": 15754
    },
    {
      "epoch": 3.8846153846153846,
      "grad_norm": 0.36328125,
      "learning_rate": 2.8828341660877744e-05,
      "loss": 0.0342,
      "step": 15756
    },
    {
      "epoch": 3.8851084812623276,
      "grad_norm": 0.31640625,
      "learning_rate": 2.8804163714172062e-05,
      "loss": 0.0378,
      "step": 15758
    },
    {
      "epoch": 3.88560157790927,
      "grad_norm": 0.3046875,
      "learning_rate": 2.877999420457056e-05,
      "loss": 0.0331,
      "step": 15760
    },
    {
      "epoch": 3.886094674556213,
      "grad_norm": 0.291015625,
      "learning_rate": 2.8755833134937336e-05,
      "loss": 0.0325,
      "step": 15762
    },
    {
      "epoch": 3.886587771203156,
      "grad_norm": 0.30859375,
      "learning_rate": 2.8731680508135682e-05,
      "loss": 0.0301,
      "step": 15764
    },
    {
      "epoch": 3.8870808678500985,
      "grad_norm": 0.3359375,
      "learning_rate": 2.8707536327027784e-05,
      "loss": 0.0384,
      "step": 15766
    },
    {
      "epoch": 3.8875739644970415,
      "grad_norm": 0.3515625,
      "learning_rate": 2.8683400594474886e-05,
      "loss": 0.0357,
      "step": 15768
    },
    {
      "epoch": 3.8880670611439845,
      "grad_norm": 0.390625,
      "learning_rate": 2.865927331333723e-05,
      "loss": 0.0312,
      "step": 15770
    },
    {
      "epoch": 3.888560157790927,
      "grad_norm": 0.330078125,
      "learning_rate": 2.8635154486473936e-05,
      "loss": 0.0351,
      "step": 15772
    },
    {
      "epoch": 3.88905325443787,
      "grad_norm": 0.35546875,
      "learning_rate": 2.8611044116743347e-05,
      "loss": 0.0313,
      "step": 15774
    },
    {
      "epoch": 3.8895463510848125,
      "grad_norm": 0.296875,
      "learning_rate": 2.8586942207002588e-05,
      "loss": 0.0339,
      "step": 15776
    },
    {
      "epoch": 3.8900394477317555,
      "grad_norm": 0.3671875,
      "learning_rate": 2.8562848760107897e-05,
      "loss": 0.0353,
      "step": 15778
    },
    {
      "epoch": 3.890532544378698,
      "grad_norm": 0.291015625,
      "learning_rate": 2.853876377891448e-05,
      "loss": 0.0355,
      "step": 15780
    },
    {
      "epoch": 3.891025641025641,
      "grad_norm": 0.296875,
      "learning_rate": 2.851468726627654e-05,
      "loss": 0.0328,
      "step": 15782
    },
    {
      "epoch": 3.891518737672584,
      "grad_norm": 0.3046875,
      "learning_rate": 2.849061922504729e-05,
      "loss": 0.0344,
      "step": 15784
    },
    {
      "epoch": 3.8920118343195265,
      "grad_norm": 0.40625,
      "learning_rate": 2.8466559658078905e-05,
      "loss": 0.0371,
      "step": 15786
    },
    {
      "epoch": 3.8925049309664694,
      "grad_norm": 0.30078125,
      "learning_rate": 2.8442508568222636e-05,
      "loss": 0.0342,
      "step": 15788
    },
    {
      "epoch": 3.8929980276134124,
      "grad_norm": 0.302734375,
      "learning_rate": 2.8418465958328587e-05,
      "loss": 0.0346,
      "step": 15790
    },
    {
      "epoch": 3.893491124260355,
      "grad_norm": 0.31640625,
      "learning_rate": 2.8394431831245993e-05,
      "loss": 0.0318,
      "step": 15792
    },
    {
      "epoch": 3.893984220907298,
      "grad_norm": 0.267578125,
      "learning_rate": 2.837040618982302e-05,
      "loss": 0.0338,
      "step": 15794
    },
    {
      "epoch": 3.894477317554241,
      "grad_norm": 0.37890625,
      "learning_rate": 2.8346389036906828e-05,
      "loss": 0.0372,
      "step": 15796
    },
    {
      "epoch": 3.8949704142011834,
      "grad_norm": 0.306640625,
      "learning_rate": 2.8322380375343616e-05,
      "loss": 0.0315,
      "step": 15798
    },
    {
      "epoch": 3.8954635108481264,
      "grad_norm": 0.314453125,
      "learning_rate": 2.8298380207978513e-05,
      "loss": 0.0347,
      "step": 15800
    },
    {
      "epoch": 3.895956607495069,
      "grad_norm": 0.2734375,
      "learning_rate": 2.8274388537655727e-05,
      "loss": 0.0332,
      "step": 15802
    },
    {
      "epoch": 3.896449704142012,
      "grad_norm": 0.330078125,
      "learning_rate": 2.8250405367218324e-05,
      "loss": 0.0347,
      "step": 15804
    },
    {
      "epoch": 3.8969428007889544,
      "grad_norm": 0.2890625,
      "learning_rate": 2.822643069950849e-05,
      "loss": 0.0315,
      "step": 15806
    },
    {
      "epoch": 3.8974358974358974,
      "grad_norm": 0.3046875,
      "learning_rate": 2.8202464537367346e-05,
      "loss": 0.031,
      "step": 15808
    },
    {
      "epoch": 3.8979289940828403,
      "grad_norm": 0.333984375,
      "learning_rate": 2.8178506883635024e-05,
      "loss": 0.0336,
      "step": 15810
    },
    {
      "epoch": 3.898422090729783,
      "grad_norm": 0.27734375,
      "learning_rate": 2.8154557741150633e-05,
      "loss": 0.0335,
      "step": 15812
    },
    {
      "epoch": 3.898915187376726,
      "grad_norm": 0.32421875,
      "learning_rate": 2.8130617112752277e-05,
      "loss": 0.0313,
      "step": 15814
    },
    {
      "epoch": 3.899408284023669,
      "grad_norm": 0.330078125,
      "learning_rate": 2.8106685001277077e-05,
      "loss": 0.0353,
      "step": 15816
    },
    {
      "epoch": 3.8999013806706113,
      "grad_norm": 0.322265625,
      "learning_rate": 2.808276140956112e-05,
      "loss": 0.0339,
      "step": 15818
    },
    {
      "epoch": 3.9003944773175543,
      "grad_norm": 0.35546875,
      "learning_rate": 2.8058846340439447e-05,
      "loss": 0.037,
      "step": 15820
    },
    {
      "epoch": 3.9008875739644973,
      "grad_norm": 0.283203125,
      "learning_rate": 2.803493979674615e-05,
      "loss": 0.0291,
      "step": 15822
    },
    {
      "epoch": 3.90138067061144,
      "grad_norm": 0.265625,
      "learning_rate": 2.8011041781314285e-05,
      "loss": 0.0316,
      "step": 15824
    },
    {
      "epoch": 3.9018737672583828,
      "grad_norm": 0.376953125,
      "learning_rate": 2.7987152296975915e-05,
      "loss": 0.0325,
      "step": 15826
    },
    {
      "epoch": 3.9023668639053253,
      "grad_norm": 0.32421875,
      "learning_rate": 2.7963271346562058e-05,
      "loss": 0.0276,
      "step": 15828
    },
    {
      "epoch": 3.9028599605522682,
      "grad_norm": 0.30078125,
      "learning_rate": 2.793939893290275e-05,
      "loss": 0.0337,
      "step": 15830
    },
    {
      "epoch": 3.9033530571992108,
      "grad_norm": 0.3125,
      "learning_rate": 2.7915535058827024e-05,
      "loss": 0.0341,
      "step": 15832
    },
    {
      "epoch": 3.9038461538461537,
      "grad_norm": 0.359375,
      "learning_rate": 2.7891679727162845e-05,
      "loss": 0.0337,
      "step": 15834
    },
    {
      "epoch": 3.9043392504930967,
      "grad_norm": 0.361328125,
      "learning_rate": 2.7867832940737216e-05,
      "loss": 0.0403,
      "step": 15836
    },
    {
      "epoch": 3.9048323471400392,
      "grad_norm": 0.279296875,
      "learning_rate": 2.7843994702376108e-05,
      "loss": 0.0322,
      "step": 15838
    },
    {
      "epoch": 3.905325443786982,
      "grad_norm": 0.306640625,
      "learning_rate": 2.782016501490451e-05,
      "loss": 0.035,
      "step": 15840
    },
    {
      "epoch": 3.905818540433925,
      "grad_norm": 0.2890625,
      "learning_rate": 2.7796343881146347e-05,
      "loss": 0.0339,
      "step": 15842
    },
    {
      "epoch": 3.9063116370808677,
      "grad_norm": 0.30078125,
      "learning_rate": 2.7772531303924576e-05,
      "loss": 0.0339,
      "step": 15844
    },
    {
      "epoch": 3.9068047337278107,
      "grad_norm": 0.333984375,
      "learning_rate": 2.7748727286061137e-05,
      "loss": 0.0347,
      "step": 15846
    },
    {
      "epoch": 3.9072978303747536,
      "grad_norm": 0.30078125,
      "learning_rate": 2.7724931830376855e-05,
      "loss": 0.0353,
      "step": 15848
    },
    {
      "epoch": 3.907790927021696,
      "grad_norm": 0.314453125,
      "learning_rate": 2.7701144939691747e-05,
      "loss": 0.0344,
      "step": 15850
    },
    {
      "epoch": 3.908284023668639,
      "grad_norm": 0.306640625,
      "learning_rate": 2.7677366616824595e-05,
      "loss": 0.0306,
      "step": 15852
    },
    {
      "epoch": 3.908777120315582,
      "grad_norm": 0.287109375,
      "learning_rate": 2.765359686459329e-05,
      "loss": 0.0343,
      "step": 15854
    },
    {
      "epoch": 3.9092702169625246,
      "grad_norm": 0.296875,
      "learning_rate": 2.7629835685814697e-05,
      "loss": 0.0334,
      "step": 15856
    },
    {
      "epoch": 3.9097633136094676,
      "grad_norm": 0.357421875,
      "learning_rate": 2.7606083083304633e-05,
      "loss": 0.0367,
      "step": 15858
    },
    {
      "epoch": 3.91025641025641,
      "grad_norm": 0.314453125,
      "learning_rate": 2.758233905987796e-05,
      "loss": 0.0339,
      "step": 15860
    },
    {
      "epoch": 3.910749506903353,
      "grad_norm": 0.279296875,
      "learning_rate": 2.7558603618348365e-05,
      "loss": 0.0353,
      "step": 15862
    },
    {
      "epoch": 3.9112426035502956,
      "grad_norm": 0.263671875,
      "learning_rate": 2.7534876761528773e-05,
      "loss": 0.0329,
      "step": 15864
    },
    {
      "epoch": 3.9117357001972386,
      "grad_norm": 0.255859375,
      "learning_rate": 2.7511158492230805e-05,
      "loss": 0.0289,
      "step": 15866
    },
    {
      "epoch": 3.9122287968441816,
      "grad_norm": 0.2890625,
      "learning_rate": 2.7487448813265347e-05,
      "loss": 0.0345,
      "step": 15868
    },
    {
      "epoch": 3.912721893491124,
      "grad_norm": 0.30859375,
      "learning_rate": 2.7463747727442034e-05,
      "loss": 0.0313,
      "step": 15870
    },
    {
      "epoch": 3.913214990138067,
      "grad_norm": 0.283203125,
      "learning_rate": 2.7440055237569606e-05,
      "loss": 0.0329,
      "step": 15872
    },
    {
      "epoch": 3.91370808678501,
      "grad_norm": 0.275390625,
      "learning_rate": 2.7416371346455792e-05,
      "loss": 0.0321,
      "step": 15874
    },
    {
      "epoch": 3.9142011834319526,
      "grad_norm": 0.302734375,
      "learning_rate": 2.7392696056907162e-05,
      "loss": 0.0331,
      "step": 15876
    },
    {
      "epoch": 3.9146942800788955,
      "grad_norm": 0.2890625,
      "learning_rate": 2.7369029371729494e-05,
      "loss": 0.0338,
      "step": 15878
    },
    {
      "epoch": 3.9151873767258385,
      "grad_norm": 0.259765625,
      "learning_rate": 2.7345371293727307e-05,
      "loss": 0.0302,
      "step": 15880
    },
    {
      "epoch": 3.915680473372781,
      "grad_norm": 0.287109375,
      "learning_rate": 2.732172182570434e-05,
      "loss": 0.0323,
      "step": 15882
    },
    {
      "epoch": 3.916173570019724,
      "grad_norm": 0.310546875,
      "learning_rate": 2.729808097046308e-05,
      "loss": 0.0343,
      "step": 15884
    },
    {
      "epoch": 3.9166666666666665,
      "grad_norm": 0.306640625,
      "learning_rate": 2.7274448730805135e-05,
      "loss": 0.0307,
      "step": 15886
    },
    {
      "epoch": 3.9171597633136095,
      "grad_norm": 0.2734375,
      "learning_rate": 2.725082510953111e-05,
      "loss": 0.0328,
      "step": 15888
    },
    {
      "epoch": 3.917652859960552,
      "grad_norm": 0.28515625,
      "learning_rate": 2.7227210109440414e-05,
      "loss": 0.0348,
      "step": 15890
    },
    {
      "epoch": 3.918145956607495,
      "grad_norm": 0.26953125,
      "learning_rate": 2.7203603733331706e-05,
      "loss": 0.0316,
      "step": 15892
    },
    {
      "epoch": 3.918639053254438,
      "grad_norm": 0.26171875,
      "learning_rate": 2.7180005984002333e-05,
      "loss": 0.031,
      "step": 15894
    },
    {
      "epoch": 3.9191321499013805,
      "grad_norm": 0.3359375,
      "learning_rate": 2.715641686424888e-05,
      "loss": 0.0352,
      "step": 15896
    },
    {
      "epoch": 3.9196252465483234,
      "grad_norm": 0.31640625,
      "learning_rate": 2.713283637686671e-05,
      "loss": 0.0356,
      "step": 15898
    },
    {
      "epoch": 3.9201183431952664,
      "grad_norm": 0.28125,
      "learning_rate": 2.710926452465026e-05,
      "loss": 0.0327,
      "step": 15900
    },
    {
      "epoch": 3.920611439842209,
      "grad_norm": 0.279296875,
      "learning_rate": 2.7085701310392974e-05,
      "loss": 0.0313,
      "step": 15902
    },
    {
      "epoch": 3.921104536489152,
      "grad_norm": 0.369140625,
      "learning_rate": 2.7062146736887116e-05,
      "loss": 0.0322,
      "step": 15904
    },
    {
      "epoch": 3.921597633136095,
      "grad_norm": 0.287109375,
      "learning_rate": 2.7038600806924164e-05,
      "loss": 0.0345,
      "step": 15906
    },
    {
      "epoch": 3.9220907297830374,
      "grad_norm": 0.2890625,
      "learning_rate": 2.7015063523294326e-05,
      "loss": 0.0322,
      "step": 15908
    },
    {
      "epoch": 3.9225838264299804,
      "grad_norm": 0.330078125,
      "learning_rate": 2.6991534888787017e-05,
      "loss": 0.034,
      "step": 15910
    },
    {
      "epoch": 3.9230769230769234,
      "grad_norm": 0.322265625,
      "learning_rate": 2.6968014906190408e-05,
      "loss": 0.0302,
      "step": 15912
    },
    {
      "epoch": 3.923570019723866,
      "grad_norm": 0.30859375,
      "learning_rate": 2.6944503578291803e-05,
      "loss": 0.0334,
      "step": 15914
    },
    {
      "epoch": 3.9240631163708084,
      "grad_norm": 0.322265625,
      "learning_rate": 2.6921000907877413e-05,
      "loss": 0.0352,
      "step": 15916
    },
    {
      "epoch": 3.9245562130177514,
      "grad_norm": 0.33203125,
      "learning_rate": 2.6897506897732437e-05,
      "loss": 0.0381,
      "step": 15918
    },
    {
      "epoch": 3.9250493096646943,
      "grad_norm": 0.279296875,
      "learning_rate": 2.687402155064107e-05,
      "loss": 0.031,
      "step": 15920
    },
    {
      "epoch": 3.925542406311637,
      "grad_norm": 0.408203125,
      "learning_rate": 2.6850544869386374e-05,
      "loss": 0.0329,
      "step": 15922
    },
    {
      "epoch": 3.92603550295858,
      "grad_norm": 0.28515625,
      "learning_rate": 2.6827076856750587e-05,
      "loss": 0.0325,
      "step": 15924
    },
    {
      "epoch": 3.926528599605523,
      "grad_norm": 0.318359375,
      "learning_rate": 2.68036175155147e-05,
      "loss": 0.0354,
      "step": 15926
    },
    {
      "epoch": 3.9270216962524653,
      "grad_norm": 0.29296875,
      "learning_rate": 2.678016684845883e-05,
      "loss": 0.0343,
      "step": 15928
    },
    {
      "epoch": 3.9275147928994083,
      "grad_norm": 0.29296875,
      "learning_rate": 2.6756724858361982e-05,
      "loss": 0.031,
      "step": 15930
    },
    {
      "epoch": 3.9280078895463513,
      "grad_norm": 0.357421875,
      "learning_rate": 2.673329154800218e-05,
      "loss": 0.0348,
      "step": 15932
    },
    {
      "epoch": 3.928500986193294,
      "grad_norm": 0.314453125,
      "learning_rate": 2.6709866920156434e-05,
      "loss": 0.0337,
      "step": 15934
    },
    {
      "epoch": 3.9289940828402368,
      "grad_norm": 0.296875,
      "learning_rate": 2.6686450977600585e-05,
      "loss": 0.0354,
      "step": 15936
    },
    {
      "epoch": 3.9294871794871797,
      "grad_norm": 0.296875,
      "learning_rate": 2.6663043723109692e-05,
      "loss": 0.0305,
      "step": 15938
    },
    {
      "epoch": 3.9299802761341223,
      "grad_norm": 0.341796875,
      "learning_rate": 2.663964515945756e-05,
      "loss": 0.032,
      "step": 15940
    },
    {
      "epoch": 3.9304733727810652,
      "grad_norm": 0.271484375,
      "learning_rate": 2.6616255289417058e-05,
      "loss": 0.0334,
      "step": 15942
    },
    {
      "epoch": 3.9309664694280078,
      "grad_norm": 0.291015625,
      "learning_rate": 2.6592874115760025e-05,
      "loss": 0.0322,
      "step": 15944
    },
    {
      "epoch": 3.9314595660749507,
      "grad_norm": 0.353515625,
      "learning_rate": 2.6569501641257267e-05,
      "loss": 0.0382,
      "step": 15946
    },
    {
      "epoch": 3.9319526627218933,
      "grad_norm": 0.302734375,
      "learning_rate": 2.6546137868678543e-05,
      "loss": 0.0345,
      "step": 15948
    },
    {
      "epoch": 3.9324457593688362,
      "grad_norm": 0.294921875,
      "learning_rate": 2.65227828007926e-05,
      "loss": 0.033,
      "step": 15950
    },
    {
      "epoch": 3.932938856015779,
      "grad_norm": 0.2890625,
      "learning_rate": 2.6499436440367165e-05,
      "loss": 0.0341,
      "step": 15952
    },
    {
      "epoch": 3.9334319526627217,
      "grad_norm": 0.296875,
      "learning_rate": 2.6476098790168857e-05,
      "loss": 0.0328,
      "step": 15954
    },
    {
      "epoch": 3.9339250493096647,
      "grad_norm": 0.287109375,
      "learning_rate": 2.6452769852963344e-05,
      "loss": 0.034,
      "step": 15956
    },
    {
      "epoch": 3.9344181459566077,
      "grad_norm": 0.28125,
      "learning_rate": 2.6429449631515245e-05,
      "loss": 0.0342,
      "step": 15958
    },
    {
      "epoch": 3.93491124260355,
      "grad_norm": 0.298828125,
      "learning_rate": 2.640613812858813e-05,
      "loss": 0.0375,
      "step": 15960
    },
    {
      "epoch": 3.935404339250493,
      "grad_norm": 0.376953125,
      "learning_rate": 2.6382835346944546e-05,
      "loss": 0.0331,
      "step": 15962
    },
    {
      "epoch": 3.935897435897436,
      "grad_norm": 0.28515625,
      "learning_rate": 2.635954128934599e-05,
      "loss": 0.032,
      "step": 15964
    },
    {
      "epoch": 3.9363905325443787,
      "grad_norm": 0.37109375,
      "learning_rate": 2.633625595855298e-05,
      "loss": 0.0357,
      "step": 15966
    },
    {
      "epoch": 3.9368836291913216,
      "grad_norm": 0.2734375,
      "learning_rate": 2.6312979357324897e-05,
      "loss": 0.033,
      "step": 15968
    },
    {
      "epoch": 3.937376725838264,
      "grad_norm": 0.314453125,
      "learning_rate": 2.6289711488420177e-05,
      "loss": 0.0342,
      "step": 15970
    },
    {
      "epoch": 3.937869822485207,
      "grad_norm": 0.28125,
      "learning_rate": 2.6266452354596183e-05,
      "loss": 0.0297,
      "step": 15972
    },
    {
      "epoch": 3.9383629191321496,
      "grad_norm": 0.283203125,
      "learning_rate": 2.6243201958609266e-05,
      "loss": 0.0317,
      "step": 15974
    },
    {
      "epoch": 3.9388560157790926,
      "grad_norm": 0.3125,
      "learning_rate": 2.621996030321472e-05,
      "loss": 0.0374,
      "step": 15976
    },
    {
      "epoch": 3.9393491124260356,
      "grad_norm": 0.283203125,
      "learning_rate": 2.6196727391166818e-05,
      "loss": 0.0318,
      "step": 15978
    },
    {
      "epoch": 3.939842209072978,
      "grad_norm": 0.263671875,
      "learning_rate": 2.6173503225218777e-05,
      "loss": 0.033,
      "step": 15980
    },
    {
      "epoch": 3.940335305719921,
      "grad_norm": 0.302734375,
      "learning_rate": 2.6150287808122832e-05,
      "loss": 0.0342,
      "step": 15982
    },
    {
      "epoch": 3.940828402366864,
      "grad_norm": 0.3046875,
      "learning_rate": 2.6127081142630083e-05,
      "loss": 0.0333,
      "step": 15984
    },
    {
      "epoch": 3.9413214990138066,
      "grad_norm": 0.34765625,
      "learning_rate": 2.6103883231490668e-05,
      "loss": 0.0332,
      "step": 15986
    },
    {
      "epoch": 3.9418145956607495,
      "grad_norm": 0.2734375,
      "learning_rate": 2.608069407745366e-05,
      "loss": 0.0321,
      "step": 15988
    },
    {
      "epoch": 3.9423076923076925,
      "grad_norm": 0.296875,
      "learning_rate": 2.605751368326713e-05,
      "loss": 0.0339,
      "step": 15990
    },
    {
      "epoch": 3.942800788954635,
      "grad_norm": 0.318359375,
      "learning_rate": 2.6034342051678074e-05,
      "loss": 0.0356,
      "step": 15992
    },
    {
      "epoch": 3.943293885601578,
      "grad_norm": 0.33203125,
      "learning_rate": 2.6011179185432444e-05,
      "loss": 0.0331,
      "step": 15994
    },
    {
      "epoch": 3.943786982248521,
      "grad_norm": 0.30078125,
      "learning_rate": 2.5988025087275213e-05,
      "loss": 0.0353,
      "step": 15996
    },
    {
      "epoch": 3.9442800788954635,
      "grad_norm": 0.306640625,
      "learning_rate": 2.596487975995019e-05,
      "loss": 0.032,
      "step": 15998
    },
    {
      "epoch": 3.9447731755424065,
      "grad_norm": 0.3359375,
      "learning_rate": 2.5941743206200332e-05,
      "loss": 0.036,
      "step": 16000
    },
    {
      "epoch": 3.945266272189349,
      "grad_norm": 0.28125,
      "learning_rate": 2.5918615428767357e-05,
      "loss": 0.0323,
      "step": 16002
    },
    {
      "epoch": 3.945759368836292,
      "grad_norm": 0.298828125,
      "learning_rate": 2.5895496430392076e-05,
      "loss": 0.0352,
      "step": 16004
    },
    {
      "epoch": 3.9462524654832345,
      "grad_norm": 0.31640625,
      "learning_rate": 2.5872386213814204e-05,
      "loss": 0.0326,
      "step": 16006
    },
    {
      "epoch": 3.9467455621301775,
      "grad_norm": 0.283203125,
      "learning_rate": 2.5849284781772453e-05,
      "loss": 0.0334,
      "step": 16008
    },
    {
      "epoch": 3.9472386587771204,
      "grad_norm": 0.3125,
      "learning_rate": 2.582619213700448e-05,
      "loss": 0.0343,
      "step": 16010
    },
    {
      "epoch": 3.947731755424063,
      "grad_norm": 0.275390625,
      "learning_rate": 2.5803108282246823e-05,
      "loss": 0.0311,
      "step": 16012
    },
    {
      "epoch": 3.948224852071006,
      "grad_norm": 0.353515625,
      "learning_rate": 2.5780033220235143e-05,
      "loss": 0.0396,
      "step": 16014
    },
    {
      "epoch": 3.948717948717949,
      "grad_norm": 0.283203125,
      "learning_rate": 2.5756966953703897e-05,
      "loss": 0.0325,
      "step": 16016
    },
    {
      "epoch": 3.9492110453648914,
      "grad_norm": 0.328125,
      "learning_rate": 2.5733909485386588e-05,
      "loss": 0.0348,
      "step": 16018
    },
    {
      "epoch": 3.9497041420118344,
      "grad_norm": 0.28515625,
      "learning_rate": 2.5710860818015657e-05,
      "loss": 0.0336,
      "step": 16020
    },
    {
      "epoch": 3.9501972386587774,
      "grad_norm": 0.326171875,
      "learning_rate": 2.5687820954322484e-05,
      "loss": 0.0344,
      "step": 16022
    },
    {
      "epoch": 3.95069033530572,
      "grad_norm": 0.310546875,
      "learning_rate": 2.5664789897037478e-05,
      "loss": 0.0362,
      "step": 16024
    },
    {
      "epoch": 3.951183431952663,
      "grad_norm": 0.328125,
      "learning_rate": 2.5641767648889835e-05,
      "loss": 0.036,
      "step": 16026
    },
    {
      "epoch": 3.9516765285996054,
      "grad_norm": 0.3515625,
      "learning_rate": 2.5618754212607953e-05,
      "loss": 0.0359,
      "step": 16028
    },
    {
      "epoch": 3.9521696252465484,
      "grad_norm": 0.3203125,
      "learning_rate": 2.5595749590918926e-05,
      "loss": 0.0337,
      "step": 16030
    },
    {
      "epoch": 3.952662721893491,
      "grad_norm": 0.3359375,
      "learning_rate": 2.5572753786549054e-05,
      "loss": 0.0374,
      "step": 16032
    },
    {
      "epoch": 3.953155818540434,
      "grad_norm": 0.30078125,
      "learning_rate": 2.5549766802223374e-05,
      "loss": 0.0332,
      "step": 16034
    },
    {
      "epoch": 3.953648915187377,
      "grad_norm": 0.33203125,
      "learning_rate": 2.5526788640666012e-05,
      "loss": 0.0353,
      "step": 16036
    },
    {
      "epoch": 3.9541420118343193,
      "grad_norm": 0.291015625,
      "learning_rate": 2.5503819304600017e-05,
      "loss": 0.0294,
      "step": 16038
    },
    {
      "epoch": 3.9546351084812623,
      "grad_norm": 0.294921875,
      "learning_rate": 2.548085879674732e-05,
      "loss": 0.0323,
      "step": 16040
    },
    {
      "epoch": 3.9551282051282053,
      "grad_norm": 0.357421875,
      "learning_rate": 2.5457907119828973e-05,
      "loss": 0.0347,
      "step": 16042
    },
    {
      "epoch": 3.955621301775148,
      "grad_norm": 0.294921875,
      "learning_rate": 2.5434964276564766e-05,
      "loss": 0.0321,
      "step": 16044
    },
    {
      "epoch": 3.956114398422091,
      "grad_norm": 0.328125,
      "learning_rate": 2.541203026967367e-05,
      "loss": 0.0293,
      "step": 16046
    },
    {
      "epoch": 3.9566074950690338,
      "grad_norm": 0.384765625,
      "learning_rate": 2.53891051018734e-05,
      "loss": 0.0337,
      "step": 16048
    },
    {
      "epoch": 3.9571005917159763,
      "grad_norm": 0.306640625,
      "learning_rate": 2.536618877588075e-05,
      "loss": 0.036,
      "step": 16050
    },
    {
      "epoch": 3.9575936883629192,
      "grad_norm": 0.3359375,
      "learning_rate": 2.5343281294411457e-05,
      "loss": 0.0312,
      "step": 16052
    },
    {
      "epoch": 3.9580867850098618,
      "grad_norm": 0.29296875,
      "learning_rate": 2.5320382660180108e-05,
      "loss": 0.0344,
      "step": 16054
    },
    {
      "epoch": 3.9585798816568047,
      "grad_norm": 0.392578125,
      "learning_rate": 2.529749287590042e-05,
      "loss": 0.0417,
      "step": 16056
    },
    {
      "epoch": 3.9590729783037473,
      "grad_norm": 0.28515625,
      "learning_rate": 2.5274611944284866e-05,
      "loss": 0.0334,
      "step": 16058
    },
    {
      "epoch": 3.9595660749506902,
      "grad_norm": 0.29296875,
      "learning_rate": 2.5251739868045064e-05,
      "loss": 0.0349,
      "step": 16060
    },
    {
      "epoch": 3.960059171597633,
      "grad_norm": 0.298828125,
      "learning_rate": 2.5228876649891398e-05,
      "loss": 0.0376,
      "step": 16062
    },
    {
      "epoch": 3.9605522682445757,
      "grad_norm": 0.458984375,
      "learning_rate": 2.5206022292533317e-05,
      "loss": 0.0368,
      "step": 16064
    },
    {
      "epoch": 3.9610453648915187,
      "grad_norm": 0.333984375,
      "learning_rate": 2.518317679867922e-05,
      "loss": 0.0353,
      "step": 16066
    },
    {
      "epoch": 3.9615384615384617,
      "grad_norm": 0.333984375,
      "learning_rate": 2.5160340171036335e-05,
      "loss": 0.0362,
      "step": 16068
    },
    {
      "epoch": 3.962031558185404,
      "grad_norm": 0.3046875,
      "learning_rate": 2.5137512412311048e-05,
      "loss": 0.0339,
      "step": 16070
    },
    {
      "epoch": 3.962524654832347,
      "grad_norm": 0.302734375,
      "learning_rate": 2.5114693525208467e-05,
      "loss": 0.0374,
      "step": 16072
    },
    {
      "epoch": 3.96301775147929,
      "grad_norm": 0.3046875,
      "learning_rate": 2.509188351243287e-05,
      "loss": 0.0341,
      "step": 16074
    },
    {
      "epoch": 3.9635108481262327,
      "grad_norm": 0.34765625,
      "learning_rate": 2.506908237668727e-05,
      "loss": 0.0347,
      "step": 16076
    },
    {
      "epoch": 3.9640039447731756,
      "grad_norm": 0.384765625,
      "learning_rate": 2.5046290120673776e-05,
      "loss": 0.0356,
      "step": 16078
    },
    {
      "epoch": 3.9644970414201186,
      "grad_norm": 0.29296875,
      "learning_rate": 2.5023506747093395e-05,
      "loss": 0.0324,
      "step": 16080
    },
    {
      "epoch": 3.964990138067061,
      "grad_norm": 0.28125,
      "learning_rate": 2.5000732258646074e-05,
      "loss": 0.0334,
      "step": 16082
    },
    {
      "epoch": 3.965483234714004,
      "grad_norm": 0.27734375,
      "learning_rate": 2.497796665803075e-05,
      "loss": 0.0333,
      "step": 16084
    },
    {
      "epoch": 3.9659763313609466,
      "grad_norm": 0.318359375,
      "learning_rate": 2.495520994794519e-05,
      "loss": 0.0343,
      "step": 16086
    },
    {
      "epoch": 3.9664694280078896,
      "grad_norm": 0.2734375,
      "learning_rate": 2.493246213108631e-05,
      "loss": 0.0334,
      "step": 16088
    },
    {
      "epoch": 3.966962524654832,
      "grad_norm": 0.279296875,
      "learning_rate": 2.4909723210149748e-05,
      "loss": 0.0311,
      "step": 16090
    },
    {
      "epoch": 3.967455621301775,
      "grad_norm": 0.283203125,
      "learning_rate": 2.4886993187830256e-05,
      "loss": 0.0325,
      "step": 16092
    },
    {
      "epoch": 3.967948717948718,
      "grad_norm": 0.30859375,
      "learning_rate": 2.486427206682144e-05,
      "loss": 0.0317,
      "step": 16094
    },
    {
      "epoch": 3.9684418145956606,
      "grad_norm": 0.28125,
      "learning_rate": 2.4841559849815878e-05,
      "loss": 0.0299,
      "step": 16096
    },
    {
      "epoch": 3.9689349112426036,
      "grad_norm": 0.310546875,
      "learning_rate": 2.481885653950514e-05,
      "loss": 0.0347,
      "step": 16098
    },
    {
      "epoch": 3.9694280078895465,
      "grad_norm": 0.328125,
      "learning_rate": 2.4796162138579616e-05,
      "loss": 0.0398,
      "step": 16100
    },
    {
      "epoch": 3.969921104536489,
      "grad_norm": 0.28125,
      "learning_rate": 2.47734766497288e-05,
      "loss": 0.0325,
      "step": 16102
    },
    {
      "epoch": 3.970414201183432,
      "grad_norm": 0.30078125,
      "learning_rate": 2.4750800075641e-05,
      "loss": 0.0312,
      "step": 16104
    },
    {
      "epoch": 3.970907297830375,
      "grad_norm": 0.28515625,
      "learning_rate": 2.4728132419003515e-05,
      "loss": 0.0298,
      "step": 16106
    },
    {
      "epoch": 3.9714003944773175,
      "grad_norm": 0.333984375,
      "learning_rate": 2.470547368250261e-05,
      "loss": 0.0341,
      "step": 16108
    },
    {
      "epoch": 3.9718934911242605,
      "grad_norm": 0.294921875,
      "learning_rate": 2.468282386882347e-05,
      "loss": 0.0306,
      "step": 16110
    },
    {
      "epoch": 3.972386587771203,
      "grad_norm": 0.326171875,
      "learning_rate": 2.4660182980650205e-05,
      "loss": 0.033,
      "step": 16112
    },
    {
      "epoch": 3.972879684418146,
      "grad_norm": 0.26171875,
      "learning_rate": 2.4637551020665905e-05,
      "loss": 0.0295,
      "step": 16114
    },
    {
      "epoch": 3.9733727810650885,
      "grad_norm": 0.361328125,
      "learning_rate": 2.4614927991552615e-05,
      "loss": 0.0345,
      "step": 16116
    },
    {
      "epoch": 3.9738658777120315,
      "grad_norm": 0.25390625,
      "learning_rate": 2.4592313895991225e-05,
      "loss": 0.0335,
      "step": 16118
    },
    {
      "epoch": 3.9743589743589745,
      "grad_norm": 0.310546875,
      "learning_rate": 2.4569708736661655e-05,
      "loss": 0.0349,
      "step": 16120
    },
    {
      "epoch": 3.974852071005917,
      "grad_norm": 0.296875,
      "learning_rate": 2.4547112516242753e-05,
      "loss": 0.0313,
      "step": 16122
    },
    {
      "epoch": 3.97534516765286,
      "grad_norm": 0.28515625,
      "learning_rate": 2.4524525237412287e-05,
      "loss": 0.0312,
      "step": 16124
    },
    {
      "epoch": 3.975838264299803,
      "grad_norm": 0.275390625,
      "learning_rate": 2.4501946902846983e-05,
      "loss": 0.0343,
      "step": 16126
    },
    {
      "epoch": 3.9763313609467454,
      "grad_norm": 0.27734375,
      "learning_rate": 2.44793775152225e-05,
      "loss": 0.0333,
      "step": 16128
    },
    {
      "epoch": 3.9768244575936884,
      "grad_norm": 0.33203125,
      "learning_rate": 2.4456817077213436e-05,
      "loss": 0.0298,
      "step": 16130
    },
    {
      "epoch": 3.9773175542406314,
      "grad_norm": 0.302734375,
      "learning_rate": 2.4434265591493355e-05,
      "loss": 0.0321,
      "step": 16132
    },
    {
      "epoch": 3.977810650887574,
      "grad_norm": 0.28125,
      "learning_rate": 2.441172306073468e-05,
      "loss": 0.0309,
      "step": 16134
    },
    {
      "epoch": 3.978303747534517,
      "grad_norm": 0.271484375,
      "learning_rate": 2.4389189487608855e-05,
      "loss": 0.0297,
      "step": 16136
    },
    {
      "epoch": 3.9787968441814594,
      "grad_norm": 0.326171875,
      "learning_rate": 2.4366664874786226e-05,
      "loss": 0.0347,
      "step": 16138
    },
    {
      "epoch": 3.9792899408284024,
      "grad_norm": 0.2890625,
      "learning_rate": 2.4344149224936084e-05,
      "loss": 0.0325,
      "step": 16140
    },
    {
      "epoch": 3.979783037475345,
      "grad_norm": 0.294921875,
      "learning_rate": 2.4321642540726686e-05,
      "loss": 0.0334,
      "step": 16142
    },
    {
      "epoch": 3.980276134122288,
      "grad_norm": 0.2734375,
      "learning_rate": 2.429914482482517e-05,
      "loss": 0.0329,
      "step": 16144
    },
    {
      "epoch": 3.980769230769231,
      "grad_norm": 0.291015625,
      "learning_rate": 2.4276656079897688e-05,
      "loss": 0.0343,
      "step": 16146
    },
    {
      "epoch": 3.9812623274161734,
      "grad_norm": 0.26953125,
      "learning_rate": 2.425417630860921e-05,
      "loss": 0.0314,
      "step": 16148
    },
    {
      "epoch": 3.9817554240631163,
      "grad_norm": 0.3359375,
      "learning_rate": 2.4231705513623748e-05,
      "loss": 0.032,
      "step": 16150
    },
    {
      "epoch": 3.9822485207100593,
      "grad_norm": 0.357421875,
      "learning_rate": 2.4209243697604222e-05,
      "loss": 0.0363,
      "step": 16152
    },
    {
      "epoch": 3.982741617357002,
      "grad_norm": 0.296875,
      "learning_rate": 2.4186790863212472e-05,
      "loss": 0.032,
      "step": 16154
    },
    {
      "epoch": 3.983234714003945,
      "grad_norm": 0.310546875,
      "learning_rate": 2.4164347013109302e-05,
      "loss": 0.0356,
      "step": 16156
    },
    {
      "epoch": 3.9837278106508878,
      "grad_norm": 0.283203125,
      "learning_rate": 2.4141912149954427e-05,
      "loss": 0.0345,
      "step": 16158
    },
    {
      "epoch": 3.9842209072978303,
      "grad_norm": 0.34375,
      "learning_rate": 2.4119486276406512e-05,
      "loss": 0.0382,
      "step": 16160
    },
    {
      "epoch": 3.9847140039447733,
      "grad_norm": 0.337890625,
      "learning_rate": 2.4097069395123094e-05,
      "loss": 0.0355,
      "step": 16162
    },
    {
      "epoch": 3.9852071005917162,
      "grad_norm": 0.306640625,
      "learning_rate": 2.4074661508760787e-05,
      "loss": 0.0338,
      "step": 16164
    },
    {
      "epoch": 3.9857001972386588,
      "grad_norm": 0.3046875,
      "learning_rate": 2.4052262619974974e-05,
      "loss": 0.0323,
      "step": 16166
    },
    {
      "epoch": 3.9861932938856017,
      "grad_norm": 0.29296875,
      "learning_rate": 2.4029872731420087e-05,
      "loss": 0.0309,
      "step": 16168
    },
    {
      "epoch": 3.9866863905325443,
      "grad_norm": 0.328125,
      "learning_rate": 2.400749184574944e-05,
      "loss": 0.0277,
      "step": 16170
    },
    {
      "epoch": 3.9871794871794872,
      "grad_norm": 0.361328125,
      "learning_rate": 2.39851199656153e-05,
      "loss": 0.0345,
      "step": 16172
    },
    {
      "epoch": 3.9876725838264298,
      "grad_norm": 0.412109375,
      "learning_rate": 2.3962757093668887e-05,
      "loss": 0.0287,
      "step": 16174
    },
    {
      "epoch": 3.9881656804733727,
      "grad_norm": 0.27734375,
      "learning_rate": 2.3940403232560228e-05,
      "loss": 0.0299,
      "step": 16176
    },
    {
      "epoch": 3.9886587771203157,
      "grad_norm": 0.2890625,
      "learning_rate": 2.3918058384938513e-05,
      "loss": 0.0348,
      "step": 16178
    },
    {
      "epoch": 3.989151873767258,
      "grad_norm": 0.3125,
      "learning_rate": 2.3895722553451615e-05,
      "loss": 0.0312,
      "step": 16180
    },
    {
      "epoch": 3.989644970414201,
      "grad_norm": 0.29296875,
      "learning_rate": 2.3873395740746552e-05,
      "loss": 0.0346,
      "step": 16182
    },
    {
      "epoch": 3.990138067061144,
      "grad_norm": 0.291015625,
      "learning_rate": 2.3851077949469114e-05,
      "loss": 0.0317,
      "step": 16184
    },
    {
      "epoch": 3.9906311637080867,
      "grad_norm": 0.333984375,
      "learning_rate": 2.382876918226409e-05,
      "loss": 0.0342,
      "step": 16186
    },
    {
      "epoch": 3.9911242603550297,
      "grad_norm": 0.3125,
      "learning_rate": 2.3806469441775236e-05,
      "loss": 0.0338,
      "step": 16188
    },
    {
      "epoch": 3.9916173570019726,
      "grad_norm": 0.287109375,
      "learning_rate": 2.3784178730645103e-05,
      "loss": 0.0322,
      "step": 16190
    },
    {
      "epoch": 3.992110453648915,
      "grad_norm": 0.365234375,
      "learning_rate": 2.376189705151539e-05,
      "loss": 0.0312,
      "step": 16192
    },
    {
      "epoch": 3.992603550295858,
      "grad_norm": 0.31640625,
      "learning_rate": 2.373962440702647e-05,
      "loss": 0.0313,
      "step": 16194
    },
    {
      "epoch": 3.9930966469428006,
      "grad_norm": 0.302734375,
      "learning_rate": 2.37173607998179e-05,
      "loss": 0.0331,
      "step": 16196
    },
    {
      "epoch": 3.9935897435897436,
      "grad_norm": 0.314453125,
      "learning_rate": 2.369510623252795e-05,
      "loss": 0.036,
      "step": 16198
    },
    {
      "epoch": 3.994082840236686,
      "grad_norm": 0.310546875,
      "learning_rate": 2.3672860707793953e-05,
      "loss": 0.0351,
      "step": 16200
    },
    {
      "epoch": 3.994575936883629,
      "grad_norm": 0.310546875,
      "learning_rate": 2.365062422825215e-05,
      "loss": 0.0302,
      "step": 16202
    },
    {
      "epoch": 3.995069033530572,
      "grad_norm": 0.34375,
      "learning_rate": 2.3628396796537588e-05,
      "loss": 0.0362,
      "step": 16204
    },
    {
      "epoch": 3.9955621301775146,
      "grad_norm": 0.3828125,
      "learning_rate": 2.3606178415284473e-05,
      "loss": 0.0318,
      "step": 16206
    },
    {
      "epoch": 3.9960552268244576,
      "grad_norm": 0.298828125,
      "learning_rate": 2.358396908712569e-05,
      "loss": 0.0319,
      "step": 16208
    },
    {
      "epoch": 3.9965483234714005,
      "grad_norm": 0.306640625,
      "learning_rate": 2.3561768814693274e-05,
      "loss": 0.0336,
      "step": 16210
    },
    {
      "epoch": 3.997041420118343,
      "grad_norm": 0.322265625,
      "learning_rate": 2.3539577600618003e-05,
      "loss": 0.0332,
      "step": 16212
    },
    {
      "epoch": 3.997534516765286,
      "grad_norm": 0.291015625,
      "learning_rate": 2.3517395447529687e-05,
      "loss": 0.0331,
      "step": 16214
    },
    {
      "epoch": 3.998027613412229,
      "grad_norm": 0.3046875,
      "learning_rate": 2.3495222358057058e-05,
      "loss": 0.0355,
      "step": 16216
    },
    {
      "epoch": 3.9985207100591715,
      "grad_norm": 0.36328125,
      "learning_rate": 2.347305833482768e-05,
      "loss": 0.0373,
      "step": 16218
    },
    {
      "epoch": 3.9990138067061145,
      "grad_norm": 0.28125,
      "learning_rate": 2.3450903380468214e-05,
      "loss": 0.0326,
      "step": 16220
    },
    {
      "epoch": 3.9995069033530575,
      "grad_norm": 0.279296875,
      "learning_rate": 2.3428757497604027e-05,
      "loss": 0.0336,
      "step": 16222
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.36328125,
      "learning_rate": 2.340662068885966e-05,
      "loss": 0.0275,
      "step": 16224
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.06094944104552269,
      "eval_runtime": 68.5538,
      "eval_samples_per_second": 232.358,
      "eval_steps_per_second": 1.823,
      "step": 16224
    },
    {
      "epoch": 4.0004930966469425,
      "grad_norm": 0.265625,
      "learning_rate": 2.3384492956858363e-05,
      "loss": 0.0223,
      "step": 16226
    },
    {
      "epoch": 4.000986193293886,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.336237430422241e-05,
      "loss": 0.0194,
      "step": 16228
    },
    {
      "epoch": 4.0014792899408285,
      "grad_norm": 0.21875,
      "learning_rate": 2.334026473357299e-05,
      "loss": 0.0194,
      "step": 16230
    },
    {
      "epoch": 4.001972386587771,
      "grad_norm": 0.2734375,
      "learning_rate": 2.3318164247530215e-05,
      "loss": 0.0216,
      "step": 16232
    },
    {
      "epoch": 4.002465483234714,
      "grad_norm": 0.279296875,
      "learning_rate": 2.3296072848713134e-05,
      "loss": 0.0211,
      "step": 16234
    },
    {
      "epoch": 4.002958579881657,
      "grad_norm": 0.234375,
      "learning_rate": 2.3273990539739633e-05,
      "loss": 0.0196,
      "step": 16236
    },
    {
      "epoch": 4.0034516765285995,
      "grad_norm": 0.259765625,
      "learning_rate": 2.3251917323226692e-05,
      "loss": 0.0246,
      "step": 16238
    },
    {
      "epoch": 4.003944773175542,
      "grad_norm": 0.29296875,
      "learning_rate": 2.3229853201790032e-05,
      "loss": 0.0261,
      "step": 16240
    },
    {
      "epoch": 4.004437869822485,
      "grad_norm": 0.240234375,
      "learning_rate": 2.3207798178044384e-05,
      "loss": 0.0227,
      "step": 16242
    },
    {
      "epoch": 4.004930966469428,
      "grad_norm": 0.26953125,
      "learning_rate": 2.318575225460341e-05,
      "loss": 0.0232,
      "step": 16244
    },
    {
      "epoch": 4.0054240631163704,
      "grad_norm": 0.30078125,
      "learning_rate": 2.316371543407967e-05,
      "loss": 0.0222,
      "step": 16246
    },
    {
      "epoch": 4.005917159763314,
      "grad_norm": 0.25390625,
      "learning_rate": 2.314168771908468e-05,
      "loss": 0.0242,
      "step": 16248
    },
    {
      "epoch": 4.006410256410256,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.311966911222876e-05,
      "loss": 0.0217,
      "step": 16250
    },
    {
      "epoch": 4.006903353057199,
      "grad_norm": 0.232421875,
      "learning_rate": 2.309765961612135e-05,
      "loss": 0.0206,
      "step": 16252
    },
    {
      "epoch": 4.007396449704142,
      "grad_norm": 0.298828125,
      "learning_rate": 2.3075659233370616e-05,
      "loss": 0.0208,
      "step": 16254
    },
    {
      "epoch": 4.007889546351085,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.305366796658375e-05,
      "loss": 0.0231,
      "step": 16256
    },
    {
      "epoch": 4.008382642998027,
      "grad_norm": 0.275390625,
      "learning_rate": 2.3031685818366842e-05,
      "loss": 0.0226,
      "step": 16258
    },
    {
      "epoch": 4.008875739644971,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.3009712791324912e-05,
      "loss": 0.0229,
      "step": 16260
    },
    {
      "epoch": 4.009368836291913,
      "grad_norm": 0.236328125,
      "learning_rate": 2.2987748888061866e-05,
      "loss": 0.0218,
      "step": 16262
    },
    {
      "epoch": 4.009861932938856,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.296579411118055e-05,
      "loss": 0.0223,
      "step": 16264
    },
    {
      "epoch": 4.010355029585799,
      "grad_norm": 0.265625,
      "learning_rate": 2.294384846328278e-05,
      "loss": 0.0212,
      "step": 16266
    },
    {
      "epoch": 4.010848126232742,
      "grad_norm": 0.26171875,
      "learning_rate": 2.2921911946969155e-05,
      "loss": 0.019,
      "step": 16268
    },
    {
      "epoch": 4.011341222879684,
      "grad_norm": 0.32421875,
      "learning_rate": 2.2899984564839317e-05,
      "loss": 0.0251,
      "step": 16270
    },
    {
      "epoch": 4.011834319526627,
      "grad_norm": 0.271484375,
      "learning_rate": 2.2878066319491786e-05,
      "loss": 0.0207,
      "step": 16272
    },
    {
      "epoch": 4.01232741617357,
      "grad_norm": 0.24609375,
      "learning_rate": 2.2856157213523977e-05,
      "loss": 0.0206,
      "step": 16274
    },
    {
      "epoch": 4.012820512820513,
      "grad_norm": 0.275390625,
      "learning_rate": 2.2834257249532266e-05,
      "loss": 0.0225,
      "step": 16276
    },
    {
      "epoch": 4.013313609467455,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.2812366430111898e-05,
      "loss": 0.0209,
      "step": 16278
    },
    {
      "epoch": 4.013806706114399,
      "grad_norm": 0.240234375,
      "learning_rate": 2.27904847578571e-05,
      "loss": 0.0208,
      "step": 16280
    },
    {
      "epoch": 4.014299802761341,
      "grad_norm": 0.248046875,
      "learning_rate": 2.276861223536092e-05,
      "loss": 0.0201,
      "step": 16282
    },
    {
      "epoch": 4.014792899408284,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.2746748865215395e-05,
      "loss": 0.0217,
      "step": 16284
    },
    {
      "epoch": 4.015285996055227,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.272489465001145e-05,
      "loss": 0.0244,
      "step": 16286
    },
    {
      "epoch": 4.01577909270217,
      "grad_norm": 0.25,
      "learning_rate": 2.2703049592338942e-05,
      "loss": 0.0205,
      "step": 16288
    },
    {
      "epoch": 4.016272189349112,
      "grad_norm": 0.275390625,
      "learning_rate": 2.268121369478662e-05,
      "loss": 0.0281,
      "step": 16290
    },
    {
      "epoch": 4.016765285996056,
      "grad_norm": 0.2353515625,
      "learning_rate": 2.2659386959942175e-05,
      "loss": 0.0167,
      "step": 16292
    },
    {
      "epoch": 4.017258382642998,
      "grad_norm": 0.23828125,
      "learning_rate": 2.263756939039219e-05,
      "loss": 0.0194,
      "step": 16294
    },
    {
      "epoch": 4.017751479289941,
      "grad_norm": 0.24609375,
      "learning_rate": 2.2615760988722167e-05,
      "loss": 0.0225,
      "step": 16296
    },
    {
      "epoch": 4.018244575936883,
      "grad_norm": 0.236328125,
      "learning_rate": 2.2593961757516558e-05,
      "loss": 0.0221,
      "step": 16298
    },
    {
      "epoch": 4.018737672583827,
      "grad_norm": 0.287109375,
      "learning_rate": 2.2572171699358626e-05,
      "loss": 0.0206,
      "step": 16300
    },
    {
      "epoch": 4.019230769230769,
      "grad_norm": 0.2578125,
      "learning_rate": 2.2550390816830645e-05,
      "loss": 0.0215,
      "step": 16302
    },
    {
      "epoch": 4.019723865877712,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.2528619112513784e-05,
      "loss": 0.0261,
      "step": 16304
    },
    {
      "epoch": 4.020216962524655,
      "grad_norm": 0.302734375,
      "learning_rate": 2.25068565889881e-05,
      "loss": 0.0216,
      "step": 16306
    },
    {
      "epoch": 4.020710059171598,
      "grad_norm": 0.267578125,
      "learning_rate": 2.2485103248832585e-05,
      "loss": 0.0251,
      "step": 16308
    },
    {
      "epoch": 4.02120315581854,
      "grad_norm": 0.25390625,
      "learning_rate": 2.2463359094625114e-05,
      "loss": 0.0191,
      "step": 16310
    },
    {
      "epoch": 4.021696252465484,
      "grad_norm": 0.28125,
      "learning_rate": 2.2441624128942495e-05,
      "loss": 0.0215,
      "step": 16312
    },
    {
      "epoch": 4.022189349112426,
      "grad_norm": 0.2490234375,
      "learning_rate": 2.241989835436048e-05,
      "loss": 0.0227,
      "step": 16314
    },
    {
      "epoch": 4.022682445759369,
      "grad_norm": 0.205078125,
      "learning_rate": 2.239818177345364e-05,
      "loss": 0.0195,
      "step": 16316
    },
    {
      "epoch": 4.023175542406312,
      "grad_norm": 0.265625,
      "learning_rate": 2.2376474388795533e-05,
      "loss": 0.0266,
      "step": 16318
    },
    {
      "epoch": 4.023668639053255,
      "grad_norm": 0.25390625,
      "learning_rate": 2.2354776202958594e-05,
      "loss": 0.0246,
      "step": 16320
    },
    {
      "epoch": 4.024161735700197,
      "grad_norm": 0.3046875,
      "learning_rate": 2.233308721851419e-05,
      "loss": 0.0215,
      "step": 16322
    },
    {
      "epoch": 4.02465483234714,
      "grad_norm": 0.25,
      "learning_rate": 2.2311407438032606e-05,
      "loss": 0.0232,
      "step": 16324
    },
    {
      "epoch": 4.025147928994083,
      "grad_norm": 0.255859375,
      "learning_rate": 2.2289736864082978e-05,
      "loss": 0.0215,
      "step": 16326
    },
    {
      "epoch": 4.0256410256410255,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.2268075499233443e-05,
      "loss": 0.0207,
      "step": 16328
    },
    {
      "epoch": 4.026134122287968,
      "grad_norm": 0.255859375,
      "learning_rate": 2.2246423346050937e-05,
      "loss": 0.0233,
      "step": 16330
    },
    {
      "epoch": 4.0266272189349115,
      "grad_norm": 0.294921875,
      "learning_rate": 2.2224780407101388e-05,
      "loss": 0.0221,
      "step": 16332
    },
    {
      "epoch": 4.027120315581854,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.2203146684949595e-05,
      "loss": 0.0199,
      "step": 16334
    },
    {
      "epoch": 4.0276134122287965,
      "grad_norm": 0.306640625,
      "learning_rate": 2.2181522182159288e-05,
      "loss": 0.0241,
      "step": 16336
    },
    {
      "epoch": 4.02810650887574,
      "grad_norm": 0.24609375,
      "learning_rate": 2.2159906901293082e-05,
      "loss": 0.0222,
      "step": 16338
    },
    {
      "epoch": 4.0285996055226825,
      "grad_norm": 0.25390625,
      "learning_rate": 2.2138300844912506e-05,
      "loss": 0.0254,
      "step": 16340
    },
    {
      "epoch": 4.029092702169625,
      "grad_norm": 0.2099609375,
      "learning_rate": 2.211670401557804e-05,
      "loss": 0.0184,
      "step": 16342
    },
    {
      "epoch": 4.029585798816568,
      "grad_norm": 0.26171875,
      "learning_rate": 2.2095116415848937e-05,
      "loss": 0.0216,
      "step": 16344
    },
    {
      "epoch": 4.030078895463511,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.207353804828356e-05,
      "loss": 0.0213,
      "step": 16346
    },
    {
      "epoch": 4.0305719921104535,
      "grad_norm": 0.291015625,
      "learning_rate": 2.205196891543897e-05,
      "loss": 0.0237,
      "step": 16348
    },
    {
      "epoch": 4.031065088757397,
      "grad_norm": 0.244140625,
      "learning_rate": 2.2030409019871277e-05,
      "loss": 0.0246,
      "step": 16350
    },
    {
      "epoch": 4.031558185404339,
      "grad_norm": 0.244140625,
      "learning_rate": 2.2008858364135443e-05,
      "loss": 0.0203,
      "step": 16352
    },
    {
      "epoch": 4.032051282051282,
      "grad_norm": 0.22265625,
      "learning_rate": 2.198731695078534e-05,
      "loss": 0.023,
      "step": 16354
    },
    {
      "epoch": 4.0325443786982245,
      "grad_norm": 0.271484375,
      "learning_rate": 2.1965784782373778e-05,
      "loss": 0.0236,
      "step": 16356
    },
    {
      "epoch": 4.033037475345168,
      "grad_norm": 0.236328125,
      "learning_rate": 2.194426186145234e-05,
      "loss": 0.0201,
      "step": 16358
    },
    {
      "epoch": 4.03353057199211,
      "grad_norm": 0.24609375,
      "learning_rate": 2.192274819057174e-05,
      "loss": 0.0228,
      "step": 16360
    },
    {
      "epoch": 4.034023668639053,
      "grad_norm": 0.263671875,
      "learning_rate": 2.190124377228139e-05,
      "loss": 0.0217,
      "step": 16362
    },
    {
      "epoch": 4.034516765285996,
      "grad_norm": 0.23828125,
      "learning_rate": 2.187974860912968e-05,
      "loss": 0.024,
      "step": 16364
    },
    {
      "epoch": 4.035009861932939,
      "grad_norm": 0.28125,
      "learning_rate": 2.1858262703663944e-05,
      "loss": 0.0208,
      "step": 16366
    },
    {
      "epoch": 4.035502958579881,
      "grad_norm": 0.2578125,
      "learning_rate": 2.1836786058430358e-05,
      "loss": 0.02,
      "step": 16368
    },
    {
      "epoch": 4.035996055226825,
      "grad_norm": 0.2470703125,
      "learning_rate": 2.181531867597406e-05,
      "loss": 0.0227,
      "step": 16370
    },
    {
      "epoch": 4.036489151873767,
      "grad_norm": 0.25390625,
      "learning_rate": 2.1793860558838974e-05,
      "loss": 0.0222,
      "step": 16372
    },
    {
      "epoch": 4.03698224852071,
      "grad_norm": 0.287109375,
      "learning_rate": 2.1772411709568108e-05,
      "loss": 0.0237,
      "step": 16374
    },
    {
      "epoch": 4.037475345167653,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.1750972130703163e-05,
      "loss": 0.0211,
      "step": 16376
    },
    {
      "epoch": 4.037968441814596,
      "grad_norm": 0.2265625,
      "learning_rate": 2.172954182478497e-05,
      "loss": 0.0187,
      "step": 16378
    },
    {
      "epoch": 4.038461538461538,
      "grad_norm": 0.255859375,
      "learning_rate": 2.170812079435305e-05,
      "loss": 0.0224,
      "step": 16380
    },
    {
      "epoch": 4.038954635108481,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.168670904194594e-05,
      "loss": 0.0193,
      "step": 16382
    },
    {
      "epoch": 4.039447731755424,
      "grad_norm": 0.271484375,
      "learning_rate": 2.1665306570101086e-05,
      "loss": 0.0176,
      "step": 16384
    },
    {
      "epoch": 4.039940828402367,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.1643913381354707e-05,
      "loss": 0.0215,
      "step": 16386
    },
    {
      "epoch": 4.040433925049309,
      "grad_norm": 0.208984375,
      "learning_rate": 2.1622529478242125e-05,
      "loss": 0.0195,
      "step": 16388
    },
    {
      "epoch": 4.040927021696253,
      "grad_norm": 0.193359375,
      "learning_rate": 2.1601154863297355e-05,
      "loss": 0.0193,
      "step": 16390
    },
    {
      "epoch": 4.041420118343195,
      "grad_norm": 0.2470703125,
      "learning_rate": 2.157978953905352e-05,
      "loss": 0.0243,
      "step": 16392
    },
    {
      "epoch": 4.041913214990138,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.155843350804243e-05,
      "loss": 0.0197,
      "step": 16394
    },
    {
      "epoch": 4.042406311637081,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.1537086772794934e-05,
      "loss": 0.0214,
      "step": 16396
    },
    {
      "epoch": 4.042899408284024,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.1515749335840773e-05,
      "loss": 0.0195,
      "step": 16398
    },
    {
      "epoch": 4.043392504930966,
      "grad_norm": 0.26171875,
      "learning_rate": 2.1494421199708447e-05,
      "loss": 0.0232,
      "step": 16400
    },
    {
      "epoch": 4.04388560157791,
      "grad_norm": 0.2890625,
      "learning_rate": 2.147310236692559e-05,
      "loss": 0.0221,
      "step": 16402
    },
    {
      "epoch": 4.044378698224852,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.145179284001848e-05,
      "loss": 0.0221,
      "step": 16404
    },
    {
      "epoch": 4.044871794871795,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.143049262151253e-05,
      "loss": 0.0209,
      "step": 16406
    },
    {
      "epoch": 4.045364891518737,
      "grad_norm": 0.244140625,
      "learning_rate": 2.1409201713931858e-05,
      "loss": 0.0211,
      "step": 16408
    },
    {
      "epoch": 4.045857988165681,
      "grad_norm": 0.2421875,
      "learning_rate": 2.1387920119799577e-05,
      "loss": 0.0234,
      "step": 16410
    },
    {
      "epoch": 4.046351084812623,
      "grad_norm": 0.25390625,
      "learning_rate": 2.1366647841637698e-05,
      "loss": 0.023,
      "step": 16412
    },
    {
      "epoch": 4.046844181459566,
      "grad_norm": 0.25390625,
      "learning_rate": 2.1345384881967036e-05,
      "loss": 0.0218,
      "step": 16414
    },
    {
      "epoch": 4.047337278106509,
      "grad_norm": 0.275390625,
      "learning_rate": 2.1324131243307466e-05,
      "loss": 0.0213,
      "step": 16416
    },
    {
      "epoch": 4.047830374753452,
      "grad_norm": 0.21875,
      "learning_rate": 2.1302886928177556e-05,
      "loss": 0.0196,
      "step": 16418
    },
    {
      "epoch": 4.048323471400394,
      "grad_norm": 0.259765625,
      "learning_rate": 2.1281651939094992e-05,
      "loss": 0.0229,
      "step": 16420
    },
    {
      "epoch": 4.048816568047338,
      "grad_norm": 0.28125,
      "learning_rate": 2.126042627857615e-05,
      "loss": 0.0196,
      "step": 16422
    },
    {
      "epoch": 4.04930966469428,
      "grad_norm": 0.23046875,
      "learning_rate": 2.123920994913643e-05,
      "loss": 0.0212,
      "step": 16424
    },
    {
      "epoch": 4.049802761341223,
      "grad_norm": 0.251953125,
      "learning_rate": 2.121800295329006e-05,
      "loss": 0.0228,
      "step": 16426
    },
    {
      "epoch": 4.050295857988166,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.1196805293550215e-05,
      "loss": 0.0203,
      "step": 16428
    },
    {
      "epoch": 4.050788954635109,
      "grad_norm": 0.26953125,
      "learning_rate": 2.1175616972428946e-05,
      "loss": 0.0215,
      "step": 16430
    },
    {
      "epoch": 4.051282051282051,
      "grad_norm": 0.27734375,
      "learning_rate": 2.1154437992437103e-05,
      "loss": 0.0229,
      "step": 16432
    },
    {
      "epoch": 4.0517751479289945,
      "grad_norm": 0.28125,
      "learning_rate": 2.1133268356084634e-05,
      "loss": 0.0208,
      "step": 16434
    },
    {
      "epoch": 4.052268244575937,
      "grad_norm": 0.2490234375,
      "learning_rate": 2.1112108065880175e-05,
      "loss": 0.0218,
      "step": 16436
    },
    {
      "epoch": 4.05276134122288,
      "grad_norm": 0.275390625,
      "learning_rate": 2.1090957124331366e-05,
      "loss": 0.0241,
      "step": 16438
    },
    {
      "epoch": 4.053254437869822,
      "grad_norm": 0.275390625,
      "learning_rate": 2.1069815533944715e-05,
      "loss": 0.0219,
      "step": 16440
    },
    {
      "epoch": 4.0537475345167655,
      "grad_norm": 0.2421875,
      "learning_rate": 2.1048683297225603e-05,
      "loss": 0.0202,
      "step": 16442
    },
    {
      "epoch": 4.054240631163708,
      "grad_norm": 0.2265625,
      "learning_rate": 2.1027560416678328e-05,
      "loss": 0.0202,
      "step": 16444
    },
    {
      "epoch": 4.054733727810651,
      "grad_norm": 0.251953125,
      "learning_rate": 2.1006446894806065e-05,
      "loss": 0.0217,
      "step": 16446
    },
    {
      "epoch": 4.055226824457594,
      "grad_norm": 0.259765625,
      "learning_rate": 2.0985342734110925e-05,
      "loss": 0.0199,
      "step": 16448
    },
    {
      "epoch": 4.0557199211045365,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.0964247937093807e-05,
      "loss": 0.0224,
      "step": 16450
    },
    {
      "epoch": 4.056213017751479,
      "grad_norm": 0.248046875,
      "learning_rate": 2.0943162506254576e-05,
      "loss": 0.0241,
      "step": 16452
    },
    {
      "epoch": 4.056706114398422,
      "grad_norm": 0.212890625,
      "learning_rate": 2.0922086444091994e-05,
      "loss": 0.0218,
      "step": 16454
    },
    {
      "epoch": 4.057199211045365,
      "grad_norm": 0.23046875,
      "learning_rate": 2.090101975310368e-05,
      "loss": 0.0233,
      "step": 16456
    },
    {
      "epoch": 4.0576923076923075,
      "grad_norm": 0.267578125,
      "learning_rate": 2.087996243578616e-05,
      "loss": 0.0207,
      "step": 16458
    },
    {
      "epoch": 4.058185404339251,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.085891449463484e-05,
      "loss": 0.0203,
      "step": 16460
    },
    {
      "epoch": 4.058678500986193,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.0837875932144045e-05,
      "loss": 0.0209,
      "step": 16462
    },
    {
      "epoch": 4.059171597633136,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.0816846750806907e-05,
      "loss": 0.021,
      "step": 16464
    },
    {
      "epoch": 4.0596646942800785,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.079582695311554e-05,
      "loss": 0.0215,
      "step": 16466
    },
    {
      "epoch": 4.060157790927022,
      "grad_norm": 0.2578125,
      "learning_rate": 2.0774816541560893e-05,
      "loss": 0.0223,
      "step": 16468
    },
    {
      "epoch": 4.060650887573964,
      "grad_norm": 0.240234375,
      "learning_rate": 2.0753815518632824e-05,
      "loss": 0.0196,
      "step": 16470
    },
    {
      "epoch": 4.061143984220907,
      "grad_norm": 0.2578125,
      "learning_rate": 2.0732823886820062e-05,
      "loss": 0.0212,
      "step": 16472
    },
    {
      "epoch": 4.06163708086785,
      "grad_norm": 0.236328125,
      "learning_rate": 2.0711841648610254e-05,
      "loss": 0.0232,
      "step": 16474
    },
    {
      "epoch": 4.062130177514793,
      "grad_norm": 0.251953125,
      "learning_rate": 2.0690868806489882e-05,
      "loss": 0.0241,
      "step": 16476
    },
    {
      "epoch": 4.062623274161735,
      "grad_norm": 0.265625,
      "learning_rate": 2.066990536294441e-05,
      "loss": 0.0213,
      "step": 16478
    },
    {
      "epoch": 4.063116370808679,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.0648951320458034e-05,
      "loss": 0.0197,
      "step": 16480
    },
    {
      "epoch": 4.063609467455621,
      "grad_norm": 0.23828125,
      "learning_rate": 2.0628006681513967e-05,
      "loss": 0.0215,
      "step": 16482
    },
    {
      "epoch": 4.064102564102564,
      "grad_norm": 0.27734375,
      "learning_rate": 2.0607071448594272e-05,
      "loss": 0.0242,
      "step": 16484
    },
    {
      "epoch": 4.064595660749507,
      "grad_norm": 0.248046875,
      "learning_rate": 2.0586145624179876e-05,
      "loss": 0.0201,
      "step": 16486
    },
    {
      "epoch": 4.06508875739645,
      "grad_norm": 0.2470703125,
      "learning_rate": 2.0565229210750613e-05,
      "loss": 0.0238,
      "step": 16488
    },
    {
      "epoch": 4.065581854043392,
      "grad_norm": 0.25390625,
      "learning_rate": 2.0544322210785205e-05,
      "loss": 0.0226,
      "step": 16490
    },
    {
      "epoch": 4.066074950690336,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.0523424626761257e-05,
      "loss": 0.0203,
      "step": 16492
    },
    {
      "epoch": 4.066568047337278,
      "grad_norm": 0.2353515625,
      "learning_rate": 2.0502536461155207e-05,
      "loss": 0.0211,
      "step": 16494
    },
    {
      "epoch": 4.067061143984221,
      "grad_norm": 0.26171875,
      "learning_rate": 2.048165771644244e-05,
      "loss": 0.022,
      "step": 16496
    },
    {
      "epoch": 4.067554240631163,
      "grad_norm": 0.2353515625,
      "learning_rate": 2.0460788395097207e-05,
      "loss": 0.0198,
      "step": 16498
    },
    {
      "epoch": 4.068047337278107,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.0439928499592643e-05,
      "loss": 0.0211,
      "step": 16500
    },
    {
      "epoch": 4.068540433925049,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.0419078032400764e-05,
      "loss": 0.0242,
      "step": 16502
    },
    {
      "epoch": 4.069033530571992,
      "grad_norm": 0.23046875,
      "learning_rate": 2.039823699599246e-05,
      "loss": 0.0189,
      "step": 16504
    },
    {
      "epoch": 4.069526627218935,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.0377405392837522e-05,
      "loss": 0.0215,
      "step": 16506
    },
    {
      "epoch": 4.070019723865878,
      "grad_norm": 0.232421875,
      "learning_rate": 2.035658322540457e-05,
      "loss": 0.0191,
      "step": 16508
    },
    {
      "epoch": 4.07051282051282,
      "grad_norm": 0.25,
      "learning_rate": 2.033577049616121e-05,
      "loss": 0.0221,
      "step": 16510
    },
    {
      "epoch": 4.071005917159764,
      "grad_norm": 0.255859375,
      "learning_rate": 2.031496720757382e-05,
      "loss": 0.021,
      "step": 16512
    },
    {
      "epoch": 4.071499013806706,
      "grad_norm": 0.2109375,
      "learning_rate": 2.0294173362107716e-05,
      "loss": 0.0232,
      "step": 16514
    },
    {
      "epoch": 4.071992110453649,
      "grad_norm": 0.23046875,
      "learning_rate": 2.0273388962227092e-05,
      "loss": 0.0207,
      "step": 16516
    },
    {
      "epoch": 4.072485207100592,
      "grad_norm": 0.24609375,
      "learning_rate": 2.025261401039501e-05,
      "loss": 0.0217,
      "step": 16518
    },
    {
      "epoch": 4.072978303747535,
      "grad_norm": 0.28515625,
      "learning_rate": 2.023184850907345e-05,
      "loss": 0.0247,
      "step": 16520
    },
    {
      "epoch": 4.073471400394477,
      "grad_norm": 0.2421875,
      "learning_rate": 2.0211092460723146e-05,
      "loss": 0.0202,
      "step": 16522
    },
    {
      "epoch": 4.07396449704142,
      "grad_norm": 0.240234375,
      "learning_rate": 2.019034586780394e-05,
      "loss": 0.0207,
      "step": 16524
    },
    {
      "epoch": 4.074457593688363,
      "grad_norm": 0.2421875,
      "learning_rate": 2.016960873277428e-05,
      "loss": 0.0202,
      "step": 16526
    },
    {
      "epoch": 4.074950690335306,
      "grad_norm": 0.263671875,
      "learning_rate": 2.0148881058091763e-05,
      "loss": 0.0214,
      "step": 16528
    },
    {
      "epoch": 4.075443786982248,
      "grad_norm": 0.283203125,
      "learning_rate": 2.012816284621264e-05,
      "loss": 0.0191,
      "step": 16530
    },
    {
      "epoch": 4.075936883629192,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.010745409959217e-05,
      "loss": 0.0216,
      "step": 16532
    },
    {
      "epoch": 4.076429980276134,
      "grad_norm": 0.2265625,
      "learning_rate": 2.0086754820684483e-05,
      "loss": 0.022,
      "step": 16534
    },
    {
      "epoch": 4.076923076923077,
      "grad_norm": 0.251953125,
      "learning_rate": 2.006606501194247e-05,
      "loss": 0.0203,
      "step": 16536
    },
    {
      "epoch": 4.07741617357002,
      "grad_norm": 0.2275390625,
      "learning_rate": 2.00453846758181e-05,
      "loss": 0.0232,
      "step": 16538
    },
    {
      "epoch": 4.077909270216963,
      "grad_norm": 0.248046875,
      "learning_rate": 2.002471381476201e-05,
      "loss": 0.0202,
      "step": 16540
    },
    {
      "epoch": 4.078402366863905,
      "grad_norm": 0.2373046875,
      "learning_rate": 2.0004052431223897e-05,
      "loss": 0.0223,
      "step": 16542
    },
    {
      "epoch": 4.0788954635108485,
      "grad_norm": 0.24609375,
      "learning_rate": 1.9983400527652197e-05,
      "loss": 0.0216,
      "step": 16544
    },
    {
      "epoch": 4.079388560157791,
      "grad_norm": 0.265625,
      "learning_rate": 1.9962758106494284e-05,
      "loss": 0.0218,
      "step": 16546
    },
    {
      "epoch": 4.079881656804734,
      "grad_norm": 0.2412109375,
      "learning_rate": 1.9942125170196436e-05,
      "loss": 0.0202,
      "step": 16548
    },
    {
      "epoch": 4.080374753451676,
      "grad_norm": 0.271484375,
      "learning_rate": 1.9921501721203684e-05,
      "loss": 0.0207,
      "step": 16550
    },
    {
      "epoch": 4.0808678500986195,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.9900887761960128e-05,
      "loss": 0.0196,
      "step": 16552
    },
    {
      "epoch": 4.081360946745562,
      "grad_norm": 0.251953125,
      "learning_rate": 1.988028329490854e-05,
      "loss": 0.0196,
      "step": 16554
    },
    {
      "epoch": 4.081854043392505,
      "grad_norm": 0.25390625,
      "learning_rate": 1.9859688322490765e-05,
      "loss": 0.0211,
      "step": 16556
    },
    {
      "epoch": 4.082347140039448,
      "grad_norm": 0.24609375,
      "learning_rate": 1.983910284714735e-05,
      "loss": 0.0203,
      "step": 16558
    },
    {
      "epoch": 4.0828402366863905,
      "grad_norm": 0.2275390625,
      "learning_rate": 1.98185268713178e-05,
      "loss": 0.0185,
      "step": 16560
    },
    {
      "epoch": 4.083333333333333,
      "grad_norm": 0.255859375,
      "learning_rate": 1.979796039744053e-05,
      "loss": 0.0203,
      "step": 16562
    },
    {
      "epoch": 4.0838264299802765,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.977740342795268e-05,
      "loss": 0.0231,
      "step": 16564
    },
    {
      "epoch": 4.084319526627219,
      "grad_norm": 0.24609375,
      "learning_rate": 1.9756855965290487e-05,
      "loss": 0.0205,
      "step": 16566
    },
    {
      "epoch": 4.0848126232741615,
      "grad_norm": 0.2451171875,
      "learning_rate": 1.973631801188882e-05,
      "loss": 0.0247,
      "step": 16568
    },
    {
      "epoch": 4.085305719921105,
      "grad_norm": 0.265625,
      "learning_rate": 1.9715789570181674e-05,
      "loss": 0.0222,
      "step": 16570
    },
    {
      "epoch": 4.085798816568047,
      "grad_norm": 0.294921875,
      "learning_rate": 1.969527064260167e-05,
      "loss": 0.0234,
      "step": 16572
    },
    {
      "epoch": 4.08629191321499,
      "grad_norm": 0.2412109375,
      "learning_rate": 1.9674761231580475e-05,
      "loss": 0.0223,
      "step": 16574
    },
    {
      "epoch": 4.0867850098619325,
      "grad_norm": 0.28125,
      "learning_rate": 1.965426133954854e-05,
      "loss": 0.0204,
      "step": 16576
    },
    {
      "epoch": 4.087278106508876,
      "grad_norm": 0.224609375,
      "learning_rate": 1.963377096893524e-05,
      "loss": 0.0214,
      "step": 16578
    },
    {
      "epoch": 4.087771203155818,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.9613290122168802e-05,
      "loss": 0.0226,
      "step": 16580
    },
    {
      "epoch": 4.088264299802761,
      "grad_norm": 0.23046875,
      "learning_rate": 1.959281880167626e-05,
      "loss": 0.0183,
      "step": 16582
    },
    {
      "epoch": 4.088757396449704,
      "grad_norm": 0.2431640625,
      "learning_rate": 1.957235700988368e-05,
      "loss": 0.0227,
      "step": 16584
    },
    {
      "epoch": 4.089250493096647,
      "grad_norm": 0.302734375,
      "learning_rate": 1.955190474921581e-05,
      "loss": 0.0219,
      "step": 16586
    },
    {
      "epoch": 4.089743589743589,
      "grad_norm": 0.291015625,
      "learning_rate": 1.9531462022096403e-05,
      "loss": 0.019,
      "step": 16588
    },
    {
      "epoch": 4.090236686390533,
      "grad_norm": 0.220703125,
      "learning_rate": 1.951102883094802e-05,
      "loss": 0.0199,
      "step": 16590
    },
    {
      "epoch": 4.090729783037475,
      "grad_norm": 0.234375,
      "learning_rate": 1.9490605178192113e-05,
      "loss": 0.022,
      "step": 16592
    },
    {
      "epoch": 4.091222879684418,
      "grad_norm": 0.208984375,
      "learning_rate": 1.9470191066249033e-05,
      "loss": 0.021,
      "step": 16594
    },
    {
      "epoch": 4.091715976331361,
      "grad_norm": 0.2333984375,
      "learning_rate": 1.944978649753787e-05,
      "loss": 0.0189,
      "step": 16596
    },
    {
      "epoch": 4.092209072978304,
      "grad_norm": 0.267578125,
      "learning_rate": 1.9429391474476797e-05,
      "loss": 0.0188,
      "step": 16598
    },
    {
      "epoch": 4.092702169625246,
      "grad_norm": 0.2255859375,
      "learning_rate": 1.9409005999482666e-05,
      "loss": 0.0197,
      "step": 16600
    },
    {
      "epoch": 4.09319526627219,
      "grad_norm": 0.28125,
      "learning_rate": 1.938863007497127e-05,
      "loss": 0.0202,
      "step": 16602
    },
    {
      "epoch": 4.093688362919132,
      "grad_norm": 0.2578125,
      "learning_rate": 1.9368263703357302e-05,
      "loss": 0.0216,
      "step": 16604
    },
    {
      "epoch": 4.094181459566075,
      "grad_norm": 0.25390625,
      "learning_rate": 1.934790688705427e-05,
      "loss": 0.0217,
      "step": 16606
    },
    {
      "epoch": 4.094674556213017,
      "grad_norm": 0.283203125,
      "learning_rate": 1.9327559628474566e-05,
      "loss": 0.0226,
      "step": 16608
    },
    {
      "epoch": 4.095167652859961,
      "grad_norm": 0.267578125,
      "learning_rate": 1.9307221930029472e-05,
      "loss": 0.0212,
      "step": 16610
    },
    {
      "epoch": 4.095660749506903,
      "grad_norm": 0.2080078125,
      "learning_rate": 1.9286893794129134e-05,
      "loss": 0.0199,
      "step": 16612
    },
    {
      "epoch": 4.096153846153846,
      "grad_norm": 0.263671875,
      "learning_rate": 1.9266575223182503e-05,
      "loss": 0.0223,
      "step": 16614
    },
    {
      "epoch": 4.096646942800789,
      "grad_norm": 0.2578125,
      "learning_rate": 1.9246266219597465e-05,
      "loss": 0.0255,
      "step": 16616
    },
    {
      "epoch": 4.097140039447732,
      "grad_norm": 0.263671875,
      "learning_rate": 1.9225966785780748e-05,
      "loss": 0.0205,
      "step": 16618
    },
    {
      "epoch": 4.097633136094674,
      "grad_norm": 0.224609375,
      "learning_rate": 1.920567692413795e-05,
      "loss": 0.0214,
      "step": 16620
    },
    {
      "epoch": 4.098126232741618,
      "grad_norm": 0.298828125,
      "learning_rate": 1.9185396637073537e-05,
      "loss": 0.0242,
      "step": 16622
    },
    {
      "epoch": 4.09861932938856,
      "grad_norm": 0.283203125,
      "learning_rate": 1.9165125926990834e-05,
      "loss": 0.0251,
      "step": 16624
    },
    {
      "epoch": 4.099112426035503,
      "grad_norm": 0.216796875,
      "learning_rate": 1.914486479629206e-05,
      "loss": 0.0201,
      "step": 16626
    },
    {
      "epoch": 4.099605522682446,
      "grad_norm": 0.3203125,
      "learning_rate": 1.912461324737822e-05,
      "loss": 0.0191,
      "step": 16628
    },
    {
      "epoch": 4.100098619329389,
      "grad_norm": 0.310546875,
      "learning_rate": 1.9104371282649257e-05,
      "loss": 0.0246,
      "step": 16630
    },
    {
      "epoch": 4.100591715976331,
      "grad_norm": 0.26171875,
      "learning_rate": 1.9084138904503966e-05,
      "loss": 0.0226,
      "step": 16632
    },
    {
      "epoch": 4.101084812623274,
      "grad_norm": 0.2392578125,
      "learning_rate": 1.9063916115339984e-05,
      "loss": 0.0204,
      "step": 16634
    },
    {
      "epoch": 4.101577909270217,
      "grad_norm": 0.267578125,
      "learning_rate": 1.9043702917553828e-05,
      "loss": 0.02,
      "step": 16636
    },
    {
      "epoch": 4.10207100591716,
      "grad_norm": 0.2294921875,
      "learning_rate": 1.9023499313540893e-05,
      "loss": 0.0207,
      "step": 16638
    },
    {
      "epoch": 4.102564102564102,
      "grad_norm": 0.255859375,
      "learning_rate": 1.9003305305695395e-05,
      "loss": 0.0247,
      "step": 16640
    },
    {
      "epoch": 4.103057199211046,
      "grad_norm": 0.21875,
      "learning_rate": 1.8983120896410468e-05,
      "loss": 0.0204,
      "step": 16642
    },
    {
      "epoch": 4.103550295857988,
      "grad_norm": 0.255859375,
      "learning_rate": 1.8962946088078037e-05,
      "loss": 0.0207,
      "step": 16644
    },
    {
      "epoch": 4.104043392504931,
      "grad_norm": 0.373046875,
      "learning_rate": 1.894278088308894e-05,
      "loss": 0.0279,
      "step": 16646
    },
    {
      "epoch": 4.104536489151874,
      "grad_norm": 0.251953125,
      "learning_rate": 1.8922625283832886e-05,
      "loss": 0.0259,
      "step": 16648
    },
    {
      "epoch": 4.105029585798817,
      "grad_norm": 0.25,
      "learning_rate": 1.8902479292698405e-05,
      "loss": 0.0199,
      "step": 16650
    },
    {
      "epoch": 4.105522682445759,
      "grad_norm": 0.271484375,
      "learning_rate": 1.8882342912072913e-05,
      "loss": 0.0177,
      "step": 16652
    },
    {
      "epoch": 4.1060157790927025,
      "grad_norm": 0.345703125,
      "learning_rate": 1.8862216144342692e-05,
      "loss": 0.0265,
      "step": 16654
    },
    {
      "epoch": 4.106508875739645,
      "grad_norm": 0.34765625,
      "learning_rate": 1.884209899189291e-05,
      "loss": 0.0262,
      "step": 16656
    },
    {
      "epoch": 4.107001972386588,
      "grad_norm": 0.271484375,
      "learning_rate": 1.882199145710747e-05,
      "loss": 0.0238,
      "step": 16658
    },
    {
      "epoch": 4.107495069033531,
      "grad_norm": 0.2392578125,
      "learning_rate": 1.880189354236932e-05,
      "loss": 0.0208,
      "step": 16660
    },
    {
      "epoch": 4.1079881656804735,
      "grad_norm": 0.265625,
      "learning_rate": 1.8781805250060125e-05,
      "loss": 0.0283,
      "step": 16662
    },
    {
      "epoch": 4.108481262327416,
      "grad_norm": 0.29296875,
      "learning_rate": 1.8761726582560467e-05,
      "loss": 0.0266,
      "step": 16664
    },
    {
      "epoch": 4.108974358974359,
      "grad_norm": 0.2373046875,
      "learning_rate": 1.874165754224979e-05,
      "loss": 0.0208,
      "step": 16666
    },
    {
      "epoch": 4.109467455621302,
      "grad_norm": 0.259765625,
      "learning_rate": 1.872159813150638e-05,
      "loss": 0.0217,
      "step": 16668
    },
    {
      "epoch": 4.1099605522682445,
      "grad_norm": 0.27734375,
      "learning_rate": 1.870154835270742e-05,
      "loss": 0.0274,
      "step": 16670
    },
    {
      "epoch": 4.110453648915187,
      "grad_norm": 0.27734375,
      "learning_rate": 1.8681508208228836e-05,
      "loss": 0.0307,
      "step": 16672
    },
    {
      "epoch": 4.1109467455621305,
      "grad_norm": 0.251953125,
      "learning_rate": 1.8661477700445606e-05,
      "loss": 0.0228,
      "step": 16674
    },
    {
      "epoch": 4.111439842209073,
      "grad_norm": 0.33203125,
      "learning_rate": 1.8641456831731386e-05,
      "loss": 0.023,
      "step": 16676
    },
    {
      "epoch": 4.1119329388560155,
      "grad_norm": 0.283203125,
      "learning_rate": 1.8621445604458765e-05,
      "loss": 0.0281,
      "step": 16678
    },
    {
      "epoch": 4.112426035502959,
      "grad_norm": 0.330078125,
      "learning_rate": 1.8601444020999215e-05,
      "loss": 0.0298,
      "step": 16680
    },
    {
      "epoch": 4.1129191321499015,
      "grad_norm": 0.2392578125,
      "learning_rate": 1.8581452083723007e-05,
      "loss": 0.0212,
      "step": 16682
    },
    {
      "epoch": 4.113412228796844,
      "grad_norm": 0.2451171875,
      "learning_rate": 1.856146979499934e-05,
      "loss": 0.0188,
      "step": 16684
    },
    {
      "epoch": 4.113905325443787,
      "grad_norm": 0.30859375,
      "learning_rate": 1.8541497157196153e-05,
      "loss": 0.0315,
      "step": 16686
    },
    {
      "epoch": 4.11439842209073,
      "grad_norm": 0.33203125,
      "learning_rate": 1.852153417268041e-05,
      "loss": 0.0281,
      "step": 16688
    },
    {
      "epoch": 4.1148915187376724,
      "grad_norm": 0.255859375,
      "learning_rate": 1.8501580843817723e-05,
      "loss": 0.0214,
      "step": 16690
    },
    {
      "epoch": 4.115384615384615,
      "grad_norm": 0.2451171875,
      "learning_rate": 1.848163717297279e-05,
      "loss": 0.0219,
      "step": 16692
    },
    {
      "epoch": 4.115877712031558,
      "grad_norm": 0.302734375,
      "learning_rate": 1.846170316250897e-05,
      "loss": 0.028,
      "step": 16694
    },
    {
      "epoch": 4.116370808678501,
      "grad_norm": 0.3203125,
      "learning_rate": 1.8441778814788578e-05,
      "loss": 0.0315,
      "step": 16696
    },
    {
      "epoch": 4.116863905325443,
      "grad_norm": 0.294921875,
      "learning_rate": 1.8421864132172785e-05,
      "loss": 0.0183,
      "step": 16698
    },
    {
      "epoch": 4.117357001972387,
      "grad_norm": 0.2431640625,
      "learning_rate": 1.8401959117021507e-05,
      "loss": 0.0263,
      "step": 16700
    },
    {
      "epoch": 4.117850098619329,
      "grad_norm": 0.287109375,
      "learning_rate": 1.8382063771693714e-05,
      "loss": 0.0326,
      "step": 16702
    },
    {
      "epoch": 4.118343195266272,
      "grad_norm": 0.3125,
      "learning_rate": 1.8362178098547013e-05,
      "loss": 0.0287,
      "step": 16704
    },
    {
      "epoch": 4.118836291913215,
      "grad_norm": 0.29296875,
      "learning_rate": 1.8342302099938057e-05,
      "loss": 0.0264,
      "step": 16706
    },
    {
      "epoch": 4.119329388560158,
      "grad_norm": 0.2490234375,
      "learning_rate": 1.83224357782222e-05,
      "loss": 0.0231,
      "step": 16708
    },
    {
      "epoch": 4.1198224852071,
      "grad_norm": 0.302734375,
      "learning_rate": 1.830257913575374e-05,
      "loss": 0.0304,
      "step": 16710
    },
    {
      "epoch": 4.120315581854044,
      "grad_norm": 0.271484375,
      "learning_rate": 1.828273217488581e-05,
      "loss": 0.0284,
      "step": 16712
    },
    {
      "epoch": 4.120808678500986,
      "grad_norm": 0.2412109375,
      "learning_rate": 1.8262894897970318e-05,
      "loss": 0.0232,
      "step": 16714
    },
    {
      "epoch": 4.121301775147929,
      "grad_norm": 0.30078125,
      "learning_rate": 1.82430673073582e-05,
      "loss": 0.0247,
      "step": 16716
    },
    {
      "epoch": 4.121794871794871,
      "grad_norm": 0.298828125,
      "learning_rate": 1.822324940539901e-05,
      "loss": 0.0314,
      "step": 16718
    },
    {
      "epoch": 4.122287968441815,
      "grad_norm": 0.30078125,
      "learning_rate": 1.820344119444142e-05,
      "loss": 0.0296,
      "step": 16720
    },
    {
      "epoch": 4.122781065088757,
      "grad_norm": 0.248046875,
      "learning_rate": 1.8183642676832703e-05,
      "loss": 0.0259,
      "step": 16722
    },
    {
      "epoch": 4.1232741617357,
      "grad_norm": 0.298828125,
      "learning_rate": 1.8163853854919134e-05,
      "loss": 0.0245,
      "step": 16724
    },
    {
      "epoch": 4.123767258382643,
      "grad_norm": 0.294921875,
      "learning_rate": 1.8144074731045822e-05,
      "loss": 0.0305,
      "step": 16726
    },
    {
      "epoch": 4.124260355029586,
      "grad_norm": 0.298828125,
      "learning_rate": 1.8124305307556632e-05,
      "loss": 0.0297,
      "step": 16728
    },
    {
      "epoch": 4.124753451676528,
      "grad_norm": 0.265625,
      "learning_rate": 1.8104545586794453e-05,
      "loss": 0.0286,
      "step": 16730
    },
    {
      "epoch": 4.125246548323472,
      "grad_norm": 0.3515625,
      "learning_rate": 1.808479557110081e-05,
      "loss": 0.0258,
      "step": 16732
    },
    {
      "epoch": 4.125739644970414,
      "grad_norm": 0.291015625,
      "learning_rate": 1.806505526281631e-05,
      "loss": 0.0276,
      "step": 16734
    },
    {
      "epoch": 4.126232741617357,
      "grad_norm": 0.322265625,
      "learning_rate": 1.8045324664280195e-05,
      "loss": 0.0302,
      "step": 16736
    },
    {
      "epoch": 4.1267258382643,
      "grad_norm": 0.296875,
      "learning_rate": 1.802560377783068e-05,
      "loss": 0.0282,
      "step": 16738
    },
    {
      "epoch": 4.127218934911243,
      "grad_norm": 0.275390625,
      "learning_rate": 1.8005892605804796e-05,
      "loss": 0.032,
      "step": 16740
    },
    {
      "epoch": 4.127712031558185,
      "grad_norm": 0.326171875,
      "learning_rate": 1.7986191150538456e-05,
      "loss": 0.0328,
      "step": 16742
    },
    {
      "epoch": 4.128205128205128,
      "grad_norm": 0.31640625,
      "learning_rate": 1.796649941436638e-05,
      "loss": 0.0326,
      "step": 16744
    },
    {
      "epoch": 4.128698224852071,
      "grad_norm": 0.306640625,
      "learning_rate": 1.7946817399622097e-05,
      "loss": 0.0284,
      "step": 16746
    },
    {
      "epoch": 4.129191321499014,
      "grad_norm": 0.2890625,
      "learning_rate": 1.792714510863812e-05,
      "loss": 0.028,
      "step": 16748
    },
    {
      "epoch": 4.129684418145956,
      "grad_norm": 0.279296875,
      "learning_rate": 1.7907482543745667e-05,
      "loss": 0.0293,
      "step": 16750
    },
    {
      "epoch": 4.1301775147929,
      "grad_norm": 0.27734375,
      "learning_rate": 1.7887829707274884e-05,
      "loss": 0.0308,
      "step": 16752
    },
    {
      "epoch": 4.130670611439842,
      "grad_norm": 0.283203125,
      "learning_rate": 1.7868186601554726e-05,
      "loss": 0.027,
      "step": 16754
    },
    {
      "epoch": 4.131163708086785,
      "grad_norm": 0.275390625,
      "learning_rate": 1.7848553228913024e-05,
      "loss": 0.0246,
      "step": 16756
    },
    {
      "epoch": 4.131656804733728,
      "grad_norm": 0.30859375,
      "learning_rate": 1.7828929591676458e-05,
      "loss": 0.0289,
      "step": 16758
    },
    {
      "epoch": 4.132149901380671,
      "grad_norm": 0.29296875,
      "learning_rate": 1.780931569217048e-05,
      "loss": 0.0308,
      "step": 16760
    },
    {
      "epoch": 4.132642998027613,
      "grad_norm": 0.287109375,
      "learning_rate": 1.7789711532719532e-05,
      "loss": 0.0281,
      "step": 16762
    },
    {
      "epoch": 4.133136094674557,
      "grad_norm": 0.263671875,
      "learning_rate": 1.7770117115646744e-05,
      "loss": 0.0281,
      "step": 16764
    },
    {
      "epoch": 4.133629191321499,
      "grad_norm": 0.337890625,
      "learning_rate": 1.775053244327419e-05,
      "loss": 0.032,
      "step": 16766
    },
    {
      "epoch": 4.134122287968442,
      "grad_norm": 0.365234375,
      "learning_rate": 1.7730957517922774e-05,
      "loss": 0.03,
      "step": 16768
    },
    {
      "epoch": 4.134615384615385,
      "grad_norm": 0.330078125,
      "learning_rate": 1.7711392341912214e-05,
      "loss": 0.0355,
      "step": 16770
    },
    {
      "epoch": 4.1351084812623276,
      "grad_norm": 0.30859375,
      "learning_rate": 1.7691836917561112e-05,
      "loss": 0.031,
      "step": 16772
    },
    {
      "epoch": 4.13560157790927,
      "grad_norm": 0.296875,
      "learning_rate": 1.7672291247186866e-05,
      "loss": 0.0296,
      "step": 16774
    },
    {
      "epoch": 4.136094674556213,
      "grad_norm": 0.298828125,
      "learning_rate": 1.7652755333105808e-05,
      "loss": 0.029,
      "step": 16776
    },
    {
      "epoch": 4.136587771203156,
      "grad_norm": 0.29296875,
      "learning_rate": 1.7633229177632982e-05,
      "loss": 0.0287,
      "step": 16778
    },
    {
      "epoch": 4.1370808678500985,
      "grad_norm": 0.349609375,
      "learning_rate": 1.761371278308237e-05,
      "loss": 0.0288,
      "step": 16780
    },
    {
      "epoch": 4.137573964497041,
      "grad_norm": 0.310546875,
      "learning_rate": 1.759420615176679e-05,
      "loss": 0.0329,
      "step": 16782
    },
    {
      "epoch": 4.1380670611439845,
      "grad_norm": 0.28125,
      "learning_rate": 1.757470928599787e-05,
      "loss": 0.0298,
      "step": 16784
    },
    {
      "epoch": 4.138560157790927,
      "grad_norm": 0.27734375,
      "learning_rate": 1.7555222188086106e-05,
      "loss": 0.0258,
      "step": 16786
    },
    {
      "epoch": 4.1390532544378695,
      "grad_norm": 0.279296875,
      "learning_rate": 1.753574486034082e-05,
      "loss": 0.0279,
      "step": 16788
    },
    {
      "epoch": 4.139546351084813,
      "grad_norm": 0.341796875,
      "learning_rate": 1.7516277305070196e-05,
      "loss": 0.028,
      "step": 16790
    },
    {
      "epoch": 4.1400394477317555,
      "grad_norm": 0.349609375,
      "learning_rate": 1.7496819524581264e-05,
      "loss": 0.0362,
      "step": 16792
    },
    {
      "epoch": 4.140532544378698,
      "grad_norm": 0.28125,
      "learning_rate": 1.747737152117983e-05,
      "loss": 0.03,
      "step": 16794
    },
    {
      "epoch": 4.141025641025641,
      "grad_norm": 0.330078125,
      "learning_rate": 1.7457933297170615e-05,
      "loss": 0.0333,
      "step": 16796
    },
    {
      "epoch": 4.141518737672584,
      "grad_norm": 0.333984375,
      "learning_rate": 1.7438504854857163e-05,
      "loss": 0.038,
      "step": 16798
    },
    {
      "epoch": 4.1420118343195265,
      "grad_norm": 0.296875,
      "learning_rate": 1.7419086196541855e-05,
      "loss": 0.0328,
      "step": 16800
    },
    {
      "epoch": 4.142504930966469,
      "grad_norm": 0.341796875,
      "learning_rate": 1.7399677324525897e-05,
      "loss": 0.0322,
      "step": 16802
    },
    {
      "epoch": 4.142998027613412,
      "grad_norm": 0.330078125,
      "learning_rate": 1.7380278241109372e-05,
      "loss": 0.0323,
      "step": 16804
    },
    {
      "epoch": 4.143491124260355,
      "grad_norm": 0.322265625,
      "learning_rate": 1.736088894859118e-05,
      "loss": 0.0353,
      "step": 16806
    },
    {
      "epoch": 4.1439842209072975,
      "grad_norm": 0.322265625,
      "learning_rate": 1.734150944926902e-05,
      "loss": 0.0328,
      "step": 16808
    },
    {
      "epoch": 4.144477317554241,
      "grad_norm": 0.283203125,
      "learning_rate": 1.7322139745439492e-05,
      "loss": 0.0295,
      "step": 16810
    },
    {
      "epoch": 4.144970414201183,
      "grad_norm": 0.3515625,
      "learning_rate": 1.7302779839398032e-05,
      "loss": 0.0293,
      "step": 16812
    },
    {
      "epoch": 4.145463510848126,
      "grad_norm": 0.359375,
      "learning_rate": 1.728342973343887e-05,
      "loss": 0.0335,
      "step": 16814
    },
    {
      "epoch": 4.145956607495069,
      "grad_norm": 0.34375,
      "learning_rate": 1.7264089429855112e-05,
      "loss": 0.0314,
      "step": 16816
    },
    {
      "epoch": 4.146449704142012,
      "grad_norm": 0.306640625,
      "learning_rate": 1.724475893093871e-05,
      "loss": 0.0314,
      "step": 16818
    },
    {
      "epoch": 4.146942800788954,
      "grad_norm": 0.279296875,
      "learning_rate": 1.722543823898043e-05,
      "loss": 0.0315,
      "step": 16820
    },
    {
      "epoch": 4.147435897435898,
      "grad_norm": 0.337890625,
      "learning_rate": 1.7206127356269828e-05,
      "loss": 0.035,
      "step": 16822
    },
    {
      "epoch": 4.14792899408284,
      "grad_norm": 0.298828125,
      "learning_rate": 1.7186826285095458e-05,
      "loss": 0.0348,
      "step": 16824
    },
    {
      "epoch": 4.148422090729783,
      "grad_norm": 0.3359375,
      "learning_rate": 1.7167535027744507e-05,
      "loss": 0.0363,
      "step": 16826
    },
    {
      "epoch": 4.148915187376726,
      "grad_norm": 0.30078125,
      "learning_rate": 1.7148253586503127e-05,
      "loss": 0.0315,
      "step": 16828
    },
    {
      "epoch": 4.149408284023669,
      "grad_norm": 0.318359375,
      "learning_rate": 1.712898196365629e-05,
      "loss": 0.0335,
      "step": 16830
    },
    {
      "epoch": 4.149901380670611,
      "grad_norm": 0.33203125,
      "learning_rate": 1.7109720161487785e-05,
      "loss": 0.0354,
      "step": 16832
    },
    {
      "epoch": 4.150394477317554,
      "grad_norm": 0.35546875,
      "learning_rate": 1.709046818228024e-05,
      "loss": 0.0334,
      "step": 16834
    },
    {
      "epoch": 4.150887573964497,
      "grad_norm": 0.298828125,
      "learning_rate": 1.7071226028315113e-05,
      "loss": 0.0308,
      "step": 16836
    },
    {
      "epoch": 4.15138067061144,
      "grad_norm": 0.322265625,
      "learning_rate": 1.7051993701872747e-05,
      "loss": 0.0332,
      "step": 16838
    },
    {
      "epoch": 4.151873767258382,
      "grad_norm": 0.33203125,
      "learning_rate": 1.70327712052322e-05,
      "loss": 0.0304,
      "step": 16840
    },
    {
      "epoch": 4.152366863905326,
      "grad_norm": 0.275390625,
      "learning_rate": 1.701355854067155e-05,
      "loss": 0.0317,
      "step": 16842
    },
    {
      "epoch": 4.152859960552268,
      "grad_norm": 0.388671875,
      "learning_rate": 1.6994355710467524e-05,
      "loss": 0.0323,
      "step": 16844
    },
    {
      "epoch": 4.153353057199211,
      "grad_norm": 0.302734375,
      "learning_rate": 1.6975162716895787e-05,
      "loss": 0.0332,
      "step": 16846
    },
    {
      "epoch": 4.153846153846154,
      "grad_norm": 0.296875,
      "learning_rate": 1.695597956223082e-05,
      "loss": 0.0351,
      "step": 16848
    },
    {
      "epoch": 4.154339250493097,
      "grad_norm": 0.294921875,
      "learning_rate": 1.693680624874593e-05,
      "loss": 0.0301,
      "step": 16850
    },
    {
      "epoch": 4.154832347140039,
      "grad_norm": 0.33984375,
      "learning_rate": 1.6917642778713293e-05,
      "loss": 0.0333,
      "step": 16852
    },
    {
      "epoch": 4.155325443786983,
      "grad_norm": 0.294921875,
      "learning_rate": 1.6898489154403817e-05,
      "loss": 0.0308,
      "step": 16854
    },
    {
      "epoch": 4.155818540433925,
      "grad_norm": 0.376953125,
      "learning_rate": 1.6879345378087396e-05,
      "loss": 0.036,
      "step": 16856
    },
    {
      "epoch": 4.156311637080868,
      "grad_norm": 0.2890625,
      "learning_rate": 1.6860211452032615e-05,
      "loss": 0.0295,
      "step": 16858
    },
    {
      "epoch": 4.15680473372781,
      "grad_norm": 0.3671875,
      "learning_rate": 1.684108737850696e-05,
      "loss": 0.0307,
      "step": 16860
    },
    {
      "epoch": 4.157297830374754,
      "grad_norm": 0.337890625,
      "learning_rate": 1.6821973159776762e-05,
      "loss": 0.0318,
      "step": 16862
    },
    {
      "epoch": 4.157790927021696,
      "grad_norm": 0.328125,
      "learning_rate": 1.6802868798107142e-05,
      "loss": 0.0339,
      "step": 16864
    },
    {
      "epoch": 4.158284023668639,
      "grad_norm": 0.29296875,
      "learning_rate": 1.6783774295762112e-05,
      "loss": 0.0331,
      "step": 16866
    },
    {
      "epoch": 4.158777120315582,
      "grad_norm": 0.361328125,
      "learning_rate": 1.6764689655004394e-05,
      "loss": 0.0343,
      "step": 16868
    },
    {
      "epoch": 4.159270216962525,
      "grad_norm": 0.388671875,
      "learning_rate": 1.674561487809574e-05,
      "loss": 0.0383,
      "step": 16870
    },
    {
      "epoch": 4.159763313609467,
      "grad_norm": 0.380859375,
      "learning_rate": 1.6726549967296522e-05,
      "loss": 0.0367,
      "step": 16872
    },
    {
      "epoch": 4.160256410256411,
      "grad_norm": 0.3203125,
      "learning_rate": 1.6707494924866075e-05,
      "loss": 0.0299,
      "step": 16874
    },
    {
      "epoch": 4.160749506903353,
      "grad_norm": 0.36328125,
      "learning_rate": 1.6688449753062518e-05,
      "loss": 0.0333,
      "step": 16876
    },
    {
      "epoch": 4.161242603550296,
      "grad_norm": 0.333984375,
      "learning_rate": 1.666941445414283e-05,
      "loss": 0.0366,
      "step": 16878
    },
    {
      "epoch": 4.161735700197239,
      "grad_norm": 0.34765625,
      "learning_rate": 1.66503890303628e-05,
      "loss": 0.0431,
      "step": 16880
    },
    {
      "epoch": 4.162228796844182,
      "grad_norm": 0.3125,
      "learning_rate": 1.6631373483976985e-05,
      "loss": 0.0328,
      "step": 16882
    },
    {
      "epoch": 4.162721893491124,
      "grad_norm": 0.34375,
      "learning_rate": 1.661236781723894e-05,
      "loss": 0.0303,
      "step": 16884
    },
    {
      "epoch": 4.1632149901380675,
      "grad_norm": 0.3359375,
      "learning_rate": 1.6593372032400856e-05,
      "loss": 0.0354,
      "step": 16886
    },
    {
      "epoch": 4.16370808678501,
      "grad_norm": 0.306640625,
      "learning_rate": 1.657438613171387e-05,
      "loss": 0.0368,
      "step": 16888
    },
    {
      "epoch": 4.164201183431953,
      "grad_norm": 0.3125,
      "learning_rate": 1.6555410117427926e-05,
      "loss": 0.0322,
      "step": 16890
    },
    {
      "epoch": 4.164694280078895,
      "grad_norm": 0.380859375,
      "learning_rate": 1.6536443991791773e-05,
      "loss": 0.03,
      "step": 16892
    },
    {
      "epoch": 4.1651873767258385,
      "grad_norm": 0.314453125,
      "learning_rate": 1.651748775705303e-05,
      "loss": 0.0326,
      "step": 16894
    },
    {
      "epoch": 4.165680473372781,
      "grad_norm": 0.3046875,
      "learning_rate": 1.6498541415458036e-05,
      "loss": 0.0341,
      "step": 16896
    },
    {
      "epoch": 4.1661735700197235,
      "grad_norm": 0.349609375,
      "learning_rate": 1.647960496925216e-05,
      "loss": 0.0309,
      "step": 16898
    },
    {
      "epoch": 4.166666666666667,
      "grad_norm": 0.30859375,
      "learning_rate": 1.6460678420679377e-05,
      "loss": 0.0342,
      "step": 16900
    },
    {
      "epoch": 4.1671597633136095,
      "grad_norm": 0.322265625,
      "learning_rate": 1.6441761771982632e-05,
      "loss": 0.0364,
      "step": 16902
    },
    {
      "epoch": 4.167652859960552,
      "grad_norm": 0.32421875,
      "learning_rate": 1.6422855025403648e-05,
      "loss": 0.0356,
      "step": 16904
    },
    {
      "epoch": 4.168145956607495,
      "grad_norm": 0.3046875,
      "learning_rate": 1.6403958183182976e-05,
      "loss": 0.0296,
      "step": 16906
    },
    {
      "epoch": 4.168639053254438,
      "grad_norm": 0.310546875,
      "learning_rate": 1.6385071247560023e-05,
      "loss": 0.0322,
      "step": 16908
    },
    {
      "epoch": 4.1691321499013805,
      "grad_norm": 0.359375,
      "learning_rate": 1.6366194220772934e-05,
      "loss": 0.0355,
      "step": 16910
    },
    {
      "epoch": 4.169625246548324,
      "grad_norm": 0.322265625,
      "learning_rate": 1.6347327105058817e-05,
      "loss": 0.0359,
      "step": 16912
    },
    {
      "epoch": 4.170118343195266,
      "grad_norm": 0.37109375,
      "learning_rate": 1.6328469902653454e-05,
      "loss": 0.0349,
      "step": 16914
    },
    {
      "epoch": 4.170611439842209,
      "grad_norm": 0.310546875,
      "learning_rate": 1.6309622615791608e-05,
      "loss": 0.0342,
      "step": 16916
    },
    {
      "epoch": 4.1711045364891515,
      "grad_norm": 0.333984375,
      "learning_rate": 1.629078524670674e-05,
      "loss": 0.0327,
      "step": 16918
    },
    {
      "epoch": 4.171597633136095,
      "grad_norm": 0.3828125,
      "learning_rate": 1.6271957797631175e-05,
      "loss": 0.0356,
      "step": 16920
    },
    {
      "epoch": 4.172090729783037,
      "grad_norm": 0.33984375,
      "learning_rate": 1.625314027079611e-05,
      "loss": 0.0374,
      "step": 16922
    },
    {
      "epoch": 4.17258382642998,
      "grad_norm": 0.2890625,
      "learning_rate": 1.6234332668431485e-05,
      "loss": 0.0311,
      "step": 16924
    },
    {
      "epoch": 4.173076923076923,
      "grad_norm": 0.337890625,
      "learning_rate": 1.621553499276617e-05,
      "loss": 0.0379,
      "step": 16926
    },
    {
      "epoch": 4.173570019723866,
      "grad_norm": 0.361328125,
      "learning_rate": 1.6196747246027688e-05,
      "loss": 0.0404,
      "step": 16928
    },
    {
      "epoch": 4.174063116370808,
      "grad_norm": 0.30859375,
      "learning_rate": 1.6177969430442606e-05,
      "loss": 0.0331,
      "step": 16930
    },
    {
      "epoch": 4.174556213017752,
      "grad_norm": 0.333984375,
      "learning_rate": 1.6159201548236124e-05,
      "loss": 0.0314,
      "step": 16932
    },
    {
      "epoch": 4.175049309664694,
      "grad_norm": 0.33984375,
      "learning_rate": 1.614044360163237e-05,
      "loss": 0.0408,
      "step": 16934
    },
    {
      "epoch": 4.175542406311637,
      "grad_norm": 0.3203125,
      "learning_rate": 1.612169559285426e-05,
      "loss": 0.0364,
      "step": 16936
    },
    {
      "epoch": 4.17603550295858,
      "grad_norm": 0.330078125,
      "learning_rate": 1.610295752412353e-05,
      "loss": 0.0371,
      "step": 16938
    },
    {
      "epoch": 4.176528599605523,
      "grad_norm": 0.388671875,
      "learning_rate": 1.608422939766079e-05,
      "loss": 0.0366,
      "step": 16940
    },
    {
      "epoch": 4.177021696252465,
      "grad_norm": 0.33984375,
      "learning_rate": 1.6065511215685336e-05,
      "loss": 0.0379,
      "step": 16942
    },
    {
      "epoch": 4.177514792899408,
      "grad_norm": 0.318359375,
      "learning_rate": 1.604680298041549e-05,
      "loss": 0.0381,
      "step": 16944
    },
    {
      "epoch": 4.178007889546351,
      "grad_norm": 0.294921875,
      "learning_rate": 1.602810469406819e-05,
      "loss": 0.0335,
      "step": 16946
    },
    {
      "epoch": 4.178500986193294,
      "grad_norm": 0.283203125,
      "learning_rate": 1.600941635885933e-05,
      "loss": 0.0306,
      "step": 16948
    },
    {
      "epoch": 4.178994082840236,
      "grad_norm": 0.328125,
      "learning_rate": 1.5990737977003577e-05,
      "loss": 0.0322,
      "step": 16950
    },
    {
      "epoch": 4.17948717948718,
      "grad_norm": 0.318359375,
      "learning_rate": 1.597206955071442e-05,
      "loss": 0.0364,
      "step": 16952
    },
    {
      "epoch": 4.179980276134122,
      "grad_norm": 0.29296875,
      "learning_rate": 1.5953411082204174e-05,
      "loss": 0.0334,
      "step": 16954
    },
    {
      "epoch": 4.180473372781065,
      "grad_norm": 0.333984375,
      "learning_rate": 1.5934762573683958e-05,
      "loss": 0.0343,
      "step": 16956
    },
    {
      "epoch": 4.180966469428008,
      "grad_norm": 0.34375,
      "learning_rate": 1.5916124027363776e-05,
      "loss": 0.034,
      "step": 16958
    },
    {
      "epoch": 4.181459566074951,
      "grad_norm": 0.328125,
      "learning_rate": 1.5897495445452338e-05,
      "loss": 0.0341,
      "step": 16960
    },
    {
      "epoch": 4.181952662721893,
      "grad_norm": 0.3125,
      "learning_rate": 1.587887683015725e-05,
      "loss": 0.0319,
      "step": 16962
    },
    {
      "epoch": 4.182445759368837,
      "grad_norm": 0.310546875,
      "learning_rate": 1.5860268183684933e-05,
      "loss": 0.0322,
      "step": 16964
    },
    {
      "epoch": 4.182938856015779,
      "grad_norm": 0.296875,
      "learning_rate": 1.584166950824061e-05,
      "loss": 0.0339,
      "step": 16966
    },
    {
      "epoch": 4.183431952662722,
      "grad_norm": 0.291015625,
      "learning_rate": 1.5823080806028333e-05,
      "loss": 0.0334,
      "step": 16968
    },
    {
      "epoch": 4.183925049309664,
      "grad_norm": 0.3203125,
      "learning_rate": 1.580450207925096e-05,
      "loss": 0.0323,
      "step": 16970
    },
    {
      "epoch": 4.184418145956608,
      "grad_norm": 0.369140625,
      "learning_rate": 1.578593333011018e-05,
      "loss": 0.0335,
      "step": 16972
    },
    {
      "epoch": 4.18491124260355,
      "grad_norm": 0.41015625,
      "learning_rate": 1.5767374560806514e-05,
      "loss": 0.0387,
      "step": 16974
    },
    {
      "epoch": 4.185404339250493,
      "grad_norm": 0.34765625,
      "learning_rate": 1.5748825773539245e-05,
      "loss": 0.0345,
      "step": 16976
    },
    {
      "epoch": 4.185897435897436,
      "grad_norm": 0.296875,
      "learning_rate": 1.57302869705065e-05,
      "loss": 0.0335,
      "step": 16978
    },
    {
      "epoch": 4.186390532544379,
      "grad_norm": 0.318359375,
      "learning_rate": 1.5711758153905266e-05,
      "loss": 0.0328,
      "step": 16980
    },
    {
      "epoch": 4.186883629191321,
      "grad_norm": 0.314453125,
      "learning_rate": 1.5693239325931296e-05,
      "loss": 0.0359,
      "step": 16982
    },
    {
      "epoch": 4.187376725838265,
      "grad_norm": 0.345703125,
      "learning_rate": 1.5674730488779166e-05,
      "loss": 0.038,
      "step": 16984
    },
    {
      "epoch": 4.187869822485207,
      "grad_norm": 0.306640625,
      "learning_rate": 1.5656231644642294e-05,
      "loss": 0.0342,
      "step": 16986
    },
    {
      "epoch": 4.18836291913215,
      "grad_norm": 0.287109375,
      "learning_rate": 1.5637742795712908e-05,
      "loss": 0.0335,
      "step": 16988
    },
    {
      "epoch": 4.188856015779093,
      "grad_norm": 0.373046875,
      "learning_rate": 1.5619263944181995e-05,
      "loss": 0.0359,
      "step": 16990
    },
    {
      "epoch": 4.189349112426036,
      "grad_norm": 0.34375,
      "learning_rate": 1.560079509223942e-05,
      "loss": 0.0383,
      "step": 16992
    },
    {
      "epoch": 4.189842209072978,
      "grad_norm": 0.318359375,
      "learning_rate": 1.5582336242073858e-05,
      "loss": 0.0404,
      "step": 16994
    },
    {
      "epoch": 4.1903353057199215,
      "grad_norm": 0.306640625,
      "learning_rate": 1.5563887395872777e-05,
      "loss": 0.0337,
      "step": 16996
    },
    {
      "epoch": 4.190828402366864,
      "grad_norm": 0.34375,
      "learning_rate": 1.5545448555822474e-05,
      "loss": 0.0353,
      "step": 16998
    },
    {
      "epoch": 4.191321499013807,
      "grad_norm": 0.337890625,
      "learning_rate": 1.552701972410805e-05,
      "loss": 0.0361,
      "step": 17000
    },
    {
      "epoch": 4.191814595660749,
      "grad_norm": 0.36328125,
      "learning_rate": 1.550860090291345e-05,
      "loss": 0.0347,
      "step": 17002
    },
    {
      "epoch": 4.1923076923076925,
      "grad_norm": 0.328125,
      "learning_rate": 1.549019209442133e-05,
      "loss": 0.0375,
      "step": 17004
    },
    {
      "epoch": 4.192800788954635,
      "grad_norm": 0.353515625,
      "learning_rate": 1.5471793300813344e-05,
      "loss": 0.034,
      "step": 17006
    },
    {
      "epoch": 4.193293885601578,
      "grad_norm": 0.34375,
      "learning_rate": 1.5453404524269766e-05,
      "loss": 0.0351,
      "step": 17008
    },
    {
      "epoch": 4.193786982248521,
      "grad_norm": 0.36328125,
      "learning_rate": 1.5435025766969803e-05,
      "loss": 0.0325,
      "step": 17010
    },
    {
      "epoch": 4.1942800788954635,
      "grad_norm": 0.294921875,
      "learning_rate": 1.5416657031091432e-05,
      "loss": 0.0346,
      "step": 17012
    },
    {
      "epoch": 4.194773175542406,
      "grad_norm": 0.314453125,
      "learning_rate": 1.5398298318811467e-05,
      "loss": 0.0387,
      "step": 17014
    },
    {
      "epoch": 4.195266272189349,
      "grad_norm": 0.33203125,
      "learning_rate": 1.5379949632305523e-05,
      "loss": 0.0369,
      "step": 17016
    },
    {
      "epoch": 4.195759368836292,
      "grad_norm": 0.322265625,
      "learning_rate": 1.5361610973747954e-05,
      "loss": 0.0365,
      "step": 17018
    },
    {
      "epoch": 4.1962524654832345,
      "grad_norm": 0.35546875,
      "learning_rate": 1.5343282345312093e-05,
      "loss": 0.039,
      "step": 17020
    },
    {
      "epoch": 4.196745562130178,
      "grad_norm": 0.34375,
      "learning_rate": 1.5324963749169917e-05,
      "loss": 0.0423,
      "step": 17022
    },
    {
      "epoch": 4.19723865877712,
      "grad_norm": 0.349609375,
      "learning_rate": 1.5306655187492302e-05,
      "loss": 0.0424,
      "step": 17024
    },
    {
      "epoch": 4.197731755424063,
      "grad_norm": 0.30078125,
      "learning_rate": 1.528835666244892e-05,
      "loss": 0.0329,
      "step": 17026
    },
    {
      "epoch": 4.1982248520710055,
      "grad_norm": 0.349609375,
      "learning_rate": 1.527006817620823e-05,
      "loss": 0.0306,
      "step": 17028
    },
    {
      "epoch": 4.198717948717949,
      "grad_norm": 0.33203125,
      "learning_rate": 1.5251789730937571e-05,
      "loss": 0.0358,
      "step": 17030
    },
    {
      "epoch": 4.199211045364891,
      "grad_norm": 0.310546875,
      "learning_rate": 1.5233521328802947e-05,
      "loss": 0.0369,
      "step": 17032
    },
    {
      "epoch": 4.199704142011834,
      "grad_norm": 0.32421875,
      "learning_rate": 1.5215262971969357e-05,
      "loss": 0.035,
      "step": 17034
    },
    {
      "epoch": 4.200197238658777,
      "grad_norm": 0.373046875,
      "learning_rate": 1.5197014662600451e-05,
      "loss": 0.0372,
      "step": 17036
    },
    {
      "epoch": 4.20069033530572,
      "grad_norm": 0.3671875,
      "learning_rate": 1.517877640285883e-05,
      "loss": 0.0392,
      "step": 17038
    },
    {
      "epoch": 4.201183431952662,
      "grad_norm": 0.298828125,
      "learning_rate": 1.5160548194905765e-05,
      "loss": 0.0339,
      "step": 17040
    },
    {
      "epoch": 4.201676528599606,
      "grad_norm": 0.3046875,
      "learning_rate": 1.514233004090141e-05,
      "loss": 0.0354,
      "step": 17042
    },
    {
      "epoch": 4.202169625246548,
      "grad_norm": 0.390625,
      "learning_rate": 1.5124121943004766e-05,
      "loss": 0.0387,
      "step": 17044
    },
    {
      "epoch": 4.202662721893491,
      "grad_norm": 0.33203125,
      "learning_rate": 1.5105923903373508e-05,
      "loss": 0.0413,
      "step": 17046
    },
    {
      "epoch": 4.203155818540434,
      "grad_norm": 0.310546875,
      "learning_rate": 1.5087735924164303e-05,
      "loss": 0.0354,
      "step": 17048
    },
    {
      "epoch": 4.203648915187377,
      "grad_norm": 0.291015625,
      "learning_rate": 1.5069558007532426e-05,
      "loss": 0.0313,
      "step": 17050
    },
    {
      "epoch": 4.204142011834319,
      "grad_norm": 0.34765625,
      "learning_rate": 1.5051390155632172e-05,
      "loss": 0.0316,
      "step": 17052
    },
    {
      "epoch": 4.204635108481263,
      "grad_norm": 0.306640625,
      "learning_rate": 1.5033232370616445e-05,
      "loss": 0.0364,
      "step": 17054
    },
    {
      "epoch": 4.205128205128205,
      "grad_norm": 0.34375,
      "learning_rate": 1.501508465463708e-05,
      "loss": 0.0357,
      "step": 17056
    },
    {
      "epoch": 4.205621301775148,
      "grad_norm": 0.36328125,
      "learning_rate": 1.4996947009844697e-05,
      "loss": 0.0342,
      "step": 17058
    },
    {
      "epoch": 4.20611439842209,
      "grad_norm": 0.32421875,
      "learning_rate": 1.4978819438388658e-05,
      "loss": 0.0363,
      "step": 17060
    },
    {
      "epoch": 4.206607495069034,
      "grad_norm": 0.31640625,
      "learning_rate": 1.4960701942417243e-05,
      "loss": 0.0374,
      "step": 17062
    },
    {
      "epoch": 4.207100591715976,
      "grad_norm": 0.27734375,
      "learning_rate": 1.4942594524077413e-05,
      "loss": 0.0388,
      "step": 17064
    },
    {
      "epoch": 4.207593688362919,
      "grad_norm": 0.384765625,
      "learning_rate": 1.4924497185515063e-05,
      "loss": 0.0352,
      "step": 17066
    },
    {
      "epoch": 4.208086785009862,
      "grad_norm": 0.30078125,
      "learning_rate": 1.4906409928874787e-05,
      "loss": 0.0352,
      "step": 17068
    },
    {
      "epoch": 4.208579881656805,
      "grad_norm": 0.34765625,
      "learning_rate": 1.4888332756300027e-05,
      "loss": 0.0368,
      "step": 17070
    },
    {
      "epoch": 4.209072978303747,
      "grad_norm": 0.337890625,
      "learning_rate": 1.487026566993307e-05,
      "loss": 0.0386,
      "step": 17072
    },
    {
      "epoch": 4.209566074950691,
      "grad_norm": 0.349609375,
      "learning_rate": 1.485220867191488e-05,
      "loss": 0.0382,
      "step": 17074
    },
    {
      "epoch": 4.210059171597633,
      "grad_norm": 0.34375,
      "learning_rate": 1.4834161764385423e-05,
      "loss": 0.0391,
      "step": 17076
    },
    {
      "epoch": 4.210552268244576,
      "grad_norm": 0.353515625,
      "learning_rate": 1.4816124949483245e-05,
      "loss": 0.0409,
      "step": 17078
    },
    {
      "epoch": 4.211045364891519,
      "grad_norm": 0.34375,
      "learning_rate": 1.4798098229345925e-05,
      "loss": 0.04,
      "step": 17080
    },
    {
      "epoch": 4.211538461538462,
      "grad_norm": 0.279296875,
      "learning_rate": 1.4780081606109642e-05,
      "loss": 0.0349,
      "step": 17082
    },
    {
      "epoch": 4.212031558185404,
      "grad_norm": 0.375,
      "learning_rate": 1.4762075081909487e-05,
      "loss": 0.0402,
      "step": 17084
    },
    {
      "epoch": 4.212524654832347,
      "grad_norm": 0.3203125,
      "learning_rate": 1.4744078658879357e-05,
      "loss": 0.0383,
      "step": 17086
    },
    {
      "epoch": 4.21301775147929,
      "grad_norm": 0.3125,
      "learning_rate": 1.4726092339151908e-05,
      "loss": 0.0413,
      "step": 17088
    },
    {
      "epoch": 4.213510848126233,
      "grad_norm": 0.357421875,
      "learning_rate": 1.4708116124858651e-05,
      "loss": 0.0355,
      "step": 17090
    },
    {
      "epoch": 4.214003944773175,
      "grad_norm": 0.318359375,
      "learning_rate": 1.4690150018129788e-05,
      "loss": 0.0358,
      "step": 17092
    },
    {
      "epoch": 4.214497041420119,
      "grad_norm": 0.357421875,
      "learning_rate": 1.467219402109451e-05,
      "loss": 0.0344,
      "step": 17094
    },
    {
      "epoch": 4.214990138067061,
      "grad_norm": 0.328125,
      "learning_rate": 1.4654248135880621e-05,
      "loss": 0.038,
      "step": 17096
    },
    {
      "epoch": 4.215483234714004,
      "grad_norm": 0.31640625,
      "learning_rate": 1.4636312364614846e-05,
      "loss": 0.0348,
      "step": 17098
    },
    {
      "epoch": 4.215976331360947,
      "grad_norm": 0.302734375,
      "learning_rate": 1.461838670942266e-05,
      "loss": 0.0386,
      "step": 17100
    },
    {
      "epoch": 4.21646942800789,
      "grad_norm": 0.33984375,
      "learning_rate": 1.4600471172428353e-05,
      "loss": 0.0382,
      "step": 17102
    },
    {
      "epoch": 4.216962524654832,
      "grad_norm": 0.294921875,
      "learning_rate": 1.458256575575503e-05,
      "loss": 0.0358,
      "step": 17104
    },
    {
      "epoch": 4.2174556213017755,
      "grad_norm": 0.3359375,
      "learning_rate": 1.4564670461524576e-05,
      "loss": 0.0359,
      "step": 17106
    },
    {
      "epoch": 4.217948717948718,
      "grad_norm": 0.34375,
      "learning_rate": 1.45467852918577e-05,
      "loss": 0.0364,
      "step": 17108
    },
    {
      "epoch": 4.218441814595661,
      "grad_norm": 0.3125,
      "learning_rate": 1.4528910248873862e-05,
      "loss": 0.0363,
      "step": 17110
    },
    {
      "epoch": 4.218934911242604,
      "grad_norm": 0.3125,
      "learning_rate": 1.4511045334691364e-05,
      "loss": 0.0381,
      "step": 17112
    },
    {
      "epoch": 4.2194280078895465,
      "grad_norm": 0.400390625,
      "learning_rate": 1.4493190551427305e-05,
      "loss": 0.0383,
      "step": 17114
    },
    {
      "epoch": 4.219921104536489,
      "grad_norm": 0.314453125,
      "learning_rate": 1.4475345901197568e-05,
      "loss": 0.0353,
      "step": 17116
    },
    {
      "epoch": 4.220414201183432,
      "grad_norm": 0.341796875,
      "learning_rate": 1.445751138611685e-05,
      "loss": 0.0343,
      "step": 17118
    },
    {
      "epoch": 4.220907297830375,
      "grad_norm": 0.3203125,
      "learning_rate": 1.4439687008298652e-05,
      "loss": 0.0364,
      "step": 17120
    },
    {
      "epoch": 4.2214003944773175,
      "grad_norm": 0.3046875,
      "learning_rate": 1.442187276985526e-05,
      "loss": 0.0378,
      "step": 17122
    },
    {
      "epoch": 4.22189349112426,
      "grad_norm": 0.3125,
      "learning_rate": 1.4404068672897741e-05,
      "loss": 0.0361,
      "step": 17124
    },
    {
      "epoch": 4.2223865877712035,
      "grad_norm": 0.30859375,
      "learning_rate": 1.4386274719535986e-05,
      "loss": 0.0361,
      "step": 17126
    },
    {
      "epoch": 4.222879684418146,
      "grad_norm": 0.326171875,
      "learning_rate": 1.4368490911878685e-05,
      "loss": 0.0369,
      "step": 17128
    },
    {
      "epoch": 4.2233727810650885,
      "grad_norm": 0.34765625,
      "learning_rate": 1.4350717252033308e-05,
      "loss": 0.0375,
      "step": 17130
    },
    {
      "epoch": 4.223865877712032,
      "grad_norm": 0.30859375,
      "learning_rate": 1.4332953742106148e-05,
      "loss": 0.0354,
      "step": 17132
    },
    {
      "epoch": 4.2243589743589745,
      "grad_norm": 0.337890625,
      "learning_rate": 1.4315200384202276e-05,
      "loss": 0.0374,
      "step": 17134
    },
    {
      "epoch": 4.224852071005917,
      "grad_norm": 0.3203125,
      "learning_rate": 1.4297457180425555e-05,
      "loss": 0.0323,
      "step": 17136
    },
    {
      "epoch": 4.22534516765286,
      "grad_norm": 0.3203125,
      "learning_rate": 1.427972413287868e-05,
      "loss": 0.0357,
      "step": 17138
    },
    {
      "epoch": 4.225838264299803,
      "grad_norm": 0.3515625,
      "learning_rate": 1.426200124366308e-05,
      "loss": 0.0348,
      "step": 17140
    },
    {
      "epoch": 4.226331360946745,
      "grad_norm": 0.310546875,
      "learning_rate": 1.4244288514879034e-05,
      "loss": 0.0369,
      "step": 17142
    },
    {
      "epoch": 4.226824457593688,
      "grad_norm": 0.310546875,
      "learning_rate": 1.4226585948625592e-05,
      "loss": 0.035,
      "step": 17144
    },
    {
      "epoch": 4.227317554240631,
      "grad_norm": 0.32421875,
      "learning_rate": 1.4208893547000612e-05,
      "loss": 0.0387,
      "step": 17146
    },
    {
      "epoch": 4.227810650887574,
      "grad_norm": 0.3515625,
      "learning_rate": 1.4191211312100739e-05,
      "loss": 0.0384,
      "step": 17148
    },
    {
      "epoch": 4.228303747534516,
      "grad_norm": 0.3359375,
      "learning_rate": 1.4173539246021428e-05,
      "loss": 0.0407,
      "step": 17150
    },
    {
      "epoch": 4.22879684418146,
      "grad_norm": 0.29296875,
      "learning_rate": 1.4155877350856939e-05,
      "loss": 0.0371,
      "step": 17152
    },
    {
      "epoch": 4.229289940828402,
      "grad_norm": 0.31640625,
      "learning_rate": 1.4138225628700242e-05,
      "loss": 0.0363,
      "step": 17154
    },
    {
      "epoch": 4.229783037475345,
      "grad_norm": 0.31640625,
      "learning_rate": 1.4120584081643196e-05,
      "loss": 0.0363,
      "step": 17156
    },
    {
      "epoch": 4.230276134122288,
      "grad_norm": 0.341796875,
      "learning_rate": 1.410295271177644e-05,
      "loss": 0.0402,
      "step": 17158
    },
    {
      "epoch": 4.230769230769231,
      "grad_norm": 0.33984375,
      "learning_rate": 1.408533152118936e-05,
      "loss": 0.0386,
      "step": 17160
    },
    {
      "epoch": 4.231262327416173,
      "grad_norm": 0.33984375,
      "learning_rate": 1.4067720511970195e-05,
      "loss": 0.0387,
      "step": 17162
    },
    {
      "epoch": 4.231755424063117,
      "grad_norm": 0.345703125,
      "learning_rate": 1.405011968620592e-05,
      "loss": 0.0374,
      "step": 17164
    },
    {
      "epoch": 4.232248520710059,
      "grad_norm": 0.34765625,
      "learning_rate": 1.4032529045982379e-05,
      "loss": 0.0351,
      "step": 17166
    },
    {
      "epoch": 4.232741617357002,
      "grad_norm": 0.33203125,
      "learning_rate": 1.4014948593384069e-05,
      "loss": 0.0379,
      "step": 17168
    },
    {
      "epoch": 4.233234714003944,
      "grad_norm": 0.365234375,
      "learning_rate": 1.3997378330494493e-05,
      "loss": 0.0363,
      "step": 17170
    },
    {
      "epoch": 4.233727810650888,
      "grad_norm": 0.34765625,
      "learning_rate": 1.3979818259395728e-05,
      "loss": 0.0343,
      "step": 17172
    },
    {
      "epoch": 4.23422090729783,
      "grad_norm": 0.310546875,
      "learning_rate": 1.3962268382168775e-05,
      "loss": 0.0344,
      "step": 17174
    },
    {
      "epoch": 4.234714003944773,
      "grad_norm": 0.3046875,
      "learning_rate": 1.3944728700893394e-05,
      "loss": 0.0368,
      "step": 17176
    },
    {
      "epoch": 4.235207100591716,
      "grad_norm": 0.28125,
      "learning_rate": 1.3927199217648124e-05,
      "loss": 0.0343,
      "step": 17178
    },
    {
      "epoch": 4.235700197238659,
      "grad_norm": 0.328125,
      "learning_rate": 1.3909679934510345e-05,
      "loss": 0.0374,
      "step": 17180
    },
    {
      "epoch": 4.236193293885601,
      "grad_norm": 0.28515625,
      "learning_rate": 1.389217085355612e-05,
      "loss": 0.0345,
      "step": 17182
    },
    {
      "epoch": 4.236686390532545,
      "grad_norm": 0.322265625,
      "learning_rate": 1.3874671976860454e-05,
      "loss": 0.0385,
      "step": 17184
    },
    {
      "epoch": 4.237179487179487,
      "grad_norm": 0.34375,
      "learning_rate": 1.385718330649698e-05,
      "loss": 0.0367,
      "step": 17186
    },
    {
      "epoch": 4.23767258382643,
      "grad_norm": 0.294921875,
      "learning_rate": 1.3839704844538281e-05,
      "loss": 0.0376,
      "step": 17188
    },
    {
      "epoch": 4.238165680473373,
      "grad_norm": 0.34765625,
      "learning_rate": 1.3822236593055605e-05,
      "loss": 0.0352,
      "step": 17190
    },
    {
      "epoch": 4.238658777120316,
      "grad_norm": 0.326171875,
      "learning_rate": 1.3804778554119036e-05,
      "loss": 0.0387,
      "step": 17192
    },
    {
      "epoch": 4.239151873767258,
      "grad_norm": 0.283203125,
      "learning_rate": 1.3787330729797487e-05,
      "loss": 0.0349,
      "step": 17194
    },
    {
      "epoch": 4.239644970414201,
      "grad_norm": 0.318359375,
      "learning_rate": 1.376989312215855e-05,
      "loss": 0.0357,
      "step": 17196
    },
    {
      "epoch": 4.240138067061144,
      "grad_norm": 0.353515625,
      "learning_rate": 1.375246573326877e-05,
      "loss": 0.0357,
      "step": 17198
    },
    {
      "epoch": 4.240631163708087,
      "grad_norm": 0.310546875,
      "learning_rate": 1.373504856519331e-05,
      "loss": 0.036,
      "step": 17200
    },
    {
      "epoch": 4.241124260355029,
      "grad_norm": 0.361328125,
      "learning_rate": 1.3717641619996268e-05,
      "loss": 0.0378,
      "step": 17202
    },
    {
      "epoch": 4.241617357001973,
      "grad_norm": 0.310546875,
      "learning_rate": 1.3700244899740421e-05,
      "loss": 0.0354,
      "step": 17204
    },
    {
      "epoch": 4.242110453648915,
      "grad_norm": 0.337890625,
      "learning_rate": 1.3682858406487387e-05,
      "loss": 0.0348,
      "step": 17206
    },
    {
      "epoch": 4.242603550295858,
      "grad_norm": 0.31640625,
      "learning_rate": 1.366548214229758e-05,
      "loss": 0.0362,
      "step": 17208
    },
    {
      "epoch": 4.243096646942801,
      "grad_norm": 0.333984375,
      "learning_rate": 1.3648116109230136e-05,
      "loss": 0.0349,
      "step": 17210
    },
    {
      "epoch": 4.243589743589744,
      "grad_norm": 0.3125,
      "learning_rate": 1.3630760309343104e-05,
      "loss": 0.0349,
      "step": 17212
    },
    {
      "epoch": 4.244082840236686,
      "grad_norm": 0.328125,
      "learning_rate": 1.3613414744693143e-05,
      "loss": 0.0363,
      "step": 17214
    },
    {
      "epoch": 4.2445759368836296,
      "grad_norm": 0.298828125,
      "learning_rate": 1.3596079417335916e-05,
      "loss": 0.0345,
      "step": 17216
    },
    {
      "epoch": 4.245069033530572,
      "grad_norm": 0.287109375,
      "learning_rate": 1.357875432932566e-05,
      "loss": 0.0342,
      "step": 17218
    },
    {
      "epoch": 4.245562130177515,
      "grad_norm": 0.32421875,
      "learning_rate": 1.3561439482715532e-05,
      "loss": 0.0339,
      "step": 17220
    },
    {
      "epoch": 4.246055226824458,
      "grad_norm": 0.30859375,
      "learning_rate": 1.3544134879557468e-05,
      "loss": 0.0384,
      "step": 17222
    },
    {
      "epoch": 4.2465483234714005,
      "grad_norm": 0.33984375,
      "learning_rate": 1.3526840521902073e-05,
      "loss": 0.0321,
      "step": 17224
    },
    {
      "epoch": 4.247041420118343,
      "grad_norm": 0.328125,
      "learning_rate": 1.350955641179893e-05,
      "loss": 0.0374,
      "step": 17226
    },
    {
      "epoch": 4.247534516765286,
      "grad_norm": 0.40625,
      "learning_rate": 1.3492282551296209e-05,
      "loss": 0.0394,
      "step": 17228
    },
    {
      "epoch": 4.248027613412229,
      "grad_norm": 0.330078125,
      "learning_rate": 1.3475018942441053e-05,
      "loss": 0.0358,
      "step": 17230
    },
    {
      "epoch": 4.2485207100591715,
      "grad_norm": 0.36328125,
      "learning_rate": 1.3457765587279214e-05,
      "loss": 0.0393,
      "step": 17232
    },
    {
      "epoch": 4.249013806706114,
      "grad_norm": 0.3515625,
      "learning_rate": 1.3440522487855344e-05,
      "loss": 0.035,
      "step": 17234
    },
    {
      "epoch": 4.2495069033530575,
      "grad_norm": 0.34765625,
      "learning_rate": 1.3423289646212856e-05,
      "loss": 0.0404,
      "step": 17236
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.32421875,
      "learning_rate": 1.3406067064393913e-05,
      "loss": 0.039,
      "step": 17238
    },
    {
      "epoch": 4.2504930966469425,
      "grad_norm": 0.384765625,
      "learning_rate": 1.3388854744439538e-05,
      "loss": 0.0375,
      "step": 17240
    },
    {
      "epoch": 4.250986193293886,
      "grad_norm": 0.349609375,
      "learning_rate": 1.3371652688389392e-05,
      "loss": 0.0353,
      "step": 17242
    },
    {
      "epoch": 4.2514792899408285,
      "grad_norm": 0.3359375,
      "learning_rate": 1.3354460898282128e-05,
      "loss": 0.0366,
      "step": 17244
    },
    {
      "epoch": 4.251972386587771,
      "grad_norm": 0.376953125,
      "learning_rate": 1.3337279376154987e-05,
      "loss": 0.0392,
      "step": 17246
    },
    {
      "epoch": 4.252465483234714,
      "grad_norm": 0.359375,
      "learning_rate": 1.3320108124044106e-05,
      "loss": 0.0385,
      "step": 17248
    },
    {
      "epoch": 4.252958579881657,
      "grad_norm": 0.330078125,
      "learning_rate": 1.3302947143984367e-05,
      "loss": 0.0368,
      "step": 17250
    },
    {
      "epoch": 4.2534516765285995,
      "grad_norm": 0.361328125,
      "learning_rate": 1.3285796438009446e-05,
      "loss": 0.0345,
      "step": 17252
    },
    {
      "epoch": 4.253944773175542,
      "grad_norm": 0.3203125,
      "learning_rate": 1.3268656008151815e-05,
      "loss": 0.0384,
      "step": 17254
    },
    {
      "epoch": 4.254437869822485,
      "grad_norm": 0.314453125,
      "learning_rate": 1.3251525856442658e-05,
      "loss": 0.0374,
      "step": 17256
    },
    {
      "epoch": 4.254930966469428,
      "grad_norm": 0.31640625,
      "learning_rate": 1.3234405984912068e-05,
      "loss": 0.0381,
      "step": 17258
    },
    {
      "epoch": 4.2554240631163704,
      "grad_norm": 0.314453125,
      "learning_rate": 1.3217296395588785e-05,
      "loss": 0.0373,
      "step": 17260
    },
    {
      "epoch": 4.255917159763314,
      "grad_norm": 0.33984375,
      "learning_rate": 1.320019709050041e-05,
      "loss": 0.0372,
      "step": 17262
    },
    {
      "epoch": 4.256410256410256,
      "grad_norm": 0.330078125,
      "learning_rate": 1.3183108071673312e-05,
      "loss": 0.0413,
      "step": 17264
    },
    {
      "epoch": 4.256903353057199,
      "grad_norm": 0.357421875,
      "learning_rate": 1.3166029341132624e-05,
      "loss": 0.0358,
      "step": 17266
    },
    {
      "epoch": 4.257396449704142,
      "grad_norm": 0.369140625,
      "learning_rate": 1.3148960900902286e-05,
      "loss": 0.0402,
      "step": 17268
    },
    {
      "epoch": 4.257889546351085,
      "grad_norm": 0.318359375,
      "learning_rate": 1.3131902753004988e-05,
      "loss": 0.0412,
      "step": 17270
    },
    {
      "epoch": 4.258382642998027,
      "grad_norm": 0.392578125,
      "learning_rate": 1.3114854899462248e-05,
      "loss": 0.04,
      "step": 17272
    },
    {
      "epoch": 4.258875739644971,
      "grad_norm": 0.416015625,
      "learning_rate": 1.3097817342294295e-05,
      "loss": 0.0358,
      "step": 17274
    },
    {
      "epoch": 4.259368836291913,
      "grad_norm": 0.359375,
      "learning_rate": 1.3080790083520167e-05,
      "loss": 0.0415,
      "step": 17276
    },
    {
      "epoch": 4.259861932938856,
      "grad_norm": 0.33203125,
      "learning_rate": 1.3063773125157718e-05,
      "loss": 0.035,
      "step": 17278
    },
    {
      "epoch": 4.260355029585799,
      "grad_norm": 0.34765625,
      "learning_rate": 1.3046766469223548e-05,
      "loss": 0.0397,
      "step": 17280
    },
    {
      "epoch": 4.260848126232742,
      "grad_norm": 0.349609375,
      "learning_rate": 1.3029770117733031e-05,
      "loss": 0.0415,
      "step": 17282
    },
    {
      "epoch": 4.261341222879684,
      "grad_norm": 0.376953125,
      "learning_rate": 1.3012784072700334e-05,
      "loss": 0.0368,
      "step": 17284
    },
    {
      "epoch": 4.261834319526627,
      "grad_norm": 0.3359375,
      "learning_rate": 1.2995808336138427e-05,
      "loss": 0.0403,
      "step": 17286
    },
    {
      "epoch": 4.26232741617357,
      "grad_norm": 0.34765625,
      "learning_rate": 1.2978842910058975e-05,
      "loss": 0.0397,
      "step": 17288
    },
    {
      "epoch": 4.262820512820513,
      "grad_norm": 0.37109375,
      "learning_rate": 1.2961887796472505e-05,
      "loss": 0.0385,
      "step": 17290
    },
    {
      "epoch": 4.263313609467455,
      "grad_norm": 0.279296875,
      "learning_rate": 1.2944942997388287e-05,
      "loss": 0.0333,
      "step": 17292
    },
    {
      "epoch": 4.263806706114399,
      "grad_norm": 0.34375,
      "learning_rate": 1.2928008514814371e-05,
      "loss": 0.04,
      "step": 17294
    },
    {
      "epoch": 4.264299802761341,
      "grad_norm": 0.314453125,
      "learning_rate": 1.2911084350757608e-05,
      "loss": 0.0371,
      "step": 17296
    },
    {
      "epoch": 4.264792899408284,
      "grad_norm": 0.33984375,
      "learning_rate": 1.289417050722358e-05,
      "loss": 0.0381,
      "step": 17298
    },
    {
      "epoch": 4.265285996055227,
      "grad_norm": 0.30078125,
      "learning_rate": 1.2877266986216684e-05,
      "loss": 0.0349,
      "step": 17300
    },
    {
      "epoch": 4.26577909270217,
      "grad_norm": 0.388671875,
      "learning_rate": 1.286037378974011e-05,
      "loss": 0.0386,
      "step": 17302
    },
    {
      "epoch": 4.266272189349112,
      "grad_norm": 0.375,
      "learning_rate": 1.2843490919795741e-05,
      "loss": 0.0368,
      "step": 17304
    },
    {
      "epoch": 4.266765285996055,
      "grad_norm": 0.4375,
      "learning_rate": 1.2826618378384315e-05,
      "loss": 0.0374,
      "step": 17306
    },
    {
      "epoch": 4.267258382642998,
      "grad_norm": 0.33203125,
      "learning_rate": 1.2809756167505328e-05,
      "loss": 0.0403,
      "step": 17308
    },
    {
      "epoch": 4.267751479289941,
      "grad_norm": 0.3203125,
      "learning_rate": 1.2792904289157048e-05,
      "loss": 0.0385,
      "step": 17310
    },
    {
      "epoch": 4.268244575936883,
      "grad_norm": 0.333984375,
      "learning_rate": 1.2776062745336504e-05,
      "loss": 0.0445,
      "step": 17312
    },
    {
      "epoch": 4.268737672583827,
      "grad_norm": 0.369140625,
      "learning_rate": 1.2759231538039529e-05,
      "loss": 0.0375,
      "step": 17314
    },
    {
      "epoch": 4.269230769230769,
      "grad_norm": 0.37109375,
      "learning_rate": 1.2742410669260717e-05,
      "loss": 0.0378,
      "step": 17316
    },
    {
      "epoch": 4.269723865877712,
      "grad_norm": 0.4765625,
      "learning_rate": 1.2725600140993388e-05,
      "loss": 0.042,
      "step": 17318
    },
    {
      "epoch": 4.270216962524655,
      "grad_norm": 0.296875,
      "learning_rate": 1.2708799955229766e-05,
      "loss": 0.034,
      "step": 17320
    },
    {
      "epoch": 4.270710059171598,
      "grad_norm": 0.341796875,
      "learning_rate": 1.2692010113960696e-05,
      "loss": 0.0386,
      "step": 17322
    },
    {
      "epoch": 4.27120315581854,
      "grad_norm": 0.373046875,
      "learning_rate": 1.2675230619175893e-05,
      "loss": 0.0367,
      "step": 17324
    },
    {
      "epoch": 4.271696252465484,
      "grad_norm": 0.3671875,
      "learning_rate": 1.2658461472863825e-05,
      "loss": 0.0328,
      "step": 17326
    },
    {
      "epoch": 4.272189349112426,
      "grad_norm": 0.322265625,
      "learning_rate": 1.2641702677011724e-05,
      "loss": 0.0394,
      "step": 17328
    },
    {
      "epoch": 4.272682445759369,
      "grad_norm": 0.30078125,
      "learning_rate": 1.2624954233605613e-05,
      "loss": 0.0369,
      "step": 17330
    },
    {
      "epoch": 4.273175542406312,
      "grad_norm": 0.384765625,
      "learning_rate": 1.2608216144630237e-05,
      "loss": 0.0386,
      "step": 17332
    },
    {
      "epoch": 4.273668639053255,
      "grad_norm": 0.318359375,
      "learning_rate": 1.2591488412069208e-05,
      "loss": 0.0387,
      "step": 17334
    },
    {
      "epoch": 4.274161735700197,
      "grad_norm": 0.333984375,
      "learning_rate": 1.257477103790482e-05,
      "loss": 0.0397,
      "step": 17336
    },
    {
      "epoch": 4.2746548323471405,
      "grad_norm": 0.337890625,
      "learning_rate": 1.2558064024118176e-05,
      "loss": 0.034,
      "step": 17338
    },
    {
      "epoch": 4.275147928994083,
      "grad_norm": 0.3046875,
      "learning_rate": 1.2541367372689172e-05,
      "loss": 0.0359,
      "step": 17340
    },
    {
      "epoch": 4.2756410256410255,
      "grad_norm": 0.341796875,
      "learning_rate": 1.2524681085596423e-05,
      "loss": 0.0375,
      "step": 17342
    },
    {
      "epoch": 4.276134122287968,
      "grad_norm": 0.33203125,
      "learning_rate": 1.2508005164817404e-05,
      "loss": 0.0359,
      "step": 17344
    },
    {
      "epoch": 4.2766272189349115,
      "grad_norm": 0.326171875,
      "learning_rate": 1.2491339612328212e-05,
      "loss": 0.0369,
      "step": 17346
    },
    {
      "epoch": 4.277120315581854,
      "grad_norm": 0.328125,
      "learning_rate": 1.24746844301039e-05,
      "loss": 0.0343,
      "step": 17348
    },
    {
      "epoch": 4.2776134122287965,
      "grad_norm": 0.326171875,
      "learning_rate": 1.2458039620118133e-05,
      "loss": 0.0368,
      "step": 17350
    },
    {
      "epoch": 4.27810650887574,
      "grad_norm": 0.3671875,
      "learning_rate": 1.244140518434348e-05,
      "loss": 0.0386,
      "step": 17352
    },
    {
      "epoch": 4.2785996055226825,
      "grad_norm": 0.291015625,
      "learning_rate": 1.2424781124751151e-05,
      "loss": 0.0327,
      "step": 17354
    },
    {
      "epoch": 4.279092702169625,
      "grad_norm": 0.27734375,
      "learning_rate": 1.2408167443311214e-05,
      "loss": 0.0322,
      "step": 17356
    },
    {
      "epoch": 4.279585798816568,
      "grad_norm": 0.3203125,
      "learning_rate": 1.2391564141992517e-05,
      "loss": 0.033,
      "step": 17358
    },
    {
      "epoch": 4.280078895463511,
      "grad_norm": 0.373046875,
      "learning_rate": 1.2374971222762565e-05,
      "loss": 0.0378,
      "step": 17360
    },
    {
      "epoch": 4.2805719921104535,
      "grad_norm": 0.349609375,
      "learning_rate": 1.2358388687587796e-05,
      "loss": 0.0351,
      "step": 17362
    },
    {
      "epoch": 4.281065088757396,
      "grad_norm": 0.37109375,
      "learning_rate": 1.2341816538433259e-05,
      "loss": 0.0364,
      "step": 17364
    },
    {
      "epoch": 4.281558185404339,
      "grad_norm": 0.380859375,
      "learning_rate": 1.2325254777262917e-05,
      "loss": 0.0347,
      "step": 17366
    },
    {
      "epoch": 4.282051282051282,
      "grad_norm": 0.322265625,
      "learning_rate": 1.230870340603939e-05,
      "loss": 0.0363,
      "step": 17368
    },
    {
      "epoch": 4.2825443786982245,
      "grad_norm": 0.369140625,
      "learning_rate": 1.2292162426724108e-05,
      "loss": 0.045,
      "step": 17370
    },
    {
      "epoch": 4.283037475345168,
      "grad_norm": 0.357421875,
      "learning_rate": 1.2275631841277268e-05,
      "loss": 0.0378,
      "step": 17372
    },
    {
      "epoch": 4.28353057199211,
      "grad_norm": 0.412109375,
      "learning_rate": 1.2259111651657862e-05,
      "loss": 0.0354,
      "step": 17374
    },
    {
      "epoch": 4.284023668639053,
      "grad_norm": 0.357421875,
      "learning_rate": 1.224260185982361e-05,
      "loss": 0.0369,
      "step": 17376
    },
    {
      "epoch": 4.284516765285996,
      "grad_norm": 0.361328125,
      "learning_rate": 1.2226102467730982e-05,
      "loss": 0.0383,
      "step": 17378
    },
    {
      "epoch": 4.285009861932939,
      "grad_norm": 0.322265625,
      "learning_rate": 1.2209613477335324e-05,
      "loss": 0.0364,
      "step": 17380
    },
    {
      "epoch": 4.285502958579881,
      "grad_norm": 0.33203125,
      "learning_rate": 1.2193134890590597e-05,
      "loss": 0.0367,
      "step": 17382
    },
    {
      "epoch": 4.285996055226825,
      "grad_norm": 0.3359375,
      "learning_rate": 1.2176666709449635e-05,
      "loss": 0.0344,
      "step": 17384
    },
    {
      "epoch": 4.286489151873767,
      "grad_norm": 0.43359375,
      "learning_rate": 1.2160208935864014e-05,
      "loss": 0.0358,
      "step": 17386
    },
    {
      "epoch": 4.28698224852071,
      "grad_norm": 0.349609375,
      "learning_rate": 1.2143761571784074e-05,
      "loss": 0.0403,
      "step": 17388
    },
    {
      "epoch": 4.287475345167653,
      "grad_norm": 0.37890625,
      "learning_rate": 1.212732461915893e-05,
      "loss": 0.0369,
      "step": 17390
    },
    {
      "epoch": 4.287968441814596,
      "grad_norm": 0.333984375,
      "learning_rate": 1.2110898079936394e-05,
      "loss": 0.0387,
      "step": 17392
    },
    {
      "epoch": 4.288461538461538,
      "grad_norm": 0.34765625,
      "learning_rate": 1.2094481956063196e-05,
      "loss": 0.037,
      "step": 17394
    },
    {
      "epoch": 4.288954635108481,
      "grad_norm": 0.3203125,
      "learning_rate": 1.207807624948466e-05,
      "loss": 0.0359,
      "step": 17396
    },
    {
      "epoch": 4.289447731755424,
      "grad_norm": 0.380859375,
      "learning_rate": 1.2061680962144983e-05,
      "loss": 0.0334,
      "step": 17398
    },
    {
      "epoch": 4.289940828402367,
      "grad_norm": 0.3125,
      "learning_rate": 1.204529609598709e-05,
      "loss": 0.0339,
      "step": 17400
    },
    {
      "epoch": 4.290433925049309,
      "grad_norm": 0.28515625,
      "learning_rate": 1.2028921652952696e-05,
      "loss": 0.0341,
      "step": 17402
    },
    {
      "epoch": 4.290927021696253,
      "grad_norm": 0.33203125,
      "learning_rate": 1.2012557634982269e-05,
      "loss": 0.0384,
      "step": 17404
    },
    {
      "epoch": 4.291420118343195,
      "grad_norm": 0.373046875,
      "learning_rate": 1.1996204044014981e-05,
      "loss": 0.036,
      "step": 17406
    },
    {
      "epoch": 4.291913214990138,
      "grad_norm": 0.314453125,
      "learning_rate": 1.1979860881988902e-05,
      "loss": 0.0352,
      "step": 17408
    },
    {
      "epoch": 4.292406311637081,
      "grad_norm": 0.314453125,
      "learning_rate": 1.196352815084073e-05,
      "loss": 0.0344,
      "step": 17410
    },
    {
      "epoch": 4.292899408284024,
      "grad_norm": 0.326171875,
      "learning_rate": 1.1947205852506005e-05,
      "loss": 0.0349,
      "step": 17412
    },
    {
      "epoch": 4.293392504930966,
      "grad_norm": 0.357421875,
      "learning_rate": 1.1930893988919011e-05,
      "loss": 0.038,
      "step": 17414
    },
    {
      "epoch": 4.29388560157791,
      "grad_norm": 0.306640625,
      "learning_rate": 1.1914592562012806e-05,
      "loss": 0.0348,
      "step": 17416
    },
    {
      "epoch": 4.294378698224852,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1898301573719195e-05,
      "loss": 0.0379,
      "step": 17418
    },
    {
      "epoch": 4.294871794871795,
      "grad_norm": 0.3125,
      "learning_rate": 1.1882021025968704e-05,
      "loss": 0.0352,
      "step": 17420
    },
    {
      "epoch": 4.295364891518737,
      "grad_norm": 0.31640625,
      "learning_rate": 1.1865750920690755e-05,
      "loss": 0.0365,
      "step": 17422
    },
    {
      "epoch": 4.295857988165681,
      "grad_norm": 0.34765625,
      "learning_rate": 1.1849491259813373e-05,
      "loss": 0.0354,
      "step": 17424
    },
    {
      "epoch": 4.296351084812623,
      "grad_norm": 0.34375,
      "learning_rate": 1.1833242045263459e-05,
      "loss": 0.0383,
      "step": 17426
    },
    {
      "epoch": 4.296844181459566,
      "grad_norm": 0.3046875,
      "learning_rate": 1.181700327896661e-05,
      "loss": 0.0373,
      "step": 17428
    },
    {
      "epoch": 4.297337278106509,
      "grad_norm": 0.306640625,
      "learning_rate": 1.1800774962847228e-05,
      "loss": 0.035,
      "step": 17430
    },
    {
      "epoch": 4.297830374753452,
      "grad_norm": 0.36328125,
      "learning_rate": 1.1784557098828453e-05,
      "loss": 0.0387,
      "step": 17432
    },
    {
      "epoch": 4.298323471400394,
      "grad_norm": 0.3046875,
      "learning_rate": 1.1768349688832203e-05,
      "loss": 0.0383,
      "step": 17434
    },
    {
      "epoch": 4.298816568047338,
      "grad_norm": 0.31640625,
      "learning_rate": 1.1752152734779143e-05,
      "loss": 0.0363,
      "step": 17436
    },
    {
      "epoch": 4.29930966469428,
      "grad_norm": 0.365234375,
      "learning_rate": 1.1735966238588691e-05,
      "loss": 0.0358,
      "step": 17438
    },
    {
      "epoch": 4.299802761341223,
      "grad_norm": 0.32421875,
      "learning_rate": 1.1719790202179026e-05,
      "loss": 0.0371,
      "step": 17440
    },
    {
      "epoch": 4.300295857988166,
      "grad_norm": 0.306640625,
      "learning_rate": 1.1703624627467135e-05,
      "loss": 0.0343,
      "step": 17442
    },
    {
      "epoch": 4.300788954635109,
      "grad_norm": 0.294921875,
      "learning_rate": 1.16874695163687e-05,
      "loss": 0.0374,
      "step": 17444
    },
    {
      "epoch": 4.301282051282051,
      "grad_norm": 0.35546875,
      "learning_rate": 1.1671324870798195e-05,
      "loss": 0.0385,
      "step": 17446
    },
    {
      "epoch": 4.3017751479289945,
      "grad_norm": 0.3046875,
      "learning_rate": 1.165519069266886e-05,
      "loss": 0.0363,
      "step": 17448
    },
    {
      "epoch": 4.302268244575937,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1639066983892687e-05,
      "loss": 0.0345,
      "step": 17450
    },
    {
      "epoch": 4.30276134122288,
      "grad_norm": 0.3125,
      "learning_rate": 1.1622953746380415e-05,
      "loss": 0.0396,
      "step": 17452
    },
    {
      "epoch": 4.303254437869822,
      "grad_norm": 0.32421875,
      "learning_rate": 1.1606850982041583e-05,
      "loss": 0.0367,
      "step": 17454
    },
    {
      "epoch": 4.3037475345167655,
      "grad_norm": 0.359375,
      "learning_rate": 1.159075869278442e-05,
      "loss": 0.0417,
      "step": 17456
    },
    {
      "epoch": 4.304240631163708,
      "grad_norm": 0.330078125,
      "learning_rate": 1.1574676880515955e-05,
      "loss": 0.0383,
      "step": 17458
    },
    {
      "epoch": 4.304733727810651,
      "grad_norm": 0.3515625,
      "learning_rate": 1.1558605547141988e-05,
      "loss": 0.0375,
      "step": 17460
    },
    {
      "epoch": 4.305226824457594,
      "grad_norm": 0.36328125,
      "learning_rate": 1.154254469456706e-05,
      "loss": 0.0373,
      "step": 17462
    },
    {
      "epoch": 4.3057199211045365,
      "grad_norm": 0.326171875,
      "learning_rate": 1.1526494324694458e-05,
      "loss": 0.0398,
      "step": 17464
    },
    {
      "epoch": 4.306213017751479,
      "grad_norm": 0.326171875,
      "learning_rate": 1.1510454439426267e-05,
      "loss": 0.0354,
      "step": 17466
    },
    {
      "epoch": 4.306706114398422,
      "grad_norm": 0.341796875,
      "learning_rate": 1.1494425040663292e-05,
      "loss": 0.0345,
      "step": 17468
    },
    {
      "epoch": 4.307199211045365,
      "grad_norm": 0.3359375,
      "learning_rate": 1.1478406130305097e-05,
      "loss": 0.0379,
      "step": 17470
    },
    {
      "epoch": 4.3076923076923075,
      "grad_norm": 0.279296875,
      "learning_rate": 1.1462397710250005e-05,
      "loss": 0.0365,
      "step": 17472
    },
    {
      "epoch": 4.308185404339251,
      "grad_norm": 0.29296875,
      "learning_rate": 1.1446399782395123e-05,
      "loss": 0.0396,
      "step": 17474
    },
    {
      "epoch": 4.308678500986193,
      "grad_norm": 0.333984375,
      "learning_rate": 1.1430412348636277e-05,
      "loss": 0.0356,
      "step": 17476
    },
    {
      "epoch": 4.309171597633136,
      "grad_norm": 0.322265625,
      "learning_rate": 1.1414435410868086e-05,
      "loss": 0.0331,
      "step": 17478
    },
    {
      "epoch": 4.3096646942800785,
      "grad_norm": 0.369140625,
      "learning_rate": 1.1398468970983888e-05,
      "loss": 0.0391,
      "step": 17480
    },
    {
      "epoch": 4.310157790927022,
      "grad_norm": 0.326171875,
      "learning_rate": 1.1382513030875808e-05,
      "loss": 0.0404,
      "step": 17482
    },
    {
      "epoch": 4.310650887573964,
      "grad_norm": 0.328125,
      "learning_rate": 1.136656759243474e-05,
      "loss": 0.0366,
      "step": 17484
    },
    {
      "epoch": 4.311143984220907,
      "grad_norm": 0.353515625,
      "learning_rate": 1.1350632657550253e-05,
      "loss": 0.0379,
      "step": 17486
    },
    {
      "epoch": 4.31163708086785,
      "grad_norm": 0.3515625,
      "learning_rate": 1.1334708228110747e-05,
      "loss": 0.037,
      "step": 17488
    },
    {
      "epoch": 4.312130177514793,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1318794306003355e-05,
      "loss": 0.0414,
      "step": 17490
    },
    {
      "epoch": 4.312623274161735,
      "grad_norm": 0.330078125,
      "learning_rate": 1.1302890893113982e-05,
      "loss": 0.0333,
      "step": 17492
    },
    {
      "epoch": 4.313116370808679,
      "grad_norm": 0.314453125,
      "learning_rate": 1.1286997991327253e-05,
      "loss": 0.037,
      "step": 17494
    },
    {
      "epoch": 4.313609467455621,
      "grad_norm": 0.314453125,
      "learning_rate": 1.1271115602526572e-05,
      "loss": 0.0328,
      "step": 17496
    },
    {
      "epoch": 4.314102564102564,
      "grad_norm": 0.3828125,
      "learning_rate": 1.1255243728594112e-05,
      "loss": 0.0342,
      "step": 17498
    },
    {
      "epoch": 4.314595660749507,
      "grad_norm": 0.318359375,
      "learning_rate": 1.1239382371410712e-05,
      "loss": 0.0386,
      "step": 17500
    },
    {
      "epoch": 4.31508875739645,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1223531532856125e-05,
      "loss": 0.0385,
      "step": 17502
    },
    {
      "epoch": 4.315581854043392,
      "grad_norm": 0.330078125,
      "learning_rate": 1.120769121480869e-05,
      "loss": 0.0354,
      "step": 17504
    },
    {
      "epoch": 4.316074950690336,
      "grad_norm": 0.3515625,
      "learning_rate": 1.11918614191456e-05,
      "loss": 0.0349,
      "step": 17506
    },
    {
      "epoch": 4.316568047337278,
      "grad_norm": 0.310546875,
      "learning_rate": 1.117604214774277e-05,
      "loss": 0.0381,
      "step": 17508
    },
    {
      "epoch": 4.317061143984221,
      "grad_norm": 0.337890625,
      "learning_rate": 1.1160233402474885e-05,
      "loss": 0.0338,
      "step": 17510
    },
    {
      "epoch": 4.317554240631163,
      "grad_norm": 0.3046875,
      "learning_rate": 1.1144435185215374e-05,
      "loss": 0.0348,
      "step": 17512
    },
    {
      "epoch": 4.318047337278107,
      "grad_norm": 0.478515625,
      "learning_rate": 1.1128647497836353e-05,
      "loss": 0.0368,
      "step": 17514
    },
    {
      "epoch": 4.318540433925049,
      "grad_norm": 0.322265625,
      "learning_rate": 1.1112870342208847e-05,
      "loss": 0.0361,
      "step": 17516
    },
    {
      "epoch": 4.319033530571992,
      "grad_norm": 0.32421875,
      "learning_rate": 1.1097103720202461e-05,
      "loss": 0.0378,
      "step": 17518
    },
    {
      "epoch": 4.319526627218935,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1081347633685657e-05,
      "loss": 0.0372,
      "step": 17520
    },
    {
      "epoch": 4.320019723865878,
      "grad_norm": 0.373046875,
      "learning_rate": 1.1065602084525618e-05,
      "loss": 0.0363,
      "step": 17522
    },
    {
      "epoch": 4.32051282051282,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1049867074588282e-05,
      "loss": 0.0366,
      "step": 17524
    },
    {
      "epoch": 4.321005917159764,
      "grad_norm": 0.31640625,
      "learning_rate": 1.1034142605738351e-05,
      "loss": 0.0349,
      "step": 17526
    },
    {
      "epoch": 4.321499013806706,
      "grad_norm": 0.330078125,
      "learning_rate": 1.1018428679839199e-05,
      "loss": 0.0366,
      "step": 17528
    },
    {
      "epoch": 4.321992110453649,
      "grad_norm": 0.322265625,
      "learning_rate": 1.1002725298753092e-05,
      "loss": 0.035,
      "step": 17530
    },
    {
      "epoch": 4.322485207100591,
      "grad_norm": 0.349609375,
      "learning_rate": 1.0987032464340896e-05,
      "loss": 0.0369,
      "step": 17532
    },
    {
      "epoch": 4.322978303747535,
      "grad_norm": 0.36328125,
      "learning_rate": 1.0971350178462391e-05,
      "loss": 0.0386,
      "step": 17534
    },
    {
      "epoch": 4.323471400394477,
      "grad_norm": 0.328125,
      "learning_rate": 1.0955678442975925e-05,
      "loss": 0.0354,
      "step": 17536
    },
    {
      "epoch": 4.32396449704142,
      "grad_norm": 0.384765625,
      "learning_rate": 1.0940017259738733e-05,
      "loss": 0.0335,
      "step": 17538
    },
    {
      "epoch": 4.324457593688363,
      "grad_norm": 0.30078125,
      "learning_rate": 1.0924366630606763e-05,
      "loss": 0.0332,
      "step": 17540
    },
    {
      "epoch": 4.324950690335306,
      "grad_norm": 0.361328125,
      "learning_rate": 1.0908726557434645e-05,
      "loss": 0.0378,
      "step": 17542
    },
    {
      "epoch": 4.325443786982248,
      "grad_norm": 0.3515625,
      "learning_rate": 1.0893097042075883e-05,
      "loss": 0.0352,
      "step": 17544
    },
    {
      "epoch": 4.325936883629192,
      "grad_norm": 0.30859375,
      "learning_rate": 1.0877478086382597e-05,
      "loss": 0.0377,
      "step": 17546
    },
    {
      "epoch": 4.326429980276134,
      "grad_norm": 0.453125,
      "learning_rate": 1.0861869692205794e-05,
      "loss": 0.0398,
      "step": 17548
    },
    {
      "epoch": 4.326923076923077,
      "grad_norm": 0.30078125,
      "learning_rate": 1.0846271861395086e-05,
      "loss": 0.0342,
      "step": 17550
    },
    {
      "epoch": 4.32741617357002,
      "grad_norm": 0.3359375,
      "learning_rate": 1.0830684595798946e-05,
      "loss": 0.0338,
      "step": 17552
    },
    {
      "epoch": 4.327909270216963,
      "grad_norm": 0.33984375,
      "learning_rate": 1.0815107897264554e-05,
      "loss": 0.0368,
      "step": 17554
    },
    {
      "epoch": 4.328402366863905,
      "grad_norm": 0.326171875,
      "learning_rate": 1.0799541767637777e-05,
      "loss": 0.0401,
      "step": 17556
    },
    {
      "epoch": 4.3288954635108485,
      "grad_norm": 0.333984375,
      "learning_rate": 1.0783986208763386e-05,
      "loss": 0.0371,
      "step": 17558
    },
    {
      "epoch": 4.329388560157791,
      "grad_norm": 0.35546875,
      "learning_rate": 1.0768441222484693e-05,
      "loss": 0.0335,
      "step": 17560
    },
    {
      "epoch": 4.329881656804734,
      "grad_norm": 0.34765625,
      "learning_rate": 1.075290681064397e-05,
      "loss": 0.0386,
      "step": 17562
    },
    {
      "epoch": 4.330374753451677,
      "grad_norm": 0.32421875,
      "learning_rate": 1.0737382975082056e-05,
      "loss": 0.0352,
      "step": 17564
    },
    {
      "epoch": 4.3308678500986195,
      "grad_norm": 0.337890625,
      "learning_rate": 1.0721869717638632e-05,
      "loss": 0.0382,
      "step": 17566
    },
    {
      "epoch": 4.331360946745562,
      "grad_norm": 0.376953125,
      "learning_rate": 1.0706367040152143e-05,
      "loss": 0.0334,
      "step": 17568
    },
    {
      "epoch": 4.331854043392505,
      "grad_norm": 0.318359375,
      "learning_rate": 1.0690874944459673e-05,
      "loss": 0.0417,
      "step": 17570
    },
    {
      "epoch": 4.332347140039448,
      "grad_norm": 0.392578125,
      "learning_rate": 1.0675393432397184e-05,
      "loss": 0.0377,
      "step": 17572
    },
    {
      "epoch": 4.3328402366863905,
      "grad_norm": 0.30859375,
      "learning_rate": 1.065992250579928e-05,
      "loss": 0.0365,
      "step": 17574
    },
    {
      "epoch": 4.333333333333333,
      "grad_norm": 0.357421875,
      "learning_rate": 1.0644462166499392e-05,
      "loss": 0.0389,
      "step": 17576
    },
    {
      "epoch": 4.3338264299802765,
      "grad_norm": 0.373046875,
      "learning_rate": 1.062901241632962e-05,
      "loss": 0.0312,
      "step": 17578
    },
    {
      "epoch": 4.334319526627219,
      "grad_norm": 0.29296875,
      "learning_rate": 1.0613573257120868e-05,
      "loss": 0.032,
      "step": 17580
    },
    {
      "epoch": 4.3348126232741615,
      "grad_norm": 0.31640625,
      "learning_rate": 1.0598144690702749e-05,
      "loss": 0.035,
      "step": 17582
    },
    {
      "epoch": 4.335305719921105,
      "grad_norm": 0.361328125,
      "learning_rate": 1.0582726718903634e-05,
      "loss": 0.0351,
      "step": 17584
    },
    {
      "epoch": 4.335798816568047,
      "grad_norm": 0.291015625,
      "learning_rate": 1.0567319343550675e-05,
      "loss": 0.033,
      "step": 17586
    },
    {
      "epoch": 4.33629191321499,
      "grad_norm": 0.306640625,
      "learning_rate": 1.0551922566469663e-05,
      "loss": 0.0373,
      "step": 17588
    },
    {
      "epoch": 4.3367850098619325,
      "grad_norm": 0.3125,
      "learning_rate": 1.0536536389485275e-05,
      "loss": 0.0347,
      "step": 17590
    },
    {
      "epoch": 4.337278106508876,
      "grad_norm": 0.337890625,
      "learning_rate": 1.0521160814420806e-05,
      "loss": 0.0355,
      "step": 17592
    },
    {
      "epoch": 4.337771203155818,
      "grad_norm": 0.3125,
      "learning_rate": 1.0505795843098366e-05,
      "loss": 0.0335,
      "step": 17594
    },
    {
      "epoch": 4.338264299802761,
      "grad_norm": 0.29296875,
      "learning_rate": 1.0490441477338786e-05,
      "loss": 0.0391,
      "step": 17596
    },
    {
      "epoch": 4.338757396449704,
      "grad_norm": 0.314453125,
      "learning_rate": 1.0475097718961658e-05,
      "loss": 0.0362,
      "step": 17598
    },
    {
      "epoch": 4.339250493096647,
      "grad_norm": 0.318359375,
      "learning_rate": 1.0459764569785314e-05,
      "loss": 0.0341,
      "step": 17600
    },
    {
      "epoch": 4.339743589743589,
      "grad_norm": 0.310546875,
      "learning_rate": 1.0444442031626744e-05,
      "loss": 0.0366,
      "step": 17602
    },
    {
      "epoch": 4.340236686390533,
      "grad_norm": 0.30859375,
      "learning_rate": 1.0429130106301854e-05,
      "loss": 0.0353,
      "step": 17604
    },
    {
      "epoch": 4.340729783037475,
      "grad_norm": 0.302734375,
      "learning_rate": 1.0413828795625125e-05,
      "loss": 0.0339,
      "step": 17606
    },
    {
      "epoch": 4.341222879684418,
      "grad_norm": 0.357421875,
      "learning_rate": 1.0398538101409872e-05,
      "loss": 0.0383,
      "step": 17608
    },
    {
      "epoch": 4.341715976331361,
      "grad_norm": 0.29296875,
      "learning_rate": 1.0383258025468124e-05,
      "loss": 0.0337,
      "step": 17610
    },
    {
      "epoch": 4.342209072978304,
      "grad_norm": 0.330078125,
      "learning_rate": 1.0367988569610653e-05,
      "loss": 0.0371,
      "step": 17612
    },
    {
      "epoch": 4.342702169625246,
      "grad_norm": 0.302734375,
      "learning_rate": 1.0352729735646982e-05,
      "loss": 0.0313,
      "step": 17614
    },
    {
      "epoch": 4.34319526627219,
      "grad_norm": 0.326171875,
      "learning_rate": 1.0337481525385362e-05,
      "loss": 0.0339,
      "step": 17616
    },
    {
      "epoch": 4.343688362919132,
      "grad_norm": 0.3515625,
      "learning_rate": 1.0322243940632814e-05,
      "loss": 0.0407,
      "step": 17618
    },
    {
      "epoch": 4.344181459566075,
      "grad_norm": 0.345703125,
      "learning_rate": 1.0307016983195039e-05,
      "loss": 0.0369,
      "step": 17620
    },
    {
      "epoch": 4.344674556213017,
      "grad_norm": 0.341796875,
      "learning_rate": 1.0291800654876527e-05,
      "loss": 0.0368,
      "step": 17622
    },
    {
      "epoch": 4.345167652859961,
      "grad_norm": 0.337890625,
      "learning_rate": 1.0276594957480502e-05,
      "loss": 0.0377,
      "step": 17624
    },
    {
      "epoch": 4.345660749506903,
      "grad_norm": 0.35546875,
      "learning_rate": 1.0261399892808932e-05,
      "loss": 0.0384,
      "step": 17626
    },
    {
      "epoch": 4.346153846153846,
      "grad_norm": 0.34375,
      "learning_rate": 1.02462154626625e-05,
      "loss": 0.0406,
      "step": 17628
    },
    {
      "epoch": 4.346646942800789,
      "grad_norm": 0.34765625,
      "learning_rate": 1.0231041668840669e-05,
      "loss": 0.0385,
      "step": 17630
    },
    {
      "epoch": 4.347140039447732,
      "grad_norm": 0.34765625,
      "learning_rate": 1.0215878513141597e-05,
      "loss": 0.0363,
      "step": 17632
    },
    {
      "epoch": 4.347633136094674,
      "grad_norm": 0.330078125,
      "learning_rate": 1.0200725997362238e-05,
      "loss": 0.0367,
      "step": 17634
    },
    {
      "epoch": 4.348126232741618,
      "grad_norm": 0.34375,
      "learning_rate": 1.0185584123298198e-05,
      "loss": 0.0338,
      "step": 17636
    },
    {
      "epoch": 4.34861932938856,
      "grad_norm": 0.37890625,
      "learning_rate": 1.017045289274391e-05,
      "loss": 0.0353,
      "step": 17638
    },
    {
      "epoch": 4.349112426035503,
      "grad_norm": 0.298828125,
      "learning_rate": 1.0155332307492482e-05,
      "loss": 0.0349,
      "step": 17640
    },
    {
      "epoch": 4.349605522682446,
      "grad_norm": 0.3046875,
      "learning_rate": 1.0140222369335816e-05,
      "loss": 0.0329,
      "step": 17642
    },
    {
      "epoch": 4.350098619329389,
      "grad_norm": 0.28515625,
      "learning_rate": 1.0125123080064502e-05,
      "loss": 0.0339,
      "step": 17644
    },
    {
      "epoch": 4.350591715976331,
      "grad_norm": 0.30859375,
      "learning_rate": 1.0110034441467908e-05,
      "loss": 0.0357,
      "step": 17646
    },
    {
      "epoch": 4.351084812623274,
      "grad_norm": 0.31640625,
      "learning_rate": 1.0094956455334137e-05,
      "loss": 0.0373,
      "step": 17648
    },
    {
      "epoch": 4.351577909270217,
      "grad_norm": 0.326171875,
      "learning_rate": 1.007988912344996e-05,
      "loss": 0.0362,
      "step": 17650
    },
    {
      "epoch": 4.35207100591716,
      "grad_norm": 0.333984375,
      "learning_rate": 1.0064832447600981e-05,
      "loss": 0.035,
      "step": 17652
    },
    {
      "epoch": 4.352564102564102,
      "grad_norm": 0.369140625,
      "learning_rate": 1.0049786429571484e-05,
      "loss": 0.0364,
      "step": 17654
    },
    {
      "epoch": 4.353057199211046,
      "grad_norm": 0.328125,
      "learning_rate": 1.0034751071144522e-05,
      "loss": 0.0356,
      "step": 17656
    },
    {
      "epoch": 4.353550295857988,
      "grad_norm": 0.30859375,
      "learning_rate": 1.0019726374101845e-05,
      "loss": 0.0354,
      "step": 17658
    },
    {
      "epoch": 4.354043392504931,
      "grad_norm": 0.29296875,
      "learning_rate": 1.0004712340223987e-05,
      "loss": 0.0341,
      "step": 17660
    },
    {
      "epoch": 4.354536489151874,
      "grad_norm": 0.357421875,
      "learning_rate": 9.989708971290213e-06,
      "loss": 0.0339,
      "step": 17662
    },
    {
      "epoch": 4.355029585798817,
      "grad_norm": 0.3203125,
      "learning_rate": 9.974716269078432e-06,
      "loss": 0.0348,
      "step": 17664
    },
    {
      "epoch": 4.355522682445759,
      "grad_norm": 0.384765625,
      "learning_rate": 9.959734235365448e-06,
      "loss": 0.034,
      "step": 17666
    },
    {
      "epoch": 4.3560157790927025,
      "grad_norm": 0.439453125,
      "learning_rate": 9.94476287192666e-06,
      "loss": 0.0353,
      "step": 17668
    },
    {
      "epoch": 4.356508875739645,
      "grad_norm": 0.28515625,
      "learning_rate": 9.929802180536274e-06,
      "loss": 0.0355,
      "step": 17670
    },
    {
      "epoch": 4.357001972386588,
      "grad_norm": 0.31640625,
      "learning_rate": 9.914852162967214e-06,
      "loss": 0.0362,
      "step": 17672
    },
    {
      "epoch": 4.357495069033531,
      "grad_norm": 0.384765625,
      "learning_rate": 9.899912820991152e-06,
      "loss": 0.0395,
      "step": 17674
    },
    {
      "epoch": 4.3579881656804735,
      "grad_norm": 0.283203125,
      "learning_rate": 9.884984156378497e-06,
      "loss": 0.0316,
      "step": 17676
    },
    {
      "epoch": 4.358481262327416,
      "grad_norm": 0.310546875,
      "learning_rate": 9.870066170898307e-06,
      "loss": 0.0372,
      "step": 17678
    },
    {
      "epoch": 4.358974358974359,
      "grad_norm": 0.3125,
      "learning_rate": 9.855158866318548e-06,
      "loss": 0.0365,
      "step": 17680
    },
    {
      "epoch": 4.359467455621302,
      "grad_norm": 0.296875,
      "learning_rate": 9.84026224440574e-06,
      "loss": 0.0304,
      "step": 17682
    },
    {
      "epoch": 4.3599605522682445,
      "grad_norm": 0.296875,
      "learning_rate": 9.825376306925238e-06,
      "loss": 0.0303,
      "step": 17684
    },
    {
      "epoch": 4.360453648915187,
      "grad_norm": 0.326171875,
      "learning_rate": 9.81050105564112e-06,
      "loss": 0.0384,
      "step": 17686
    },
    {
      "epoch": 4.3609467455621305,
      "grad_norm": 0.337890625,
      "learning_rate": 9.795636492316184e-06,
      "loss": 0.0387,
      "step": 17688
    },
    {
      "epoch": 4.361439842209073,
      "grad_norm": 0.326171875,
      "learning_rate": 9.780782618711981e-06,
      "loss": 0.0359,
      "step": 17690
    },
    {
      "epoch": 4.3619329388560155,
      "grad_norm": 0.28515625,
      "learning_rate": 9.765939436588712e-06,
      "loss": 0.0299,
      "step": 17692
    },
    {
      "epoch": 4.362426035502959,
      "grad_norm": 0.341796875,
      "learning_rate": 9.751106947705468e-06,
      "loss": 0.0356,
      "step": 17694
    },
    {
      "epoch": 4.3629191321499015,
      "grad_norm": 0.287109375,
      "learning_rate": 9.73628515381989e-06,
      "loss": 0.0332,
      "step": 17696
    },
    {
      "epoch": 4.363412228796844,
      "grad_norm": 0.314453125,
      "learning_rate": 9.72147405668853e-06,
      "loss": 0.0373,
      "step": 17698
    },
    {
      "epoch": 4.363905325443787,
      "grad_norm": 0.306640625,
      "learning_rate": 9.706673658066533e-06,
      "loss": 0.0364,
      "step": 17700
    },
    {
      "epoch": 4.36439842209073,
      "grad_norm": 0.333984375,
      "learning_rate": 9.691883959707838e-06,
      "loss": 0.0364,
      "step": 17702
    },
    {
      "epoch": 4.3648915187376724,
      "grad_norm": 0.29296875,
      "learning_rate": 9.677104963365125e-06,
      "loss": 0.0359,
      "step": 17704
    },
    {
      "epoch": 4.365384615384615,
      "grad_norm": 0.38671875,
      "learning_rate": 9.662336670789729e-06,
      "loss": 0.0389,
      "step": 17706
    },
    {
      "epoch": 4.365877712031558,
      "grad_norm": 0.34765625,
      "learning_rate": 9.647579083731861e-06,
      "loss": 0.038,
      "step": 17708
    },
    {
      "epoch": 4.366370808678501,
      "grad_norm": 0.310546875,
      "learning_rate": 9.6328322039403e-06,
      "loss": 0.0394,
      "step": 17710
    },
    {
      "epoch": 4.366863905325443,
      "grad_norm": 0.32421875,
      "learning_rate": 9.618096033162705e-06,
      "loss": 0.0363,
      "step": 17712
    },
    {
      "epoch": 4.367357001972387,
      "grad_norm": 0.333984375,
      "learning_rate": 9.603370573145343e-06,
      "loss": 0.036,
      "step": 17714
    },
    {
      "epoch": 4.367850098619329,
      "grad_norm": 0.302734375,
      "learning_rate": 9.588655825633275e-06,
      "loss": 0.0324,
      "step": 17716
    },
    {
      "epoch": 4.368343195266272,
      "grad_norm": 0.384765625,
      "learning_rate": 9.573951792370295e-06,
      "loss": 0.0336,
      "step": 17718
    },
    {
      "epoch": 4.368836291913215,
      "grad_norm": 0.294921875,
      "learning_rate": 9.559258475098876e-06,
      "loss": 0.0341,
      "step": 17720
    },
    {
      "epoch": 4.369329388560158,
      "grad_norm": 0.3671875,
      "learning_rate": 9.544575875560314e-06,
      "loss": 0.0342,
      "step": 17722
    },
    {
      "epoch": 4.3698224852071,
      "grad_norm": 0.375,
      "learning_rate": 9.529903995494516e-06,
      "loss": 0.0333,
      "step": 17724
    },
    {
      "epoch": 4.370315581854044,
      "grad_norm": 0.302734375,
      "learning_rate": 9.515242836640248e-06,
      "loss": 0.0335,
      "step": 17726
    },
    {
      "epoch": 4.370808678500986,
      "grad_norm": 0.333984375,
      "learning_rate": 9.500592400734897e-06,
      "loss": 0.0366,
      "step": 17728
    },
    {
      "epoch": 4.371301775147929,
      "grad_norm": 0.37109375,
      "learning_rate": 9.485952689514622e-06,
      "loss": 0.0383,
      "step": 17730
    },
    {
      "epoch": 4.371794871794872,
      "grad_norm": 0.296875,
      "learning_rate": 9.471323704714341e-06,
      "loss": 0.0366,
      "step": 17732
    },
    {
      "epoch": 4.372287968441815,
      "grad_norm": 0.3046875,
      "learning_rate": 9.456705448067604e-06,
      "loss": 0.0325,
      "step": 17734
    },
    {
      "epoch": 4.372781065088757,
      "grad_norm": 0.380859375,
      "learning_rate": 9.442097921306848e-06,
      "loss": 0.036,
      "step": 17736
    },
    {
      "epoch": 4.3732741617357,
      "grad_norm": 0.326171875,
      "learning_rate": 9.427501126163052e-06,
      "loss": 0.0391,
      "step": 17738
    },
    {
      "epoch": 4.373767258382643,
      "grad_norm": 0.3125,
      "learning_rate": 9.412915064366112e-06,
      "loss": 0.0368,
      "step": 17740
    },
    {
      "epoch": 4.374260355029586,
      "grad_norm": 0.34375,
      "learning_rate": 9.39833973764448e-06,
      "loss": 0.0365,
      "step": 17742
    },
    {
      "epoch": 4.374753451676528,
      "grad_norm": 0.2890625,
      "learning_rate": 9.38377514772546e-06,
      "loss": 0.0323,
      "step": 17744
    },
    {
      "epoch": 4.375246548323472,
      "grad_norm": 0.296875,
      "learning_rate": 9.369221296335006e-06,
      "loss": 0.037,
      "step": 17746
    },
    {
      "epoch": 4.375739644970414,
      "grad_norm": 0.3046875,
      "learning_rate": 9.354678185197852e-06,
      "loss": 0.0371,
      "step": 17748
    },
    {
      "epoch": 4.376232741617357,
      "grad_norm": 0.3046875,
      "learning_rate": 9.340145816037449e-06,
      "loss": 0.0379,
      "step": 17750
    },
    {
      "epoch": 4.3767258382643,
      "grad_norm": 0.298828125,
      "learning_rate": 9.32562419057591e-06,
      "loss": 0.0329,
      "step": 17752
    },
    {
      "epoch": 4.377218934911243,
      "grad_norm": 0.306640625,
      "learning_rate": 9.311113310534203e-06,
      "loss": 0.0319,
      "step": 17754
    },
    {
      "epoch": 4.377712031558185,
      "grad_norm": 0.2734375,
      "learning_rate": 9.296613177631897e-06,
      "loss": 0.0307,
      "step": 17756
    },
    {
      "epoch": 4.378205128205128,
      "grad_norm": 0.2734375,
      "learning_rate": 9.28212379358735e-06,
      "loss": 0.0319,
      "step": 17758
    },
    {
      "epoch": 4.378698224852071,
      "grad_norm": 0.3203125,
      "learning_rate": 9.267645160117632e-06,
      "loss": 0.0346,
      "step": 17760
    },
    {
      "epoch": 4.379191321499014,
      "grad_norm": 0.3359375,
      "learning_rate": 9.253177278938551e-06,
      "loss": 0.0385,
      "step": 17762
    },
    {
      "epoch": 4.379684418145956,
      "grad_norm": 0.36328125,
      "learning_rate": 9.238720151764624e-06,
      "loss": 0.034,
      "step": 17764
    },
    {
      "epoch": 4.3801775147929,
      "grad_norm": 0.416015625,
      "learning_rate": 9.224273780309101e-06,
      "loss": 0.0365,
      "step": 17766
    },
    {
      "epoch": 4.380670611439842,
      "grad_norm": 0.28515625,
      "learning_rate": 9.20983816628399e-06,
      "loss": 0.0373,
      "step": 17768
    },
    {
      "epoch": 4.381163708086785,
      "grad_norm": 0.302734375,
      "learning_rate": 9.195413311399948e-06,
      "loss": 0.0347,
      "step": 17770
    },
    {
      "epoch": 4.381656804733728,
      "grad_norm": 0.29296875,
      "learning_rate": 9.180999217366404e-06,
      "loss": 0.0329,
      "step": 17772
    },
    {
      "epoch": 4.382149901380671,
      "grad_norm": 0.30078125,
      "learning_rate": 9.166595885891527e-06,
      "loss": 0.0342,
      "step": 17774
    },
    {
      "epoch": 4.382642998027613,
      "grad_norm": 0.326171875,
      "learning_rate": 9.152203318682184e-06,
      "loss": 0.0374,
      "step": 17776
    },
    {
      "epoch": 4.383136094674557,
      "grad_norm": 0.34375,
      "learning_rate": 9.13782151744399e-06,
      "loss": 0.0321,
      "step": 17778
    },
    {
      "epoch": 4.383629191321499,
      "grad_norm": 0.330078125,
      "learning_rate": 9.123450483881245e-06,
      "loss": 0.0328,
      "step": 17780
    },
    {
      "epoch": 4.384122287968442,
      "grad_norm": 0.3671875,
      "learning_rate": 9.109090219697026e-06,
      "loss": 0.0357,
      "step": 17782
    },
    {
      "epoch": 4.384615384615385,
      "grad_norm": 0.37109375,
      "learning_rate": 9.094740726593065e-06,
      "loss": 0.0392,
      "step": 17784
    },
    {
      "epoch": 4.3851084812623276,
      "grad_norm": 0.37890625,
      "learning_rate": 9.080402006269884e-06,
      "loss": 0.0352,
      "step": 17786
    },
    {
      "epoch": 4.38560157790927,
      "grad_norm": 0.34375,
      "learning_rate": 9.066074060426688e-06,
      "loss": 0.0344,
      "step": 17788
    },
    {
      "epoch": 4.386094674556213,
      "grad_norm": 0.275390625,
      "learning_rate": 9.051756890761431e-06,
      "loss": 0.0297,
      "step": 17790
    },
    {
      "epoch": 4.386587771203156,
      "grad_norm": 0.2890625,
      "learning_rate": 9.037450498970768e-06,
      "loss": 0.0317,
      "step": 17792
    },
    {
      "epoch": 4.3870808678500985,
      "grad_norm": 0.314453125,
      "learning_rate": 9.023154886750096e-06,
      "loss": 0.0357,
      "step": 17794
    },
    {
      "epoch": 4.387573964497041,
      "grad_norm": 0.3046875,
      "learning_rate": 9.008870055793517e-06,
      "loss": 0.0315,
      "step": 17796
    },
    {
      "epoch": 4.3880670611439845,
      "grad_norm": 0.326171875,
      "learning_rate": 8.994596007793887e-06,
      "loss": 0.0343,
      "step": 17798
    },
    {
      "epoch": 4.388560157790927,
      "grad_norm": 0.318359375,
      "learning_rate": 8.98033274444272e-06,
      "loss": 0.0384,
      "step": 17800
    },
    {
      "epoch": 4.3890532544378695,
      "grad_norm": 0.35546875,
      "learning_rate": 8.96608026743031e-06,
      "loss": 0.0348,
      "step": 17802
    },
    {
      "epoch": 4.389546351084813,
      "grad_norm": 0.287109375,
      "learning_rate": 8.951838578445649e-06,
      "loss": 0.035,
      "step": 17804
    },
    {
      "epoch": 4.3900394477317555,
      "grad_norm": 0.287109375,
      "learning_rate": 8.937607679176474e-06,
      "loss": 0.0331,
      "step": 17806
    },
    {
      "epoch": 4.390532544378698,
      "grad_norm": 0.31640625,
      "learning_rate": 8.923387571309216e-06,
      "loss": 0.0373,
      "step": 17808
    },
    {
      "epoch": 4.391025641025641,
      "grad_norm": 0.302734375,
      "learning_rate": 8.909178256529038e-06,
      "loss": 0.0345,
      "step": 17810
    },
    {
      "epoch": 4.391518737672584,
      "grad_norm": 0.314453125,
      "learning_rate": 8.894979736519837e-06,
      "loss": 0.035,
      "step": 17812
    },
    {
      "epoch": 4.3920118343195265,
      "grad_norm": 0.3125,
      "learning_rate": 8.88079201296419e-06,
      "loss": 0.0349,
      "step": 17814
    },
    {
      "epoch": 4.392504930966469,
      "grad_norm": 0.330078125,
      "learning_rate": 8.866615087543429e-06,
      "loss": 0.0329,
      "step": 17816
    },
    {
      "epoch": 4.392998027613412,
      "grad_norm": 0.326171875,
      "learning_rate": 8.852448961937598e-06,
      "loss": 0.0346,
      "step": 17818
    },
    {
      "epoch": 4.393491124260355,
      "grad_norm": 0.328125,
      "learning_rate": 8.83829363782548e-06,
      "loss": 0.0358,
      "step": 17820
    },
    {
      "epoch": 4.3939842209072975,
      "grad_norm": 0.279296875,
      "learning_rate": 8.82414911688454e-06,
      "loss": 0.0368,
      "step": 17822
    },
    {
      "epoch": 4.394477317554241,
      "grad_norm": 0.30078125,
      "learning_rate": 8.810015400790994e-06,
      "loss": 0.0361,
      "step": 17824
    },
    {
      "epoch": 4.394970414201183,
      "grad_norm": 0.30859375,
      "learning_rate": 8.795892491219781e-06,
      "loss": 0.0333,
      "step": 17826
    },
    {
      "epoch": 4.395463510848126,
      "grad_norm": 0.2578125,
      "learning_rate": 8.781780389844484e-06,
      "loss": 0.032,
      "step": 17828
    },
    {
      "epoch": 4.395956607495069,
      "grad_norm": 0.310546875,
      "learning_rate": 8.767679098337545e-06,
      "loss": 0.0349,
      "step": 17830
    },
    {
      "epoch": 4.396449704142012,
      "grad_norm": 0.33203125,
      "learning_rate": 8.75358861837e-06,
      "loss": 0.0355,
      "step": 17832
    },
    {
      "epoch": 4.396942800788954,
      "grad_norm": 0.337890625,
      "learning_rate": 8.739508951611652e-06,
      "loss": 0.0352,
      "step": 17834
    },
    {
      "epoch": 4.397435897435898,
      "grad_norm": 0.3046875,
      "learning_rate": 8.72544009973103e-06,
      "loss": 0.035,
      "step": 17836
    },
    {
      "epoch": 4.39792899408284,
      "grad_norm": 0.275390625,
      "learning_rate": 8.711382064395368e-06,
      "loss": 0.0305,
      "step": 17838
    },
    {
      "epoch": 4.398422090729783,
      "grad_norm": 0.349609375,
      "learning_rate": 8.697334847270633e-06,
      "loss": 0.0393,
      "step": 17840
    },
    {
      "epoch": 4.398915187376726,
      "grad_norm": 0.306640625,
      "learning_rate": 8.68329845002146e-06,
      "loss": 0.0356,
      "step": 17842
    },
    {
      "epoch": 4.399408284023669,
      "grad_norm": 0.33203125,
      "learning_rate": 8.669272874311296e-06,
      "loss": 0.0338,
      "step": 17844
    },
    {
      "epoch": 4.399901380670611,
      "grad_norm": 0.310546875,
      "learning_rate": 8.65525812180219e-06,
      "loss": 0.0341,
      "step": 17846
    },
    {
      "epoch": 4.400394477317554,
      "grad_norm": 0.337890625,
      "learning_rate": 8.641254194155034e-06,
      "loss": 0.0328,
      "step": 17848
    },
    {
      "epoch": 4.400887573964497,
      "grad_norm": 0.330078125,
      "learning_rate": 8.627261093029327e-06,
      "loss": 0.0365,
      "step": 17850
    },
    {
      "epoch": 4.40138067061144,
      "grad_norm": 0.3125,
      "learning_rate": 8.613278820083326e-06,
      "loss": 0.0386,
      "step": 17852
    },
    {
      "epoch": 4.401873767258382,
      "grad_norm": 0.33984375,
      "learning_rate": 8.599307376974053e-06,
      "loss": 0.0377,
      "step": 17854
    },
    {
      "epoch": 4.402366863905326,
      "grad_norm": 0.349609375,
      "learning_rate": 8.585346765357138e-06,
      "loss": 0.0385,
      "step": 17856
    },
    {
      "epoch": 4.402859960552268,
      "grad_norm": 0.296875,
      "learning_rate": 8.57139698688706e-06,
      "loss": 0.0351,
      "step": 17858
    },
    {
      "epoch": 4.403353057199211,
      "grad_norm": 0.353515625,
      "learning_rate": 8.55745804321687e-06,
      "loss": 0.0354,
      "step": 17860
    },
    {
      "epoch": 4.403846153846154,
      "grad_norm": 0.302734375,
      "learning_rate": 8.543529935998484e-06,
      "loss": 0.0356,
      "step": 17862
    },
    {
      "epoch": 4.404339250493097,
      "grad_norm": 0.32421875,
      "learning_rate": 8.529612666882414e-06,
      "loss": 0.0312,
      "step": 17864
    },
    {
      "epoch": 4.404832347140039,
      "grad_norm": 0.27734375,
      "learning_rate": 8.51570623751795e-06,
      "loss": 0.033,
      "step": 17866
    },
    {
      "epoch": 4.405325443786983,
      "grad_norm": 0.2734375,
      "learning_rate": 8.501810649553099e-06,
      "loss": 0.0312,
      "step": 17868
    },
    {
      "epoch": 4.405818540433925,
      "grad_norm": 0.267578125,
      "learning_rate": 8.487925904634509e-06,
      "loss": 0.033,
      "step": 17870
    },
    {
      "epoch": 4.406311637080868,
      "grad_norm": 0.33984375,
      "learning_rate": 8.474052004407673e-06,
      "loss": 0.0356,
      "step": 17872
    },
    {
      "epoch": 4.40680473372781,
      "grad_norm": 0.310546875,
      "learning_rate": 8.460188950516656e-06,
      "loss": 0.0343,
      "step": 17874
    },
    {
      "epoch": 4.407297830374754,
      "grad_norm": 0.294921875,
      "learning_rate": 8.446336744604378e-06,
      "loss": 0.0318,
      "step": 17876
    },
    {
      "epoch": 4.407790927021696,
      "grad_norm": 0.330078125,
      "learning_rate": 8.432495388312345e-06,
      "loss": 0.033,
      "step": 17878
    },
    {
      "epoch": 4.408284023668639,
      "grad_norm": 0.29296875,
      "learning_rate": 8.418664883280858e-06,
      "loss": 0.0323,
      "step": 17880
    },
    {
      "epoch": 4.408777120315582,
      "grad_norm": 0.2890625,
      "learning_rate": 8.404845231148918e-06,
      "loss": 0.0334,
      "step": 17882
    },
    {
      "epoch": 4.409270216962525,
      "grad_norm": 0.28125,
      "learning_rate": 8.391036433554188e-06,
      "loss": 0.0344,
      "step": 17884
    },
    {
      "epoch": 4.409763313609467,
      "grad_norm": 0.27734375,
      "learning_rate": 8.377238492133155e-06,
      "loss": 0.0322,
      "step": 17886
    },
    {
      "epoch": 4.410256410256411,
      "grad_norm": 0.310546875,
      "learning_rate": 8.363451408520883e-06,
      "loss": 0.0348,
      "step": 17888
    },
    {
      "epoch": 4.410749506903353,
      "grad_norm": 0.33984375,
      "learning_rate": 8.34967518435129e-06,
      "loss": 0.0344,
      "step": 17890
    },
    {
      "epoch": 4.411242603550296,
      "grad_norm": 0.296875,
      "learning_rate": 8.335909821256881e-06,
      "loss": 0.0285,
      "step": 17892
    },
    {
      "epoch": 4.411735700197239,
      "grad_norm": 0.287109375,
      "learning_rate": 8.32215532086894e-06,
      "loss": 0.0303,
      "step": 17894
    },
    {
      "epoch": 4.412228796844182,
      "grad_norm": 0.2890625,
      "learning_rate": 8.308411684817452e-06,
      "loss": 0.0335,
      "step": 17896
    },
    {
      "epoch": 4.412721893491124,
      "grad_norm": 0.314453125,
      "learning_rate": 8.294678914731124e-06,
      "loss": 0.0331,
      "step": 17898
    },
    {
      "epoch": 4.4132149901380675,
      "grad_norm": 0.38671875,
      "learning_rate": 8.28095701223739e-06,
      "loss": 0.0359,
      "step": 17900
    },
    {
      "epoch": 4.41370808678501,
      "grad_norm": 0.3125,
      "learning_rate": 8.26724597896229e-06,
      "loss": 0.0337,
      "step": 17902
    },
    {
      "epoch": 4.414201183431953,
      "grad_norm": 0.3125,
      "learning_rate": 8.253545816530762e-06,
      "loss": 0.0336,
      "step": 17904
    },
    {
      "epoch": 4.414694280078895,
      "grad_norm": 0.310546875,
      "learning_rate": 8.23985652656628e-06,
      "loss": 0.0397,
      "step": 17906
    },
    {
      "epoch": 4.4151873767258385,
      "grad_norm": 0.310546875,
      "learning_rate": 8.226178110691118e-06,
      "loss": 0.0381,
      "step": 17908
    },
    {
      "epoch": 4.415680473372781,
      "grad_norm": 0.34375,
      "learning_rate": 8.212510570526255e-06,
      "loss": 0.0328,
      "step": 17910
    },
    {
      "epoch": 4.4161735700197235,
      "grad_norm": 0.3046875,
      "learning_rate": 8.198853907691362e-06,
      "loss": 0.0379,
      "step": 17912
    },
    {
      "epoch": 4.416666666666667,
      "grad_norm": 0.328125,
      "learning_rate": 8.185208123804844e-06,
      "loss": 0.035,
      "step": 17914
    },
    {
      "epoch": 4.4171597633136095,
      "grad_norm": 0.283203125,
      "learning_rate": 8.171573220483763e-06,
      "loss": 0.0339,
      "step": 17916
    },
    {
      "epoch": 4.417652859960552,
      "grad_norm": 0.326171875,
      "learning_rate": 8.157949199343994e-06,
      "loss": 0.0318,
      "step": 17918
    },
    {
      "epoch": 4.418145956607495,
      "grad_norm": 0.34375,
      "learning_rate": 8.144336062000013e-06,
      "loss": 0.0358,
      "step": 17920
    },
    {
      "epoch": 4.418639053254438,
      "grad_norm": 0.302734375,
      "learning_rate": 8.13073381006506e-06,
      "loss": 0.0314,
      "step": 17922
    },
    {
      "epoch": 4.4191321499013805,
      "grad_norm": 0.294921875,
      "learning_rate": 8.117142445151083e-06,
      "loss": 0.0304,
      "step": 17924
    },
    {
      "epoch": 4.419625246548323,
      "grad_norm": 0.27734375,
      "learning_rate": 8.103561968868733e-06,
      "loss": 0.0319,
      "step": 17926
    },
    {
      "epoch": 4.420118343195266,
      "grad_norm": 0.345703125,
      "learning_rate": 8.089992382827371e-06,
      "loss": 0.029,
      "step": 17928
    },
    {
      "epoch": 4.420611439842209,
      "grad_norm": 0.28125,
      "learning_rate": 8.076433688635077e-06,
      "loss": 0.0339,
      "step": 17930
    },
    {
      "epoch": 4.4211045364891515,
      "grad_norm": 0.33203125,
      "learning_rate": 8.06288588789864e-06,
      "loss": 0.0317,
      "step": 17932
    },
    {
      "epoch": 4.421597633136095,
      "grad_norm": 0.314453125,
      "learning_rate": 8.049348982223526e-06,
      "loss": 0.0345,
      "step": 17934
    },
    {
      "epoch": 4.422090729783037,
      "grad_norm": 0.31640625,
      "learning_rate": 8.035822973213947e-06,
      "loss": 0.0358,
      "step": 17936
    },
    {
      "epoch": 4.42258382642998,
      "grad_norm": 0.2890625,
      "learning_rate": 8.0223078624728e-06,
      "loss": 0.0329,
      "step": 17938
    },
    {
      "epoch": 4.423076923076923,
      "grad_norm": 0.310546875,
      "learning_rate": 8.008803651601704e-06,
      "loss": 0.0344,
      "step": 17940
    },
    {
      "epoch": 4.423570019723866,
      "grad_norm": 0.30859375,
      "learning_rate": 7.99531034220099e-06,
      "loss": 0.0347,
      "step": 17942
    },
    {
      "epoch": 4.424063116370808,
      "grad_norm": 0.30859375,
      "learning_rate": 7.981827935869701e-06,
      "loss": 0.0348,
      "step": 17944
    },
    {
      "epoch": 4.424556213017752,
      "grad_norm": 0.37109375,
      "learning_rate": 7.96835643420557e-06,
      "loss": 0.0329,
      "step": 17946
    },
    {
      "epoch": 4.425049309664694,
      "grad_norm": 0.267578125,
      "learning_rate": 7.954895838805032e-06,
      "loss": 0.0325,
      "step": 17948
    },
    {
      "epoch": 4.425542406311637,
      "grad_norm": 0.326171875,
      "learning_rate": 7.941446151263243e-06,
      "loss": 0.0355,
      "step": 17950
    },
    {
      "epoch": 4.42603550295858,
      "grad_norm": 0.37109375,
      "learning_rate": 7.928007373174073e-06,
      "loss": 0.0376,
      "step": 17952
    },
    {
      "epoch": 4.426528599605523,
      "grad_norm": 0.28515625,
      "learning_rate": 7.914579506130104e-06,
      "loss": 0.0327,
      "step": 17954
    },
    {
      "epoch": 4.427021696252465,
      "grad_norm": 0.291015625,
      "learning_rate": 7.901162551722595e-06,
      "loss": 0.0346,
      "step": 17956
    },
    {
      "epoch": 4.427514792899409,
      "grad_norm": 0.3046875,
      "learning_rate": 7.887756511541544e-06,
      "loss": 0.0323,
      "step": 17958
    },
    {
      "epoch": 4.428007889546351,
      "grad_norm": 0.28515625,
      "learning_rate": 7.874361387175633e-06,
      "loss": 0.0344,
      "step": 17960
    },
    {
      "epoch": 4.428500986193294,
      "grad_norm": 0.283203125,
      "learning_rate": 7.86097718021227e-06,
      "loss": 0.0316,
      "step": 17962
    },
    {
      "epoch": 4.428994082840236,
      "grad_norm": 0.31640625,
      "learning_rate": 7.847603892237543e-06,
      "loss": 0.0317,
      "step": 17964
    },
    {
      "epoch": 4.42948717948718,
      "grad_norm": 0.287109375,
      "learning_rate": 7.834241524836262e-06,
      "loss": 0.0342,
      "step": 17966
    },
    {
      "epoch": 4.429980276134122,
      "grad_norm": 0.326171875,
      "learning_rate": 7.820890079591946e-06,
      "loss": 0.0323,
      "step": 17968
    },
    {
      "epoch": 4.430473372781065,
      "grad_norm": 0.322265625,
      "learning_rate": 7.807549558086824e-06,
      "loss": 0.0306,
      "step": 17970
    },
    {
      "epoch": 4.430966469428008,
      "grad_norm": 0.294921875,
      "learning_rate": 7.794219961901817e-06,
      "loss": 0.0324,
      "step": 17972
    },
    {
      "epoch": 4.431459566074951,
      "grad_norm": 0.330078125,
      "learning_rate": 7.780901292616549e-06,
      "loss": 0.0341,
      "step": 17974
    },
    {
      "epoch": 4.431952662721893,
      "grad_norm": 0.267578125,
      "learning_rate": 7.767593551809382e-06,
      "loss": 0.0295,
      "step": 17976
    },
    {
      "epoch": 4.432445759368837,
      "grad_norm": 0.3515625,
      "learning_rate": 7.75429674105731e-06,
      "loss": 0.0355,
      "step": 17978
    },
    {
      "epoch": 4.432938856015779,
      "grad_norm": 0.35546875,
      "learning_rate": 7.741010861936138e-06,
      "loss": 0.032,
      "step": 17980
    },
    {
      "epoch": 4.433431952662722,
      "grad_norm": 0.302734375,
      "learning_rate": 7.727735916020274e-06,
      "loss": 0.0341,
      "step": 17982
    },
    {
      "epoch": 4.433925049309664,
      "grad_norm": 0.34765625,
      "learning_rate": 7.714471904882881e-06,
      "loss": 0.0316,
      "step": 17984
    },
    {
      "epoch": 4.434418145956608,
      "grad_norm": 0.369140625,
      "learning_rate": 7.701218830095825e-06,
      "loss": 0.0357,
      "step": 17986
    },
    {
      "epoch": 4.43491124260355,
      "grad_norm": 0.337890625,
      "learning_rate": 7.68797669322967e-06,
      "loss": 0.0369,
      "step": 17988
    },
    {
      "epoch": 4.435404339250493,
      "grad_norm": 0.294921875,
      "learning_rate": 7.674745495853686e-06,
      "loss": 0.0336,
      "step": 17990
    },
    {
      "epoch": 4.435897435897436,
      "grad_norm": 0.333984375,
      "learning_rate": 7.661525239535828e-06,
      "loss": 0.036,
      "step": 17992
    },
    {
      "epoch": 4.436390532544379,
      "grad_norm": 0.328125,
      "learning_rate": 7.648315925842809e-06,
      "loss": 0.0361,
      "step": 17994
    },
    {
      "epoch": 4.436883629191321,
      "grad_norm": 0.369140625,
      "learning_rate": 7.635117556339955e-06,
      "loss": 0.0346,
      "step": 17996
    },
    {
      "epoch": 4.437376725838265,
      "grad_norm": 0.27734375,
      "learning_rate": 7.621930132591371e-06,
      "loss": 0.0313,
      "step": 17998
    },
    {
      "epoch": 4.437869822485207,
      "grad_norm": 0.34375,
      "learning_rate": 7.60875365615984e-06,
      "loss": 0.0304,
      "step": 18000
    },
    {
      "epoch": 4.43836291913215,
      "grad_norm": 0.3125,
      "learning_rate": 7.595588128606846e-06,
      "loss": 0.0343,
      "step": 18002
    },
    {
      "epoch": 4.438856015779093,
      "grad_norm": 0.2890625,
      "learning_rate": 7.582433551492574e-06,
      "loss": 0.0352,
      "step": 18004
    },
    {
      "epoch": 4.439349112426036,
      "grad_norm": 0.30859375,
      "learning_rate": 7.569289926375933e-06,
      "loss": 0.0321,
      "step": 18006
    },
    {
      "epoch": 4.439842209072978,
      "grad_norm": 0.31640625,
      "learning_rate": 7.556157254814511e-06,
      "loss": 0.0325,
      "step": 18008
    },
    {
      "epoch": 4.4403353057199215,
      "grad_norm": 0.298828125,
      "learning_rate": 7.5430355383645625e-06,
      "loss": 0.0324,
      "step": 18010
    },
    {
      "epoch": 4.440828402366864,
      "grad_norm": 0.30859375,
      "learning_rate": 7.529924778581143e-06,
      "loss": 0.0332,
      "step": 18012
    },
    {
      "epoch": 4.441321499013807,
      "grad_norm": 0.330078125,
      "learning_rate": 7.516824977017911e-06,
      "loss": 0.0309,
      "step": 18014
    },
    {
      "epoch": 4.441814595660749,
      "grad_norm": 0.3515625,
      "learning_rate": 7.5037361352272904e-06,
      "loss": 0.0324,
      "step": 18016
    },
    {
      "epoch": 4.4423076923076925,
      "grad_norm": 0.30078125,
      "learning_rate": 7.490658254760363e-06,
      "loss": 0.0345,
      "step": 18018
    },
    {
      "epoch": 4.442800788954635,
      "grad_norm": 0.30859375,
      "learning_rate": 7.477591337166934e-06,
      "loss": 0.031,
      "step": 18020
    },
    {
      "epoch": 4.443293885601578,
      "grad_norm": 0.28515625,
      "learning_rate": 7.46453538399553e-06,
      "loss": 0.032,
      "step": 18022
    },
    {
      "epoch": 4.443786982248521,
      "grad_norm": 0.294921875,
      "learning_rate": 7.451490396793304e-06,
      "loss": 0.0307,
      "step": 18024
    },
    {
      "epoch": 4.4442800788954635,
      "grad_norm": 0.326171875,
      "learning_rate": 7.4384563771062176e-06,
      "loss": 0.034,
      "step": 18026
    },
    {
      "epoch": 4.444773175542406,
      "grad_norm": 0.287109375,
      "learning_rate": 7.425433326478837e-06,
      "loss": 0.0313,
      "step": 18028
    },
    {
      "epoch": 4.445266272189349,
      "grad_norm": 0.345703125,
      "learning_rate": 7.412421246454482e-06,
      "loss": 0.0349,
      "step": 18030
    },
    {
      "epoch": 4.445759368836292,
      "grad_norm": 0.310546875,
      "learning_rate": 7.399420138575142e-06,
      "loss": 0.0341,
      "step": 18032
    },
    {
      "epoch": 4.4462524654832345,
      "grad_norm": 0.32421875,
      "learning_rate": 7.386430004381539e-06,
      "loss": 0.0325,
      "step": 18034
    },
    {
      "epoch": 4.446745562130178,
      "grad_norm": 0.318359375,
      "learning_rate": 7.373450845413077e-06,
      "loss": 0.0304,
      "step": 18036
    },
    {
      "epoch": 4.44723865877712,
      "grad_norm": 0.298828125,
      "learning_rate": 7.360482663207824e-06,
      "loss": 0.0383,
      "step": 18038
    },
    {
      "epoch": 4.447731755424063,
      "grad_norm": 0.267578125,
      "learning_rate": 7.34752545930264e-06,
      "loss": 0.0323,
      "step": 18040
    },
    {
      "epoch": 4.4482248520710055,
      "grad_norm": 0.291015625,
      "learning_rate": 7.334579235232975e-06,
      "loss": 0.0311,
      "step": 18042
    },
    {
      "epoch": 4.448717948717949,
      "grad_norm": 0.306640625,
      "learning_rate": 7.321643992533056e-06,
      "loss": 0.0304,
      "step": 18044
    },
    {
      "epoch": 4.449211045364891,
      "grad_norm": 0.3515625,
      "learning_rate": 7.308719732735769e-06,
      "loss": 0.0323,
      "step": 18046
    },
    {
      "epoch": 4.449704142011834,
      "grad_norm": 0.33203125,
      "learning_rate": 7.295806457372711e-06,
      "loss": 0.0337,
      "step": 18048
    },
    {
      "epoch": 4.450197238658777,
      "grad_norm": 0.34765625,
      "learning_rate": 7.282904167974203e-06,
      "loss": 0.0332,
      "step": 18050
    },
    {
      "epoch": 4.45069033530572,
      "grad_norm": 0.318359375,
      "learning_rate": 7.270012866069176e-06,
      "loss": 0.0362,
      "step": 18052
    },
    {
      "epoch": 4.451183431952662,
      "grad_norm": 0.28515625,
      "learning_rate": 7.257132553185408e-06,
      "loss": 0.0289,
      "step": 18054
    },
    {
      "epoch": 4.451676528599606,
      "grad_norm": 0.294921875,
      "learning_rate": 7.244263230849202e-06,
      "loss": 0.0334,
      "step": 18056
    },
    {
      "epoch": 4.452169625246548,
      "grad_norm": 0.318359375,
      "learning_rate": 7.231404900585714e-06,
      "loss": 0.0354,
      "step": 18058
    },
    {
      "epoch": 4.452662721893491,
      "grad_norm": 0.279296875,
      "learning_rate": 7.218557563918693e-06,
      "loss": 0.035,
      "step": 18060
    },
    {
      "epoch": 4.453155818540434,
      "grad_norm": 0.337890625,
      "learning_rate": 7.2057212223706205e-06,
      "loss": 0.0305,
      "step": 18062
    },
    {
      "epoch": 4.453648915187377,
      "grad_norm": 0.291015625,
      "learning_rate": 7.192895877462691e-06,
      "loss": 0.0336,
      "step": 18064
    },
    {
      "epoch": 4.454142011834319,
      "grad_norm": 0.265625,
      "learning_rate": 7.180081530714744e-06,
      "loss": 0.034,
      "step": 18066
    },
    {
      "epoch": 4.454635108481263,
      "grad_norm": 0.349609375,
      "learning_rate": 7.16727818364541e-06,
      "loss": 0.038,
      "step": 18068
    },
    {
      "epoch": 4.455128205128205,
      "grad_norm": 0.345703125,
      "learning_rate": 7.154485837771885e-06,
      "loss": 0.0339,
      "step": 18070
    },
    {
      "epoch": 4.455621301775148,
      "grad_norm": 0.322265625,
      "learning_rate": 7.141704494610202e-06,
      "loss": 0.0321,
      "step": 18072
    },
    {
      "epoch": 4.45611439842209,
      "grad_norm": 0.298828125,
      "learning_rate": 7.128934155674971e-06,
      "loss": 0.0304,
      "step": 18074
    },
    {
      "epoch": 4.456607495069034,
      "grad_norm": 0.34765625,
      "learning_rate": 7.1161748224795825e-06,
      "loss": 0.033,
      "step": 18076
    },
    {
      "epoch": 4.457100591715976,
      "grad_norm": 0.390625,
      "learning_rate": 7.103426496536081e-06,
      "loss": 0.0342,
      "step": 18078
    },
    {
      "epoch": 4.457593688362919,
      "grad_norm": 0.28515625,
      "learning_rate": 7.09068917935517e-06,
      "loss": 0.0349,
      "step": 18080
    },
    {
      "epoch": 4.458086785009862,
      "grad_norm": 0.26953125,
      "learning_rate": 7.077962872446375e-06,
      "loss": 0.0341,
      "step": 18082
    },
    {
      "epoch": 4.458579881656805,
      "grad_norm": 0.3046875,
      "learning_rate": 7.0652475773177464e-06,
      "loss": 0.0325,
      "step": 18084
    },
    {
      "epoch": 4.459072978303747,
      "grad_norm": 0.283203125,
      "learning_rate": 7.052543295476199e-06,
      "loss": 0.0291,
      "step": 18086
    },
    {
      "epoch": 4.459566074950691,
      "grad_norm": 0.328125,
      "learning_rate": 7.039850028427208e-06,
      "loss": 0.0321,
      "step": 18088
    },
    {
      "epoch": 4.460059171597633,
      "grad_norm": 0.287109375,
      "learning_rate": 7.027167777675015e-06,
      "loss": 0.0311,
      "step": 18090
    },
    {
      "epoch": 4.460552268244576,
      "grad_norm": 0.2890625,
      "learning_rate": 7.014496544722526e-06,
      "loss": 0.0338,
      "step": 18092
    },
    {
      "epoch": 4.461045364891519,
      "grad_norm": 0.28515625,
      "learning_rate": 7.001836331071365e-06,
      "loss": 0.0313,
      "step": 18094
    },
    {
      "epoch": 4.461538461538462,
      "grad_norm": 0.30859375,
      "learning_rate": 6.989187138221853e-06,
      "loss": 0.0309,
      "step": 18096
    },
    {
      "epoch": 4.462031558185404,
      "grad_norm": 0.298828125,
      "learning_rate": 6.976548967672936e-06,
      "loss": 0.033,
      "step": 18098
    },
    {
      "epoch": 4.462524654832347,
      "grad_norm": 0.298828125,
      "learning_rate": 6.96392182092237e-06,
      "loss": 0.0349,
      "step": 18100
    },
    {
      "epoch": 4.46301775147929,
      "grad_norm": 0.333984375,
      "learning_rate": 6.951305699466504e-06,
      "loss": 0.031,
      "step": 18102
    },
    {
      "epoch": 4.463510848126233,
      "grad_norm": 0.29296875,
      "learning_rate": 6.93870060480043e-06,
      "loss": 0.0385,
      "step": 18104
    },
    {
      "epoch": 4.464003944773175,
      "grad_norm": 0.28125,
      "learning_rate": 6.92610653841792e-06,
      "loss": 0.0332,
      "step": 18106
    },
    {
      "epoch": 4.464497041420119,
      "grad_norm": 0.345703125,
      "learning_rate": 6.913523501811448e-06,
      "loss": 0.0348,
      "step": 18108
    },
    {
      "epoch": 4.464990138067061,
      "grad_norm": 0.33203125,
      "learning_rate": 6.900951496472163e-06,
      "loss": 0.0326,
      "step": 18110
    },
    {
      "epoch": 4.465483234714004,
      "grad_norm": 0.30859375,
      "learning_rate": 6.888390523889931e-06,
      "loss": 0.0343,
      "step": 18112
    },
    {
      "epoch": 4.465976331360947,
      "grad_norm": 0.283203125,
      "learning_rate": 6.875840585553306e-06,
      "loss": 0.0322,
      "step": 18114
    },
    {
      "epoch": 4.46646942800789,
      "grad_norm": 0.34765625,
      "learning_rate": 6.863301682949486e-06,
      "loss": 0.033,
      "step": 18116
    },
    {
      "epoch": 4.466962524654832,
      "grad_norm": 0.28125,
      "learning_rate": 6.850773817564426e-06,
      "loss": 0.0335,
      "step": 18118
    },
    {
      "epoch": 4.4674556213017755,
      "grad_norm": 0.34375,
      "learning_rate": 6.838256990882752e-06,
      "loss": 0.0352,
      "step": 18120
    },
    {
      "epoch": 4.467948717948718,
      "grad_norm": 0.349609375,
      "learning_rate": 6.8257512043877755e-06,
      "loss": 0.0362,
      "step": 18122
    },
    {
      "epoch": 4.468441814595661,
      "grad_norm": 0.349609375,
      "learning_rate": 6.813256459561501e-06,
      "loss": 0.0365,
      "step": 18124
    },
    {
      "epoch": 4.468934911242604,
      "grad_norm": 0.318359375,
      "learning_rate": 6.800772757884621e-06,
      "loss": 0.0317,
      "step": 18126
    },
    {
      "epoch": 4.4694280078895465,
      "grad_norm": 0.30859375,
      "learning_rate": 6.788300100836542e-06,
      "loss": 0.0305,
      "step": 18128
    },
    {
      "epoch": 4.469921104536489,
      "grad_norm": 0.275390625,
      "learning_rate": 6.775838489895314e-06,
      "loss": 0.0353,
      "step": 18130
    },
    {
      "epoch": 4.470414201183432,
      "grad_norm": 0.29296875,
      "learning_rate": 6.763387926537723e-06,
      "loss": 0.0345,
      "step": 18132
    },
    {
      "epoch": 4.470907297830375,
      "grad_norm": 0.275390625,
      "learning_rate": 6.750948412239222e-06,
      "loss": 0.0332,
      "step": 18134
    },
    {
      "epoch": 4.4714003944773175,
      "grad_norm": 0.3046875,
      "learning_rate": 6.738519948473976e-06,
      "loss": 0.0326,
      "step": 18136
    },
    {
      "epoch": 4.47189349112426,
      "grad_norm": 0.2734375,
      "learning_rate": 6.726102536714829e-06,
      "loss": 0.031,
      "step": 18138
    },
    {
      "epoch": 4.4723865877712035,
      "grad_norm": 0.349609375,
      "learning_rate": 6.713696178433293e-06,
      "loss": 0.0318,
      "step": 18140
    },
    {
      "epoch": 4.472879684418146,
      "grad_norm": 0.2890625,
      "learning_rate": 6.701300875099614e-06,
      "loss": 0.0356,
      "step": 18142
    },
    {
      "epoch": 4.4733727810650885,
      "grad_norm": 0.318359375,
      "learning_rate": 6.688916628182706e-06,
      "loss": 0.0313,
      "step": 18144
    },
    {
      "epoch": 4.473865877712032,
      "grad_norm": 0.296875,
      "learning_rate": 6.67654343915014e-06,
      "loss": 0.0323,
      "step": 18146
    },
    {
      "epoch": 4.4743589743589745,
      "grad_norm": 0.271484375,
      "learning_rate": 6.664181309468232e-06,
      "loss": 0.0315,
      "step": 18148
    },
    {
      "epoch": 4.474852071005917,
      "grad_norm": 0.30859375,
      "learning_rate": 6.6518302406019524e-06,
      "loss": 0.0357,
      "step": 18150
    },
    {
      "epoch": 4.4753451676528595,
      "grad_norm": 0.322265625,
      "learning_rate": 6.639490234014989e-06,
      "loss": 0.0339,
      "step": 18152
    },
    {
      "epoch": 4.475838264299803,
      "grad_norm": 0.3671875,
      "learning_rate": 6.627161291169692e-06,
      "loss": 0.0312,
      "step": 18154
    },
    {
      "epoch": 4.476331360946745,
      "grad_norm": 0.349609375,
      "learning_rate": 6.614843413527106e-06,
      "loss": 0.0322,
      "step": 18156
    },
    {
      "epoch": 4.476824457593688,
      "grad_norm": 0.333984375,
      "learning_rate": 6.602536602546983e-06,
      "loss": 0.0337,
      "step": 18158
    },
    {
      "epoch": 4.477317554240631,
      "grad_norm": 0.30078125,
      "learning_rate": 6.590240859687702e-06,
      "loss": 0.0317,
      "step": 18160
    },
    {
      "epoch": 4.477810650887574,
      "grad_norm": 0.26171875,
      "learning_rate": 6.5779561864064535e-06,
      "loss": 0.0313,
      "step": 18162
    },
    {
      "epoch": 4.478303747534516,
      "grad_norm": 0.2890625,
      "learning_rate": 6.565682584158983e-06,
      "loss": 0.0349,
      "step": 18164
    },
    {
      "epoch": 4.47879684418146,
      "grad_norm": 0.306640625,
      "learning_rate": 6.553420054399794e-06,
      "loss": 0.0322,
      "step": 18166
    },
    {
      "epoch": 4.479289940828402,
      "grad_norm": 0.341796875,
      "learning_rate": 6.54116859858207e-06,
      "loss": 0.0305,
      "step": 18168
    },
    {
      "epoch": 4.479783037475345,
      "grad_norm": 0.34375,
      "learning_rate": 6.528928218157671e-06,
      "loss": 0.0339,
      "step": 18170
    },
    {
      "epoch": 4.480276134122288,
      "grad_norm": 0.30859375,
      "learning_rate": 6.516698914577168e-06,
      "loss": 0.0347,
      "step": 18172
    },
    {
      "epoch": 4.480769230769231,
      "grad_norm": 0.310546875,
      "learning_rate": 6.5044806892897605e-06,
      "loss": 0.0352,
      "step": 18174
    },
    {
      "epoch": 4.481262327416173,
      "grad_norm": 0.3203125,
      "learning_rate": 6.492273543743432e-06,
      "loss": 0.0355,
      "step": 18176
    },
    {
      "epoch": 4.481755424063117,
      "grad_norm": 0.29296875,
      "learning_rate": 6.48007747938475e-06,
      "loss": 0.0335,
      "step": 18178
    },
    {
      "epoch": 4.482248520710059,
      "grad_norm": 0.28515625,
      "learning_rate": 6.467892497659045e-06,
      "loss": 0.0335,
      "step": 18180
    },
    {
      "epoch": 4.482741617357002,
      "grad_norm": 0.30859375,
      "learning_rate": 6.455718600010285e-06,
      "loss": 0.0307,
      "step": 18182
    },
    {
      "epoch": 4.483234714003945,
      "grad_norm": 0.34375,
      "learning_rate": 6.443555787881161e-06,
      "loss": 0.0346,
      "step": 18184
    },
    {
      "epoch": 4.483727810650888,
      "grad_norm": 0.294921875,
      "learning_rate": 6.431404062713042e-06,
      "loss": 0.0321,
      "step": 18186
    },
    {
      "epoch": 4.48422090729783,
      "grad_norm": 0.28515625,
      "learning_rate": 6.419263425945932e-06,
      "loss": 0.0323,
      "step": 18188
    },
    {
      "epoch": 4.484714003944773,
      "grad_norm": 0.337890625,
      "learning_rate": 6.407133879018623e-06,
      "loss": 0.0339,
      "step": 18190
    },
    {
      "epoch": 4.485207100591716,
      "grad_norm": 0.302734375,
      "learning_rate": 6.3950154233684775e-06,
      "loss": 0.0325,
      "step": 18192
    },
    {
      "epoch": 4.485700197238659,
      "grad_norm": 0.291015625,
      "learning_rate": 6.3829080604316585e-06,
      "loss": 0.0334,
      "step": 18194
    },
    {
      "epoch": 4.486193293885601,
      "grad_norm": 0.2890625,
      "learning_rate": 6.370811791642906e-06,
      "loss": 0.0286,
      "step": 18196
    },
    {
      "epoch": 4.486686390532545,
      "grad_norm": 0.330078125,
      "learning_rate": 6.3587266184357176e-06,
      "loss": 0.0335,
      "step": 18198
    },
    {
      "epoch": 4.487179487179487,
      "grad_norm": 0.26953125,
      "learning_rate": 6.34665254224226e-06,
      "loss": 0.0333,
      "step": 18200
    },
    {
      "epoch": 4.48767258382643,
      "grad_norm": 0.330078125,
      "learning_rate": 6.334589564493343e-06,
      "loss": 0.0344,
      "step": 18202
    },
    {
      "epoch": 4.488165680473373,
      "grad_norm": 0.294921875,
      "learning_rate": 6.322537686618557e-06,
      "loss": 0.0314,
      "step": 18204
    },
    {
      "epoch": 4.488658777120316,
      "grad_norm": 0.32421875,
      "learning_rate": 6.310496910046049e-06,
      "loss": 0.0339,
      "step": 18206
    },
    {
      "epoch": 4.489151873767258,
      "grad_norm": 0.30859375,
      "learning_rate": 6.298467236202787e-06,
      "loss": 0.0326,
      "step": 18208
    },
    {
      "epoch": 4.489644970414201,
      "grad_norm": 0.328125,
      "learning_rate": 6.2864486665142996e-06,
      "loss": 0.0374,
      "step": 18210
    },
    {
      "epoch": 4.490138067061144,
      "grad_norm": 0.27734375,
      "learning_rate": 6.274441202404879e-06,
      "loss": 0.0305,
      "step": 18212
    },
    {
      "epoch": 4.490631163708087,
      "grad_norm": 0.2890625,
      "learning_rate": 6.2624448452974885e-06,
      "loss": 0.0314,
      "step": 18214
    },
    {
      "epoch": 4.491124260355029,
      "grad_norm": 0.267578125,
      "learning_rate": 6.250459596613711e-06,
      "loss": 0.0296,
      "step": 18216
    },
    {
      "epoch": 4.491617357001973,
      "grad_norm": 0.287109375,
      "learning_rate": 6.238485457773935e-06,
      "loss": 0.0315,
      "step": 18218
    },
    {
      "epoch": 4.492110453648915,
      "grad_norm": 0.2890625,
      "learning_rate": 6.226522430197102e-06,
      "loss": 0.0309,
      "step": 18220
    },
    {
      "epoch": 4.492603550295858,
      "grad_norm": 0.302734375,
      "learning_rate": 6.214570515300944e-06,
      "loss": 0.0318,
      "step": 18222
    },
    {
      "epoch": 4.493096646942801,
      "grad_norm": 0.296875,
      "learning_rate": 6.202629714501807e-06,
      "loss": 0.031,
      "step": 18224
    },
    {
      "epoch": 4.493589743589744,
      "grad_norm": 0.294921875,
      "learning_rate": 6.1907000292147464e-06,
      "loss": 0.0332,
      "step": 18226
    },
    {
      "epoch": 4.494082840236686,
      "grad_norm": 0.271484375,
      "learning_rate": 6.178781460853511e-06,
      "loss": 0.0269,
      "step": 18228
    },
    {
      "epoch": 4.4945759368836296,
      "grad_norm": 0.271484375,
      "learning_rate": 6.16687401083047e-06,
      "loss": 0.0311,
      "step": 18230
    },
    {
      "epoch": 4.495069033530572,
      "grad_norm": 0.27734375,
      "learning_rate": 6.154977680556783e-06,
      "loss": 0.0312,
      "step": 18232
    },
    {
      "epoch": 4.495562130177515,
      "grad_norm": 0.310546875,
      "learning_rate": 6.14309247144218e-06,
      "loss": 0.0371,
      "step": 18234
    },
    {
      "epoch": 4.496055226824458,
      "grad_norm": 0.291015625,
      "learning_rate": 6.131218384895176e-06,
      "loss": 0.0334,
      "step": 18236
    },
    {
      "epoch": 4.4965483234714005,
      "grad_norm": 0.314453125,
      "learning_rate": 6.11935542232287e-06,
      "loss": 0.0347,
      "step": 18238
    },
    {
      "epoch": 4.497041420118343,
      "grad_norm": 0.373046875,
      "learning_rate": 6.107503585131113e-06,
      "loss": 0.0322,
      "step": 18240
    },
    {
      "epoch": 4.497534516765286,
      "grad_norm": 0.287109375,
      "learning_rate": 6.095662874724417e-06,
      "loss": 0.0343,
      "step": 18242
    },
    {
      "epoch": 4.498027613412229,
      "grad_norm": 0.318359375,
      "learning_rate": 6.083833292505958e-06,
      "loss": 0.0337,
      "step": 18244
    },
    {
      "epoch": 4.4985207100591715,
      "grad_norm": 0.310546875,
      "learning_rate": 6.072014839877627e-06,
      "loss": 0.0326,
      "step": 18246
    },
    {
      "epoch": 4.499013806706114,
      "grad_norm": 0.322265625,
      "learning_rate": 6.060207518239946e-06,
      "loss": 0.0344,
      "step": 18248
    },
    {
      "epoch": 4.4995069033530575,
      "grad_norm": 0.361328125,
      "learning_rate": 6.048411328992187e-06,
      "loss": 0.0406,
      "step": 18250
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.333984375,
      "learning_rate": 6.036626273532231e-06,
      "loss": 0.0387,
      "step": 18252
    },
    {
      "epoch": 4.5004930966469425,
      "grad_norm": 0.275390625,
      "learning_rate": 6.024852353256682e-06,
      "loss": 0.0327,
      "step": 18254
    },
    {
      "epoch": 4.500986193293886,
      "grad_norm": 0.263671875,
      "learning_rate": 6.013089569560825e-06,
      "loss": 0.0323,
      "step": 18256
    },
    {
      "epoch": 4.5014792899408285,
      "grad_norm": 0.333984375,
      "learning_rate": 6.001337923838602e-06,
      "loss": 0.0344,
      "step": 18258
    },
    {
      "epoch": 4.501972386587771,
      "grad_norm": 0.30859375,
      "learning_rate": 5.989597417482684e-06,
      "loss": 0.0375,
      "step": 18260
    },
    {
      "epoch": 4.5024654832347135,
      "grad_norm": 0.302734375,
      "learning_rate": 5.977868051884317e-06,
      "loss": 0.0321,
      "step": 18262
    },
    {
      "epoch": 4.502958579881657,
      "grad_norm": 0.267578125,
      "learning_rate": 5.966149828433576e-06,
      "loss": 0.033,
      "step": 18264
    },
    {
      "epoch": 4.5034516765285995,
      "grad_norm": 0.322265625,
      "learning_rate": 5.954442748519073e-06,
      "loss": 0.037,
      "step": 18266
    },
    {
      "epoch": 4.503944773175542,
      "grad_norm": 0.3046875,
      "learning_rate": 5.9427468135282e-06,
      "loss": 0.0316,
      "step": 18268
    },
    {
      "epoch": 4.504437869822485,
      "grad_norm": 0.30078125,
      "learning_rate": 5.931062024846967e-06,
      "loss": 0.0326,
      "step": 18270
    },
    {
      "epoch": 4.504930966469428,
      "grad_norm": 0.296875,
      "learning_rate": 5.919388383860103e-06,
      "loss": 0.0323,
      "step": 18272
    },
    {
      "epoch": 4.5054240631163704,
      "grad_norm": 0.3046875,
      "learning_rate": 5.907725891950999e-06,
      "loss": 0.0318,
      "step": 18274
    },
    {
      "epoch": 4.505917159763314,
      "grad_norm": 0.267578125,
      "learning_rate": 5.8960745505017175e-06,
      "loss": 0.0342,
      "step": 18276
    },
    {
      "epoch": 4.506410256410256,
      "grad_norm": 0.37890625,
      "learning_rate": 5.884434360893021e-06,
      "loss": 0.0332,
      "step": 18278
    },
    {
      "epoch": 4.506903353057199,
      "grad_norm": 0.36328125,
      "learning_rate": 5.872805324504316e-06,
      "loss": 0.0336,
      "step": 18280
    },
    {
      "epoch": 4.507396449704142,
      "grad_norm": 0.26953125,
      "learning_rate": 5.8611874427137225e-06,
      "loss": 0.0347,
      "step": 18282
    },
    {
      "epoch": 4.507889546351085,
      "grad_norm": 0.29296875,
      "learning_rate": 5.849580716898017e-06,
      "loss": 0.0307,
      "step": 18284
    },
    {
      "epoch": 4.508382642998027,
      "grad_norm": 0.322265625,
      "learning_rate": 5.8379851484326566e-06,
      "loss": 0.0322,
      "step": 18286
    },
    {
      "epoch": 4.508875739644971,
      "grad_norm": 0.32421875,
      "learning_rate": 5.826400738691806e-06,
      "loss": 0.0335,
      "step": 18288
    },
    {
      "epoch": 4.509368836291913,
      "grad_norm": 0.296875,
      "learning_rate": 5.814827489048258e-06,
      "loss": 0.0324,
      "step": 18290
    },
    {
      "epoch": 4.509861932938856,
      "grad_norm": 0.3125,
      "learning_rate": 5.803265400873514e-06,
      "loss": 0.0284,
      "step": 18292
    },
    {
      "epoch": 4.510355029585799,
      "grad_norm": 0.2890625,
      "learning_rate": 5.7917144755377576e-06,
      "loss": 0.0316,
      "step": 18294
    },
    {
      "epoch": 4.510848126232742,
      "grad_norm": 0.4140625,
      "learning_rate": 5.780174714409825e-06,
      "loss": 0.0343,
      "step": 18296
    },
    {
      "epoch": 4.511341222879684,
      "grad_norm": 0.2890625,
      "learning_rate": 5.7686461188572345e-06,
      "loss": 0.0326,
      "step": 18298
    },
    {
      "epoch": 4.511834319526627,
      "grad_norm": 0.35546875,
      "learning_rate": 5.757128690246205e-06,
      "loss": 0.0347,
      "step": 18300
    },
    {
      "epoch": 4.51232741617357,
      "grad_norm": 0.306640625,
      "learning_rate": 5.7456224299415996e-06,
      "loss": 0.0325,
      "step": 18302
    },
    {
      "epoch": 4.512820512820513,
      "grad_norm": 0.314453125,
      "learning_rate": 5.734127339306994e-06,
      "loss": 0.0333,
      "step": 18304
    },
    {
      "epoch": 4.513313609467455,
      "grad_norm": 0.43359375,
      "learning_rate": 5.7226434197046206e-06,
      "loss": 0.0355,
      "step": 18306
    },
    {
      "epoch": 4.513806706114399,
      "grad_norm": 0.365234375,
      "learning_rate": 5.71117067249538e-06,
      "loss": 0.0317,
      "step": 18308
    },
    {
      "epoch": 4.514299802761341,
      "grad_norm": 0.263671875,
      "learning_rate": 5.6997090990388525e-06,
      "loss": 0.0308,
      "step": 18310
    },
    {
      "epoch": 4.514792899408284,
      "grad_norm": 0.291015625,
      "learning_rate": 5.688258700693294e-06,
      "loss": 0.0323,
      "step": 18312
    },
    {
      "epoch": 4.515285996055227,
      "grad_norm": 0.287109375,
      "learning_rate": 5.676819478815654e-06,
      "loss": 0.0311,
      "step": 18314
    },
    {
      "epoch": 4.51577909270217,
      "grad_norm": 0.298828125,
      "learning_rate": 5.665391434761535e-06,
      "loss": 0.0316,
      "step": 18316
    },
    {
      "epoch": 4.516272189349112,
      "grad_norm": 0.28515625,
      "learning_rate": 5.653974569885223e-06,
      "loss": 0.0315,
      "step": 18318
    },
    {
      "epoch": 4.516765285996055,
      "grad_norm": 0.30078125,
      "learning_rate": 5.642568885539701e-06,
      "loss": 0.031,
      "step": 18320
    },
    {
      "epoch": 4.517258382642998,
      "grad_norm": 0.32421875,
      "learning_rate": 5.631174383076587e-06,
      "loss": 0.0313,
      "step": 18322
    },
    {
      "epoch": 4.517751479289941,
      "grad_norm": 0.3046875,
      "learning_rate": 5.61979106384618e-06,
      "loss": 0.0313,
      "step": 18324
    },
    {
      "epoch": 4.518244575936883,
      "grad_norm": 0.283203125,
      "learning_rate": 5.608418929197512e-06,
      "loss": 0.0351,
      "step": 18326
    },
    {
      "epoch": 4.518737672583827,
      "grad_norm": 0.291015625,
      "learning_rate": 5.597057980478193e-06,
      "loss": 0.0317,
      "step": 18328
    },
    {
      "epoch": 4.519230769230769,
      "grad_norm": 0.328125,
      "learning_rate": 5.585708219034591e-06,
      "loss": 0.0334,
      "step": 18330
    },
    {
      "epoch": 4.519723865877712,
      "grad_norm": 0.31640625,
      "learning_rate": 5.574369646211697e-06,
      "loss": 0.0341,
      "step": 18332
    },
    {
      "epoch": 4.520216962524655,
      "grad_norm": 0.275390625,
      "learning_rate": 5.563042263353213e-06,
      "loss": 0.0334,
      "step": 18334
    },
    {
      "epoch": 4.520710059171598,
      "grad_norm": 0.283203125,
      "learning_rate": 5.551726071801511e-06,
      "loss": 0.0316,
      "step": 18336
    },
    {
      "epoch": 4.52120315581854,
      "grad_norm": 0.296875,
      "learning_rate": 5.5404210728975616e-06,
      "loss": 0.0318,
      "step": 18338
    },
    {
      "epoch": 4.521696252465484,
      "grad_norm": 0.275390625,
      "learning_rate": 5.529127267981138e-06,
      "loss": 0.0312,
      "step": 18340
    },
    {
      "epoch": 4.522189349112426,
      "grad_norm": 0.32421875,
      "learning_rate": 5.51784465839058e-06,
      "loss": 0.0307,
      "step": 18342
    },
    {
      "epoch": 4.522682445759369,
      "grad_norm": 0.30078125,
      "learning_rate": 5.506573245462943e-06,
      "loss": 0.0333,
      "step": 18344
    },
    {
      "epoch": 4.523175542406312,
      "grad_norm": 0.28515625,
      "learning_rate": 5.495313030533966e-06,
      "loss": 0.0311,
      "step": 18346
    },
    {
      "epoch": 4.523668639053255,
      "grad_norm": 0.337890625,
      "learning_rate": 5.484064014938051e-06,
      "loss": 0.0344,
      "step": 18348
    },
    {
      "epoch": 4.524161735700197,
      "grad_norm": 0.306640625,
      "learning_rate": 5.472826200008263e-06,
      "loss": 0.0354,
      "step": 18350
    },
    {
      "epoch": 4.5246548323471405,
      "grad_norm": 0.3671875,
      "learning_rate": 5.461599587076316e-06,
      "loss": 0.0324,
      "step": 18352
    },
    {
      "epoch": 4.525147928994083,
      "grad_norm": 0.265625,
      "learning_rate": 5.450384177472678e-06,
      "loss": 0.0314,
      "step": 18354
    },
    {
      "epoch": 4.5256410256410255,
      "grad_norm": 0.298828125,
      "learning_rate": 5.439179972526387e-06,
      "loss": 0.033,
      "step": 18356
    },
    {
      "epoch": 4.526134122287968,
      "grad_norm": 0.328125,
      "learning_rate": 5.427986973565258e-06,
      "loss": 0.0326,
      "step": 18358
    },
    {
      "epoch": 4.5266272189349115,
      "grad_norm": 0.306640625,
      "learning_rate": 5.416805181915685e-06,
      "loss": 0.0339,
      "step": 18360
    },
    {
      "epoch": 4.527120315581854,
      "grad_norm": 0.3203125,
      "learning_rate": 5.405634598902776e-06,
      "loss": 0.035,
      "step": 18362
    },
    {
      "epoch": 4.5276134122287965,
      "grad_norm": 0.255859375,
      "learning_rate": 5.394475225850337e-06,
      "loss": 0.0298,
      "step": 18364
    },
    {
      "epoch": 4.52810650887574,
      "grad_norm": 0.322265625,
      "learning_rate": 5.383327064080757e-06,
      "loss": 0.0305,
      "step": 18366
    },
    {
      "epoch": 4.5285996055226825,
      "grad_norm": 0.345703125,
      "learning_rate": 5.372190114915221e-06,
      "loss": 0.0334,
      "step": 18368
    },
    {
      "epoch": 4.529092702169625,
      "grad_norm": 0.296875,
      "learning_rate": 5.361064379673464e-06,
      "loss": 0.033,
      "step": 18370
    },
    {
      "epoch": 4.529585798816568,
      "grad_norm": 0.2890625,
      "learning_rate": 5.349949859673998e-06,
      "loss": 0.0295,
      "step": 18372
    },
    {
      "epoch": 4.530078895463511,
      "grad_norm": 0.31640625,
      "learning_rate": 5.338846556233923e-06,
      "loss": 0.0372,
      "step": 18374
    },
    {
      "epoch": 4.5305719921104535,
      "grad_norm": 0.318359375,
      "learning_rate": 5.3277544706690555e-06,
      "loss": 0.035,
      "step": 18376
    },
    {
      "epoch": 4.531065088757396,
      "grad_norm": 0.267578125,
      "learning_rate": 5.316673604293876e-06,
      "loss": 0.0333,
      "step": 18378
    },
    {
      "epoch": 4.531558185404339,
      "grad_norm": 0.328125,
      "learning_rate": 5.305603958421479e-06,
      "loss": 0.0355,
      "step": 18380
    },
    {
      "epoch": 4.532051282051282,
      "grad_norm": 0.298828125,
      "learning_rate": 5.294545534363759e-06,
      "loss": 0.0336,
      "step": 18382
    },
    {
      "epoch": 4.5325443786982245,
      "grad_norm": 0.294921875,
      "learning_rate": 5.2834983334311365e-06,
      "loss": 0.0338,
      "step": 18384
    },
    {
      "epoch": 4.533037475345168,
      "grad_norm": 0.3125,
      "learning_rate": 5.272462356932805e-06,
      "loss": 0.0322,
      "step": 18386
    },
    {
      "epoch": 4.53353057199211,
      "grad_norm": 0.318359375,
      "learning_rate": 5.261437606176567e-06,
      "loss": 0.0309,
      "step": 18388
    },
    {
      "epoch": 4.534023668639053,
      "grad_norm": 0.31640625,
      "learning_rate": 5.25042408246893e-06,
      "loss": 0.0319,
      "step": 18390
    },
    {
      "epoch": 4.534516765285996,
      "grad_norm": 0.328125,
      "learning_rate": 5.239421787115062e-06,
      "loss": 0.0302,
      "step": 18392
    },
    {
      "epoch": 4.535009861932939,
      "grad_norm": 0.26953125,
      "learning_rate": 5.228430721418764e-06,
      "loss": 0.0278,
      "step": 18394
    },
    {
      "epoch": 4.535502958579881,
      "grad_norm": 0.3046875,
      "learning_rate": 5.217450886682584e-06,
      "loss": 0.0354,
      "step": 18396
    },
    {
      "epoch": 4.535996055226825,
      "grad_norm": 0.28515625,
      "learning_rate": 5.206482284207647e-06,
      "loss": 0.0342,
      "step": 18398
    },
    {
      "epoch": 4.536489151873767,
      "grad_norm": 0.33984375,
      "learning_rate": 5.195524915293848e-06,
      "loss": 0.0329,
      "step": 18400
    },
    {
      "epoch": 4.53698224852071,
      "grad_norm": 0.314453125,
      "learning_rate": 5.184578781239646e-06,
      "loss": 0.0309,
      "step": 18402
    },
    {
      "epoch": 4.537475345167653,
      "grad_norm": 0.283203125,
      "learning_rate": 5.1736438833422495e-06,
      "loss": 0.0309,
      "step": 18404
    },
    {
      "epoch": 4.537968441814596,
      "grad_norm": 0.28515625,
      "learning_rate": 5.162720222897488e-06,
      "loss": 0.0321,
      "step": 18406
    },
    {
      "epoch": 4.538461538461538,
      "grad_norm": 0.271484375,
      "learning_rate": 5.151807801199892e-06,
      "loss": 0.0308,
      "step": 18408
    },
    {
      "epoch": 4.538954635108482,
      "grad_norm": 0.333984375,
      "learning_rate": 5.14090661954264e-06,
      "loss": 0.0329,
      "step": 18410
    },
    {
      "epoch": 4.539447731755424,
      "grad_norm": 0.328125,
      "learning_rate": 5.130016679217564e-06,
      "loss": 0.0357,
      "step": 18412
    },
    {
      "epoch": 4.539940828402367,
      "grad_norm": 0.34765625,
      "learning_rate": 5.11913798151521e-06,
      "loss": 0.0328,
      "step": 18414
    },
    {
      "epoch": 4.540433925049309,
      "grad_norm": 0.3203125,
      "learning_rate": 5.108270527724745e-06,
      "loss": 0.0315,
      "step": 18416
    },
    {
      "epoch": 4.540927021696253,
      "grad_norm": 0.298828125,
      "learning_rate": 5.097414319134042e-06,
      "loss": 0.0348,
      "step": 18418
    },
    {
      "epoch": 4.541420118343195,
      "grad_norm": 0.291015625,
      "learning_rate": 5.086569357029591e-06,
      "loss": 0.0312,
      "step": 18420
    },
    {
      "epoch": 4.541913214990138,
      "grad_norm": 0.2734375,
      "learning_rate": 5.075735642696611e-06,
      "loss": 0.0334,
      "step": 18422
    },
    {
      "epoch": 4.542406311637081,
      "grad_norm": 0.302734375,
      "learning_rate": 5.064913177418939e-06,
      "loss": 0.0333,
      "step": 18424
    },
    {
      "epoch": 4.542899408284024,
      "grad_norm": 0.310546875,
      "learning_rate": 5.054101962479119e-06,
      "loss": 0.0353,
      "step": 18426
    },
    {
      "epoch": 4.543392504930966,
      "grad_norm": 0.291015625,
      "learning_rate": 5.043301999158323e-06,
      "loss": 0.0331,
      "step": 18428
    },
    {
      "epoch": 4.543885601577909,
      "grad_norm": 0.275390625,
      "learning_rate": 5.032513288736407e-06,
      "loss": 0.0304,
      "step": 18430
    },
    {
      "epoch": 4.544378698224852,
      "grad_norm": 0.30859375,
      "learning_rate": 5.021735832491892e-06,
      "loss": 0.0311,
      "step": 18432
    },
    {
      "epoch": 4.544871794871795,
      "grad_norm": 0.2734375,
      "learning_rate": 5.010969631701978e-06,
      "loss": 0.031,
      "step": 18434
    },
    {
      "epoch": 4.545364891518737,
      "grad_norm": 0.291015625,
      "learning_rate": 5.000214687642513e-06,
      "loss": 0.0336,
      "step": 18436
    },
    {
      "epoch": 4.545857988165681,
      "grad_norm": 0.291015625,
      "learning_rate": 4.989471001588009e-06,
      "loss": 0.0296,
      "step": 18438
    },
    {
      "epoch": 4.546351084812623,
      "grad_norm": 0.318359375,
      "learning_rate": 4.97873857481167e-06,
      "loss": 0.0334,
      "step": 18440
    },
    {
      "epoch": 4.546844181459566,
      "grad_norm": 0.291015625,
      "learning_rate": 4.968017408585357e-06,
      "loss": 0.0329,
      "step": 18442
    },
    {
      "epoch": 4.547337278106509,
      "grad_norm": 0.3046875,
      "learning_rate": 4.957307504179553e-06,
      "loss": 0.0365,
      "step": 18444
    },
    {
      "epoch": 4.547830374753452,
      "grad_norm": 0.33984375,
      "learning_rate": 4.946608862863456e-06,
      "loss": 0.0312,
      "step": 18446
    },
    {
      "epoch": 4.548323471400394,
      "grad_norm": 0.3671875,
      "learning_rate": 4.935921485904926e-06,
      "loss": 0.0354,
      "step": 18448
    },
    {
      "epoch": 4.548816568047338,
      "grad_norm": 0.306640625,
      "learning_rate": 4.925245374570464e-06,
      "loss": 0.0321,
      "step": 18450
    },
    {
      "epoch": 4.54930966469428,
      "grad_norm": 0.271484375,
      "learning_rate": 4.9145805301252565e-06,
      "loss": 0.0314,
      "step": 18452
    },
    {
      "epoch": 4.549802761341223,
      "grad_norm": 0.298828125,
      "learning_rate": 4.903926953833149e-06,
      "loss": 0.032,
      "step": 18454
    },
    {
      "epoch": 4.550295857988166,
      "grad_norm": 0.337890625,
      "learning_rate": 4.893284646956642e-06,
      "loss": 0.0356,
      "step": 18456
    },
    {
      "epoch": 4.550788954635109,
      "grad_norm": 0.30078125,
      "learning_rate": 4.882653610756937e-06,
      "loss": 0.0315,
      "step": 18458
    },
    {
      "epoch": 4.551282051282051,
      "grad_norm": 0.26953125,
      "learning_rate": 4.872033846493818e-06,
      "loss": 0.0309,
      "step": 18460
    },
    {
      "epoch": 4.5517751479289945,
      "grad_norm": 0.298828125,
      "learning_rate": 4.861425355425819e-06,
      "loss": 0.0319,
      "step": 18462
    },
    {
      "epoch": 4.552268244575937,
      "grad_norm": 0.373046875,
      "learning_rate": 4.850828138810104e-06,
      "loss": 0.0327,
      "step": 18464
    },
    {
      "epoch": 4.55276134122288,
      "grad_norm": 0.259765625,
      "learning_rate": 4.8402421979025e-06,
      "loss": 0.0322,
      "step": 18466
    },
    {
      "epoch": 4.553254437869823,
      "grad_norm": 0.27734375,
      "learning_rate": 4.829667533957494e-06,
      "loss": 0.0366,
      "step": 18468
    },
    {
      "epoch": 4.5537475345167655,
      "grad_norm": 0.318359375,
      "learning_rate": 4.819104148228249e-06,
      "loss": 0.0305,
      "step": 18470
    },
    {
      "epoch": 4.554240631163708,
      "grad_norm": 0.263671875,
      "learning_rate": 4.808552041966608e-06,
      "loss": 0.0304,
      "step": 18472
    },
    {
      "epoch": 4.554733727810651,
      "grad_norm": 0.283203125,
      "learning_rate": 4.798011216423004e-06,
      "loss": 0.0309,
      "step": 18474
    },
    {
      "epoch": 4.555226824457594,
      "grad_norm": 0.3125,
      "learning_rate": 4.787481672846616e-06,
      "loss": 0.0331,
      "step": 18476
    },
    {
      "epoch": 4.5557199211045365,
      "grad_norm": 0.26953125,
      "learning_rate": 4.776963412485258e-06,
      "loss": 0.0327,
      "step": 18478
    },
    {
      "epoch": 4.556213017751479,
      "grad_norm": 0.267578125,
      "learning_rate": 4.766456436585376e-06,
      "loss": 0.0323,
      "step": 18480
    },
    {
      "epoch": 4.556706114398422,
      "grad_norm": 0.328125,
      "learning_rate": 4.755960746392129e-06,
      "loss": 0.033,
      "step": 18482
    },
    {
      "epoch": 4.557199211045365,
      "grad_norm": 0.287109375,
      "learning_rate": 4.745476343149313e-06,
      "loss": 0.031,
      "step": 18484
    },
    {
      "epoch": 4.5576923076923075,
      "grad_norm": 0.29296875,
      "learning_rate": 4.735003228099399e-06,
      "loss": 0.0341,
      "step": 18486
    },
    {
      "epoch": 4.55818540433925,
      "grad_norm": 0.28125,
      "learning_rate": 4.724541402483462e-06,
      "loss": 0.034,
      "step": 18488
    },
    {
      "epoch": 4.558678500986193,
      "grad_norm": 0.271484375,
      "learning_rate": 4.714090867541343e-06,
      "loss": 0.031,
      "step": 18490
    },
    {
      "epoch": 4.559171597633136,
      "grad_norm": 0.31640625,
      "learning_rate": 4.7036516245114514e-06,
      "loss": 0.0342,
      "step": 18492
    },
    {
      "epoch": 4.5596646942800785,
      "grad_norm": 0.349609375,
      "learning_rate": 4.693223674630909e-06,
      "loss": 0.0302,
      "step": 18494
    },
    {
      "epoch": 4.560157790927022,
      "grad_norm": 0.333984375,
      "learning_rate": 4.6828070191354935e-06,
      "loss": 0.0314,
      "step": 18496
    },
    {
      "epoch": 4.560650887573964,
      "grad_norm": 0.3125,
      "learning_rate": 4.672401659259629e-06,
      "loss": 0.0332,
      "step": 18498
    },
    {
      "epoch": 4.561143984220907,
      "grad_norm": 0.296875,
      "learning_rate": 4.6620075962364175e-06,
      "loss": 0.0335,
      "step": 18500
    },
    {
      "epoch": 4.56163708086785,
      "grad_norm": 0.2890625,
      "learning_rate": 4.651624831297574e-06,
      "loss": 0.0306,
      "step": 18502
    },
    {
      "epoch": 4.562130177514793,
      "grad_norm": 0.310546875,
      "learning_rate": 4.641253365673581e-06,
      "loss": 0.0306,
      "step": 18504
    },
    {
      "epoch": 4.562623274161735,
      "grad_norm": 0.32421875,
      "learning_rate": 4.630893200593445e-06,
      "loss": 0.0336,
      "step": 18506
    },
    {
      "epoch": 4.563116370808679,
      "grad_norm": 0.330078125,
      "learning_rate": 4.62054433728496e-06,
      "loss": 0.0339,
      "step": 18508
    },
    {
      "epoch": 4.563609467455621,
      "grad_norm": 0.30859375,
      "learning_rate": 4.610206776974502e-06,
      "loss": 0.0275,
      "step": 18510
    },
    {
      "epoch": 4.564102564102564,
      "grad_norm": 0.302734375,
      "learning_rate": 4.599880520887123e-06,
      "loss": 0.0314,
      "step": 18512
    },
    {
      "epoch": 4.564595660749507,
      "grad_norm": 0.283203125,
      "learning_rate": 4.589565570246556e-06,
      "loss": 0.0322,
      "step": 18514
    },
    {
      "epoch": 4.56508875739645,
      "grad_norm": 0.330078125,
      "learning_rate": 4.579261926275158e-06,
      "loss": 0.0351,
      "step": 18516
    },
    {
      "epoch": 4.565581854043392,
      "grad_norm": 0.326171875,
      "learning_rate": 4.568969590194005e-06,
      "loss": 0.0314,
      "step": 18518
    },
    {
      "epoch": 4.566074950690336,
      "grad_norm": 0.302734375,
      "learning_rate": 4.558688563222746e-06,
      "loss": 0.0321,
      "step": 18520
    },
    {
      "epoch": 4.566568047337278,
      "grad_norm": 0.283203125,
      "learning_rate": 4.5484188465797935e-06,
      "loss": 0.0323,
      "step": 18522
    },
    {
      "epoch": 4.567061143984221,
      "grad_norm": 0.30859375,
      "learning_rate": 4.538160441482131e-06,
      "loss": 0.0322,
      "step": 18524
    },
    {
      "epoch": 4.567554240631163,
      "grad_norm": 0.296875,
      "learning_rate": 4.527913349145441e-06,
      "loss": 0.0316,
      "step": 18526
    },
    {
      "epoch": 4.568047337278107,
      "grad_norm": 0.302734375,
      "learning_rate": 4.5176775707840735e-06,
      "loss": 0.0335,
      "step": 18528
    },
    {
      "epoch": 4.568540433925049,
      "grad_norm": 0.28515625,
      "learning_rate": 4.507453107611015e-06,
      "loss": 0.0335,
      "step": 18530
    },
    {
      "epoch": 4.569033530571992,
      "grad_norm": 0.31640625,
      "learning_rate": 4.49723996083794e-06,
      "loss": 0.0306,
      "step": 18532
    },
    {
      "epoch": 4.569526627218935,
      "grad_norm": 0.330078125,
      "learning_rate": 4.487038131675114e-06,
      "loss": 0.0305,
      "step": 18534
    },
    {
      "epoch": 4.570019723865878,
      "grad_norm": 0.357421875,
      "learning_rate": 4.476847621331581e-06,
      "loss": 0.0341,
      "step": 18536
    },
    {
      "epoch": 4.57051282051282,
      "grad_norm": 0.28515625,
      "learning_rate": 4.466668431014931e-06,
      "loss": 0.0333,
      "step": 18538
    },
    {
      "epoch": 4.571005917159764,
      "grad_norm": 0.38671875,
      "learning_rate": 4.456500561931454e-06,
      "loss": 0.0386,
      "step": 18540
    },
    {
      "epoch": 4.571499013806706,
      "grad_norm": 0.279296875,
      "learning_rate": 4.446344015286108e-06,
      "loss": 0.0367,
      "step": 18542
    },
    {
      "epoch": 4.571992110453649,
      "grad_norm": 0.30859375,
      "learning_rate": 4.436198792282498e-06,
      "loss": 0.0308,
      "step": 18544
    },
    {
      "epoch": 4.572485207100591,
      "grad_norm": 0.33984375,
      "learning_rate": 4.426064894122905e-06,
      "loss": 0.0323,
      "step": 18546
    },
    {
      "epoch": 4.572978303747535,
      "grad_norm": 0.294921875,
      "learning_rate": 4.415942322008215e-06,
      "loss": 0.0337,
      "step": 18548
    },
    {
      "epoch": 4.573471400394477,
      "grad_norm": 0.32421875,
      "learning_rate": 4.405831077138056e-06,
      "loss": 0.0343,
      "step": 18550
    },
    {
      "epoch": 4.57396449704142,
      "grad_norm": 0.314453125,
      "learning_rate": 4.3957311607106256e-06,
      "loss": 0.0306,
      "step": 18552
    },
    {
      "epoch": 4.574457593688363,
      "grad_norm": 0.283203125,
      "learning_rate": 4.385642573922843e-06,
      "loss": 0.0335,
      "step": 18554
    },
    {
      "epoch": 4.574950690335306,
      "grad_norm": 0.32421875,
      "learning_rate": 4.375565317970265e-06,
      "loss": 0.0333,
      "step": 18556
    },
    {
      "epoch": 4.575443786982248,
      "grad_norm": 0.28515625,
      "learning_rate": 4.36549939404709e-06,
      "loss": 0.0315,
      "step": 18558
    },
    {
      "epoch": 4.575936883629192,
      "grad_norm": 0.2734375,
      "learning_rate": 4.355444803346198e-06,
      "loss": 0.0328,
      "step": 18560
    },
    {
      "epoch": 4.576429980276134,
      "grad_norm": 0.3125,
      "learning_rate": 4.345401547059091e-06,
      "loss": 0.0332,
      "step": 18562
    },
    {
      "epoch": 4.576923076923077,
      "grad_norm": 0.2890625,
      "learning_rate": 4.335369626375985e-06,
      "loss": 0.0356,
      "step": 18564
    },
    {
      "epoch": 4.57741617357002,
      "grad_norm": 0.30078125,
      "learning_rate": 4.3253490424856935e-06,
      "loss": 0.0305,
      "step": 18566
    },
    {
      "epoch": 4.577909270216963,
      "grad_norm": 0.341796875,
      "learning_rate": 4.315339796575724e-06,
      "loss": 0.0355,
      "step": 18568
    },
    {
      "epoch": 4.578402366863905,
      "grad_norm": 0.373046875,
      "learning_rate": 4.305341889832215e-06,
      "loss": 0.0362,
      "step": 18570
    },
    {
      "epoch": 4.5788954635108485,
      "grad_norm": 0.322265625,
      "learning_rate": 4.295355323439986e-06,
      "loss": 0.0347,
      "step": 18572
    },
    {
      "epoch": 4.579388560157791,
      "grad_norm": 0.361328125,
      "learning_rate": 4.285380098582514e-06,
      "loss": 0.0368,
      "step": 18574
    },
    {
      "epoch": 4.579881656804734,
      "grad_norm": 0.291015625,
      "learning_rate": 4.275416216441863e-06,
      "loss": 0.0324,
      "step": 18576
    },
    {
      "epoch": 4.580374753451677,
      "grad_norm": 0.279296875,
      "learning_rate": 4.265463678198889e-06,
      "loss": 0.0354,
      "step": 18578
    },
    {
      "epoch": 4.5808678500986195,
      "grad_norm": 0.2734375,
      "learning_rate": 4.255522485032959e-06,
      "loss": 0.0329,
      "step": 18580
    },
    {
      "epoch": 4.581360946745562,
      "grad_norm": 0.314453125,
      "learning_rate": 4.2455926381221975e-06,
      "loss": 0.0355,
      "step": 18582
    },
    {
      "epoch": 4.581854043392505,
      "grad_norm": 0.3046875,
      "learning_rate": 4.23567413864332e-06,
      "loss": 0.0314,
      "step": 18584
    },
    {
      "epoch": 4.582347140039448,
      "grad_norm": 0.2734375,
      "learning_rate": 4.22576698777174e-06,
      "loss": 0.0313,
      "step": 18586
    },
    {
      "epoch": 4.5828402366863905,
      "grad_norm": 0.265625,
      "learning_rate": 4.215871186681508e-06,
      "loss": 0.0324,
      "step": 18588
    },
    {
      "epoch": 4.583333333333333,
      "grad_norm": 0.28125,
      "learning_rate": 4.205986736545331e-06,
      "loss": 0.0293,
      "step": 18590
    },
    {
      "epoch": 4.5838264299802765,
      "grad_norm": 0.37109375,
      "learning_rate": 4.196113638534582e-06,
      "loss": 0.0318,
      "step": 18592
    },
    {
      "epoch": 4.584319526627219,
      "grad_norm": 0.412109375,
      "learning_rate": 4.1862518938192375e-06,
      "loss": 0.0341,
      "step": 18594
    },
    {
      "epoch": 4.5848126232741615,
      "grad_norm": 0.3046875,
      "learning_rate": 4.176401503568028e-06,
      "loss": 0.0366,
      "step": 18596
    },
    {
      "epoch": 4.585305719921105,
      "grad_norm": 0.263671875,
      "learning_rate": 4.166562468948243e-06,
      "loss": 0.0304,
      "step": 18598
    },
    {
      "epoch": 4.585798816568047,
      "grad_norm": 0.306640625,
      "learning_rate": 4.156734791125861e-06,
      "loss": 0.0339,
      "step": 18600
    },
    {
      "epoch": 4.58629191321499,
      "grad_norm": 0.33203125,
      "learning_rate": 4.146918471265537e-06,
      "loss": 0.0365,
      "step": 18602
    },
    {
      "epoch": 4.5867850098619325,
      "grad_norm": 0.341796875,
      "learning_rate": 4.137113510530544e-06,
      "loss": 0.0362,
      "step": 18604
    },
    {
      "epoch": 4.587278106508876,
      "grad_norm": 0.2890625,
      "learning_rate": 4.127319910082839e-06,
      "loss": 0.0297,
      "step": 18606
    },
    {
      "epoch": 4.587771203155818,
      "grad_norm": 0.28125,
      "learning_rate": 4.117537671082983e-06,
      "loss": 0.0322,
      "step": 18608
    },
    {
      "epoch": 4.588264299802761,
      "grad_norm": 0.310546875,
      "learning_rate": 4.107766794690282e-06,
      "loss": 0.0332,
      "step": 18610
    },
    {
      "epoch": 4.588757396449704,
      "grad_norm": 0.33203125,
      "learning_rate": 4.0980072820625995e-06,
      "loss": 0.0303,
      "step": 18612
    },
    {
      "epoch": 4.589250493096647,
      "grad_norm": 0.30859375,
      "learning_rate": 4.088259134356498e-06,
      "loss": 0.0359,
      "step": 18614
    },
    {
      "epoch": 4.589743589743589,
      "grad_norm": 0.291015625,
      "learning_rate": 4.078522352727199e-06,
      "loss": 0.0348,
      "step": 18616
    },
    {
      "epoch": 4.590236686390533,
      "grad_norm": 0.28125,
      "learning_rate": 4.068796938328556e-06,
      "loss": 0.0347,
      "step": 18618
    },
    {
      "epoch": 4.590729783037475,
      "grad_norm": 0.302734375,
      "learning_rate": 4.059082892313082e-06,
      "loss": 0.0312,
      "step": 18620
    },
    {
      "epoch": 4.591222879684418,
      "grad_norm": 0.26171875,
      "learning_rate": 4.0493802158319435e-06,
      "loss": 0.0289,
      "step": 18622
    },
    {
      "epoch": 4.591715976331361,
      "grad_norm": 0.291015625,
      "learning_rate": 4.039688910035e-06,
      "loss": 0.0316,
      "step": 18624
    },
    {
      "epoch": 4.592209072978304,
      "grad_norm": 0.3515625,
      "learning_rate": 4.030008976070665e-06,
      "loss": 0.0347,
      "step": 18626
    },
    {
      "epoch": 4.592702169625246,
      "grad_norm": 0.294921875,
      "learning_rate": 4.020340415086099e-06,
      "loss": 0.0324,
      "step": 18628
    },
    {
      "epoch": 4.59319526627219,
      "grad_norm": 0.29296875,
      "learning_rate": 4.0106832282270854e-06,
      "loss": 0.0331,
      "step": 18630
    },
    {
      "epoch": 4.593688362919132,
      "grad_norm": 0.34375,
      "learning_rate": 4.0010374166380315e-06,
      "loss": 0.035,
      "step": 18632
    },
    {
      "epoch": 4.594181459566075,
      "grad_norm": 0.291015625,
      "learning_rate": 3.991402981462045e-06,
      "loss": 0.0328,
      "step": 18634
    },
    {
      "epoch": 4.594674556213018,
      "grad_norm": 0.32421875,
      "learning_rate": 3.981779923840845e-06,
      "loss": 0.0311,
      "step": 18636
    },
    {
      "epoch": 4.595167652859961,
      "grad_norm": 0.29296875,
      "learning_rate": 3.97216824491482e-06,
      "loss": 0.035,
      "step": 18638
    },
    {
      "epoch": 4.595660749506903,
      "grad_norm": 0.345703125,
      "learning_rate": 3.9625679458230145e-06,
      "loss": 0.0345,
      "step": 18640
    },
    {
      "epoch": 4.596153846153846,
      "grad_norm": 0.251953125,
      "learning_rate": 3.952979027703107e-06,
      "loss": 0.0322,
      "step": 18642
    },
    {
      "epoch": 4.596646942800789,
      "grad_norm": 0.296875,
      "learning_rate": 3.9434014916914455e-06,
      "loss": 0.0342,
      "step": 18644
    },
    {
      "epoch": 4.597140039447732,
      "grad_norm": 0.271484375,
      "learning_rate": 3.933835338923009e-06,
      "loss": 0.0333,
      "step": 18646
    },
    {
      "epoch": 4.597633136094674,
      "grad_norm": 0.33203125,
      "learning_rate": 3.9242805705314576e-06,
      "loss": 0.0327,
      "step": 18648
    },
    {
      "epoch": 4.598126232741618,
      "grad_norm": 0.2890625,
      "learning_rate": 3.914737187649076e-06,
      "loss": 0.0296,
      "step": 18650
    },
    {
      "epoch": 4.59861932938856,
      "grad_norm": 0.296875,
      "learning_rate": 3.905205191406802e-06,
      "loss": 0.031,
      "step": 18652
    },
    {
      "epoch": 4.599112426035503,
      "grad_norm": 0.2890625,
      "learning_rate": 3.895684582934256e-06,
      "loss": 0.0316,
      "step": 18654
    },
    {
      "epoch": 4.599605522682445,
      "grad_norm": 0.35546875,
      "learning_rate": 3.886175363359646e-06,
      "loss": 0.029,
      "step": 18656
    },
    {
      "epoch": 4.600098619329389,
      "grad_norm": 0.326171875,
      "learning_rate": 3.876677533809892e-06,
      "loss": 0.0328,
      "step": 18658
    },
    {
      "epoch": 4.600591715976331,
      "grad_norm": 0.326171875,
      "learning_rate": 3.867191095410539e-06,
      "loss": 0.0329,
      "step": 18660
    },
    {
      "epoch": 4.601084812623274,
      "grad_norm": 0.28125,
      "learning_rate": 3.857716049285765e-06,
      "loss": 0.0324,
      "step": 18662
    },
    {
      "epoch": 4.601577909270217,
      "grad_norm": 0.3984375,
      "learning_rate": 3.848252396558438e-06,
      "loss": 0.0313,
      "step": 18664
    },
    {
      "epoch": 4.60207100591716,
      "grad_norm": 0.30078125,
      "learning_rate": 3.8388001383500405e-06,
      "loss": 0.0343,
      "step": 18666
    },
    {
      "epoch": 4.602564102564102,
      "grad_norm": 0.28515625,
      "learning_rate": 3.8293592757807415e-06,
      "loss": 0.0328,
      "step": 18668
    },
    {
      "epoch": 4.603057199211046,
      "grad_norm": 0.310546875,
      "learning_rate": 3.819929809969292e-06,
      "loss": 0.036,
      "step": 18670
    },
    {
      "epoch": 4.603550295857988,
      "grad_norm": 0.27734375,
      "learning_rate": 3.810511742033174e-06,
      "loss": 0.0323,
      "step": 18672
    },
    {
      "epoch": 4.604043392504931,
      "grad_norm": 0.291015625,
      "learning_rate": 3.8011050730884624e-06,
      "loss": 0.0315,
      "step": 18674
    },
    {
      "epoch": 4.604536489151874,
      "grad_norm": 0.283203125,
      "learning_rate": 3.79170980424991e-06,
      "loss": 0.0319,
      "step": 18676
    },
    {
      "epoch": 4.605029585798817,
      "grad_norm": 0.271484375,
      "learning_rate": 3.782325936630893e-06,
      "loss": 0.0335,
      "step": 18678
    },
    {
      "epoch": 4.605522682445759,
      "grad_norm": 0.333984375,
      "learning_rate": 3.7729534713434765e-06,
      "loss": 0.033,
      "step": 18680
    },
    {
      "epoch": 4.6060157790927025,
      "grad_norm": 0.2734375,
      "learning_rate": 3.7635924094983397e-06,
      "loss": 0.0342,
      "step": 18682
    },
    {
      "epoch": 4.606508875739645,
      "grad_norm": 0.3203125,
      "learning_rate": 3.7542427522047953e-06,
      "loss": 0.0318,
      "step": 18684
    },
    {
      "epoch": 4.607001972386588,
      "grad_norm": 0.294921875,
      "learning_rate": 3.7449045005708783e-06,
      "loss": 0.0307,
      "step": 18686
    },
    {
      "epoch": 4.607495069033531,
      "grad_norm": 0.3359375,
      "learning_rate": 3.7355776557031706e-06,
      "loss": 0.0323,
      "step": 18688
    },
    {
      "epoch": 4.6079881656804735,
      "grad_norm": 0.29296875,
      "learning_rate": 3.7262622187070105e-06,
      "loss": 0.0326,
      "step": 18690
    },
    {
      "epoch": 4.608481262327416,
      "grad_norm": 0.3203125,
      "learning_rate": 3.7169581906862817e-06,
      "loss": 0.0339,
      "step": 18692
    },
    {
      "epoch": 4.608974358974359,
      "grad_norm": 0.2890625,
      "learning_rate": 3.7076655727435905e-06,
      "loss": 0.0306,
      "step": 18694
    },
    {
      "epoch": 4.609467455621302,
      "grad_norm": 0.3046875,
      "learning_rate": 3.6983843659801674e-06,
      "loss": 0.0338,
      "step": 18696
    },
    {
      "epoch": 4.6099605522682445,
      "grad_norm": 0.32421875,
      "learning_rate": 3.689114571495844e-06,
      "loss": 0.0344,
      "step": 18698
    },
    {
      "epoch": 4.610453648915187,
      "grad_norm": 0.3125,
      "learning_rate": 3.679856190389208e-06,
      "loss": 0.0359,
      "step": 18700
    },
    {
      "epoch": 4.6109467455621305,
      "grad_norm": 0.2734375,
      "learning_rate": 3.670609223757371e-06,
      "loss": 0.0288,
      "step": 18702
    },
    {
      "epoch": 4.611439842209073,
      "grad_norm": 0.306640625,
      "learning_rate": 3.6613736726962e-06,
      "loss": 0.0314,
      "step": 18704
    },
    {
      "epoch": 4.6119329388560155,
      "grad_norm": 0.28515625,
      "learning_rate": 3.6521495383001204e-06,
      "loss": 0.0325,
      "step": 18706
    },
    {
      "epoch": 4.612426035502959,
      "grad_norm": 0.279296875,
      "learning_rate": 3.6429368216622573e-06,
      "loss": 0.0345,
      "step": 18708
    },
    {
      "epoch": 4.6129191321499015,
      "grad_norm": 0.365234375,
      "learning_rate": 3.6337355238743818e-06,
      "loss": 0.0324,
      "step": 18710
    },
    {
      "epoch": 4.613412228796844,
      "grad_norm": 0.306640625,
      "learning_rate": 3.6245456460268666e-06,
      "loss": 0.0327,
      "step": 18712
    },
    {
      "epoch": 4.6139053254437865,
      "grad_norm": 0.287109375,
      "learning_rate": 3.615367189208807e-06,
      "loss": 0.0312,
      "step": 18714
    },
    {
      "epoch": 4.61439842209073,
      "grad_norm": 0.412109375,
      "learning_rate": 3.606200154507855e-06,
      "loss": 0.0343,
      "step": 18716
    },
    {
      "epoch": 4.6148915187376724,
      "grad_norm": 0.25390625,
      "learning_rate": 3.5970445430103973e-06,
      "loss": 0.0319,
      "step": 18718
    },
    {
      "epoch": 4.615384615384615,
      "grad_norm": 0.30859375,
      "learning_rate": 3.587900355801399e-06,
      "loss": 0.0331,
      "step": 18720
    },
    {
      "epoch": 4.615877712031558,
      "grad_norm": 0.267578125,
      "learning_rate": 3.578767593964505e-06,
      "loss": 0.031,
      "step": 18722
    },
    {
      "epoch": 4.616370808678501,
      "grad_norm": 0.33203125,
      "learning_rate": 3.5696462585820044e-06,
      "loss": 0.0339,
      "step": 18724
    },
    {
      "epoch": 4.616863905325443,
      "grad_norm": 0.28515625,
      "learning_rate": 3.5605363507348112e-06,
      "loss": 0.0306,
      "step": 18726
    },
    {
      "epoch": 4.617357001972387,
      "grad_norm": 0.345703125,
      "learning_rate": 3.5514378715025165e-06,
      "loss": 0.0306,
      "step": 18728
    },
    {
      "epoch": 4.617850098619329,
      "grad_norm": 0.3125,
      "learning_rate": 3.542350821963314e-06,
      "loss": 0.0311,
      "step": 18730
    },
    {
      "epoch": 4.618343195266272,
      "grad_norm": 0.296875,
      "learning_rate": 3.5332752031941197e-06,
      "loss": 0.0346,
      "step": 18732
    },
    {
      "epoch": 4.618836291913215,
      "grad_norm": 0.267578125,
      "learning_rate": 3.5242110162703956e-06,
      "loss": 0.0319,
      "step": 18734
    },
    {
      "epoch": 4.619329388560158,
      "grad_norm": 0.294921875,
      "learning_rate": 3.515158262266327e-06,
      "loss": 0.0287,
      "step": 18736
    },
    {
      "epoch": 4.6198224852071,
      "grad_norm": 0.359375,
      "learning_rate": 3.5061169422547113e-06,
      "loss": 0.0306,
      "step": 18738
    },
    {
      "epoch": 4.620315581854044,
      "grad_norm": 0.33984375,
      "learning_rate": 3.4970870573069693e-06,
      "loss": 0.0314,
      "step": 18740
    },
    {
      "epoch": 4.620808678500986,
      "grad_norm": 0.314453125,
      "learning_rate": 3.4880686084932444e-06,
      "loss": 0.0373,
      "step": 18742
    },
    {
      "epoch": 4.621301775147929,
      "grad_norm": 0.359375,
      "learning_rate": 3.479061596882205e-06,
      "loss": 0.0309,
      "step": 18744
    },
    {
      "epoch": 4.621794871794872,
      "grad_norm": 0.298828125,
      "learning_rate": 3.4700660235413075e-06,
      "loss": 0.0368,
      "step": 18746
    },
    {
      "epoch": 4.622287968441815,
      "grad_norm": 0.294921875,
      "learning_rate": 3.4610818895365217e-06,
      "loss": 0.0327,
      "step": 18748
    },
    {
      "epoch": 4.622781065088757,
      "grad_norm": 0.314453125,
      "learning_rate": 3.45210919593254e-06,
      "loss": 0.0331,
      "step": 18750
    },
    {
      "epoch": 4.6232741617357,
      "grad_norm": 0.3046875,
      "learning_rate": 3.4431479437926683e-06,
      "loss": 0.0273,
      "step": 18752
    },
    {
      "epoch": 4.623767258382643,
      "grad_norm": 0.291015625,
      "learning_rate": 3.434198134178879e-06,
      "loss": 0.0316,
      "step": 18754
    },
    {
      "epoch": 4.624260355029586,
      "grad_norm": 0.294921875,
      "learning_rate": 3.4252597681517674e-06,
      "loss": 0.0325,
      "step": 18756
    },
    {
      "epoch": 4.624753451676528,
      "grad_norm": 0.306640625,
      "learning_rate": 3.4163328467705536e-06,
      "loss": 0.0309,
      "step": 18758
    },
    {
      "epoch": 4.625246548323472,
      "grad_norm": 0.2890625,
      "learning_rate": 3.40741737109318e-06,
      "loss": 0.0282,
      "step": 18760
    },
    {
      "epoch": 4.625739644970414,
      "grad_norm": 0.29296875,
      "learning_rate": 3.398513342176135e-06,
      "loss": 0.0331,
      "step": 18762
    },
    {
      "epoch": 4.626232741617357,
      "grad_norm": 0.357421875,
      "learning_rate": 3.389620761074619e-06,
      "loss": 0.0309,
      "step": 18764
    },
    {
      "epoch": 4.6267258382643,
      "grad_norm": 0.2734375,
      "learning_rate": 3.3807396288424444e-06,
      "loss": 0.034,
      "step": 18766
    },
    {
      "epoch": 4.627218934911243,
      "grad_norm": 0.333984375,
      "learning_rate": 3.371869946532069e-06,
      "loss": 0.0326,
      "step": 18768
    },
    {
      "epoch": 4.627712031558185,
      "grad_norm": 0.294921875,
      "learning_rate": 3.3630117151946194e-06,
      "loss": 0.0321,
      "step": 18770
    },
    {
      "epoch": 4.628205128205128,
      "grad_norm": 0.34375,
      "learning_rate": 3.354164935879822e-06,
      "loss": 0.0328,
      "step": 18772
    },
    {
      "epoch": 4.628698224852071,
      "grad_norm": 0.302734375,
      "learning_rate": 3.3453296096361053e-06,
      "loss": 0.0361,
      "step": 18774
    },
    {
      "epoch": 4.629191321499014,
      "grad_norm": 0.302734375,
      "learning_rate": 3.3365057375104535e-06,
      "loss": 0.0354,
      "step": 18776
    },
    {
      "epoch": 4.629684418145956,
      "grad_norm": 0.2890625,
      "learning_rate": 3.3276933205485862e-06,
      "loss": 0.0333,
      "step": 18778
    },
    {
      "epoch": 4.6301775147929,
      "grad_norm": 0.279296875,
      "learning_rate": 3.3188923597948007e-06,
      "loss": 0.0317,
      "step": 18780
    },
    {
      "epoch": 4.630670611439842,
      "grad_norm": 0.2578125,
      "learning_rate": 3.3101028562920744e-06,
      "loss": 0.0301,
      "step": 18782
    },
    {
      "epoch": 4.631163708086785,
      "grad_norm": 0.33984375,
      "learning_rate": 3.301324811081996e-06,
      "loss": 0.0318,
      "step": 18784
    },
    {
      "epoch": 4.631656804733728,
      "grad_norm": 0.29296875,
      "learning_rate": 3.2925582252048338e-06,
      "loss": 0.033,
      "step": 18786
    },
    {
      "epoch": 4.632149901380671,
      "grad_norm": 0.275390625,
      "learning_rate": 3.2838030996994782e-06,
      "loss": 0.0325,
      "step": 18788
    },
    {
      "epoch": 4.632642998027613,
      "grad_norm": 0.330078125,
      "learning_rate": 3.2750594356034336e-06,
      "loss": 0.0334,
      "step": 18790
    },
    {
      "epoch": 4.633136094674557,
      "grad_norm": 0.28515625,
      "learning_rate": 3.2663272339528926e-06,
      "loss": 0.0341,
      "step": 18792
    },
    {
      "epoch": 4.633629191321499,
      "grad_norm": 0.28515625,
      "learning_rate": 3.2576064957826613e-06,
      "loss": 0.029,
      "step": 18794
    },
    {
      "epoch": 4.634122287968442,
      "grad_norm": 0.29296875,
      "learning_rate": 3.248897222126213e-06,
      "loss": 0.0326,
      "step": 18796
    },
    {
      "epoch": 4.634615384615385,
      "grad_norm": 0.314453125,
      "learning_rate": 3.240199414015621e-06,
      "loss": 0.0341,
      "step": 18798
    },
    {
      "epoch": 4.6351084812623276,
      "grad_norm": 0.337890625,
      "learning_rate": 3.23151307248164e-06,
      "loss": 0.0302,
      "step": 18800
    },
    {
      "epoch": 4.63560157790927,
      "grad_norm": 0.302734375,
      "learning_rate": 3.222838198553646e-06,
      "loss": 0.0331,
      "step": 18802
    },
    {
      "epoch": 4.6360946745562135,
      "grad_norm": 0.314453125,
      "learning_rate": 3.214174793259661e-06,
      "loss": 0.0344,
      "step": 18804
    },
    {
      "epoch": 4.636587771203156,
      "grad_norm": 0.279296875,
      "learning_rate": 3.2055228576263528e-06,
      "loss": 0.03,
      "step": 18806
    },
    {
      "epoch": 4.6370808678500985,
      "grad_norm": 0.26953125,
      "learning_rate": 3.1968823926790014e-06,
      "loss": 0.0345,
      "step": 18808
    },
    {
      "epoch": 4.637573964497041,
      "grad_norm": 0.30859375,
      "learning_rate": 3.1882533994415766e-06,
      "loss": 0.035,
      "step": 18810
    },
    {
      "epoch": 4.6380670611439845,
      "grad_norm": 0.40234375,
      "learning_rate": 3.1796358789366485e-06,
      "loss": 0.0336,
      "step": 18812
    },
    {
      "epoch": 4.638560157790927,
      "grad_norm": 0.306640625,
      "learning_rate": 3.171029832185435e-06,
      "loss": 0.0325,
      "step": 18814
    },
    {
      "epoch": 4.6390532544378695,
      "grad_norm": 0.408203125,
      "learning_rate": 3.1624352602078187e-06,
      "loss": 0.0351,
      "step": 18816
    },
    {
      "epoch": 4.639546351084813,
      "grad_norm": 0.373046875,
      "learning_rate": 3.1538521640223084e-06,
      "loss": 0.0326,
      "step": 18818
    },
    {
      "epoch": 4.6400394477317555,
      "grad_norm": 0.279296875,
      "learning_rate": 3.145280544646012e-06,
      "loss": 0.0297,
      "step": 18820
    },
    {
      "epoch": 4.640532544378698,
      "grad_norm": 0.3203125,
      "learning_rate": 3.1367204030947618e-06,
      "loss": 0.0332,
      "step": 18822
    },
    {
      "epoch": 4.641025641025641,
      "grad_norm": 0.2890625,
      "learning_rate": 3.1281717403829457e-06,
      "loss": 0.0296,
      "step": 18824
    },
    {
      "epoch": 4.641518737672584,
      "grad_norm": 0.30078125,
      "learning_rate": 3.119634557523643e-06,
      "loss": 0.0348,
      "step": 18826
    },
    {
      "epoch": 4.6420118343195265,
      "grad_norm": 0.275390625,
      "learning_rate": 3.111108855528555e-06,
      "loss": 0.0344,
      "step": 18828
    },
    {
      "epoch": 4.642504930966469,
      "grad_norm": 0.27734375,
      "learning_rate": 3.102594635408029e-06,
      "loss": 0.0318,
      "step": 18830
    },
    {
      "epoch": 4.642998027613412,
      "grad_norm": 0.2734375,
      "learning_rate": 3.0940918981710585e-06,
      "loss": 0.0338,
      "step": 18832
    },
    {
      "epoch": 4.643491124260355,
      "grad_norm": 0.29296875,
      "learning_rate": 3.085600644825226e-06,
      "loss": 0.0309,
      "step": 18834
    },
    {
      "epoch": 4.6439842209072975,
      "grad_norm": 0.291015625,
      "learning_rate": 3.077120876376838e-06,
      "loss": 0.0324,
      "step": 18836
    },
    {
      "epoch": 4.644477317554241,
      "grad_norm": 0.33203125,
      "learning_rate": 3.0686525938307566e-06,
      "loss": 0.0316,
      "step": 18838
    },
    {
      "epoch": 4.644970414201183,
      "grad_norm": 0.27734375,
      "learning_rate": 3.060195798190546e-06,
      "loss": 0.0321,
      "step": 18840
    },
    {
      "epoch": 4.645463510848126,
      "grad_norm": 0.30859375,
      "learning_rate": 3.0517504904583714e-06,
      "loss": 0.0291,
      "step": 18842
    },
    {
      "epoch": 4.645956607495069,
      "grad_norm": 0.314453125,
      "learning_rate": 3.0433166716350657e-06,
      "loss": 0.0299,
      "step": 18844
    },
    {
      "epoch": 4.646449704142012,
      "grad_norm": 0.25390625,
      "learning_rate": 3.0348943427200737e-06,
      "loss": 0.0309,
      "step": 18846
    },
    {
      "epoch": 4.646942800788954,
      "grad_norm": 0.33984375,
      "learning_rate": 3.0264835047114637e-06,
      "loss": 0.0351,
      "step": 18848
    },
    {
      "epoch": 4.647435897435898,
      "grad_norm": 0.322265625,
      "learning_rate": 3.018084158606005e-06,
      "loss": 0.0307,
      "step": 18850
    },
    {
      "epoch": 4.64792899408284,
      "grad_norm": 0.2890625,
      "learning_rate": 3.0096963053990345e-06,
      "loss": 0.0326,
      "step": 18852
    },
    {
      "epoch": 4.648422090729783,
      "grad_norm": 0.322265625,
      "learning_rate": 3.001319946084602e-06,
      "loss": 0.0324,
      "step": 18854
    },
    {
      "epoch": 4.648915187376726,
      "grad_norm": 0.29296875,
      "learning_rate": 2.992955081655313e-06,
      "loss": 0.0284,
      "step": 18856
    },
    {
      "epoch": 4.649408284023669,
      "grad_norm": 0.302734375,
      "learning_rate": 2.984601713102464e-06,
      "loss": 0.0343,
      "step": 18858
    },
    {
      "epoch": 4.649901380670611,
      "grad_norm": 0.35546875,
      "learning_rate": 2.9762598414159848e-06,
      "loss": 0.0351,
      "step": 18860
    },
    {
      "epoch": 4.650394477317555,
      "grad_norm": 0.2734375,
      "learning_rate": 2.9679294675844073e-06,
      "loss": 0.0312,
      "step": 18862
    },
    {
      "epoch": 4.650887573964497,
      "grad_norm": 0.28515625,
      "learning_rate": 2.9596105925949745e-06,
      "loss": 0.0307,
      "step": 18864
    },
    {
      "epoch": 4.65138067061144,
      "grad_norm": 0.265625,
      "learning_rate": 2.9513032174334654e-06,
      "loss": 0.0346,
      "step": 18866
    },
    {
      "epoch": 4.651873767258382,
      "grad_norm": 0.287109375,
      "learning_rate": 2.9430073430843917e-06,
      "loss": 0.0324,
      "step": 18868
    },
    {
      "epoch": 4.652366863905326,
      "grad_norm": 0.3046875,
      "learning_rate": 2.9347229705308344e-06,
      "loss": 0.0343,
      "step": 18870
    },
    {
      "epoch": 4.652859960552268,
      "grad_norm": 0.30859375,
      "learning_rate": 2.926450100754552e-06,
      "loss": 0.0375,
      "step": 18872
    },
    {
      "epoch": 4.653353057199211,
      "grad_norm": 0.298828125,
      "learning_rate": 2.918188734735927e-06,
      "loss": 0.0327,
      "step": 18874
    },
    {
      "epoch": 4.653846153846154,
      "grad_norm": 0.34765625,
      "learning_rate": 2.9099388734539543e-06,
      "loss": 0.0339,
      "step": 18876
    },
    {
      "epoch": 4.654339250493097,
      "grad_norm": 0.28125,
      "learning_rate": 2.90170051788633e-06,
      "loss": 0.0314,
      "step": 18878
    },
    {
      "epoch": 4.654832347140039,
      "grad_norm": 0.298828125,
      "learning_rate": 2.8934736690092945e-06,
      "loss": 0.0348,
      "step": 18880
    },
    {
      "epoch": 4.655325443786982,
      "grad_norm": 0.287109375,
      "learning_rate": 2.885258327797835e-06,
      "loss": 0.0329,
      "step": 18882
    },
    {
      "epoch": 4.655818540433925,
      "grad_norm": 0.32421875,
      "learning_rate": 2.8770544952254617e-06,
      "loss": 0.0341,
      "step": 18884
    },
    {
      "epoch": 4.656311637080868,
      "grad_norm": 0.28515625,
      "learning_rate": 2.8688621722643973e-06,
      "loss": 0.0317,
      "step": 18886
    },
    {
      "epoch": 4.65680473372781,
      "grad_norm": 0.30859375,
      "learning_rate": 2.8606813598854864e-06,
      "loss": 0.0351,
      "step": 18888
    },
    {
      "epoch": 4.657297830374754,
      "grad_norm": 0.267578125,
      "learning_rate": 2.8525120590581768e-06,
      "loss": 0.0326,
      "step": 18890
    },
    {
      "epoch": 4.657790927021696,
      "grad_norm": 0.310546875,
      "learning_rate": 2.8443542707506043e-06,
      "loss": 0.0313,
      "step": 18892
    },
    {
      "epoch": 4.658284023668639,
      "grad_norm": 0.337890625,
      "learning_rate": 2.836207995929485e-06,
      "loss": 0.032,
      "step": 18894
    },
    {
      "epoch": 4.658777120315582,
      "grad_norm": 0.28515625,
      "learning_rate": 2.8280732355602247e-06,
      "loss": 0.0298,
      "step": 18896
    },
    {
      "epoch": 4.659270216962525,
      "grad_norm": 0.287109375,
      "learning_rate": 2.8199499906068184e-06,
      "loss": 0.0325,
      "step": 18898
    },
    {
      "epoch": 4.659763313609467,
      "grad_norm": 0.2890625,
      "learning_rate": 2.8118382620319183e-06,
      "loss": 0.0315,
      "step": 18900
    },
    {
      "epoch": 4.660256410256411,
      "grad_norm": 0.296875,
      "learning_rate": 2.803738050796811e-06,
      "loss": 0.0293,
      "step": 18902
    },
    {
      "epoch": 4.660749506903353,
      "grad_norm": 0.28125,
      "learning_rate": 2.7956493578614297e-06,
      "loss": 0.0319,
      "step": 18904
    },
    {
      "epoch": 4.661242603550296,
      "grad_norm": 0.287109375,
      "learning_rate": 2.7875721841843285e-06,
      "loss": 0.0327,
      "step": 18906
    },
    {
      "epoch": 4.661735700197239,
      "grad_norm": 0.2734375,
      "learning_rate": 2.779506530722653e-06,
      "loss": 0.0299,
      "step": 18908
    },
    {
      "epoch": 4.662228796844182,
      "grad_norm": 0.2890625,
      "learning_rate": 2.7714523984322947e-06,
      "loss": 0.032,
      "step": 18910
    },
    {
      "epoch": 4.662721893491124,
      "grad_norm": 0.28125,
      "learning_rate": 2.763409788267668e-06,
      "loss": 0.0321,
      "step": 18912
    },
    {
      "epoch": 4.6632149901380675,
      "grad_norm": 0.287109375,
      "learning_rate": 2.7553787011818877e-06,
      "loss": 0.0333,
      "step": 18914
    },
    {
      "epoch": 4.66370808678501,
      "grad_norm": 0.37890625,
      "learning_rate": 2.7473591381266708e-06,
      "loss": 0.0356,
      "step": 18916
    },
    {
      "epoch": 4.664201183431953,
      "grad_norm": 0.349609375,
      "learning_rate": 2.73935110005239e-06,
      "loss": 0.0349,
      "step": 18918
    },
    {
      "epoch": 4.664694280078895,
      "grad_norm": 0.341796875,
      "learning_rate": 2.7313545879080417e-06,
      "loss": 0.0335,
      "step": 18920
    },
    {
      "epoch": 4.6651873767258385,
      "grad_norm": 0.337890625,
      "learning_rate": 2.723369602641235e-06,
      "loss": 0.035,
      "step": 18922
    },
    {
      "epoch": 4.665680473372781,
      "grad_norm": 0.287109375,
      "learning_rate": 2.7153961451982677e-06,
      "loss": 0.0354,
      "step": 18924
    },
    {
      "epoch": 4.6661735700197235,
      "grad_norm": 0.3984375,
      "learning_rate": 2.7074342165240298e-06,
      "loss": 0.04,
      "step": 18926
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 0.296875,
      "learning_rate": 2.699483817562032e-06,
      "loss": 0.0306,
      "step": 18928
    },
    {
      "epoch": 4.6671597633136095,
      "grad_norm": 0.265625,
      "learning_rate": 2.6915449492544765e-06,
      "loss": 0.0342,
      "step": 18930
    },
    {
      "epoch": 4.667652859960552,
      "grad_norm": 0.3828125,
      "learning_rate": 2.6836176125421332e-06,
      "loss": 0.0306,
      "step": 18932
    },
    {
      "epoch": 4.668145956607495,
      "grad_norm": 0.34375,
      "learning_rate": 2.67570180836445e-06,
      "loss": 0.0357,
      "step": 18934
    },
    {
      "epoch": 4.668639053254438,
      "grad_norm": 0.275390625,
      "learning_rate": 2.667797537659489e-06,
      "loss": 0.0299,
      "step": 18936
    },
    {
      "epoch": 4.6691321499013805,
      "grad_norm": 0.259765625,
      "learning_rate": 2.659904801363966e-06,
      "loss": 0.0312,
      "step": 18938
    },
    {
      "epoch": 4.669625246548323,
      "grad_norm": 0.263671875,
      "learning_rate": 2.6520236004131893e-06,
      "loss": 0.0313,
      "step": 18940
    },
    {
      "epoch": 4.670118343195266,
      "grad_norm": 0.31640625,
      "learning_rate": 2.644153935741145e-06,
      "loss": 0.0322,
      "step": 18942
    },
    {
      "epoch": 4.670611439842209,
      "grad_norm": 0.31640625,
      "learning_rate": 2.63629580828042e-06,
      "loss": 0.0294,
      "step": 18944
    },
    {
      "epoch": 4.6711045364891515,
      "grad_norm": 0.349609375,
      "learning_rate": 2.6284492189622478e-06,
      "loss": 0.0352,
      "step": 18946
    },
    {
      "epoch": 4.671597633136095,
      "grad_norm": 0.28515625,
      "learning_rate": 2.6206141687165065e-06,
      "loss": 0.0354,
      "step": 18948
    },
    {
      "epoch": 4.672090729783037,
      "grad_norm": 0.29296875,
      "learning_rate": 2.6127906584716754e-06,
      "loss": 0.031,
      "step": 18950
    },
    {
      "epoch": 4.67258382642998,
      "grad_norm": 0.291015625,
      "learning_rate": 2.604978689154891e-06,
      "loss": 0.0339,
      "step": 18952
    },
    {
      "epoch": 4.673076923076923,
      "grad_norm": 0.3125,
      "learning_rate": 2.597178261691924e-06,
      "loss": 0.0325,
      "step": 18954
    },
    {
      "epoch": 4.673570019723866,
      "grad_norm": 0.287109375,
      "learning_rate": 2.5893893770071566e-06,
      "loss": 0.0301,
      "step": 18956
    },
    {
      "epoch": 4.674063116370808,
      "grad_norm": 0.267578125,
      "learning_rate": 2.5816120360236174e-06,
      "loss": 0.0341,
      "step": 18958
    },
    {
      "epoch": 4.674556213017752,
      "grad_norm": 0.302734375,
      "learning_rate": 2.5738462396629692e-06,
      "loss": 0.0336,
      "step": 18960
    },
    {
      "epoch": 4.675049309664694,
      "grad_norm": 0.3125,
      "learning_rate": 2.566091988845498e-06,
      "loss": 0.0331,
      "step": 18962
    },
    {
      "epoch": 4.675542406311637,
      "grad_norm": 0.265625,
      "learning_rate": 2.558349284490125e-06,
      "loss": 0.0297,
      "step": 18964
    },
    {
      "epoch": 4.67603550295858,
      "grad_norm": 0.2890625,
      "learning_rate": 2.5506181275144035e-06,
      "loss": 0.0327,
      "step": 18966
    },
    {
      "epoch": 4.676528599605523,
      "grad_norm": 0.2734375,
      "learning_rate": 2.5428985188345356e-06,
      "loss": 0.0292,
      "step": 18968
    },
    {
      "epoch": 4.677021696252465,
      "grad_norm": 0.29296875,
      "learning_rate": 2.535190459365311e-06,
      "loss": 0.0325,
      "step": 18970
    },
    {
      "epoch": 4.677514792899409,
      "grad_norm": 0.279296875,
      "learning_rate": 2.527493950020199e-06,
      "loss": 0.031,
      "step": 18972
    },
    {
      "epoch": 4.678007889546351,
      "grad_norm": 0.2890625,
      "learning_rate": 2.5198089917112588e-06,
      "loss": 0.0339,
      "step": 18974
    },
    {
      "epoch": 4.678500986193294,
      "grad_norm": 0.265625,
      "learning_rate": 2.512135585349218e-06,
      "loss": 0.0334,
      "step": 18976
    },
    {
      "epoch": 4.678994082840236,
      "grad_norm": 0.326171875,
      "learning_rate": 2.5044737318434153e-06,
      "loss": 0.0328,
      "step": 18978
    },
    {
      "epoch": 4.67948717948718,
      "grad_norm": 0.32421875,
      "learning_rate": 2.4968234321018246e-06,
      "loss": 0.0306,
      "step": 18980
    },
    {
      "epoch": 4.679980276134122,
      "grad_norm": 0.3046875,
      "learning_rate": 2.489184687031043e-06,
      "loss": 0.0348,
      "step": 18982
    },
    {
      "epoch": 4.680473372781065,
      "grad_norm": 0.28515625,
      "learning_rate": 2.4815574975363023e-06,
      "loss": 0.0372,
      "step": 18984
    },
    {
      "epoch": 4.680966469428008,
      "grad_norm": 0.267578125,
      "learning_rate": 2.4739418645214783e-06,
      "loss": 0.0306,
      "step": 18986
    },
    {
      "epoch": 4.681459566074951,
      "grad_norm": 0.357421875,
      "learning_rate": 2.4663377888890503e-06,
      "loss": 0.0339,
      "step": 18988
    },
    {
      "epoch": 4.681952662721893,
      "grad_norm": 0.294921875,
      "learning_rate": 2.4587452715401638e-06,
      "loss": 0.0307,
      "step": 18990
    },
    {
      "epoch": 4.682445759368837,
      "grad_norm": 0.318359375,
      "learning_rate": 2.451164313374554e-06,
      "loss": 0.033,
      "step": 18992
    },
    {
      "epoch": 4.682938856015779,
      "grad_norm": 0.283203125,
      "learning_rate": 2.4435949152906145e-06,
      "loss": 0.0319,
      "step": 18994
    },
    {
      "epoch": 4.683431952662722,
      "grad_norm": 0.326171875,
      "learning_rate": 2.4360370781853716e-06,
      "loss": 0.0316,
      "step": 18996
    },
    {
      "epoch": 4.683925049309664,
      "grad_norm": 0.287109375,
      "learning_rate": 2.428490802954453e-06,
      "loss": 0.0289,
      "step": 18998
    },
    {
      "epoch": 4.684418145956608,
      "grad_norm": 0.302734375,
      "learning_rate": 2.4209560904921545e-06,
      "loss": 0.0336,
      "step": 19000
    },
    {
      "epoch": 4.68491124260355,
      "grad_norm": 0.318359375,
      "learning_rate": 2.4134329416913517e-06,
      "loss": 0.0296,
      "step": 19002
    },
    {
      "epoch": 4.685404339250493,
      "grad_norm": 0.28515625,
      "learning_rate": 2.405921357443619e-06,
      "loss": 0.0319,
      "step": 19004
    },
    {
      "epoch": 4.685897435897436,
      "grad_norm": 0.29296875,
      "learning_rate": 2.3984213386391006e-06,
      "loss": 0.0317,
      "step": 19006
    },
    {
      "epoch": 4.686390532544379,
      "grad_norm": 0.279296875,
      "learning_rate": 2.390932886166575e-06,
      "loss": 0.0283,
      "step": 19008
    },
    {
      "epoch": 4.686883629191321,
      "grad_norm": 0.263671875,
      "learning_rate": 2.3834560009135087e-06,
      "loss": 0.03,
      "step": 19010
    },
    {
      "epoch": 4.687376725838265,
      "grad_norm": 0.28515625,
      "learning_rate": 2.375990683765894e-06,
      "loss": 0.032,
      "step": 19012
    },
    {
      "epoch": 4.687869822485207,
      "grad_norm": 0.32421875,
      "learning_rate": 2.3685369356084675e-06,
      "loss": 0.0317,
      "step": 19014
    },
    {
      "epoch": 4.68836291913215,
      "grad_norm": 0.314453125,
      "learning_rate": 2.3610947573245e-06,
      "loss": 0.0313,
      "step": 19016
    },
    {
      "epoch": 4.688856015779093,
      "grad_norm": 0.3046875,
      "learning_rate": 2.3536641497959754e-06,
      "loss": 0.0324,
      "step": 19018
    },
    {
      "epoch": 4.689349112426036,
      "grad_norm": 0.2734375,
      "learning_rate": 2.3462451139034226e-06,
      "loss": 0.0292,
      "step": 19020
    },
    {
      "epoch": 4.689842209072978,
      "grad_norm": 0.294921875,
      "learning_rate": 2.338837650526049e-06,
      "loss": 0.0343,
      "step": 19022
    },
    {
      "epoch": 4.6903353057199215,
      "grad_norm": 0.265625,
      "learning_rate": 2.331441760541697e-06,
      "loss": 0.0267,
      "step": 19024
    },
    {
      "epoch": 4.690828402366864,
      "grad_norm": 0.2734375,
      "learning_rate": 2.324057444826788e-06,
      "loss": 0.0289,
      "step": 19026
    },
    {
      "epoch": 4.691321499013807,
      "grad_norm": 0.2578125,
      "learning_rate": 2.3166847042564445e-06,
      "loss": 0.0299,
      "step": 19028
    },
    {
      "epoch": 4.69181459566075,
      "grad_norm": 0.259765625,
      "learning_rate": 2.3093235397043333e-06,
      "loss": 0.03,
      "step": 19030
    },
    {
      "epoch": 4.6923076923076925,
      "grad_norm": 0.314453125,
      "learning_rate": 2.3019739520428464e-06,
      "loss": 0.0311,
      "step": 19032
    },
    {
      "epoch": 4.692800788954635,
      "grad_norm": 0.2890625,
      "learning_rate": 2.2946359421429086e-06,
      "loss": 0.0295,
      "step": 19034
    },
    {
      "epoch": 4.693293885601578,
      "grad_norm": 0.34375,
      "learning_rate": 2.287309510874125e-06,
      "loss": 0.0348,
      "step": 19036
    },
    {
      "epoch": 4.693786982248521,
      "grad_norm": 0.25390625,
      "learning_rate": 2.279994659104734e-06,
      "loss": 0.0305,
      "step": 19038
    },
    {
      "epoch": 4.6942800788954635,
      "grad_norm": 0.275390625,
      "learning_rate": 2.2726913877015534e-06,
      "loss": 0.0313,
      "step": 19040
    },
    {
      "epoch": 4.694773175542406,
      "grad_norm": 0.2734375,
      "learning_rate": 2.2653996975301016e-06,
      "loss": 0.0334,
      "step": 19042
    },
    {
      "epoch": 4.695266272189349,
      "grad_norm": 0.28515625,
      "learning_rate": 2.258119589454455e-06,
      "loss": 0.0326,
      "step": 19044
    },
    {
      "epoch": 4.695759368836292,
      "grad_norm": 0.3359375,
      "learning_rate": 2.250851064337367e-06,
      "loss": 0.0332,
      "step": 19046
    },
    {
      "epoch": 4.6962524654832345,
      "grad_norm": 0.30078125,
      "learning_rate": 2.2435941230401824e-06,
      "loss": 0.0327,
      "step": 19048
    },
    {
      "epoch": 4.696745562130177,
      "grad_norm": 0.306640625,
      "learning_rate": 2.236348766422891e-06,
      "loss": 0.034,
      "step": 19050
    },
    {
      "epoch": 4.69723865877712,
      "grad_norm": 0.322265625,
      "learning_rate": 2.229114995344128e-06,
      "loss": 0.0324,
      "step": 19052
    },
    {
      "epoch": 4.697731755424063,
      "grad_norm": 0.279296875,
      "learning_rate": 2.221892810661108e-06,
      "loss": 0.0312,
      "step": 19054
    },
    {
      "epoch": 4.6982248520710055,
      "grad_norm": 0.30078125,
      "learning_rate": 2.214682213229724e-06,
      "loss": 0.032,
      "step": 19056
    },
    {
      "epoch": 4.698717948717949,
      "grad_norm": 0.2890625,
      "learning_rate": 2.207483203904448e-06,
      "loss": 0.0327,
      "step": 19058
    },
    {
      "epoch": 4.699211045364891,
      "grad_norm": 0.28125,
      "learning_rate": 2.200295783538431e-06,
      "loss": 0.0311,
      "step": 19060
    },
    {
      "epoch": 4.699704142011834,
      "grad_norm": 0.384765625,
      "learning_rate": 2.193119952983391e-06,
      "loss": 0.0325,
      "step": 19062
    },
    {
      "epoch": 4.700197238658777,
      "grad_norm": 0.294921875,
      "learning_rate": 2.1859557130897267e-06,
      "loss": 0.0328,
      "step": 19064
    },
    {
      "epoch": 4.70069033530572,
      "grad_norm": 0.314453125,
      "learning_rate": 2.178803064706436e-06,
      "loss": 0.0366,
      "step": 19066
    },
    {
      "epoch": 4.701183431952662,
      "grad_norm": 0.306640625,
      "learning_rate": 2.1716620086811523e-06,
      "loss": 0.0345,
      "step": 19068
    },
    {
      "epoch": 4.701676528599606,
      "grad_norm": 0.32421875,
      "learning_rate": 2.164532545860121e-06,
      "loss": 0.0316,
      "step": 19070
    },
    {
      "epoch": 4.702169625246548,
      "grad_norm": 0.298828125,
      "learning_rate": 2.1574146770882097e-06,
      "loss": 0.0304,
      "step": 19072
    },
    {
      "epoch": 4.702662721893491,
      "grad_norm": 0.357421875,
      "learning_rate": 2.150308403208967e-06,
      "loss": 0.0325,
      "step": 19074
    },
    {
      "epoch": 4.703155818540434,
      "grad_norm": 0.357421875,
      "learning_rate": 2.1432137250644857e-06,
      "loss": 0.0318,
      "step": 19076
    },
    {
      "epoch": 4.703648915187377,
      "grad_norm": 0.259765625,
      "learning_rate": 2.1361306434955486e-06,
      "loss": 0.032,
      "step": 19078
    },
    {
      "epoch": 4.704142011834319,
      "grad_norm": 0.326171875,
      "learning_rate": 2.1290591593415176e-06,
      "loss": 0.0359,
      "step": 19080
    },
    {
      "epoch": 4.704635108481263,
      "grad_norm": 0.287109375,
      "learning_rate": 2.121999273440434e-06,
      "loss": 0.033,
      "step": 19082
    },
    {
      "epoch": 4.705128205128205,
      "grad_norm": 0.28515625,
      "learning_rate": 2.1149509866289054e-06,
      "loss": 0.0299,
      "step": 19084
    },
    {
      "epoch": 4.705621301775148,
      "grad_norm": 0.27734375,
      "learning_rate": 2.1079142997422084e-06,
      "loss": 0.0278,
      "step": 19086
    },
    {
      "epoch": 4.706114398422091,
      "grad_norm": 0.34375,
      "learning_rate": 2.100889213614243e-06,
      "loss": 0.0298,
      "step": 19088
    },
    {
      "epoch": 4.706607495069034,
      "grad_norm": 0.3046875,
      "learning_rate": 2.0938757290774878e-06,
      "loss": 0.0324,
      "step": 19090
    },
    {
      "epoch": 4.707100591715976,
      "grad_norm": 0.2412109375,
      "learning_rate": 2.086873846963089e-06,
      "loss": 0.0303,
      "step": 19092
    },
    {
      "epoch": 4.707593688362919,
      "grad_norm": 0.271484375,
      "learning_rate": 2.0798835681008264e-06,
      "loss": 0.0277,
      "step": 19094
    },
    {
      "epoch": 4.708086785009862,
      "grad_norm": 0.2451171875,
      "learning_rate": 2.072904893319072e-06,
      "loss": 0.0288,
      "step": 19096
    },
    {
      "epoch": 4.708579881656805,
      "grad_norm": 0.28515625,
      "learning_rate": 2.0659378234448525e-06,
      "loss": 0.0325,
      "step": 19098
    },
    {
      "epoch": 4.709072978303747,
      "grad_norm": 0.326171875,
      "learning_rate": 2.0589823593037847e-06,
      "loss": 0.036,
      "step": 19100
    },
    {
      "epoch": 4.709566074950691,
      "grad_norm": 0.28515625,
      "learning_rate": 2.052038501720155e-06,
      "loss": 0.0319,
      "step": 19102
    },
    {
      "epoch": 4.710059171597633,
      "grad_norm": 0.318359375,
      "learning_rate": 2.045106251516826e-06,
      "loss": 0.0319,
      "step": 19104
    },
    {
      "epoch": 4.710552268244576,
      "grad_norm": 0.310546875,
      "learning_rate": 2.0381856095153086e-06,
      "loss": 0.0319,
      "step": 19106
    },
    {
      "epoch": 4.711045364891518,
      "grad_norm": 0.298828125,
      "learning_rate": 2.0312765765357455e-06,
      "loss": 0.0346,
      "step": 19108
    },
    {
      "epoch": 4.711538461538462,
      "grad_norm": 0.279296875,
      "learning_rate": 2.024379153396905e-06,
      "loss": 0.0307,
      "step": 19110
    },
    {
      "epoch": 4.712031558185404,
      "grad_norm": 0.314453125,
      "learning_rate": 2.0174933409161544e-06,
      "loss": 0.0304,
      "step": 19112
    },
    {
      "epoch": 4.712524654832347,
      "grad_norm": 0.26953125,
      "learning_rate": 2.0106191399095088e-06,
      "loss": 0.0286,
      "step": 19114
    },
    {
      "epoch": 4.71301775147929,
      "grad_norm": 0.28515625,
      "learning_rate": 2.0037565511915936e-06,
      "loss": 0.0321,
      "step": 19116
    },
    {
      "epoch": 4.713510848126233,
      "grad_norm": 0.32421875,
      "learning_rate": 1.9969055755756803e-06,
      "loss": 0.0325,
      "step": 19118
    },
    {
      "epoch": 4.714003944773175,
      "grad_norm": 0.296875,
      "learning_rate": 1.9900662138736204e-06,
      "loss": 0.0345,
      "step": 19120
    },
    {
      "epoch": 4.714497041420119,
      "grad_norm": 0.333984375,
      "learning_rate": 1.9832384668959427e-06,
      "loss": 0.0338,
      "step": 19122
    },
    {
      "epoch": 4.714990138067061,
      "grad_norm": 0.41796875,
      "learning_rate": 1.9764223354517553e-06,
      "loss": 0.0343,
      "step": 19124
    },
    {
      "epoch": 4.715483234714004,
      "grad_norm": 0.28125,
      "learning_rate": 1.9696178203488124e-06,
      "loss": 0.032,
      "step": 19126
    },
    {
      "epoch": 4.715976331360947,
      "grad_norm": 0.279296875,
      "learning_rate": 1.9628249223935023e-06,
      "loss": 0.0323,
      "step": 19128
    },
    {
      "epoch": 4.71646942800789,
      "grad_norm": 0.32421875,
      "learning_rate": 1.9560436423907923e-06,
      "loss": 0.0306,
      "step": 19130
    },
    {
      "epoch": 4.716962524654832,
      "grad_norm": 0.2578125,
      "learning_rate": 1.9492739811443286e-06,
      "loss": 0.026,
      "step": 19132
    },
    {
      "epoch": 4.7174556213017755,
      "grad_norm": 0.29296875,
      "learning_rate": 1.942515939456335e-06,
      "loss": 0.033,
      "step": 19134
    },
    {
      "epoch": 4.717948717948718,
      "grad_norm": 0.283203125,
      "learning_rate": 1.935769518127706e-06,
      "loss": 0.0349,
      "step": 19136
    },
    {
      "epoch": 4.718441814595661,
      "grad_norm": 0.298828125,
      "learning_rate": 1.9290347179578894e-06,
      "loss": 0.0336,
      "step": 19138
    },
    {
      "epoch": 4.718934911242604,
      "grad_norm": 0.318359375,
      "learning_rate": 1.9223115397450254e-06,
      "loss": 0.0329,
      "step": 19140
    },
    {
      "epoch": 4.7194280078895465,
      "grad_norm": 0.28125,
      "learning_rate": 1.915599984285832e-06,
      "loss": 0.0311,
      "step": 19142
    },
    {
      "epoch": 4.719921104536489,
      "grad_norm": 0.2890625,
      "learning_rate": 1.908900052375673e-06,
      "loss": 0.0326,
      "step": 19144
    },
    {
      "epoch": 4.720414201183432,
      "grad_norm": 0.30859375,
      "learning_rate": 1.9022117448085353e-06,
      "loss": 0.0365,
      "step": 19146
    },
    {
      "epoch": 4.720907297830375,
      "grad_norm": 0.32421875,
      "learning_rate": 1.8955350623770184e-06,
      "loss": 0.0325,
      "step": 19148
    },
    {
      "epoch": 4.7214003944773175,
      "grad_norm": 0.3125,
      "learning_rate": 1.888870005872334e-06,
      "loss": 0.0319,
      "step": 19150
    },
    {
      "epoch": 4.72189349112426,
      "grad_norm": 0.27734375,
      "learning_rate": 1.882216576084339e-06,
      "loss": 0.0316,
      "step": 19152
    },
    {
      "epoch": 4.7223865877712035,
      "grad_norm": 0.28125,
      "learning_rate": 1.8755747738015028e-06,
      "loss": 0.0305,
      "step": 19154
    },
    {
      "epoch": 4.722879684418146,
      "grad_norm": 0.291015625,
      "learning_rate": 1.8689445998109067e-06,
      "loss": 0.031,
      "step": 19156
    },
    {
      "epoch": 4.7233727810650885,
      "grad_norm": 0.294921875,
      "learning_rate": 1.8623260548982668e-06,
      "loss": 0.0322,
      "step": 19158
    },
    {
      "epoch": 4.723865877712032,
      "grad_norm": 0.30859375,
      "learning_rate": 1.8557191398479223e-06,
      "loss": 0.0301,
      "step": 19160
    },
    {
      "epoch": 4.7243589743589745,
      "grad_norm": 0.275390625,
      "learning_rate": 1.8491238554428248e-06,
      "loss": 0.0301,
      "step": 19162
    },
    {
      "epoch": 4.724852071005917,
      "grad_norm": 0.310546875,
      "learning_rate": 1.842540202464571e-06,
      "loss": 0.0321,
      "step": 19164
    },
    {
      "epoch": 4.7253451676528595,
      "grad_norm": 0.26171875,
      "learning_rate": 1.835968181693315e-06,
      "loss": 0.0333,
      "step": 19166
    },
    {
      "epoch": 4.725838264299803,
      "grad_norm": 0.28515625,
      "learning_rate": 1.8294077939079336e-06,
      "loss": 0.0344,
      "step": 19168
    },
    {
      "epoch": 4.726331360946745,
      "grad_norm": 0.259765625,
      "learning_rate": 1.822859039885838e-06,
      "loss": 0.0343,
      "step": 19170
    },
    {
      "epoch": 4.726824457593688,
      "grad_norm": 0.337890625,
      "learning_rate": 1.8163219204030856e-06,
      "loss": 0.0324,
      "step": 19172
    },
    {
      "epoch": 4.727317554240631,
      "grad_norm": 0.26953125,
      "learning_rate": 1.809796436234379e-06,
      "loss": 0.0283,
      "step": 19174
    },
    {
      "epoch": 4.727810650887574,
      "grad_norm": 0.27734375,
      "learning_rate": 1.8032825881530213e-06,
      "loss": 0.0315,
      "step": 19176
    },
    {
      "epoch": 4.728303747534516,
      "grad_norm": 0.2890625,
      "learning_rate": 1.7967803769309399e-06,
      "loss": 0.0333,
      "step": 19178
    },
    {
      "epoch": 4.72879684418146,
      "grad_norm": 0.3515625,
      "learning_rate": 1.7902898033386738e-06,
      "loss": 0.0332,
      "step": 19180
    },
    {
      "epoch": 4.729289940828402,
      "grad_norm": 0.28125,
      "learning_rate": 1.783810868145408e-06,
      "loss": 0.0309,
      "step": 19182
    },
    {
      "epoch": 4.729783037475345,
      "grad_norm": 0.34375,
      "learning_rate": 1.7773435721189058e-06,
      "loss": 0.0321,
      "step": 19184
    },
    {
      "epoch": 4.730276134122288,
      "grad_norm": 0.287109375,
      "learning_rate": 1.7708879160256097e-06,
      "loss": 0.0335,
      "step": 19186
    },
    {
      "epoch": 4.730769230769231,
      "grad_norm": 0.3125,
      "learning_rate": 1.7644439006305303e-06,
      "loss": 0.0328,
      "step": 19188
    },
    {
      "epoch": 4.731262327416173,
      "grad_norm": 0.283203125,
      "learning_rate": 1.758011526697334e-06,
      "loss": 0.0309,
      "step": 19190
    },
    {
      "epoch": 4.731755424063117,
      "grad_norm": 0.287109375,
      "learning_rate": 1.7515907949882782e-06,
      "loss": 0.0294,
      "step": 19192
    },
    {
      "epoch": 4.732248520710059,
      "grad_norm": 0.267578125,
      "learning_rate": 1.745181706264254e-06,
      "loss": 0.0312,
      "step": 19194
    },
    {
      "epoch": 4.732741617357002,
      "grad_norm": 0.28125,
      "learning_rate": 1.7387842612847983e-06,
      "loss": 0.0325,
      "step": 19196
    },
    {
      "epoch": 4.733234714003945,
      "grad_norm": 0.259765625,
      "learning_rate": 1.7323984608080267e-06,
      "loss": 0.0312,
      "step": 19198
    },
    {
      "epoch": 4.733727810650888,
      "grad_norm": 0.298828125,
      "learning_rate": 1.726024305590679e-06,
      "loss": 0.0327,
      "step": 19200
    },
    {
      "epoch": 4.73422090729783,
      "grad_norm": 0.296875,
      "learning_rate": 1.71966179638815e-06,
      "loss": 0.0309,
      "step": 19202
    },
    {
      "epoch": 4.734714003944773,
      "grad_norm": 0.31640625,
      "learning_rate": 1.7133109339544263e-06,
      "loss": 0.0315,
      "step": 19204
    },
    {
      "epoch": 4.735207100591716,
      "grad_norm": 0.3046875,
      "learning_rate": 1.7069717190421275e-06,
      "loss": 0.035,
      "step": 19206
    },
    {
      "epoch": 4.735700197238659,
      "grad_norm": 0.283203125,
      "learning_rate": 1.700644152402464e-06,
      "loss": 0.0318,
      "step": 19208
    },
    {
      "epoch": 4.736193293885601,
      "grad_norm": 0.28515625,
      "learning_rate": 1.694328234785314e-06,
      "loss": 0.0312,
      "step": 19210
    },
    {
      "epoch": 4.736686390532545,
      "grad_norm": 0.3359375,
      "learning_rate": 1.6880239669391229e-06,
      "loss": 0.0335,
      "step": 19212
    },
    {
      "epoch": 4.737179487179487,
      "grad_norm": 0.2578125,
      "learning_rate": 1.6817313496110155e-06,
      "loss": 0.0289,
      "step": 19214
    },
    {
      "epoch": 4.73767258382643,
      "grad_norm": 0.267578125,
      "learning_rate": 1.6754503835466727e-06,
      "loss": 0.029,
      "step": 19216
    },
    {
      "epoch": 4.738165680473373,
      "grad_norm": 0.279296875,
      "learning_rate": 1.669181069490422e-06,
      "loss": 0.0321,
      "step": 19218
    },
    {
      "epoch": 4.738658777120316,
      "grad_norm": 0.23046875,
      "learning_rate": 1.6629234081852463e-06,
      "loss": 0.028,
      "step": 19220
    },
    {
      "epoch": 4.739151873767258,
      "grad_norm": 0.28515625,
      "learning_rate": 1.6566774003726637e-06,
      "loss": 0.0319,
      "step": 19222
    },
    {
      "epoch": 4.739644970414201,
      "grad_norm": 0.27734375,
      "learning_rate": 1.6504430467929044e-06,
      "loss": 0.0347,
      "step": 19224
    },
    {
      "epoch": 4.740138067061144,
      "grad_norm": 0.29296875,
      "learning_rate": 1.644220348184755e-06,
      "loss": 0.0324,
      "step": 19226
    },
    {
      "epoch": 4.740631163708087,
      "grad_norm": 0.265625,
      "learning_rate": 1.6380093052856483e-06,
      "loss": 0.0317,
      "step": 19228
    },
    {
      "epoch": 4.741124260355029,
      "grad_norm": 0.37109375,
      "learning_rate": 1.6318099188316061e-06,
      "loss": 0.0334,
      "step": 19230
    },
    {
      "epoch": 4.741617357001973,
      "grad_norm": 0.271484375,
      "learning_rate": 1.6256221895573187e-06,
      "loss": 0.0331,
      "step": 19232
    },
    {
      "epoch": 4.742110453648915,
      "grad_norm": 0.310546875,
      "learning_rate": 1.6194461181960552e-06,
      "loss": 0.0348,
      "step": 19234
    },
    {
      "epoch": 4.742603550295858,
      "grad_norm": 0.314453125,
      "learning_rate": 1.6132817054797079e-06,
      "loss": 0.029,
      "step": 19236
    },
    {
      "epoch": 4.743096646942801,
      "grad_norm": 0.26171875,
      "learning_rate": 1.6071289521388033e-06,
      "loss": 0.0281,
      "step": 19238
    },
    {
      "epoch": 4.743589743589744,
      "grad_norm": 0.267578125,
      "learning_rate": 1.6009878589024696e-06,
      "loss": 0.0317,
      "step": 19240
    },
    {
      "epoch": 4.744082840236686,
      "grad_norm": 0.30078125,
      "learning_rate": 1.5948584264984689e-06,
      "loss": 0.0308,
      "step": 19242
    },
    {
      "epoch": 4.7445759368836296,
      "grad_norm": 0.28515625,
      "learning_rate": 1.5887406556531648e-06,
      "loss": 0.0333,
      "step": 19244
    },
    {
      "epoch": 4.745069033530572,
      "grad_norm": 0.287109375,
      "learning_rate": 1.582634547091555e-06,
      "loss": 0.0287,
      "step": 19246
    },
    {
      "epoch": 4.745562130177515,
      "grad_norm": 0.27734375,
      "learning_rate": 1.5765401015372382e-06,
      "loss": 0.0304,
      "step": 19248
    },
    {
      "epoch": 4.746055226824458,
      "grad_norm": 0.287109375,
      "learning_rate": 1.5704573197124484e-06,
      "loss": 0.0342,
      "step": 19250
    },
    {
      "epoch": 4.7465483234714005,
      "grad_norm": 0.2421875,
      "learning_rate": 1.5643862023380307e-06,
      "loss": 0.0275,
      "step": 19252
    },
    {
      "epoch": 4.747041420118343,
      "grad_norm": 0.28515625,
      "learning_rate": 1.5583267501334208e-06,
      "loss": 0.0297,
      "step": 19254
    },
    {
      "epoch": 4.7475345167652865,
      "grad_norm": 0.34375,
      "learning_rate": 1.5522789638167445e-06,
      "loss": 0.0318,
      "step": 19256
    },
    {
      "epoch": 4.748027613412229,
      "grad_norm": 0.2734375,
      "learning_rate": 1.5462428441046616e-06,
      "loss": 0.034,
      "step": 19258
    },
    {
      "epoch": 4.7485207100591715,
      "grad_norm": 0.298828125,
      "learning_rate": 1.5402183917125002e-06,
      "loss": 0.0338,
      "step": 19260
    },
    {
      "epoch": 4.749013806706114,
      "grad_norm": 0.3046875,
      "learning_rate": 1.534205607354189e-06,
      "loss": 0.0306,
      "step": 19262
    },
    {
      "epoch": 4.7495069033530575,
      "grad_norm": 0.2890625,
      "learning_rate": 1.5282044917422689e-06,
      "loss": 0.0359,
      "step": 19264
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.283203125,
      "learning_rate": 1.5222150455879159e-06,
      "loss": 0.032,
      "step": 19266
    },
    {
      "epoch": 4.7504930966469425,
      "grad_norm": 0.3125,
      "learning_rate": 1.5162372696009175e-06,
      "loss": 0.0314,
      "step": 19268
    },
    {
      "epoch": 4.750986193293886,
      "grad_norm": 0.26171875,
      "learning_rate": 1.5102711644896627e-06,
      "loss": 0.0341,
      "step": 19270
    },
    {
      "epoch": 4.7514792899408285,
      "grad_norm": 0.28125,
      "learning_rate": 1.504316730961175e-06,
      "loss": 0.0315,
      "step": 19272
    },
    {
      "epoch": 4.751972386587771,
      "grad_norm": 0.287109375,
      "learning_rate": 1.4983739697210675e-06,
      "loss": 0.0346,
      "step": 19274
    },
    {
      "epoch": 4.7524654832347135,
      "grad_norm": 0.34765625,
      "learning_rate": 1.4924428814736213e-06,
      "loss": 0.0329,
      "step": 19276
    },
    {
      "epoch": 4.752958579881657,
      "grad_norm": 0.302734375,
      "learning_rate": 1.4865234669216854e-06,
      "loss": 0.0335,
      "step": 19278
    },
    {
      "epoch": 4.7534516765285995,
      "grad_norm": 0.306640625,
      "learning_rate": 1.480615726766743e-06,
      "loss": 0.0309,
      "step": 19280
    },
    {
      "epoch": 4.753944773175542,
      "grad_norm": 0.314453125,
      "learning_rate": 1.474719661708901e-06,
      "loss": 0.0323,
      "step": 19282
    },
    {
      "epoch": 4.754437869822485,
      "grad_norm": 0.267578125,
      "learning_rate": 1.468835272446889e-06,
      "loss": 0.0307,
      "step": 19284
    },
    {
      "epoch": 4.754930966469428,
      "grad_norm": 0.291015625,
      "learning_rate": 1.462962559678005e-06,
      "loss": 0.0314,
      "step": 19286
    },
    {
      "epoch": 4.7554240631163704,
      "grad_norm": 0.279296875,
      "learning_rate": 1.4571015240982256e-06,
      "loss": 0.0335,
      "step": 19288
    },
    {
      "epoch": 4.755917159763314,
      "grad_norm": 0.291015625,
      "learning_rate": 1.4512521664021061e-06,
      "loss": 0.0323,
      "step": 19290
    },
    {
      "epoch": 4.756410256410256,
      "grad_norm": 0.32421875,
      "learning_rate": 1.445414487282837e-06,
      "loss": 0.0324,
      "step": 19292
    },
    {
      "epoch": 4.756903353057199,
      "grad_norm": 0.275390625,
      "learning_rate": 1.4395884874321975e-06,
      "loss": 0.0262,
      "step": 19294
    },
    {
      "epoch": 4.757396449704142,
      "grad_norm": 0.283203125,
      "learning_rate": 1.4337741675406246e-06,
      "loss": 0.0313,
      "step": 19296
    },
    {
      "epoch": 4.757889546351085,
      "grad_norm": 0.3125,
      "learning_rate": 1.4279715282971228e-06,
      "loss": 0.0312,
      "step": 19298
    },
    {
      "epoch": 4.758382642998027,
      "grad_norm": 0.3203125,
      "learning_rate": 1.4221805703893753e-06,
      "loss": 0.0322,
      "step": 19300
    },
    {
      "epoch": 4.758875739644971,
      "grad_norm": 0.24609375,
      "learning_rate": 1.4164012945035998e-06,
      "loss": 0.0292,
      "step": 19302
    },
    {
      "epoch": 4.759368836291913,
      "grad_norm": 0.296875,
      "learning_rate": 1.4106337013246928e-06,
      "loss": 0.032,
      "step": 19304
    },
    {
      "epoch": 4.759861932938856,
      "grad_norm": 0.314453125,
      "learning_rate": 1.404877791536141e-06,
      "loss": 0.0329,
      "step": 19306
    },
    {
      "epoch": 4.760355029585799,
      "grad_norm": 0.283203125,
      "learning_rate": 1.3991335658200655e-06,
      "loss": 0.0297,
      "step": 19308
    },
    {
      "epoch": 4.760848126232742,
      "grad_norm": 0.291015625,
      "learning_rate": 1.3934010248571772e-06,
      "loss": 0.0306,
      "step": 19310
    },
    {
      "epoch": 4.761341222879684,
      "grad_norm": 0.33203125,
      "learning_rate": 1.3876801693268105e-06,
      "loss": 0.0312,
      "step": 19312
    },
    {
      "epoch": 4.761834319526627,
      "grad_norm": 0.26953125,
      "learning_rate": 1.381970999906934e-06,
      "loss": 0.03,
      "step": 19314
    },
    {
      "epoch": 4.76232741617357,
      "grad_norm": 0.302734375,
      "learning_rate": 1.3762735172740959e-06,
      "loss": 0.0311,
      "step": 19316
    },
    {
      "epoch": 4.762820512820513,
      "grad_norm": 0.279296875,
      "learning_rate": 1.370587722103489e-06,
      "loss": 0.0306,
      "step": 19318
    },
    {
      "epoch": 4.763313609467455,
      "grad_norm": 0.279296875,
      "learning_rate": 1.364913615068919e-06,
      "loss": 0.032,
      "step": 19320
    },
    {
      "epoch": 4.763806706114399,
      "grad_norm": 0.27734375,
      "learning_rate": 1.3592511968427923e-06,
      "loss": 0.03,
      "step": 19322
    },
    {
      "epoch": 4.764299802761341,
      "grad_norm": 0.35546875,
      "learning_rate": 1.353600468096139e-06,
      "loss": 0.0334,
      "step": 19324
    },
    {
      "epoch": 4.764792899408284,
      "grad_norm": 0.3515625,
      "learning_rate": 1.3479614294986009e-06,
      "loss": 0.0332,
      "step": 19326
    },
    {
      "epoch": 4.765285996055227,
      "grad_norm": 0.29296875,
      "learning_rate": 1.3423340817184437e-06,
      "loss": 0.0273,
      "step": 19328
    },
    {
      "epoch": 4.76577909270217,
      "grad_norm": 0.328125,
      "learning_rate": 1.3367184254225118e-06,
      "loss": 0.0342,
      "step": 19330
    },
    {
      "epoch": 4.766272189349112,
      "grad_norm": 0.2890625,
      "learning_rate": 1.3311144612763393e-06,
      "loss": 0.0342,
      "step": 19332
    },
    {
      "epoch": 4.766765285996055,
      "grad_norm": 0.248046875,
      "learning_rate": 1.325522189943984e-06,
      "loss": 0.0291,
      "step": 19334
    },
    {
      "epoch": 4.767258382642998,
      "grad_norm": 0.306640625,
      "learning_rate": 1.3199416120881713e-06,
      "loss": 0.034,
      "step": 19336
    },
    {
      "epoch": 4.767751479289941,
      "grad_norm": 0.3125,
      "learning_rate": 1.314372728370239e-06,
      "loss": 0.0325,
      "step": 19338
    },
    {
      "epoch": 4.768244575936883,
      "grad_norm": 0.25390625,
      "learning_rate": 1.3088155394501366e-06,
      "loss": 0.0315,
      "step": 19340
    },
    {
      "epoch": 4.768737672583827,
      "grad_norm": 0.25390625,
      "learning_rate": 1.3032700459864156e-06,
      "loss": 0.0325,
      "step": 19342
    },
    {
      "epoch": 4.769230769230769,
      "grad_norm": 0.2890625,
      "learning_rate": 1.2977362486362277e-06,
      "loss": 0.0279,
      "step": 19344
    },
    {
      "epoch": 4.769723865877712,
      "grad_norm": 0.28515625,
      "learning_rate": 1.292214148055393e-06,
      "loss": 0.0339,
      "step": 19346
    },
    {
      "epoch": 4.770216962524655,
      "grad_norm": 0.322265625,
      "learning_rate": 1.2867037448982877e-06,
      "loss": 0.0302,
      "step": 19348
    },
    {
      "epoch": 4.770710059171598,
      "grad_norm": 0.28515625,
      "learning_rate": 1.2812050398179453e-06,
      "loss": 0.0303,
      "step": 19350
    },
    {
      "epoch": 4.77120315581854,
      "grad_norm": 0.291015625,
      "learning_rate": 1.2757180334659668e-06,
      "loss": 0.0313,
      "step": 19352
    },
    {
      "epoch": 4.771696252465484,
      "grad_norm": 0.326171875,
      "learning_rate": 1.270242726492621e-06,
      "loss": 0.0327,
      "step": 19354
    },
    {
      "epoch": 4.772189349112426,
      "grad_norm": 0.29296875,
      "learning_rate": 1.264779119546744e-06,
      "loss": 0.0344,
      "step": 19356
    },
    {
      "epoch": 4.772682445759369,
      "grad_norm": 0.259765625,
      "learning_rate": 1.2593272132757961e-06,
      "loss": 0.0313,
      "step": 19358
    },
    {
      "epoch": 4.773175542406312,
      "grad_norm": 0.26953125,
      "learning_rate": 1.2538870083258824e-06,
      "loss": 0.0307,
      "step": 19360
    },
    {
      "epoch": 4.773668639053255,
      "grad_norm": 0.328125,
      "learning_rate": 1.2484585053416765e-06,
      "loss": 0.0297,
      "step": 19362
    },
    {
      "epoch": 4.774161735700197,
      "grad_norm": 0.33984375,
      "learning_rate": 1.2430417049665077e-06,
      "loss": 0.0363,
      "step": 19364
    },
    {
      "epoch": 4.7746548323471405,
      "grad_norm": 0.337890625,
      "learning_rate": 1.237636607842274e-06,
      "loss": 0.0363,
      "step": 19366
    },
    {
      "epoch": 4.775147928994083,
      "grad_norm": 0.283203125,
      "learning_rate": 1.2322432146095297e-06,
      "loss": 0.0328,
      "step": 19368
    },
    {
      "epoch": 4.7756410256410255,
      "grad_norm": 0.28515625,
      "learning_rate": 1.226861525907419e-06,
      "loss": 0.0332,
      "step": 19370
    },
    {
      "epoch": 4.776134122287968,
      "grad_norm": 0.322265625,
      "learning_rate": 1.221491542373676e-06,
      "loss": 0.0296,
      "step": 19372
    },
    {
      "epoch": 4.7766272189349115,
      "grad_norm": 0.310546875,
      "learning_rate": 1.2161332646447143e-06,
      "loss": 0.031,
      "step": 19374
    },
    {
      "epoch": 4.777120315581854,
      "grad_norm": 0.376953125,
      "learning_rate": 1.2107866933554813e-06,
      "loss": 0.0315,
      "step": 19376
    },
    {
      "epoch": 4.7776134122287965,
      "grad_norm": 0.302734375,
      "learning_rate": 1.2054518291396033e-06,
      "loss": 0.0312,
      "step": 19378
    },
    {
      "epoch": 4.77810650887574,
      "grad_norm": 0.30078125,
      "learning_rate": 1.200128672629275e-06,
      "loss": 0.0341,
      "step": 19380
    },
    {
      "epoch": 4.7785996055226825,
      "grad_norm": 0.28515625,
      "learning_rate": 1.1948172244553357e-06,
      "loss": 0.0325,
      "step": 19382
    },
    {
      "epoch": 4.779092702169625,
      "grad_norm": 0.31640625,
      "learning_rate": 1.1895174852472157e-06,
      "loss": 0.0306,
      "step": 19384
    },
    {
      "epoch": 4.779585798816568,
      "grad_norm": 0.3203125,
      "learning_rate": 1.1842294556329458e-06,
      "loss": 0.0338,
      "step": 19386
    },
    {
      "epoch": 4.780078895463511,
      "grad_norm": 0.291015625,
      "learning_rate": 1.1789531362392136e-06,
      "loss": 0.0307,
      "step": 19388
    },
    {
      "epoch": 4.7805719921104535,
      "grad_norm": 0.26953125,
      "learning_rate": 1.1736885276912747e-06,
      "loss": 0.0308,
      "step": 19390
    },
    {
      "epoch": 4.781065088757396,
      "grad_norm": 0.30859375,
      "learning_rate": 1.1684356306130295e-06,
      "loss": 0.0309,
      "step": 19392
    },
    {
      "epoch": 4.781558185404339,
      "grad_norm": 0.3125,
      "learning_rate": 1.1631944456269583e-06,
      "loss": 0.0314,
      "step": 19394
    },
    {
      "epoch": 4.782051282051282,
      "grad_norm": 0.30078125,
      "learning_rate": 1.1579649733541752e-06,
      "loss": 0.034,
      "step": 19396
    },
    {
      "epoch": 4.7825443786982245,
      "grad_norm": 0.369140625,
      "learning_rate": 1.152747214414418e-06,
      "loss": 0.0338,
      "step": 19398
    },
    {
      "epoch": 4.783037475345168,
      "grad_norm": 0.302734375,
      "learning_rate": 1.147541169426003e-06,
      "loss": 0.0299,
      "step": 19400
    },
    {
      "epoch": 4.78353057199211,
      "grad_norm": 0.3359375,
      "learning_rate": 1.1423468390058812e-06,
      "loss": 0.0345,
      "step": 19402
    },
    {
      "epoch": 4.784023668639053,
      "grad_norm": 0.345703125,
      "learning_rate": 1.1371642237696044e-06,
      "loss": 0.0295,
      "step": 19404
    },
    {
      "epoch": 4.784516765285996,
      "grad_norm": 0.33984375,
      "learning_rate": 1.1319933243313596e-06,
      "loss": 0.0312,
      "step": 19406
    },
    {
      "epoch": 4.785009861932939,
      "grad_norm": 0.318359375,
      "learning_rate": 1.1268341413039008e-06,
      "loss": 0.0364,
      "step": 19408
    },
    {
      "epoch": 4.785502958579881,
      "grad_norm": 0.2578125,
      "learning_rate": 1.1216866752986389e-06,
      "loss": 0.0285,
      "step": 19410
    },
    {
      "epoch": 4.785996055226825,
      "grad_norm": 0.267578125,
      "learning_rate": 1.1165509269255748e-06,
      "loss": 0.0295,
      "step": 19412
    },
    {
      "epoch": 4.786489151873767,
      "grad_norm": 0.26953125,
      "learning_rate": 1.1114268967933105e-06,
      "loss": 0.03,
      "step": 19414
    },
    {
      "epoch": 4.78698224852071,
      "grad_norm": 0.279296875,
      "learning_rate": 1.1063145855090939e-06,
      "loss": 0.0325,
      "step": 19416
    },
    {
      "epoch": 4.787475345167653,
      "grad_norm": 0.271484375,
      "learning_rate": 1.1012139936787402e-06,
      "loss": 0.0323,
      "step": 19418
    },
    {
      "epoch": 4.787968441814596,
      "grad_norm": 0.3125,
      "learning_rate": 1.0961251219067214e-06,
      "loss": 0.0329,
      "step": 19420
    },
    {
      "epoch": 4.788461538461538,
      "grad_norm": 0.287109375,
      "learning_rate": 1.0910479707960663e-06,
      "loss": 0.035,
      "step": 19422
    },
    {
      "epoch": 4.788954635108482,
      "grad_norm": 0.322265625,
      "learning_rate": 1.0859825409484713e-06,
      "loss": 0.0304,
      "step": 19424
    },
    {
      "epoch": 4.789447731755424,
      "grad_norm": 0.259765625,
      "learning_rate": 1.080928832964212e-06,
      "loss": 0.0346,
      "step": 19426
    },
    {
      "epoch": 4.789940828402367,
      "grad_norm": 0.2890625,
      "learning_rate": 1.0758868474421868e-06,
      "loss": 0.0326,
      "step": 19428
    },
    {
      "epoch": 4.790433925049309,
      "grad_norm": 0.296875,
      "learning_rate": 1.0708565849798846e-06,
      "loss": 0.0295,
      "step": 19430
    },
    {
      "epoch": 4.790927021696253,
      "grad_norm": 0.318359375,
      "learning_rate": 1.0658380461734175e-06,
      "loss": 0.0342,
      "step": 19432
    },
    {
      "epoch": 4.791420118343195,
      "grad_norm": 0.275390625,
      "learning_rate": 1.0608312316175317e-06,
      "loss": 0.0305,
      "step": 19434
    },
    {
      "epoch": 4.791913214990138,
      "grad_norm": 0.287109375,
      "learning_rate": 1.055836141905553e-06,
      "loss": 0.0325,
      "step": 19436
    },
    {
      "epoch": 4.792406311637081,
      "grad_norm": 0.2890625,
      "learning_rate": 1.0508527776294186e-06,
      "loss": 0.0325,
      "step": 19438
    },
    {
      "epoch": 4.792899408284024,
      "grad_norm": 0.2890625,
      "learning_rate": 1.0458811393796896e-06,
      "loss": 0.0299,
      "step": 19440
    },
    {
      "epoch": 4.793392504930966,
      "grad_norm": 0.29296875,
      "learning_rate": 1.0409212277455394e-06,
      "loss": 0.0302,
      "step": 19442
    },
    {
      "epoch": 4.793885601577909,
      "grad_norm": 0.33984375,
      "learning_rate": 1.0359730433147308e-06,
      "loss": 0.0317,
      "step": 19444
    },
    {
      "epoch": 4.794378698224852,
      "grad_norm": 0.337890625,
      "learning_rate": 1.0310365866736727e-06,
      "loss": 0.0302,
      "step": 19446
    },
    {
      "epoch": 4.794871794871795,
      "grad_norm": 0.26953125,
      "learning_rate": 1.0261118584073526e-06,
      "loss": 0.0314,
      "step": 19448
    },
    {
      "epoch": 4.795364891518737,
      "grad_norm": 0.302734375,
      "learning_rate": 1.0211988590993704e-06,
      "loss": 0.0323,
      "step": 19450
    },
    {
      "epoch": 4.795857988165681,
      "grad_norm": 0.298828125,
      "learning_rate": 1.0162975893319492e-06,
      "loss": 0.0314,
      "step": 19452
    },
    {
      "epoch": 4.796351084812623,
      "grad_norm": 0.3046875,
      "learning_rate": 1.0114080496859136e-06,
      "loss": 0.0314,
      "step": 19454
    },
    {
      "epoch": 4.796844181459566,
      "grad_norm": 0.291015625,
      "learning_rate": 1.0065302407407106e-06,
      "loss": 0.0333,
      "step": 19456
    },
    {
      "epoch": 4.797337278106509,
      "grad_norm": 0.3046875,
      "learning_rate": 1.0016641630743894e-06,
      "loss": 0.0303,
      "step": 19458
    },
    {
      "epoch": 4.797830374753452,
      "grad_norm": 0.29296875,
      "learning_rate": 9.968098172635887e-07,
      "loss": 0.0309,
      "step": 19460
    },
    {
      "epoch": 4.798323471400394,
      "grad_norm": 0.287109375,
      "learning_rate": 9.919672038835925e-07,
      "loss": 0.0317,
      "step": 19462
    },
    {
      "epoch": 4.798816568047338,
      "grad_norm": 0.267578125,
      "learning_rate": 9.871363235082865e-07,
      "loss": 0.0276,
      "step": 19464
    },
    {
      "epoch": 4.79930966469428,
      "grad_norm": 0.341796875,
      "learning_rate": 9.823171767101236e-07,
      "loss": 0.0317,
      "step": 19466
    },
    {
      "epoch": 4.799802761341223,
      "grad_norm": 0.2578125,
      "learning_rate": 9.77509764060225e-07,
      "loss": 0.0288,
      "step": 19468
    },
    {
      "epoch": 4.800295857988166,
      "grad_norm": 0.345703125,
      "learning_rate": 9.727140861282902e-07,
      "loss": 0.0328,
      "step": 19470
    },
    {
      "epoch": 4.800788954635109,
      "grad_norm": 0.37890625,
      "learning_rate": 9.679301434826426e-07,
      "loss": 0.0329,
      "step": 19472
    },
    {
      "epoch": 4.801282051282051,
      "grad_norm": 0.357421875,
      "learning_rate": 9.631579366901843e-07,
      "loss": 0.0338,
      "step": 19474
    },
    {
      "epoch": 4.8017751479289945,
      "grad_norm": 0.294921875,
      "learning_rate": 9.58397466316463e-07,
      "loss": 0.0315,
      "step": 19476
    },
    {
      "epoch": 4.802268244575937,
      "grad_norm": 0.27734375,
      "learning_rate": 9.536487329256271e-07,
      "loss": 0.03,
      "step": 19478
    },
    {
      "epoch": 4.80276134122288,
      "grad_norm": 0.287109375,
      "learning_rate": 9.489117370804157e-07,
      "loss": 0.031,
      "step": 19480
    },
    {
      "epoch": 4.803254437869823,
      "grad_norm": 0.279296875,
      "learning_rate": 9.441864793422017e-07,
      "loss": 0.0356,
      "step": 19482
    },
    {
      "epoch": 4.8037475345167655,
      "grad_norm": 0.30859375,
      "learning_rate": 9.394729602709374e-07,
      "loss": 0.0302,
      "step": 19484
    },
    {
      "epoch": 4.804240631163708,
      "grad_norm": 0.283203125,
      "learning_rate": 9.34771180425209e-07,
      "loss": 0.0339,
      "step": 19486
    },
    {
      "epoch": 4.804733727810651,
      "grad_norm": 0.287109375,
      "learning_rate": 9.300811403622156e-07,
      "loss": 0.0303,
      "step": 19488
    },
    {
      "epoch": 4.805226824457594,
      "grad_norm": 0.26953125,
      "learning_rate": 9.254028406377346e-07,
      "loss": 0.0316,
      "step": 19490
    },
    {
      "epoch": 4.8057199211045365,
      "grad_norm": 0.302734375,
      "learning_rate": 9.207362818061783e-07,
      "loss": 0.0358,
      "step": 19492
    },
    {
      "epoch": 4.806213017751479,
      "grad_norm": 0.318359375,
      "learning_rate": 9.160814644205595e-07,
      "loss": 0.0354,
      "step": 19494
    },
    {
      "epoch": 4.806706114398422,
      "grad_norm": 0.30859375,
      "learning_rate": 9.11438389032504e-07,
      "loss": 0.0346,
      "step": 19496
    },
    {
      "epoch": 4.807199211045365,
      "grad_norm": 0.296875,
      "learning_rate": 9.068070561922382e-07,
      "loss": 0.0313,
      "step": 19498
    },
    {
      "epoch": 4.8076923076923075,
      "grad_norm": 0.337890625,
      "learning_rate": 9.021874664486008e-07,
      "loss": 0.0333,
      "step": 19500
    },
    {
      "epoch": 4.80818540433925,
      "grad_norm": 0.361328125,
      "learning_rate": 8.975796203490428e-07,
      "loss": 0.0293,
      "step": 19502
    },
    {
      "epoch": 4.808678500986193,
      "grad_norm": 0.26171875,
      "learning_rate": 8.929835184396162e-07,
      "loss": 0.0304,
      "step": 19504
    },
    {
      "epoch": 4.809171597633136,
      "grad_norm": 0.3046875,
      "learning_rate": 8.883991612649967e-07,
      "loss": 0.0322,
      "step": 19506
    },
    {
      "epoch": 4.8096646942800785,
      "grad_norm": 0.31640625,
      "learning_rate": 8.838265493684272e-07,
      "loss": 0.034,
      "step": 19508
    },
    {
      "epoch": 4.810157790927022,
      "grad_norm": 0.29296875,
      "learning_rate": 8.792656832918078e-07,
      "loss": 0.0317,
      "step": 19510
    },
    {
      "epoch": 4.810650887573964,
      "grad_norm": 0.283203125,
      "learning_rate": 8.747165635756283e-07,
      "loss": 0.0326,
      "step": 19512
    },
    {
      "epoch": 4.811143984220907,
      "grad_norm": 0.306640625,
      "learning_rate": 8.701791907589796e-07,
      "loss": 0.0305,
      "step": 19514
    },
    {
      "epoch": 4.81163708086785,
      "grad_norm": 0.37890625,
      "learning_rate": 8.65653565379565e-07,
      "loss": 0.0316,
      "step": 19516
    },
    {
      "epoch": 4.812130177514793,
      "grad_norm": 0.31640625,
      "learning_rate": 8.611396879736999e-07,
      "loss": 0.0307,
      "step": 19518
    },
    {
      "epoch": 4.812623274161735,
      "grad_norm": 0.2734375,
      "learning_rate": 8.56637559076312e-07,
      "loss": 0.0326,
      "step": 19520
    },
    {
      "epoch": 4.813116370808679,
      "grad_norm": 0.310546875,
      "learning_rate": 8.521471792209079e-07,
      "loss": 0.0312,
      "step": 19522
    },
    {
      "epoch": 4.813609467455621,
      "grad_norm": 0.28515625,
      "learning_rate": 8.476685489396397e-07,
      "loss": 0.0308,
      "step": 19524
    },
    {
      "epoch": 4.814102564102564,
      "grad_norm": 0.345703125,
      "learning_rate": 8.432016687632271e-07,
      "loss": 0.0327,
      "step": 19526
    },
    {
      "epoch": 4.814595660749507,
      "grad_norm": 0.404296875,
      "learning_rate": 8.387465392210581e-07,
      "loss": 0.0328,
      "step": 19528
    },
    {
      "epoch": 4.81508875739645,
      "grad_norm": 0.341796875,
      "learning_rate": 8.343031608410656e-07,
      "loss": 0.0337,
      "step": 19530
    },
    {
      "epoch": 4.815581854043392,
      "grad_norm": 0.25390625,
      "learning_rate": 8.298715341498175e-07,
      "loss": 0.0314,
      "step": 19532
    },
    {
      "epoch": 4.816074950690336,
      "grad_norm": 0.29296875,
      "learning_rate": 8.254516596724826e-07,
      "loss": 0.0349,
      "step": 19534
    },
    {
      "epoch": 4.816568047337278,
      "grad_norm": 0.2890625,
      "learning_rate": 8.210435379328418e-07,
      "loss": 0.0317,
      "step": 19536
    },
    {
      "epoch": 4.817061143984221,
      "grad_norm": 0.2578125,
      "learning_rate": 8.166471694532996e-07,
      "loss": 0.0322,
      "step": 19538
    },
    {
      "epoch": 4.817554240631163,
      "grad_norm": 0.3203125,
      "learning_rate": 8.12262554754828e-07,
      "loss": 0.0329,
      "step": 19540
    },
    {
      "epoch": 4.818047337278107,
      "grad_norm": 0.30859375,
      "learning_rate": 8.078896943570446e-07,
      "loss": 0.0332,
      "step": 19542
    },
    {
      "epoch": 4.818540433925049,
      "grad_norm": 0.416015625,
      "learning_rate": 8.035285887781463e-07,
      "loss": 0.0325,
      "step": 19544
    },
    {
      "epoch": 4.819033530571992,
      "grad_norm": 0.3203125,
      "learning_rate": 7.991792385349528e-07,
      "loss": 0.0343,
      "step": 19546
    },
    {
      "epoch": 4.819526627218935,
      "grad_norm": 0.283203125,
      "learning_rate": 7.948416441428852e-07,
      "loss": 0.03,
      "step": 19548
    },
    {
      "epoch": 4.820019723865878,
      "grad_norm": 0.3359375,
      "learning_rate": 7.905158061159656e-07,
      "loss": 0.0308,
      "step": 19550
    },
    {
      "epoch": 4.82051282051282,
      "grad_norm": 0.2734375,
      "learning_rate": 7.862017249668507e-07,
      "loss": 0.0307,
      "step": 19552
    },
    {
      "epoch": 4.821005917159764,
      "grad_norm": 0.291015625,
      "learning_rate": 7.818994012067649e-07,
      "loss": 0.0342,
      "step": 19554
    },
    {
      "epoch": 4.821499013806706,
      "grad_norm": 0.26953125,
      "learning_rate": 7.776088353455668e-07,
      "loss": 0.0278,
      "step": 19556
    },
    {
      "epoch": 4.821992110453649,
      "grad_norm": 0.291015625,
      "learning_rate": 7.733300278917055e-07,
      "loss": 0.0338,
      "step": 19558
    },
    {
      "epoch": 4.822485207100591,
      "grad_norm": 0.3359375,
      "learning_rate": 7.690629793522419e-07,
      "loss": 0.0335,
      "step": 19560
    },
    {
      "epoch": 4.822978303747535,
      "grad_norm": 0.28515625,
      "learning_rate": 7.648076902328494e-07,
      "loss": 0.0314,
      "step": 19562
    },
    {
      "epoch": 4.823471400394477,
      "grad_norm": 0.275390625,
      "learning_rate": 7.605641610378133e-07,
      "loss": 0.0325,
      "step": 19564
    },
    {
      "epoch": 4.82396449704142,
      "grad_norm": 0.294921875,
      "learning_rate": 7.563323922699983e-07,
      "loss": 0.0336,
      "step": 19566
    },
    {
      "epoch": 4.824457593688363,
      "grad_norm": 0.296875,
      "learning_rate": 7.52112384430903e-07,
      "loss": 0.0357,
      "step": 19568
    },
    {
      "epoch": 4.824950690335306,
      "grad_norm": 0.302734375,
      "learning_rate": 7.479041380206275e-07,
      "loss": 0.0345,
      "step": 19570
    },
    {
      "epoch": 4.825443786982248,
      "grad_norm": 0.361328125,
      "learning_rate": 7.437076535378507e-07,
      "loss": 0.032,
      "step": 19572
    },
    {
      "epoch": 4.825936883629192,
      "grad_norm": 0.28515625,
      "learning_rate": 7.395229314798968e-07,
      "loss": 0.0324,
      "step": 19574
    },
    {
      "epoch": 4.826429980276134,
      "grad_norm": 0.271484375,
      "learning_rate": 7.353499723426804e-07,
      "loss": 0.0326,
      "step": 19576
    },
    {
      "epoch": 4.826923076923077,
      "grad_norm": 0.283203125,
      "learning_rate": 7.31188776620706e-07,
      "loss": 0.0308,
      "step": 19578
    },
    {
      "epoch": 4.82741617357002,
      "grad_norm": 0.28515625,
      "learning_rate": 7.270393448071233e-07,
      "loss": 0.0307,
      "step": 19580
    },
    {
      "epoch": 4.827909270216963,
      "grad_norm": 0.287109375,
      "learning_rate": 7.229016773936281e-07,
      "loss": 0.0301,
      "step": 19582
    },
    {
      "epoch": 4.828402366863905,
      "grad_norm": 0.255859375,
      "learning_rate": 7.187757748705948e-07,
      "loss": 0.0304,
      "step": 19584
    },
    {
      "epoch": 4.8288954635108485,
      "grad_norm": 0.345703125,
      "learning_rate": 7.146616377269432e-07,
      "loss": 0.031,
      "step": 19586
    },
    {
      "epoch": 4.829388560157791,
      "grad_norm": 0.267578125,
      "learning_rate": 7.105592664502169e-07,
      "loss": 0.0312,
      "step": 19588
    },
    {
      "epoch": 4.829881656804734,
      "grad_norm": 0.296875,
      "learning_rate": 7.064686615265826e-07,
      "loss": 0.0305,
      "step": 19590
    },
    {
      "epoch": 4.830374753451677,
      "grad_norm": 0.28125,
      "learning_rate": 7.023898234407967e-07,
      "loss": 0.0308,
      "step": 19592
    },
    {
      "epoch": 4.8308678500986195,
      "grad_norm": 0.314453125,
      "learning_rate": 6.98322752676217e-07,
      "loss": 0.0308,
      "step": 19594
    },
    {
      "epoch": 4.831360946745562,
      "grad_norm": 0.271484375,
      "learning_rate": 6.94267449714836e-07,
      "loss": 0.0337,
      "step": 19596
    },
    {
      "epoch": 4.831854043392505,
      "grad_norm": 0.296875,
      "learning_rate": 6.902239150372025e-07,
      "loss": 0.0309,
      "step": 19598
    },
    {
      "epoch": 4.832347140039448,
      "grad_norm": 0.333984375,
      "learning_rate": 6.86192149122522e-07,
      "loss": 0.0336,
      "step": 19600
    },
    {
      "epoch": 4.8328402366863905,
      "grad_norm": 0.298828125,
      "learning_rate": 6.821721524485569e-07,
      "loss": 0.0291,
      "step": 19602
    },
    {
      "epoch": 4.833333333333333,
      "grad_norm": 0.28515625,
      "learning_rate": 6.781639254917149e-07,
      "loss": 0.0342,
      "step": 19604
    },
    {
      "epoch": 4.8338264299802765,
      "grad_norm": 0.298828125,
      "learning_rate": 6.741674687269827e-07,
      "loss": 0.035,
      "step": 19606
    },
    {
      "epoch": 4.834319526627219,
      "grad_norm": 0.302734375,
      "learning_rate": 6.701827826279816e-07,
      "loss": 0.0316,
      "step": 19608
    },
    {
      "epoch": 4.8348126232741615,
      "grad_norm": 0.296875,
      "learning_rate": 6.662098676669005e-07,
      "loss": 0.0328,
      "step": 19610
    },
    {
      "epoch": 4.835305719921105,
      "grad_norm": 0.29296875,
      "learning_rate": 6.622487243145625e-07,
      "loss": 0.0344,
      "step": 19612
    },
    {
      "epoch": 4.835798816568047,
      "grad_norm": 0.314453125,
      "learning_rate": 6.582993530403813e-07,
      "loss": 0.0314,
      "step": 19614
    },
    {
      "epoch": 4.83629191321499,
      "grad_norm": 0.30859375,
      "learning_rate": 6.543617543123826e-07,
      "loss": 0.0345,
      "step": 19616
    },
    {
      "epoch": 4.8367850098619325,
      "grad_norm": 0.330078125,
      "learning_rate": 6.504359285971817e-07,
      "loss": 0.0338,
      "step": 19618
    },
    {
      "epoch": 4.837278106508876,
      "grad_norm": 0.275390625,
      "learning_rate": 6.465218763600289e-07,
      "loss": 0.0332,
      "step": 19620
    },
    {
      "epoch": 4.837771203155818,
      "grad_norm": 0.34375,
      "learning_rate": 6.426195980647532e-07,
      "loss": 0.0323,
      "step": 19622
    },
    {
      "epoch": 4.838264299802761,
      "grad_norm": 0.3125,
      "learning_rate": 6.387290941737845e-07,
      "loss": 0.0366,
      "step": 19624
    },
    {
      "epoch": 4.838757396449704,
      "grad_norm": 0.318359375,
      "learning_rate": 6.348503651481985e-07,
      "loss": 0.033,
      "step": 19626
    },
    {
      "epoch": 4.839250493096647,
      "grad_norm": 0.291015625,
      "learning_rate": 6.309834114476165e-07,
      "loss": 0.0326,
      "step": 19628
    },
    {
      "epoch": 4.839743589743589,
      "grad_norm": 0.337890625,
      "learning_rate": 6.271282335303052e-07,
      "loss": 0.0321,
      "step": 19630
    },
    {
      "epoch": 4.840236686390533,
      "grad_norm": 0.341796875,
      "learning_rate": 6.232848318531326e-07,
      "loss": 0.0315,
      "step": 19632
    },
    {
      "epoch": 4.840729783037475,
      "grad_norm": 0.3125,
      "learning_rate": 6.194532068715453e-07,
      "loss": 0.0312,
      "step": 19634
    },
    {
      "epoch": 4.841222879684418,
      "grad_norm": 0.2890625,
      "learning_rate": 6.156333590396246e-07,
      "loss": 0.036,
      "step": 19636
    },
    {
      "epoch": 4.841715976331361,
      "grad_norm": 0.298828125,
      "learning_rate": 6.118252888100528e-07,
      "loss": 0.0316,
      "step": 19638
    },
    {
      "epoch": 4.842209072978304,
      "grad_norm": 0.314453125,
      "learning_rate": 6.080289966340802e-07,
      "loss": 0.033,
      "step": 19640
    },
    {
      "epoch": 4.842702169625246,
      "grad_norm": 0.314453125,
      "learning_rate": 6.042444829616245e-07,
      "loss": 0.0351,
      "step": 19642
    },
    {
      "epoch": 4.84319526627219,
      "grad_norm": 0.341796875,
      "learning_rate": 6.004717482411381e-07,
      "loss": 0.0332,
      "step": 19644
    },
    {
      "epoch": 4.843688362919132,
      "grad_norm": 0.298828125,
      "learning_rate": 5.967107929197302e-07,
      "loss": 0.0334,
      "step": 19646
    },
    {
      "epoch": 4.844181459566075,
      "grad_norm": 0.2890625,
      "learning_rate": 5.929616174430885e-07,
      "loss": 0.0297,
      "step": 19648
    },
    {
      "epoch": 4.844674556213018,
      "grad_norm": 0.275390625,
      "learning_rate": 5.892242222555133e-07,
      "loss": 0.0327,
      "step": 19650
    },
    {
      "epoch": 4.845167652859961,
      "grad_norm": 0.267578125,
      "learning_rate": 5.854986077999169e-07,
      "loss": 0.0322,
      "step": 19652
    },
    {
      "epoch": 4.845660749506903,
      "grad_norm": 0.283203125,
      "learning_rate": 5.817847745177796e-07,
      "loss": 0.0327,
      "step": 19654
    },
    {
      "epoch": 4.846153846153846,
      "grad_norm": 0.275390625,
      "learning_rate": 5.78082722849238e-07,
      "loss": 0.0292,
      "step": 19656
    },
    {
      "epoch": 4.846646942800789,
      "grad_norm": 0.33203125,
      "learning_rate": 5.743924532329858e-07,
      "loss": 0.0345,
      "step": 19658
    },
    {
      "epoch": 4.847140039447732,
      "grad_norm": 0.287109375,
      "learning_rate": 5.70713966106351e-07,
      "loss": 0.0315,
      "step": 19660
    },
    {
      "epoch": 4.847633136094674,
      "grad_norm": 0.287109375,
      "learning_rate": 5.670472619052514e-07,
      "loss": 0.0309,
      "step": 19662
    },
    {
      "epoch": 4.848126232741618,
      "grad_norm": 0.314453125,
      "learning_rate": 5.633923410642283e-07,
      "loss": 0.0326,
      "step": 19664
    },
    {
      "epoch": 4.84861932938856,
      "grad_norm": 0.3515625,
      "learning_rate": 5.597492040163799e-07,
      "loss": 0.0292,
      "step": 19666
    },
    {
      "epoch": 4.849112426035503,
      "grad_norm": 0.30078125,
      "learning_rate": 5.561178511934607e-07,
      "loss": 0.0313,
      "step": 19668
    },
    {
      "epoch": 4.849605522682445,
      "grad_norm": 0.28125,
      "learning_rate": 5.52498283025793e-07,
      "loss": 0.03,
      "step": 19670
    },
    {
      "epoch": 4.850098619329389,
      "grad_norm": 0.298828125,
      "learning_rate": 5.488904999423339e-07,
      "loss": 0.0297,
      "step": 19672
    },
    {
      "epoch": 4.850591715976331,
      "grad_norm": 0.2578125,
      "learning_rate": 5.45294502370608e-07,
      "loss": 0.0298,
      "step": 19674
    },
    {
      "epoch": 4.851084812623274,
      "grad_norm": 0.330078125,
      "learning_rate": 5.417102907367632e-07,
      "loss": 0.0323,
      "step": 19676
    },
    {
      "epoch": 4.851577909270217,
      "grad_norm": 0.337890625,
      "learning_rate": 5.3813786546556e-07,
      "loss": 0.0337,
      "step": 19678
    },
    {
      "epoch": 4.85207100591716,
      "grad_norm": 0.27734375,
      "learning_rate": 5.345772269803484e-07,
      "loss": 0.0327,
      "step": 19680
    },
    {
      "epoch": 4.852564102564102,
      "grad_norm": 0.33984375,
      "learning_rate": 5.310283757030688e-07,
      "loss": 0.0336,
      "step": 19682
    },
    {
      "epoch": 4.853057199211046,
      "grad_norm": 0.2275390625,
      "learning_rate": 5.274913120542957e-07,
      "loss": 0.0288,
      "step": 19684
    },
    {
      "epoch": 4.853550295857988,
      "grad_norm": 0.2451171875,
      "learning_rate": 5.239660364531829e-07,
      "loss": 0.0317,
      "step": 19686
    },
    {
      "epoch": 4.854043392504931,
      "grad_norm": 0.3203125,
      "learning_rate": 5.204525493174961e-07,
      "loss": 0.0337,
      "step": 19688
    },
    {
      "epoch": 4.854536489151874,
      "grad_norm": 0.369140625,
      "learning_rate": 5.169508510636023e-07,
      "loss": 0.0346,
      "step": 19690
    },
    {
      "epoch": 4.855029585798817,
      "grad_norm": 0.287109375,
      "learning_rate": 5.134609421064918e-07,
      "loss": 0.0334,
      "step": 19692
    },
    {
      "epoch": 4.855522682445759,
      "grad_norm": 0.33984375,
      "learning_rate": 5.099828228597004e-07,
      "loss": 0.0349,
      "step": 19694
    },
    {
      "epoch": 4.8560157790927025,
      "grad_norm": 0.265625,
      "learning_rate": 5.065164937354428e-07,
      "loss": 0.0361,
      "step": 19696
    },
    {
      "epoch": 4.856508875739645,
      "grad_norm": 0.306640625,
      "learning_rate": 5.030619551444682e-07,
      "loss": 0.0324,
      "step": 19698
    },
    {
      "epoch": 4.857001972386588,
      "grad_norm": 0.275390625,
      "learning_rate": 4.996192074961714e-07,
      "loss": 0.0322,
      "step": 19700
    },
    {
      "epoch": 4.857495069033531,
      "grad_norm": 0.328125,
      "learning_rate": 4.961882511985483e-07,
      "loss": 0.0346,
      "step": 19702
    },
    {
      "epoch": 4.8579881656804735,
      "grad_norm": 0.275390625,
      "learning_rate": 4.927690866581736e-07,
      "loss": 0.0338,
      "step": 19704
    },
    {
      "epoch": 4.858481262327416,
      "grad_norm": 0.365234375,
      "learning_rate": 4.893617142802342e-07,
      "loss": 0.0295,
      "step": 19706
    },
    {
      "epoch": 4.858974358974359,
      "grad_norm": 0.279296875,
      "learning_rate": 4.859661344685296e-07,
      "loss": 0.0298,
      "step": 19708
    },
    {
      "epoch": 4.859467455621302,
      "grad_norm": 0.248046875,
      "learning_rate": 4.825823476254488e-07,
      "loss": 0.0298,
      "step": 19710
    },
    {
      "epoch": 4.8599605522682445,
      "grad_norm": 0.296875,
      "learning_rate": 4.792103541520044e-07,
      "loss": 0.0309,
      "step": 19712
    },
    {
      "epoch": 4.860453648915187,
      "grad_norm": 0.33984375,
      "learning_rate": 4.758501544477767e-07,
      "loss": 0.0328,
      "step": 19714
    },
    {
      "epoch": 4.8609467455621305,
      "grad_norm": 0.265625,
      "learning_rate": 4.7250174891098067e-07,
      "loss": 0.0316,
      "step": 19716
    },
    {
      "epoch": 4.861439842209073,
      "grad_norm": 0.326171875,
      "learning_rate": 4.6916513793840986e-07,
      "loss": 0.03,
      "step": 19718
    },
    {
      "epoch": 4.8619329388560155,
      "grad_norm": 0.328125,
      "learning_rate": 4.6584032192549254e-07,
      "loss": 0.0314,
      "step": 19720
    },
    {
      "epoch": 4.862426035502959,
      "grad_norm": 0.318359375,
      "learning_rate": 4.625273012662246e-07,
      "loss": 0.0303,
      "step": 19722
    },
    {
      "epoch": 4.8629191321499015,
      "grad_norm": 0.328125,
      "learning_rate": 4.5922607635321414e-07,
      "loss": 0.0311,
      "step": 19724
    },
    {
      "epoch": 4.863412228796844,
      "grad_norm": 0.291015625,
      "learning_rate": 4.559366475776705e-07,
      "loss": 0.029,
      "step": 19726
    },
    {
      "epoch": 4.8639053254437865,
      "grad_norm": 0.251953125,
      "learning_rate": 4.5265901532942636e-07,
      "loss": 0.0306,
      "step": 19728
    },
    {
      "epoch": 4.86439842209073,
      "grad_norm": 0.3984375,
      "learning_rate": 4.4939317999688204e-07,
      "loss": 0.0351,
      "step": 19730
    },
    {
      "epoch": 4.8648915187376724,
      "grad_norm": 0.2890625,
      "learning_rate": 4.461391419670724e-07,
      "loss": 0.0308,
      "step": 19732
    },
    {
      "epoch": 4.865384615384615,
      "grad_norm": 0.25390625,
      "learning_rate": 4.4289690162561124e-07,
      "loss": 0.0323,
      "step": 19734
    },
    {
      "epoch": 4.865877712031558,
      "grad_norm": 0.287109375,
      "learning_rate": 4.3966645935671346e-07,
      "loss": 0.0337,
      "step": 19736
    },
    {
      "epoch": 4.866370808678501,
      "grad_norm": 0.3203125,
      "learning_rate": 4.364478155432283e-07,
      "loss": 0.0301,
      "step": 19738
    },
    {
      "epoch": 4.866863905325443,
      "grad_norm": 0.306640625,
      "learning_rate": 4.3324097056656187e-07,
      "loss": 0.0328,
      "step": 19740
    },
    {
      "epoch": 4.867357001972387,
      "grad_norm": 0.294921875,
      "learning_rate": 4.3004592480674346e-07,
      "loss": 0.03,
      "step": 19742
    },
    {
      "epoch": 4.867850098619329,
      "grad_norm": 0.328125,
      "learning_rate": 4.268626786424146e-07,
      "loss": 0.031,
      "step": 19744
    },
    {
      "epoch": 4.868343195266272,
      "grad_norm": 0.294921875,
      "learning_rate": 4.236912324508069e-07,
      "loss": 0.0327,
      "step": 19746
    },
    {
      "epoch": 4.868836291913215,
      "grad_norm": 0.30859375,
      "learning_rate": 4.20531586607753e-07,
      "loss": 0.0259,
      "step": 19748
    },
    {
      "epoch": 4.869329388560158,
      "grad_norm": 0.345703125,
      "learning_rate": 4.173837414876869e-07,
      "loss": 0.0359,
      "step": 19750
    },
    {
      "epoch": 4.8698224852071,
      "grad_norm": 0.3046875,
      "learning_rate": 4.1424769746364334e-07,
      "loss": 0.0326,
      "step": 19752
    },
    {
      "epoch": 4.870315581854044,
      "grad_norm": 0.25390625,
      "learning_rate": 4.1112345490725845e-07,
      "loss": 0.0273,
      "step": 19754
    },
    {
      "epoch": 4.870808678500986,
      "grad_norm": 0.2734375,
      "learning_rate": 4.080110141887805e-07,
      "loss": 0.0319,
      "step": 19756
    },
    {
      "epoch": 4.871301775147929,
      "grad_norm": 0.400390625,
      "learning_rate": 4.049103756770478e-07,
      "loss": 0.0326,
      "step": 19758
    },
    {
      "epoch": 4.871794871794872,
      "grad_norm": 0.294921875,
      "learning_rate": 4.0182153973949974e-07,
      "loss": 0.0281,
      "step": 19760
    },
    {
      "epoch": 4.872287968441815,
      "grad_norm": 0.328125,
      "learning_rate": 3.987445067421991e-07,
      "loss": 0.029,
      "step": 19762
    },
    {
      "epoch": 4.872781065088757,
      "grad_norm": 0.265625,
      "learning_rate": 3.9567927704975416e-07,
      "loss": 0.0304,
      "step": 19764
    },
    {
      "epoch": 4.8732741617357,
      "grad_norm": 0.294921875,
      "learning_rate": 3.9262585102545215e-07,
      "loss": 0.0317,
      "step": 19766
    },
    {
      "epoch": 4.873767258382643,
      "grad_norm": 0.322265625,
      "learning_rate": 3.895842290311147e-07,
      "loss": 0.0328,
      "step": 19768
    },
    {
      "epoch": 4.874260355029586,
      "grad_norm": 0.326171875,
      "learning_rate": 3.8655441142720905e-07,
      "loss": 0.0321,
      "step": 19770
    },
    {
      "epoch": 4.874753451676528,
      "grad_norm": 0.26171875,
      "learning_rate": 3.8353639857275914e-07,
      "loss": 0.0304,
      "step": 19772
    },
    {
      "epoch": 4.875246548323472,
      "grad_norm": 0.30859375,
      "learning_rate": 3.805301908254455e-07,
      "loss": 0.0336,
      "step": 19774
    },
    {
      "epoch": 4.875739644970414,
      "grad_norm": 0.3046875,
      "learning_rate": 3.775357885415054e-07,
      "loss": 0.0347,
      "step": 19776
    },
    {
      "epoch": 4.876232741617357,
      "grad_norm": 0.318359375,
      "learning_rate": 3.745531920757994e-07,
      "loss": 0.0306,
      "step": 19778
    },
    {
      "epoch": 4.8767258382643,
      "grad_norm": 0.31640625,
      "learning_rate": 3.7158240178176707e-07,
      "loss": 0.0321,
      "step": 19780
    },
    {
      "epoch": 4.877218934911243,
      "grad_norm": 0.3125,
      "learning_rate": 3.6862341801148226e-07,
      "loss": 0.0331,
      "step": 19782
    },
    {
      "epoch": 4.877712031558185,
      "grad_norm": 0.283203125,
      "learning_rate": 3.6567624111558675e-07,
      "loss": 0.0304,
      "step": 19784
    },
    {
      "epoch": 4.878205128205128,
      "grad_norm": 0.275390625,
      "learning_rate": 3.6274087144335665e-07,
      "loss": 0.0342,
      "step": 19786
    },
    {
      "epoch": 4.878698224852071,
      "grad_norm": 0.373046875,
      "learning_rate": 3.598173093426249e-07,
      "loss": 0.0342,
      "step": 19788
    },
    {
      "epoch": 4.879191321499014,
      "grad_norm": 0.314453125,
      "learning_rate": 3.569055551598588e-07,
      "loss": 0.0308,
      "step": 19790
    },
    {
      "epoch": 4.879684418145956,
      "grad_norm": 0.298828125,
      "learning_rate": 3.5400560924012673e-07,
      "loss": 0.0324,
      "step": 19792
    },
    {
      "epoch": 4.8801775147929,
      "grad_norm": 0.294921875,
      "learning_rate": 3.51117471927076e-07,
      "loss": 0.0316,
      "step": 19794
    },
    {
      "epoch": 4.880670611439842,
      "grad_norm": 0.36328125,
      "learning_rate": 3.4824114356297734e-07,
      "loss": 0.0309,
      "step": 19796
    },
    {
      "epoch": 4.881163708086785,
      "grad_norm": 0.271484375,
      "learning_rate": 3.4537662448869134e-07,
      "loss": 0.0308,
      "step": 19798
    },
    {
      "epoch": 4.881656804733728,
      "grad_norm": 0.306640625,
      "learning_rate": 3.4252391504366877e-07,
      "loss": 0.0288,
      "step": 19800
    },
    {
      "epoch": 4.882149901380671,
      "grad_norm": 0.2734375,
      "learning_rate": 3.3968301556598356e-07,
      "loss": 0.0304,
      "step": 19802
    },
    {
      "epoch": 4.882642998027613,
      "grad_norm": 0.353515625,
      "learning_rate": 3.3685392639228875e-07,
      "loss": 0.0296,
      "step": 19804
    },
    {
      "epoch": 4.883136094674557,
      "grad_norm": 0.283203125,
      "learning_rate": 3.340366478578494e-07,
      "loss": 0.0352,
      "step": 19806
    },
    {
      "epoch": 4.883629191321499,
      "grad_norm": 0.291015625,
      "learning_rate": 3.312311802965429e-07,
      "loss": 0.0315,
      "step": 19808
    },
    {
      "epoch": 4.884122287968442,
      "grad_norm": 0.3046875,
      "learning_rate": 3.284375240408144e-07,
      "loss": 0.0281,
      "step": 19810
    },
    {
      "epoch": 4.884615384615385,
      "grad_norm": 0.3359375,
      "learning_rate": 3.2565567942172136e-07,
      "loss": 0.0324,
      "step": 19812
    },
    {
      "epoch": 4.8851084812623276,
      "grad_norm": 0.296875,
      "learning_rate": 3.228856467689556e-07,
      "loss": 0.0359,
      "step": 19814
    },
    {
      "epoch": 4.88560157790927,
      "grad_norm": 0.28125,
      "learning_rate": 3.2012742641076564e-07,
      "loss": 0.0315,
      "step": 19816
    },
    {
      "epoch": 4.8860946745562135,
      "grad_norm": 0.2734375,
      "learning_rate": 3.1738101867401227e-07,
      "loss": 0.0306,
      "step": 19818
    },
    {
      "epoch": 4.886587771203156,
      "grad_norm": 0.271484375,
      "learning_rate": 3.146464238841573e-07,
      "loss": 0.0281,
      "step": 19820
    },
    {
      "epoch": 4.8870808678500985,
      "grad_norm": 0.3203125,
      "learning_rate": 3.1192364236528606e-07,
      "loss": 0.0361,
      "step": 19822
    },
    {
      "epoch": 4.887573964497041,
      "grad_norm": 0.34375,
      "learning_rate": 3.092126744400514e-07,
      "loss": 0.0339,
      "step": 19824
    },
    {
      "epoch": 4.8880670611439845,
      "grad_norm": 0.35546875,
      "learning_rate": 3.065135204296965e-07,
      "loss": 0.0294,
      "step": 19826
    },
    {
      "epoch": 4.888560157790927,
      "grad_norm": 0.318359375,
      "learning_rate": 3.0382618065413205e-07,
      "loss": 0.0331,
      "step": 19828
    },
    {
      "epoch": 4.8890532544378695,
      "grad_norm": 0.318359375,
      "learning_rate": 3.011506554317811e-07,
      "loss": 0.0293,
      "step": 19830
    },
    {
      "epoch": 4.889546351084813,
      "grad_norm": 0.28515625,
      "learning_rate": 2.984869450797345e-07,
      "loss": 0.0319,
      "step": 19832
    },
    {
      "epoch": 4.8900394477317555,
      "grad_norm": 0.349609375,
      "learning_rate": 2.958350499136398e-07,
      "loss": 0.0335,
      "step": 19834
    },
    {
      "epoch": 4.890532544378698,
      "grad_norm": 0.2734375,
      "learning_rate": 2.9319497024776764e-07,
      "loss": 0.0334,
      "step": 19836
    },
    {
      "epoch": 4.891025641025641,
      "grad_norm": 0.29296875,
      "learning_rate": 2.9056670639499015e-07,
      "loss": 0.031,
      "step": 19838
    },
    {
      "epoch": 4.891518737672584,
      "grad_norm": 0.302734375,
      "learning_rate": 2.879502586667582e-07,
      "loss": 0.0326,
      "step": 19840
    },
    {
      "epoch": 4.8920118343195265,
      "grad_norm": 0.40625,
      "learning_rate": 2.853456273731459e-07,
      "loss": 0.0354,
      "step": 19842
    },
    {
      "epoch": 4.892504930966469,
      "grad_norm": 0.291015625,
      "learning_rate": 2.827528128228063e-07,
      "loss": 0.0322,
      "step": 19844
    },
    {
      "epoch": 4.892998027613412,
      "grad_norm": 0.283203125,
      "learning_rate": 2.8017181532300486e-07,
      "loss": 0.0326,
      "step": 19846
    },
    {
      "epoch": 4.893491124260355,
      "grad_norm": 0.296875,
      "learning_rate": 2.7760263517961903e-07,
      "loss": 0.03,
      "step": 19848
    },
    {
      "epoch": 4.8939842209072975,
      "grad_norm": 0.25,
      "learning_rate": 2.750452726970942e-07,
      "loss": 0.032,
      "step": 19850
    },
    {
      "epoch": 4.894477317554241,
      "grad_norm": 0.33984375,
      "learning_rate": 2.724997281784991e-07,
      "loss": 0.035,
      "step": 19852
    },
    {
      "epoch": 4.894970414201183,
      "grad_norm": 0.2734375,
      "learning_rate": 2.6996600192549237e-07,
      "loss": 0.0298,
      "step": 19854
    },
    {
      "epoch": 4.895463510848126,
      "grad_norm": 0.318359375,
      "learning_rate": 2.6744409423833384e-07,
      "loss": 0.0328,
      "step": 19856
    },
    {
      "epoch": 4.895956607495069,
      "grad_norm": 0.255859375,
      "learning_rate": 2.6493400541588444e-07,
      "loss": 0.0313,
      "step": 19858
    },
    {
      "epoch": 4.896449704142012,
      "grad_norm": 0.322265625,
      "learning_rate": 2.624357357556062e-07,
      "loss": 0.033,
      "step": 19860
    },
    {
      "epoch": 4.896942800788954,
      "grad_norm": 0.279296875,
      "learning_rate": 2.5994928555356234e-07,
      "loss": 0.0298,
      "step": 19862
    },
    {
      "epoch": 4.897435897435898,
      "grad_norm": 0.287109375,
      "learning_rate": 2.5747465510439494e-07,
      "loss": 0.0293,
      "step": 19864
    },
    {
      "epoch": 4.89792899408284,
      "grad_norm": 0.31640625,
      "learning_rate": 2.550118447013805e-07,
      "loss": 0.0319,
      "step": 19866
    },
    {
      "epoch": 4.898422090729783,
      "grad_norm": 0.28125,
      "learning_rate": 2.5256085463636337e-07,
      "loss": 0.0316,
      "step": 19868
    },
    {
      "epoch": 4.898915187376726,
      "grad_norm": 0.28125,
      "learning_rate": 2.501216851998001e-07,
      "loss": 0.0296,
      "step": 19870
    },
    {
      "epoch": 4.899408284023669,
      "grad_norm": 0.3125,
      "learning_rate": 2.476943366807483e-07,
      "loss": 0.0334,
      "step": 19872
    },
    {
      "epoch": 4.899901380670611,
      "grad_norm": 0.328125,
      "learning_rate": 2.4527880936687787e-07,
      "loss": 0.0321,
      "step": 19874
    },
    {
      "epoch": 4.900394477317555,
      "grad_norm": 0.34765625,
      "learning_rate": 2.4287510354441543e-07,
      "loss": 0.0349,
      "step": 19876
    },
    {
      "epoch": 4.900887573964497,
      "grad_norm": 0.287109375,
      "learning_rate": 2.4048321949822205e-07,
      "loss": 0.0278,
      "step": 19878
    },
    {
      "epoch": 4.90138067061144,
      "grad_norm": 0.259765625,
      "learning_rate": 2.3810315751175982e-07,
      "loss": 0.03,
      "step": 19880
    },
    {
      "epoch": 4.901873767258382,
      "grad_norm": 0.34765625,
      "learning_rate": 2.357349178670698e-07,
      "loss": 0.031,
      "step": 19882
    },
    {
      "epoch": 4.902366863905326,
      "grad_norm": 0.28125,
      "learning_rate": 2.3337850084480528e-07,
      "loss": 0.0261,
      "step": 19884
    },
    {
      "epoch": 4.902859960552268,
      "grad_norm": 0.275390625,
      "learning_rate": 2.3103390672422064e-07,
      "loss": 0.0321,
      "step": 19886
    },
    {
      "epoch": 4.903353057199211,
      "grad_norm": 0.30078125,
      "learning_rate": 2.287011357831492e-07,
      "loss": 0.0324,
      "step": 19888
    },
    {
      "epoch": 4.903846153846154,
      "grad_norm": 0.359375,
      "learning_rate": 2.2638018829804764e-07,
      "loss": 0.032,
      "step": 19890
    },
    {
      "epoch": 4.904339250493097,
      "grad_norm": 0.34375,
      "learning_rate": 2.2407106454395145e-07,
      "loss": 0.038,
      "step": 19892
    },
    {
      "epoch": 4.904832347140039,
      "grad_norm": 0.259765625,
      "learning_rate": 2.217737647945195e-07,
      "loss": 0.0304,
      "step": 19894
    },
    {
      "epoch": 4.905325443786982,
      "grad_norm": 0.28515625,
      "learning_rate": 2.194882893219785e-07,
      "loss": 0.0332,
      "step": 19896
    },
    {
      "epoch": 4.905818540433925,
      "grad_norm": 0.26953125,
      "learning_rate": 2.172146383971896e-07,
      "loss": 0.0321,
      "step": 19898
    },
    {
      "epoch": 4.906311637080868,
      "grad_norm": 0.294921875,
      "learning_rate": 2.1495281228957053e-07,
      "loss": 0.0322,
      "step": 19900
    },
    {
      "epoch": 4.90680473372781,
      "grad_norm": 0.3125,
      "learning_rate": 2.127028112671847e-07,
      "loss": 0.0328,
      "step": 19902
    },
    {
      "epoch": 4.907297830374754,
      "grad_norm": 0.283203125,
      "learning_rate": 2.1046463559664109e-07,
      "loss": 0.0335,
      "step": 19904
    },
    {
      "epoch": 4.907790927021696,
      "grad_norm": 0.296875,
      "learning_rate": 2.0823828554320524e-07,
      "loss": 0.0328,
      "step": 19906
    },
    {
      "epoch": 4.908284023668639,
      "grad_norm": 0.310546875,
      "learning_rate": 2.0602376137068835e-07,
      "loss": 0.0292,
      "step": 19908
    },
    {
      "epoch": 4.908777120315582,
      "grad_norm": 0.28515625,
      "learning_rate": 2.0382106334153605e-07,
      "loss": 0.0326,
      "step": 19910
    },
    {
      "epoch": 4.909270216962525,
      "grad_norm": 0.287109375,
      "learning_rate": 2.01630191716784e-07,
      "loss": 0.032,
      "step": 19912
    },
    {
      "epoch": 4.909763313609467,
      "grad_norm": 0.349609375,
      "learning_rate": 1.994511467560467e-07,
      "loss": 0.0351,
      "step": 19914
    },
    {
      "epoch": 4.910256410256411,
      "grad_norm": 0.31640625,
      "learning_rate": 1.9728392871757318e-07,
      "loss": 0.0323,
      "step": 19916
    },
    {
      "epoch": 4.910749506903353,
      "grad_norm": 0.2734375,
      "learning_rate": 1.951285378581802e-07,
      "loss": 0.0337,
      "step": 19918
    },
    {
      "epoch": 4.911242603550296,
      "grad_norm": 0.26171875,
      "learning_rate": 1.9298497443329678e-07,
      "loss": 0.0312,
      "step": 19920
    },
    {
      "epoch": 4.911735700197239,
      "grad_norm": 0.2490234375,
      "learning_rate": 1.9085323869694195e-07,
      "loss": 0.0276,
      "step": 19922
    },
    {
      "epoch": 4.912228796844182,
      "grad_norm": 0.287109375,
      "learning_rate": 1.8873333090173585e-07,
      "loss": 0.0327,
      "step": 19924
    },
    {
      "epoch": 4.912721893491124,
      "grad_norm": 0.310546875,
      "learning_rate": 1.8662525129891084e-07,
      "loss": 0.0298,
      "step": 19926
    },
    {
      "epoch": 4.9132149901380675,
      "grad_norm": 0.279296875,
      "learning_rate": 1.8452900013827823e-07,
      "loss": 0.0314,
      "step": 19928
    },
    {
      "epoch": 4.91370808678501,
      "grad_norm": 0.267578125,
      "learning_rate": 1.824445776682504e-07,
      "loss": 0.0305,
      "step": 19930
    },
    {
      "epoch": 4.914201183431953,
      "grad_norm": 0.294921875,
      "learning_rate": 1.803719841358631e-07,
      "loss": 0.0313,
      "step": 19932
    },
    {
      "epoch": 4.914694280078895,
      "grad_norm": 0.2734375,
      "learning_rate": 1.7831121978670873e-07,
      "loss": 0.032,
      "step": 19934
    },
    {
      "epoch": 4.9151873767258385,
      "grad_norm": 0.25,
      "learning_rate": 1.7626228486500306e-07,
      "loss": 0.0287,
      "step": 19936
    },
    {
      "epoch": 4.915680473372781,
      "grad_norm": 0.283203125,
      "learning_rate": 1.742251796135741e-07,
      "loss": 0.0309,
      "step": 19938
    },
    {
      "epoch": 4.9161735700197235,
      "grad_norm": 0.296875,
      "learning_rate": 1.721999042738065e-07,
      "loss": 0.033,
      "step": 19940
    },
    {
      "epoch": 4.916666666666667,
      "grad_norm": 0.296875,
      "learning_rate": 1.701864590857194e-07,
      "loss": 0.0292,
      "step": 19942
    },
    {
      "epoch": 4.9171597633136095,
      "grad_norm": 0.265625,
      "learning_rate": 1.6818484428792191e-07,
      "loss": 0.031,
      "step": 19944
    },
    {
      "epoch": 4.917652859960552,
      "grad_norm": 0.28125,
      "learning_rate": 1.661950601176021e-07,
      "loss": 0.0331,
      "step": 19946
    },
    {
      "epoch": 4.918145956607495,
      "grad_norm": 0.2734375,
      "learning_rate": 1.6421710681057134e-07,
      "loss": 0.0303,
      "step": 19948
    },
    {
      "epoch": 4.918639053254438,
      "grad_norm": 0.251953125,
      "learning_rate": 1.62250984601231e-07,
      "loss": 0.03,
      "step": 19950
    },
    {
      "epoch": 4.9191321499013805,
      "grad_norm": 0.333984375,
      "learning_rate": 1.602966937225614e-07,
      "loss": 0.0336,
      "step": 19952
    },
    {
      "epoch": 4.919625246548323,
      "grad_norm": 0.298828125,
      "learning_rate": 1.5835423440617725e-07,
      "loss": 0.0342,
      "step": 19954
    },
    {
      "epoch": 4.920118343195266,
      "grad_norm": 0.265625,
      "learning_rate": 1.5642360688225e-07,
      "loss": 0.0313,
      "step": 19956
    },
    {
      "epoch": 4.920611439842209,
      "grad_norm": 0.267578125,
      "learning_rate": 1.545048113795855e-07,
      "loss": 0.0299,
      "step": 19958
    },
    {
      "epoch": 4.9211045364891515,
      "grad_norm": 0.33984375,
      "learning_rate": 1.5259784812556854e-07,
      "loss": 0.0307,
      "step": 19960
    },
    {
      "epoch": 4.921597633136095,
      "grad_norm": 0.271484375,
      "learning_rate": 1.5070271734618501e-07,
      "loss": 0.0326,
      "step": 19962
    },
    {
      "epoch": 4.922090729783037,
      "grad_norm": 0.28515625,
      "learning_rate": 1.4881941926601084e-07,
      "loss": 0.0307,
      "step": 19964
    },
    {
      "epoch": 4.92258382642998,
      "grad_norm": 0.32421875,
      "learning_rate": 1.4694795410823415e-07,
      "loss": 0.0321,
      "step": 19966
    },
    {
      "epoch": 4.923076923076923,
      "grad_norm": 0.30859375,
      "learning_rate": 1.4508832209463307e-07,
      "loss": 0.0287,
      "step": 19968
    },
    {
      "epoch": 4.923570019723866,
      "grad_norm": 0.298828125,
      "learning_rate": 1.4324052344558692e-07,
      "loss": 0.0318,
      "step": 19970
    },
    {
      "epoch": 4.924063116370808,
      "grad_norm": 0.31640625,
      "learning_rate": 1.4140455838006495e-07,
      "loss": 0.0333,
      "step": 19972
    },
    {
      "epoch": 4.924556213017752,
      "grad_norm": 0.322265625,
      "learning_rate": 1.3958042711563757e-07,
      "loss": 0.0365,
      "step": 19974
    },
    {
      "epoch": 4.925049309664694,
      "grad_norm": 0.259765625,
      "learning_rate": 1.3776812986848743e-07,
      "loss": 0.0292,
      "step": 19976
    },
    {
      "epoch": 4.925542406311637,
      "grad_norm": 0.380859375,
      "learning_rate": 1.3596766685336494e-07,
      "loss": 0.0309,
      "step": 19978
    },
    {
      "epoch": 4.92603550295858,
      "grad_norm": 0.271484375,
      "learning_rate": 1.3417903828364386e-07,
      "loss": 0.0309,
      "step": 19980
    },
    {
      "epoch": 4.926528599605523,
      "grad_norm": 0.2890625,
      "learning_rate": 1.3240224437129912e-07,
      "loss": 0.0336,
      "step": 19982
    },
    {
      "epoch": 4.927021696252465,
      "grad_norm": 0.279296875,
      "learning_rate": 1.3063728532686227e-07,
      "loss": 0.0326,
      "step": 19984
    },
    {
      "epoch": 4.927514792899409,
      "grad_norm": 0.28125,
      "learning_rate": 1.2888416135951044e-07,
      "loss": 0.0295,
      "step": 19986
    },
    {
      "epoch": 4.928007889546351,
      "grad_norm": 0.337890625,
      "learning_rate": 1.2714287267699963e-07,
      "loss": 0.0328,
      "step": 19988
    },
    {
      "epoch": 4.928500986193294,
      "grad_norm": 0.30078125,
      "learning_rate": 1.2541341948567598e-07,
      "loss": 0.0323,
      "step": 19990
    },
    {
      "epoch": 4.928994082840236,
      "grad_norm": 0.294921875,
      "learning_rate": 1.2369580199048657e-07,
      "loss": 0.0338,
      "step": 19992
    },
    {
      "epoch": 4.92948717948718,
      "grad_norm": 0.28125,
      "learning_rate": 1.2199002039497976e-07,
      "loss": 0.0292,
      "step": 19994
    },
    {
      "epoch": 4.929980276134122,
      "grad_norm": 0.337890625,
      "learning_rate": 1.20296074901316e-07,
      "loss": 0.0305,
      "step": 19996
    },
    {
      "epoch": 4.930473372781065,
      "grad_norm": 0.26171875,
      "learning_rate": 1.1861396571021255e-07,
      "loss": 0.0319,
      "step": 19998
    },
    {
      "epoch": 4.930966469428008,
      "grad_norm": 0.283203125,
      "learning_rate": 1.1694369302102104e-07,
      "loss": 0.0306,
      "step": 20000
    },
    {
      "epoch": 4.931459566074951,
      "grad_norm": 0.337890625,
      "learning_rate": 1.1528525703168313e-07,
      "loss": 0.0363,
      "step": 20002
    },
    {
      "epoch": 4.931952662721893,
      "grad_norm": 0.29296875,
      "learning_rate": 1.1363865793871942e-07,
      "loss": 0.0327,
      "step": 20004
    },
    {
      "epoch": 4.932445759368837,
      "grad_norm": 0.28515625,
      "learning_rate": 1.1200389593727378e-07,
      "loss": 0.0314,
      "step": 20006
    },
    {
      "epoch": 4.932938856015779,
      "grad_norm": 0.291015625,
      "learning_rate": 1.1038097122106905e-07,
      "loss": 0.0324,
      "step": 20008
    },
    {
      "epoch": 4.933431952662722,
      "grad_norm": 0.287109375,
      "learning_rate": 1.0876988398242915e-07,
      "loss": 0.0314,
      "step": 20010
    },
    {
      "epoch": 4.933925049309664,
      "grad_norm": 0.267578125,
      "learning_rate": 1.0717063441229025e-07,
      "loss": 0.0322,
      "step": 20012
    },
    {
      "epoch": 4.934418145956608,
      "grad_norm": 0.275390625,
      "learning_rate": 1.055832227001452e-07,
      "loss": 0.0327,
      "step": 20014
    },
    {
      "epoch": 4.93491124260355,
      "grad_norm": 0.279296875,
      "learning_rate": 1.0400764903414351e-07,
      "loss": 0.0357,
      "step": 20016
    },
    {
      "epoch": 4.935404339250493,
      "grad_norm": 0.380859375,
      "learning_rate": 1.0244391360096916e-07,
      "loss": 0.0313,
      "step": 20018
    },
    {
      "epoch": 4.935897435897436,
      "grad_norm": 0.291015625,
      "learning_rate": 1.0089201658595171e-07,
      "loss": 0.0305,
      "step": 20020
    },
    {
      "epoch": 4.936390532544379,
      "grad_norm": 0.34765625,
      "learning_rate": 9.935195817299958e-08,
      "loss": 0.0343,
      "step": 20022
    },
    {
      "epoch": 4.936883629191321,
      "grad_norm": 0.263671875,
      "learning_rate": 9.782373854461125e-08,
      "loss": 0.0314,
      "step": 20024
    },
    {
      "epoch": 4.937376725838265,
      "grad_norm": 0.3125,
      "learning_rate": 9.630735788188628e-08,
      "loss": 0.0328,
      "step": 20026
    },
    {
      "epoch": 4.937869822485207,
      "grad_norm": 0.267578125,
      "learning_rate": 9.480281636453647e-08,
      "loss": 0.0282,
      "step": 20028
    },
    {
      "epoch": 4.93836291913215,
      "grad_norm": 0.26953125,
      "learning_rate": 9.331011417085256e-08,
      "loss": 0.0301,
      "step": 20030
    },
    {
      "epoch": 4.938856015779093,
      "grad_norm": 0.291015625,
      "learning_rate": 9.182925147772636e-08,
      "loss": 0.0357,
      "step": 20032
    },
    {
      "epoch": 4.939349112426036,
      "grad_norm": 0.28125,
      "learning_rate": 9.036022846065084e-08,
      "loss": 0.0304,
      "step": 20034
    },
    {
      "epoch": 4.939842209072978,
      "grad_norm": 0.26171875,
      "learning_rate": 8.890304529370897e-08,
      "loss": 0.0318,
      "step": 20036
    },
    {
      "epoch": 4.9403353057199215,
      "grad_norm": 0.302734375,
      "learning_rate": 8.745770214959592e-08,
      "loss": 0.0326,
      "step": 20038
    },
    {
      "epoch": 4.940828402366864,
      "grad_norm": 0.2890625,
      "learning_rate": 8.602419919957472e-08,
      "loss": 0.0317,
      "step": 20040
    },
    {
      "epoch": 4.941321499013807,
      "grad_norm": 0.3359375,
      "learning_rate": 8.460253661354279e-08,
      "loss": 0.0317,
      "step": 20042
    },
    {
      "epoch": 4.94181459566075,
      "grad_norm": 0.263671875,
      "learning_rate": 8.319271455996536e-08,
      "loss": 0.0304,
      "step": 20044
    },
    {
      "epoch": 4.9423076923076925,
      "grad_norm": 0.296875,
      "learning_rate": 8.17947332059088e-08,
      "loss": 0.0324,
      "step": 20046
    },
    {
      "epoch": 4.942800788954635,
      "grad_norm": 0.31640625,
      "learning_rate": 8.04085927170628e-08,
      "loss": 0.0339,
      "step": 20048
    },
    {
      "epoch": 4.943293885601578,
      "grad_norm": 0.328125,
      "learning_rate": 7.903429325766265e-08,
      "loss": 0.0316,
      "step": 20050
    },
    {
      "epoch": 4.943786982248521,
      "grad_norm": 0.28515625,
      "learning_rate": 7.767183499058917e-08,
      "loss": 0.0337,
      "step": 20052
    },
    {
      "epoch": 4.9442800788954635,
      "grad_norm": 0.287109375,
      "learning_rate": 7.632121807730208e-08,
      "loss": 0.0305,
      "step": 20054
    },
    {
      "epoch": 4.944773175542406,
      "grad_norm": 0.318359375,
      "learning_rate": 7.498244267785114e-08,
      "loss": 0.0343,
      "step": 20056
    },
    {
      "epoch": 4.945266272189349,
      "grad_norm": 0.275390625,
      "learning_rate": 7.365550895088724e-08,
      "loss": 0.0305,
      "step": 20058
    },
    {
      "epoch": 4.945759368836292,
      "grad_norm": 0.28125,
      "learning_rate": 7.234041705367345e-08,
      "loss": 0.0335,
      "step": 20060
    },
    {
      "epoch": 4.9462524654832345,
      "grad_norm": 0.30859375,
      "learning_rate": 7.103716714204068e-08,
      "loss": 0.0312,
      "step": 20062
    },
    {
      "epoch": 4.946745562130177,
      "grad_norm": 0.279296875,
      "learning_rate": 6.974575937042094e-08,
      "loss": 0.0316,
      "step": 20064
    },
    {
      "epoch": 4.94723865877712,
      "grad_norm": 0.302734375,
      "learning_rate": 6.846619389188069e-08,
      "loss": 0.0328,
      "step": 20066
    },
    {
      "epoch": 4.947731755424063,
      "grad_norm": 0.267578125,
      "learning_rate": 6.719847085804309e-08,
      "loss": 0.0296,
      "step": 20068
    },
    {
      "epoch": 4.9482248520710055,
      "grad_norm": 0.359375,
      "learning_rate": 6.594259041914353e-08,
      "loss": 0.0376,
      "step": 20070
    },
    {
      "epoch": 4.948717948717949,
      "grad_norm": 0.26953125,
      "learning_rate": 6.469855272399628e-08,
      "loss": 0.031,
      "step": 20072
    },
    {
      "epoch": 4.949211045364891,
      "grad_norm": 0.337890625,
      "learning_rate": 6.346635792005006e-08,
      "loss": 0.0335,
      "step": 20074
    },
    {
      "epoch": 4.949704142011834,
      "grad_norm": 0.291015625,
      "learning_rate": 6.224600615331034e-08,
      "loss": 0.0323,
      "step": 20076
    },
    {
      "epoch": 4.950197238658777,
      "grad_norm": 0.318359375,
      "learning_rate": 6.103749756839472e-08,
      "loss": 0.0331,
      "step": 20078
    },
    {
      "epoch": 4.95069033530572,
      "grad_norm": 0.3046875,
      "learning_rate": 5.9840832308522e-08,
      "loss": 0.0344,
      "step": 20080
    },
    {
      "epoch": 4.951183431952662,
      "grad_norm": 0.328125,
      "learning_rate": 5.865601051551206e-08,
      "loss": 0.0342,
      "step": 20082
    },
    {
      "epoch": 4.951676528599606,
      "grad_norm": 0.326171875,
      "learning_rate": 5.748303232976371e-08,
      "loss": 0.0341,
      "step": 20084
    },
    {
      "epoch": 4.952169625246548,
      "grad_norm": 0.306640625,
      "learning_rate": 5.632189789027687e-08,
      "loss": 0.0322,
      "step": 20086
    },
    {
      "epoch": 4.952662721893491,
      "grad_norm": 0.322265625,
      "learning_rate": 5.517260733467477e-08,
      "loss": 0.036,
      "step": 20088
    },
    {
      "epoch": 4.953155818540434,
      "grad_norm": 0.2890625,
      "learning_rate": 5.40351607991263e-08,
      "loss": 0.0316,
      "step": 20090
    },
    {
      "epoch": 4.953648915187377,
      "grad_norm": 0.31640625,
      "learning_rate": 5.29095584184347e-08,
      "loss": 0.0336,
      "step": 20092
    },
    {
      "epoch": 4.954142011834319,
      "grad_norm": 0.296875,
      "learning_rate": 5.17958003260044e-08,
      "loss": 0.0281,
      "step": 20094
    },
    {
      "epoch": 4.954635108481263,
      "grad_norm": 0.287109375,
      "learning_rate": 5.069388665380759e-08,
      "loss": 0.031,
      "step": 20096
    },
    {
      "epoch": 4.955128205128205,
      "grad_norm": 0.359375,
      "learning_rate": 4.960381753242871e-08,
      "loss": 0.0331,
      "step": 20098
    },
    {
      "epoch": 4.955621301775148,
      "grad_norm": 0.28515625,
      "learning_rate": 4.852559309105331e-08,
      "loss": 0.0305,
      "step": 20100
    },
    {
      "epoch": 4.956114398422091,
      "grad_norm": 0.328125,
      "learning_rate": 4.745921345744586e-08,
      "loss": 0.0279,
      "step": 20102
    },
    {
      "epoch": 4.956607495069034,
      "grad_norm": 0.3671875,
      "learning_rate": 4.640467875799415e-08,
      "loss": 0.032,
      "step": 20104
    },
    {
      "epoch": 4.957100591715976,
      "grad_norm": 0.291015625,
      "learning_rate": 4.536198911765377e-08,
      "loss": 0.0344,
      "step": 20106
    },
    {
      "epoch": 4.957593688362919,
      "grad_norm": 0.3046875,
      "learning_rate": 4.433114465999255e-08,
      "loss": 0.0296,
      "step": 20108
    },
    {
      "epoch": 4.958086785009862,
      "grad_norm": 0.294921875,
      "learning_rate": 4.3312145507168337e-08,
      "loss": 0.0327,
      "step": 20110
    },
    {
      "epoch": 4.958579881656805,
      "grad_norm": 0.365234375,
      "learning_rate": 4.230499177994007e-08,
      "loss": 0.0393,
      "step": 20112
    },
    {
      "epoch": 4.959072978303747,
      "grad_norm": 0.27734375,
      "learning_rate": 4.130968359766785e-08,
      "loss": 0.032,
      "step": 20114
    },
    {
      "epoch": 4.959566074950691,
      "grad_norm": 0.28515625,
      "learning_rate": 4.032622107829065e-08,
      "loss": 0.0331,
      "step": 20116
    },
    {
      "epoch": 4.960059171597633,
      "grad_norm": 0.302734375,
      "learning_rate": 3.935460433837079e-08,
      "loss": 0.036,
      "step": 20118
    },
    {
      "epoch": 4.960552268244576,
      "grad_norm": 0.42578125,
      "learning_rate": 3.83948334930273e-08,
      "loss": 0.0348,
      "step": 20120
    },
    {
      "epoch": 4.961045364891518,
      "grad_norm": 0.328125,
      "learning_rate": 3.744690865601363e-08,
      "loss": 0.0336,
      "step": 20122
    },
    {
      "epoch": 4.961538461538462,
      "grad_norm": 0.3125,
      "learning_rate": 3.651082993966215e-08,
      "loss": 0.0345,
      "step": 20124
    },
    {
      "epoch": 4.962031558185404,
      "grad_norm": 0.30859375,
      "learning_rate": 3.558659745490634e-08,
      "loss": 0.0324,
      "step": 20126
    },
    {
      "epoch": 4.962524654832347,
      "grad_norm": 0.310546875,
      "learning_rate": 3.467421131125859e-08,
      "loss": 0.0356,
      "step": 20128
    },
    {
      "epoch": 4.96301775147929,
      "grad_norm": 0.310546875,
      "learning_rate": 3.377367161686573e-08,
      "loss": 0.0326,
      "step": 20130
    },
    {
      "epoch": 4.963510848126233,
      "grad_norm": 0.330078125,
      "learning_rate": 3.28849784784202e-08,
      "loss": 0.0332,
      "step": 20132
    },
    {
      "epoch": 4.964003944773175,
      "grad_norm": 0.3828125,
      "learning_rate": 3.200813200124886e-08,
      "loss": 0.034,
      "step": 20134
    },
    {
      "epoch": 4.964497041420119,
      "grad_norm": 0.287109375,
      "learning_rate": 3.114313228926857e-08,
      "loss": 0.0308,
      "step": 20136
    },
    {
      "epoch": 4.964990138067061,
      "grad_norm": 0.28125,
      "learning_rate": 3.0289979444986236e-08,
      "loss": 0.032,
      "step": 20138
    },
    {
      "epoch": 4.965483234714004,
      "grad_norm": 0.27734375,
      "learning_rate": 2.9448673569487663e-08,
      "loss": 0.0321,
      "step": 20140
    },
    {
      "epoch": 4.965976331360947,
      "grad_norm": 0.326171875,
      "learning_rate": 2.8619214762504176e-08,
      "loss": 0.0328,
      "step": 20142
    },
    {
      "epoch": 4.96646942800789,
      "grad_norm": 0.259765625,
      "learning_rate": 2.7801603122301624e-08,
      "loss": 0.0318,
      "step": 20144
    },
    {
      "epoch": 4.966962524654832,
      "grad_norm": 0.28515625,
      "learning_rate": 2.6995838745780267e-08,
      "loss": 0.0296,
      "step": 20146
    },
    {
      "epoch": 4.9674556213017755,
      "grad_norm": 0.283203125,
      "learning_rate": 2.620192172843039e-08,
      "loss": 0.0311,
      "step": 20148
    },
    {
      "epoch": 4.967948717948718,
      "grad_norm": 0.279296875,
      "learning_rate": 2.541985216433229e-08,
      "loss": 0.0303,
      "step": 20150
    },
    {
      "epoch": 4.968441814595661,
      "grad_norm": 0.26953125,
      "learning_rate": 2.4649630146178494e-08,
      "loss": 0.0285,
      "step": 20152
    },
    {
      "epoch": 4.968934911242604,
      "grad_norm": 0.30078125,
      "learning_rate": 2.3891255765229325e-08,
      "loss": 0.0332,
      "step": 20154
    },
    {
      "epoch": 4.9694280078895465,
      "grad_norm": 0.328125,
      "learning_rate": 2.3144729111357343e-08,
      "loss": 0.0381,
      "step": 20156
    },
    {
      "epoch": 4.969921104536489,
      "grad_norm": 0.275390625,
      "learning_rate": 2.241005027303622e-08,
      "loss": 0.0312,
      "step": 20158
    },
    {
      "epoch": 4.970414201183432,
      "grad_norm": 0.2890625,
      "learning_rate": 2.1687219337329645e-08,
      "loss": 0.0299,
      "step": 20160
    },
    {
      "epoch": 4.970907297830375,
      "grad_norm": 0.283203125,
      "learning_rate": 2.0976236389891325e-08,
      "loss": 0.0288,
      "step": 20162
    },
    {
      "epoch": 4.9714003944773175,
      "grad_norm": 0.31640625,
      "learning_rate": 2.0277101514987184e-08,
      "loss": 0.0324,
      "step": 20164
    },
    {
      "epoch": 4.97189349112426,
      "grad_norm": 0.28125,
      "learning_rate": 1.958981479546207e-08,
      "loss": 0.0291,
      "step": 20166
    },
    {
      "epoch": 4.9723865877712035,
      "grad_norm": 0.3046875,
      "learning_rate": 1.8914376312750836e-08,
      "loss": 0.0316,
      "step": 20168
    },
    {
      "epoch": 4.972879684418146,
      "grad_norm": 0.2578125,
      "learning_rate": 1.8250786146922772e-08,
      "loss": 0.0284,
      "step": 20170
    },
    {
      "epoch": 4.9733727810650885,
      "grad_norm": 0.353515625,
      "learning_rate": 1.7599044376592766e-08,
      "loss": 0.0331,
      "step": 20172
    },
    {
      "epoch": 4.973865877712032,
      "grad_norm": 0.25390625,
      "learning_rate": 1.695915107901014e-08,
      "loss": 0.0318,
      "step": 20174
    },
    {
      "epoch": 4.9743589743589745,
      "grad_norm": 0.29296875,
      "learning_rate": 1.633110633000312e-08,
      "loss": 0.0331,
      "step": 20176
    },
    {
      "epoch": 4.974852071005917,
      "grad_norm": 0.291015625,
      "learning_rate": 1.5714910204001066e-08,
      "loss": 0.0299,
      "step": 20178
    },
    {
      "epoch": 4.9753451676528595,
      "grad_norm": 0.28125,
      "learning_rate": 1.511056277402334e-08,
      "loss": 0.0299,
      "step": 20180
    },
    {
      "epoch": 4.975838264299803,
      "grad_norm": 0.265625,
      "learning_rate": 1.451806411167933e-08,
      "loss": 0.0326,
      "step": 20182
    },
    {
      "epoch": 4.976331360946745,
      "grad_norm": 0.267578125,
      "learning_rate": 1.3937414287201745e-08,
      "loss": 0.032,
      "step": 20184
    },
    {
      "epoch": 4.976824457593688,
      "grad_norm": 0.314453125,
      "learning_rate": 1.336861336938e-08,
      "loss": 0.0285,
      "step": 20186
    },
    {
      "epoch": 4.977317554240631,
      "grad_norm": 0.306640625,
      "learning_rate": 1.2811661425637944e-08,
      "loss": 0.0306,
      "step": 20188
    },
    {
      "epoch": 4.977810650887574,
      "grad_norm": 0.26953125,
      "learning_rate": 1.226655852196723e-08,
      "loss": 0.0296,
      "step": 20190
    },
    {
      "epoch": 4.978303747534516,
      "grad_norm": 0.2578125,
      "learning_rate": 1.1733304722971738e-08,
      "loss": 0.028,
      "step": 20192
    },
    {
      "epoch": 4.97879684418146,
      "grad_norm": 0.314453125,
      "learning_rate": 1.1211900091845362e-08,
      "loss": 0.0331,
      "step": 20194
    },
    {
      "epoch": 4.979289940828402,
      "grad_norm": 0.28125,
      "learning_rate": 1.0702344690372013e-08,
      "loss": 0.0311,
      "step": 20196
    },
    {
      "epoch": 4.979783037475345,
      "grad_norm": 0.283203125,
      "learning_rate": 1.0204638578947823e-08,
      "loss": 0.0318,
      "step": 20198
    },
    {
      "epoch": 4.980276134122288,
      "grad_norm": 0.267578125,
      "learning_rate": 9.718781816536738e-09,
      "loss": 0.0317,
      "step": 20200
    },
    {
      "epoch": 4.980769230769231,
      "grad_norm": 0.28515625,
      "learning_rate": 9.244774460726024e-09,
      "loss": 0.033,
      "step": 20202
    },
    {
      "epoch": 4.981262327416173,
      "grad_norm": 0.267578125,
      "learning_rate": 8.782616567681867e-09,
      "loss": 0.03,
      "step": 20204
    },
    {
      "epoch": 4.981755424063117,
      "grad_norm": 0.32421875,
      "learning_rate": 8.332308192182669e-09,
      "loss": 0.0308,
      "step": 20206
    },
    {
      "epoch": 4.982248520710059,
      "grad_norm": 0.341796875,
      "learning_rate": 7.89384938758575e-09,
      "loss": 0.0348,
      "step": 20208
    },
    {
      "epoch": 4.982741617357002,
      "grad_norm": 0.27734375,
      "learning_rate": 7.467240205849545e-09,
      "loss": 0.0307,
      "step": 20210
    },
    {
      "epoch": 4.983234714003945,
      "grad_norm": 0.271484375,
      "learning_rate": 7.052480697533614e-09,
      "loss": 0.0342,
      "step": 20212
    },
    {
      "epoch": 4.983727810650888,
      "grad_norm": 0.279296875,
      "learning_rate": 6.6495709117875286e-09,
      "loss": 0.0331,
      "step": 20214
    },
    {
      "epoch": 4.98422090729783,
      "grad_norm": 0.328125,
      "learning_rate": 6.25851089636198e-09,
      "loss": 0.0365,
      "step": 20216
    },
    {
      "epoch": 4.984714003944773,
      "grad_norm": 0.337890625,
      "learning_rate": 5.879300697597678e-09,
      "loss": 0.034,
      "step": 20218
    },
    {
      "epoch": 4.985207100591716,
      "grad_norm": 0.296875,
      "learning_rate": 5.511940360425349e-09,
      "loss": 0.0324,
      "step": 20220
    },
    {
      "epoch": 4.985700197238659,
      "grad_norm": 0.302734375,
      "learning_rate": 5.15642992838794e-09,
      "loss": 0.031,
      "step": 20222
    },
    {
      "epoch": 4.986193293885601,
      "grad_norm": 0.2734375,
      "learning_rate": 4.812769443618415e-09,
      "loss": 0.0297,
      "step": 20224
    },
    {
      "epoch": 4.986686390532545,
      "grad_norm": 0.287109375,
      "learning_rate": 4.480958946828651e-09,
      "loss": 0.0266,
      "step": 20226
    },
    {
      "epoch": 4.987179487179487,
      "grad_norm": 0.3125,
      "learning_rate": 4.16099847735385e-09,
      "loss": 0.0331,
      "step": 20228
    },
    {
      "epoch": 4.98767258382643,
      "grad_norm": 0.400390625,
      "learning_rate": 3.85288807310813e-09,
      "loss": 0.0274,
      "step": 20230
    },
    {
      "epoch": 4.988165680473373,
      "grad_norm": 0.267578125,
      "learning_rate": 3.556627770595622e-09,
      "loss": 0.0286,
      "step": 20232
    },
    {
      "epoch": 4.988658777120316,
      "grad_norm": 0.283203125,
      "learning_rate": 3.2722176049437836e-09,
      "loss": 0.0333,
      "step": 20234
    },
    {
      "epoch": 4.989151873767258,
      "grad_norm": 0.287109375,
      "learning_rate": 2.999657609825679e-09,
      "loss": 0.0298,
      "step": 20236
    },
    {
      "epoch": 4.989644970414201,
      "grad_norm": 0.287109375,
      "learning_rate": 2.7389478175710025e-09,
      "loss": 0.033,
      "step": 20238
    },
    {
      "epoch": 4.990138067061144,
      "grad_norm": 0.2890625,
      "learning_rate": 2.490088259066159e-09,
      "loss": 0.0304,
      "step": 20240
    },
    {
      "epoch": 4.990631163708087,
      "grad_norm": 0.326171875,
      "learning_rate": 2.2530789637986716e-09,
      "loss": 0.0329,
      "step": 20242
    },
    {
      "epoch": 4.991124260355029,
      "grad_norm": 0.298828125,
      "learning_rate": 2.027919959857183e-09,
      "loss": 0.0323,
      "step": 20244
    },
    {
      "epoch": 4.991617357001973,
      "grad_norm": 0.27734375,
      "learning_rate": 1.8146112739203525e-09,
      "loss": 0.0306,
      "step": 20246
    },
    {
      "epoch": 4.992110453648915,
      "grad_norm": 0.33984375,
      "learning_rate": 1.6131529312790605e-09,
      "loss": 0.0299,
      "step": 20248
    },
    {
      "epoch": 4.992603550295858,
      "grad_norm": 0.3125,
      "learning_rate": 1.423544955792e-09,
      "loss": 0.0301,
      "step": 20250
    },
    {
      "epoch": 4.993096646942801,
      "grad_norm": 0.283203125,
      "learning_rate": 1.2457873699300848e-09,
      "loss": 0.0318,
      "step": 20252
    },
    {
      "epoch": 4.993589743589744,
      "grad_norm": 0.310546875,
      "learning_rate": 1.0798801947764502e-09,
      "loss": 0.0346,
      "step": 20254
    },
    {
      "epoch": 4.994082840236686,
      "grad_norm": 0.30859375,
      "learning_rate": 9.258234499709417e-10,
      "loss": 0.0335,
      "step": 20256
    },
    {
      "epoch": 4.9945759368836296,
      "grad_norm": 0.326171875,
      "learning_rate": 7.836171537878301e-10,
      "loss": 0.029,
      "step": 20258
    },
    {
      "epoch": 4.995069033530572,
      "grad_norm": 0.33984375,
      "learning_rate": 6.532613230580964e-10,
      "loss": 0.0346,
      "step": 20260
    },
    {
      "epoch": 4.995562130177515,
      "grad_norm": 0.373046875,
      "learning_rate": 5.347559732471474e-10,
      "loss": 0.0307,
      "step": 20262
    },
    {
      "epoch": 4.996055226824458,
      "grad_norm": 0.287109375,
      "learning_rate": 4.2810111839930447e-10,
      "loss": 0.0305,
      "step": 20264
    },
    {
      "epoch": 4.9965483234714005,
      "grad_norm": 0.296875,
      "learning_rate": 3.3329677113780323e-10,
      "loss": 0.0321,
      "step": 20266
    },
    {
      "epoch": 4.997041420118343,
      "grad_norm": 0.306640625,
      "learning_rate": 2.50342942709203e-10,
      "loss": 0.0318,
      "step": 20268
    },
    {
      "epoch": 4.9975345167652865,
      "grad_norm": 0.275390625,
      "learning_rate": 1.7923964293897754e-10,
      "loss": 0.0316,
      "step": 20270
    },
    {
      "epoch": 4.998027613412229,
      "grad_norm": 0.2890625,
      "learning_rate": 1.1998688026482185e-10,
      "loss": 0.0336,
      "step": 20272
    },
    {
      "epoch": 4.9985207100591715,
      "grad_norm": 0.349609375,
      "learning_rate": 7.258466170334544e-11,
      "loss": 0.0354,
      "step": 20274
    },
    {
      "epoch": 4.999013806706114,
      "grad_norm": 0.27734375,
      "learning_rate": 3.7032992861174565e-11,
      "loss": 0.0314,
      "step": 20276
    },
    {
      "epoch": 4.9995069033530575,
      "grad_norm": 0.26953125,
      "learning_rate": 1.3331877957156735e-11,
      "loss": 0.0323,
      "step": 20278
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.34765625,
      "learning_rate": 1.4813198001562e-12,
      "loss": 0.0257,
      "step": 20280
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.060740407556295395,
      "eval_runtime": 69.8761,
      "eval_samples_per_second": 227.961,
      "eval_steps_per_second": 1.789,
      "step": 20280
    },
    {
      "epoch": 5.0,
      "step": 20280,
      "total_flos": 7.48763679670272e+20,
      "train_loss": 0.06658803844747282,
      "train_runtime": 14605.9054,
      "train_samples_per_second": 177.725,
      "train_steps_per_second": 1.388
    }
  ],
  "logging_steps": 2,
  "max_steps": 20280,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 4056,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.48763679670272e+20,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}