{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9999335474581903,
  "eval_steps": 500,
  "global_step": 16929,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 7.78125,
      "learning_rate": 1.5e-05,
      "loss": 1.4908,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.0625,
      "learning_rate": 3e-05,
      "loss": 1.5469,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.625,
      "learning_rate": 4.5e-05,
      "loss": 1.4631,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.375,
      "learning_rate": 6e-05,
      "loss": 1.4561,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.21875,
      "learning_rate": 5.99999870552063e-05,
      "loss": 1.452,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.125,
      "learning_rate": 5.999994822083636e-05,
      "loss": 1.4263,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6875,
      "learning_rate": 5.9999883496923705e-05,
      "loss": 1.4167,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5625,
      "learning_rate": 5.9999792883524176e-05,
      "loss": 1.4179,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.375,
      "learning_rate": 5.999967638071599e-05,
      "loss": 1.3843,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.84375,
      "learning_rate": 5.9999533988599675e-05,
      "loss": 1.4072,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6875,
      "learning_rate": 5.999936570729811e-05,
      "loss": 1.3193,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.75,
      "learning_rate": 5.999917153695653e-05,
      "loss": 1.3554,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.15625,
      "learning_rate": 5.999895147774249e-05,
      "loss": 1.3453,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6875,
      "learning_rate": 5.9998705529845914e-05,
      "loss": 1.3736,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.953125,
      "learning_rate": 5.9998433693479035e-05,
      "loss": 1.3863,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.84375,
      "learning_rate": 5.9998135968876455e-05,
      "loss": 1.313,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.15625,
      "learning_rate": 5.999781235629511e-05,
      "loss": 1.4022,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.84375,
      "learning_rate": 5.999746285601425e-05,
      "loss": 1.4179,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.84375,
      "learning_rate": 5.9997087468335514e-05,
      "loss": 1.4068,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.65625,
      "learning_rate": 5.999668619358285e-05,
      "loss": 1.4107,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.375,
      "learning_rate": 5.999625903210254e-05,
      "loss": 1.355,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.53125,
      "learning_rate": 5.9995805984263235e-05,
      "loss": 1.4045,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.09375,
      "learning_rate": 5.99953270504559e-05,
      "loss": 1.335,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.125,
      "learning_rate": 5.9994822231093855e-05,
      "loss": 1.3568,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5625,
      "learning_rate": 5.999429152661274e-05,
      "loss": 1.3344,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3125,
      "learning_rate": 5.9993734937470554e-05,
      "loss": 1.3172,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5625,
      "learning_rate": 5.999315246414764e-05,
      "loss": 1.3562,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.125,
      "learning_rate": 5.9992544107146624e-05,
      "loss": 1.3601,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.4375,
      "learning_rate": 5.9991909866992556e-05,
      "loss": 1.3281,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.0625,
      "learning_rate": 5.999124974423274e-05,
      "loss": 1.4124,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.0625,
      "learning_rate": 5.999056373943688e-05,
      "loss": 1.3428,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.4375,
      "learning_rate": 5.9989851853196974e-05,
      "loss": 1.3556,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.75,
      "learning_rate": 5.9989114086127384e-05,
      "loss": 1.4002,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.0,
      "learning_rate": 5.998835043886477e-05,
      "loss": 1.4018,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.9375,
      "learning_rate": 5.998756091206816e-05,
      "loss": 1.3767,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.28125,
      "learning_rate": 5.998674550641891e-05,
      "loss": 1.4117,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.4375,
      "learning_rate": 5.9985904222620705e-05,
      "loss": 1.4244,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.75,
      "learning_rate": 5.9985037061399545e-05,
      "loss": 1.2954,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.875,
      "learning_rate": 5.9984144023503796e-05,
      "loss": 1.3381,
      "step": 195
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.53125,
      "learning_rate": 5.9983225109704137e-05,
      "loss": 1.3754,
      "step": 200
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.65625,
      "learning_rate": 5.9982280320793565e-05,
      "loss": 1.3988,
      "step": 205
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.21875,
      "learning_rate": 5.998130965758742e-05,
      "loss": 1.3583,
      "step": 210
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.71875,
      "learning_rate": 5.9980313120923394e-05,
      "loss": 1.3655,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.875,
      "learning_rate": 5.997929071166146e-05,
      "loss": 1.3826,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.625,
      "learning_rate": 5.997824243068395e-05,
      "loss": 1.365,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.125,
      "learning_rate": 5.9977168278895516e-05,
      "loss": 1.3666,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.65625,
      "learning_rate": 5.9976068257223146e-05,
      "loss": 1.3921,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.53125,
      "learning_rate": 5.9974942366616116e-05,
      "loss": 1.3138,
      "step": 240
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.03125,
      "learning_rate": 5.9973790608046095e-05,
      "loss": 1.3864,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.21875,
      "learning_rate": 5.9972612982507006e-05,
      "loss": 1.2755,
      "step": 250
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.9375,
      "learning_rate": 5.997140949101513e-05,
      "loss": 1.3516,
      "step": 255
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.78125,
      "learning_rate": 5.997018013460906e-05,
      "loss": 1.3214,
      "step": 260
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.75,
      "learning_rate": 5.996892491434972e-05,
      "loss": 1.3606,
      "step": 265
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.34375,
      "learning_rate": 5.9967643831320346e-05,
      "loss": 1.357,
      "step": 270
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.46875,
      "learning_rate": 5.99663368866265e-05,
      "loss": 1.4333,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.46875,
      "learning_rate": 5.996500408139605e-05,
      "loss": 1.4344,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.8125,
      "learning_rate": 5.9963645416779187e-05,
      "loss": 1.3336,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.0625,
      "learning_rate": 5.9962260893948426e-05,
      "loss": 1.2713,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.8125,
      "learning_rate": 5.99608505140986e-05,
      "loss": 1.407,
      "step": 295
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.96875,
      "learning_rate": 5.995941427844681e-05,
      "loss": 1.3888,
      "step": 300
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.03125,
      "learning_rate": 5.9957952188232554e-05,
      "loss": 1.3372,
      "step": 305
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.625,
      "learning_rate": 5.9956464244717565e-05,
      "loss": 1.3894,
      "step": 310
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.34375,
      "learning_rate": 5.995495044918594e-05,
      "loss": 1.35,
      "step": 315
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.53125,
      "learning_rate": 5.995341080294405e-05,
      "loss": 1.2879,
      "step": 320
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.53125,
      "learning_rate": 5.995184530732058e-05,
      "loss": 1.3465,
      "step": 325
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.8125,
      "learning_rate": 5.995025396366655e-05,
      "loss": 1.3824,
      "step": 330
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.625,
      "learning_rate": 5.994863677335526e-05,
      "loss": 1.3866,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.09375,
      "learning_rate": 5.994699373778232e-05,
      "loss": 1.3199,
      "step": 340
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.59375,
      "learning_rate": 5.994532485836565e-05,
      "loss": 1.309,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.5625,
      "learning_rate": 5.994363013654547e-05,
      "loss": 1.353,
      "step": 350
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.65625,
      "learning_rate": 5.99419095737843e-05,
      "loss": 1.3219,
      "step": 355
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.65625,
      "learning_rate": 5.994016317156696e-05,
      "loss": 1.3854,
      "step": 360
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.03125,
      "learning_rate": 5.993839093140058e-05,
      "loss": 1.4163,
      "step": 365
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.4375,
      "learning_rate": 5.993659285481457e-05,
      "loss": 1.3182,
      "step": 370
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.78125,
      "learning_rate": 5.993476894336065e-05,
      "loss": 1.3262,
      "step": 375
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.0,
      "learning_rate": 5.9932919198612836e-05,
      "loss": 1.2898,
      "step": 380
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.4375,
      "learning_rate": 5.9931043622167415e-05,
      "loss": 1.2814,
      "step": 385
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.1875,
      "learning_rate": 5.9929142215643e-05,
      "loss": 1.3117,
      "step": 390
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.375,
      "learning_rate": 5.9927214980680465e-05,
      "loss": 1.2959,
      "step": 395
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.5,
      "learning_rate": 5.9925261918943e-05,
      "loss": 1.4071,
      "step": 400
    },
    {
      "epoch": 0.07,
      "grad_norm": 5.21875,
      "learning_rate": 5.992328303211607e-05,
      "loss": 1.3745,
      "step": 405
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.0625,
      "learning_rate": 5.9921278321907416e-05,
      "loss": 1.3082,
      "step": 410
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.1875,
      "learning_rate": 5.991924779004709e-05,
      "loss": 1.3505,
      "step": 415
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.5,
      "learning_rate": 5.9917191438287385e-05,
      "loss": 1.3175,
      "step": 420
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.375,
      "learning_rate": 5.9915109268402945e-05,
      "loss": 1.2956,
      "step": 425
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.5,
      "learning_rate": 5.991300128219062e-05,
      "loss": 1.3669,
      "step": 430
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.96875,
      "learning_rate": 5.99108674814696e-05,
      "loss": 1.3332,
      "step": 435
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.125,
      "learning_rate": 5.99087078680813e-05,
      "loss": 1.3186,
      "step": 440
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.84375,
      "learning_rate": 5.990652244388945e-05,
      "loss": 1.3829,
      "step": 445
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.96875,
      "learning_rate": 5.9904311210780055e-05,
      "loss": 1.3311,
      "step": 450
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.53125,
      "learning_rate": 5.990207417066134e-05,
      "loss": 1.2819,
      "step": 455
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.4375,
      "learning_rate": 5.989981132546388e-05,
      "loss": 1.3235,
      "step": 460
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.375,
      "learning_rate": 5.989752267714045e-05,
      "loss": 1.303,
      "step": 465
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.75,
      "learning_rate": 5.989520822766614e-05,
      "loss": 1.3631,
      "step": 470
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.09375,
      "learning_rate": 5.989286797903829e-05,
      "loss": 1.2994,
      "step": 475
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.34375,
      "learning_rate": 5.989050193327649e-05,
      "loss": 1.3148,
      "step": 480
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.75,
      "learning_rate": 5.9888110092422605e-05,
      "loss": 1.4402,
      "step": 485
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.3125,
      "learning_rate": 5.988569245854077e-05,
      "loss": 1.3807,
      "step": 490
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.0,
      "learning_rate": 5.988324903371736e-05,
      "loss": 1.338,
      "step": 495
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.40625,
      "learning_rate": 5.988077982006103e-05,
      "loss": 1.3091,
      "step": 500
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.875,
      "learning_rate": 5.987828481970267e-05,
      "loss": 1.2977,
      "step": 505
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.875,
      "learning_rate": 5.9875764034795424e-05,
      "loss": 1.3424,
      "step": 510
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.09375,
      "learning_rate": 5.9873217467514714e-05,
      "loss": 1.3303,
      "step": 515
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.75,
      "learning_rate": 5.987064512005817e-05,
      "loss": 1.387,
      "step": 520
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.5625,
      "learning_rate": 5.986804699464571e-05,
      "loss": 1.345,
      "step": 525
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.1875,
      "learning_rate": 5.9865423093519465e-05,
      "loss": 1.3771,
      "step": 530
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.375,
      "learning_rate": 5.986277341894384e-05,
      "loss": 1.3298,
      "step": 535
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.8125,
      "learning_rate": 5.986009797320546e-05,
      "loss": 1.2843,
      "step": 540
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.375,
      "learning_rate": 5.985739675861319e-05,
      "loss": 1.3001,
      "step": 545
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.1875,
      "learning_rate": 5.985466977749816e-05,
      "loss": 1.401,
      "step": 550
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.5625,
      "learning_rate": 5.985191703221371e-05,
      "loss": 1.3484,
      "step": 555
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.46875,
      "learning_rate": 5.9849138525135416e-05,
      "loss": 1.3024,
      "step": 560
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.25,
      "learning_rate": 5.984633425866109e-05,
      "loss": 1.4023,
      "step": 565
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.40625,
      "learning_rate": 5.9843504235210785e-05,
      "loss": 1.342,
      "step": 570
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.828125,
      "learning_rate": 5.984064845722676e-05,
      "loss": 1.3334,
      "step": 575
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.8125,
      "learning_rate": 5.9837766927173525e-05,
      "loss": 1.2903,
      "step": 580
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.65625,
      "learning_rate": 5.983485964753779e-05,
      "loss": 1.3846,
      "step": 585
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.03125,
      "learning_rate": 5.9831926620828506e-05,
      "loss": 1.3149,
      "step": 590
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.859375,
      "learning_rate": 5.9828967849576834e-05,
      "loss": 1.3479,
      "step": 595
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.5625,
      "learning_rate": 5.982598333633614e-05,
      "loss": 1.4663,
      "step": 600
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.28125,
      "learning_rate": 5.982297308368204e-05,
      "loss": 1.4407,
      "step": 605
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.15625,
      "learning_rate": 5.9819937094212325e-05,
      "loss": 1.2985,
      "step": 610
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.90625,
      "learning_rate": 5.981687537054701e-05,
      "loss": 1.3391,
      "step": 615
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.65625,
      "learning_rate": 5.981378791532833e-05,
      "loss": 1.2657,
      "step": 620
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.6875,
      "learning_rate": 5.9810674731220716e-05,
      "loss": 1.3232,
      "step": 625
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.46875,
      "learning_rate": 5.980753582091079e-05,
      "loss": 1.412,
      "step": 630
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.4375,
      "learning_rate": 5.980437118710741e-05,
      "loss": 1.2918,
      "step": 635
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.875,
      "learning_rate": 5.980118083254159e-05,
      "loss": 1.3751,
      "step": 640
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.5625,
      "learning_rate": 5.9797964759966574e-05,
      "loss": 1.3412,
      "step": 645
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.1875,
      "learning_rate": 5.9794722972157785e-05,
      "loss": 1.3715,
      "step": 650
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.625,
      "learning_rate": 5.979145547191284e-05,
      "loss": 1.2992,
      "step": 655
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.21875,
      "learning_rate": 5.978816226205155e-05,
      "loss": 1.33,
      "step": 660
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.84375,
      "learning_rate": 5.978484334541591e-05,
      "loss": 1.3168,
      "step": 665
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.34375,
      "learning_rate": 5.97814987248701e-05,
      "loss": 1.3821,
      "step": 670
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.875,
      "learning_rate": 5.977812840330049e-05,
      "loss": 1.2717,
      "step": 675
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.84375,
      "learning_rate": 5.9774732383615594e-05,
      "loss": 1.4104,
      "step": 680
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.6875,
      "learning_rate": 5.977131066874615e-05,
      "loss": 1.3501,
      "step": 685
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.25,
      "learning_rate": 5.9767863261645055e-05,
      "loss": 1.2627,
      "step": 690
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.75,
      "learning_rate": 5.976439016528736e-05,
      "loss": 1.2337,
      "step": 695
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.6875,
      "learning_rate": 5.97608913826703e-05,
      "loss": 1.4027,
      "step": 700
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.0,
      "learning_rate": 5.97573669168133e-05,
      "loss": 1.3029,
      "step": 705
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.40625,
      "learning_rate": 5.9753816770757895e-05,
      "loss": 1.3413,
      "step": 710
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.59375,
      "learning_rate": 5.9750240947567835e-05,
      "loss": 1.3196,
      "step": 715
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.78125,
      "learning_rate": 5.974663945032899e-05,
      "loss": 1.3558,
      "step": 720
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.0625,
      "learning_rate": 5.974301228214941e-05,
      "loss": 1.3365,
      "step": 725
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.5625,
      "learning_rate": 5.9739359446159284e-05,
      "loss": 1.3516,
      "step": 730
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.78125,
      "learning_rate": 5.973568094551097e-05,
      "loss": 1.3054,
      "step": 735
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.34375,
      "learning_rate": 5.9731976783378975e-05,
      "loss": 1.3778,
      "step": 740
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.0,
      "learning_rate": 5.972824696295991e-05,
      "loss": 1.2534,
      "step": 745
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.25,
      "learning_rate": 5.9724491487472566e-05,
      "loss": 1.3194,
      "step": 750
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.8125,
      "learning_rate": 5.972071036015789e-05,
      "loss": 1.2788,
      "step": 755
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.09375,
      "learning_rate": 5.9716903584278915e-05,
      "loss": 1.298,
      "step": 760
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.15625,
      "learning_rate": 5.971307116312085e-05,
      "loss": 1.4129,
      "step": 765
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.125,
      "learning_rate": 5.9709213099991016e-05,
      "loss": 1.3477,
      "step": 770
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.34375,
      "learning_rate": 5.970532939821888e-05,
      "loss": 1.3616,
      "step": 775
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.6875,
      "learning_rate": 5.970142006115601e-05,
      "loss": 1.3422,
      "step": 780
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.03125,
      "learning_rate": 5.969748509217611e-05,
      "loss": 1.29,
      "step": 785
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.71875,
      "learning_rate": 5.969352449467501e-05,
      "loss": 1.2633,
      "step": 790
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.875,
      "learning_rate": 5.9689538272070664e-05,
      "loss": 1.3293,
      "step": 795
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.1875,
      "learning_rate": 5.96855264278031e-05,
      "loss": 1.3215,
      "step": 800
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.03125,
      "learning_rate": 5.9681488965334503e-05,
      "loss": 1.398,
      "step": 805
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.4375,
      "learning_rate": 5.967742588814915e-05,
      "loss": 1.3202,
      "step": 810
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.0,
      "learning_rate": 5.9673337199753406e-05,
      "loss": 1.4,
      "step": 815
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.125,
      "learning_rate": 5.9669222903675756e-05,
      "loss": 1.337,
      "step": 820
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.90625,
      "learning_rate": 5.966508300346678e-05,
      "loss": 1.3281,
      "step": 825
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.40625,
      "learning_rate": 5.966091750269918e-05,
      "loss": 1.3512,
      "step": 830
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.46875,
      "learning_rate": 5.96567264049677e-05,
      "loss": 1.3155,
      "step": 835
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.5625,
      "learning_rate": 5.96525097138892e-05,
      "loss": 1.4283,
      "step": 840
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.1875,
      "learning_rate": 5.964826743310263e-05,
      "loss": 1.3494,
      "step": 845
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.9375,
      "learning_rate": 5.964399956626902e-05,
      "loss": 1.3876,
      "step": 850
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.21875,
      "learning_rate": 5.963970611707149e-05,
      "loss": 1.3442,
      "step": 855
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.78125,
      "learning_rate": 5.963538708921522e-05,
      "loss": 1.3384,
      "step": 860
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.09375,
      "learning_rate": 5.9631042486427474e-05,
      "loss": 1.3665,
      "step": 865
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.28125,
      "learning_rate": 5.962667231245757e-05,
      "loss": 1.2918,
      "step": 870
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.78125,
      "learning_rate": 5.962227657107694e-05,
      "loss": 1.3723,
      "step": 875
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.46875,
      "learning_rate": 5.961785526607901e-05,
      "loss": 1.3158,
      "step": 880
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.40625,
      "learning_rate": 5.9613408401279334e-05,
      "loss": 1.3147,
      "step": 885
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.53125,
      "learning_rate": 5.960893598051549e-05,
      "loss": 1.3635,
      "step": 890
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.65625,
      "learning_rate": 5.9604438007647093e-05,
      "loss": 1.3384,
      "step": 895
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.5,
      "learning_rate": 5.959991448655586e-05,
      "loss": 1.2457,
      "step": 900
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.1875,
      "learning_rate": 5.959536542114551e-05,
      "loss": 1.3478,
      "step": 905
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.125,
      "learning_rate": 5.9590790815341845e-05,
      "loss": 1.3477,
      "step": 910
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.53125,
      "learning_rate": 5.958619067309266e-05,
      "loss": 1.3543,
      "step": 915
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.875,
      "learning_rate": 5.958156499836784e-05,
      "loss": 1.2512,
      "step": 920
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.5,
      "learning_rate": 5.957691379515926e-05,
      "loss": 1.3305,
      "step": 925
    },
    {
      "epoch": 0.16,
      "grad_norm": 4.5625,
      "learning_rate": 5.9572237067480845e-05,
      "loss": 1.3761,
      "step": 930
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.25,
      "learning_rate": 5.9567534819368555e-05,
      "loss": 1.3712,
      "step": 935
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.9375,
      "learning_rate": 5.956280705488036e-05,
      "loss": 1.2846,
      "step": 940
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.28125,
      "learning_rate": 5.955805377809627e-05,
      "loss": 1.2998,
      "step": 945
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.09375,
      "learning_rate": 5.955327499311827e-05,
      "loss": 1.4294,
      "step": 950
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.90625,
      "learning_rate": 5.954847070407041e-05,
      "loss": 1.3406,
      "step": 955
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.09375,
      "learning_rate": 5.954364091509871e-05,
      "loss": 1.3176,
      "step": 960
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.3125,
      "learning_rate": 5.953878563037122e-05,
      "loss": 1.3496,
      "step": 965
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.09375,
      "learning_rate": 5.953390485407799e-05,
      "loss": 1.4221,
      "step": 970
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.09375,
      "learning_rate": 5.952899859043105e-05,
      "loss": 1.2472,
      "step": 975
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.28125,
      "learning_rate": 5.952406684366444e-05,
      "loss": 1.2978,
      "step": 980
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.0625,
      "learning_rate": 5.9519109618034204e-05,
      "loss": 1.3261,
      "step": 985
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.78125,
      "learning_rate": 5.9514126917818336e-05,
      "loss": 1.3967,
      "step": 990
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.46875,
      "learning_rate": 5.950911874731685e-05,
      "loss": 1.3451,
      "step": 995
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.40625,
      "learning_rate": 5.9504085110851734e-05,
      "loss": 1.3612,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.8125,
      "learning_rate": 5.949902601276695e-05,
      "loss": 1.294,
      "step": 1005
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.53125,
      "learning_rate": 5.94939414574284e-05,
      "loss": 1.339,
      "step": 1010
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.71875,
      "learning_rate": 5.948883144922402e-05,
      "loss": 1.2529,
      "step": 1015
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.59375,
      "learning_rate": 5.948369599256366e-05,
      "loss": 1.2141,
      "step": 1020
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.9375,
      "learning_rate": 5.947853509187916e-05,
      "loss": 1.3487,
      "step": 1025
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.34375,
      "learning_rate": 5.9473348751624284e-05,
      "loss": 1.3342,
      "step": 1030
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.9375,
      "learning_rate": 5.94681369762748e-05,
      "loss": 1.3191,
      "step": 1035
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.34375,
      "learning_rate": 5.9462899770328374e-05,
      "loss": 1.3673,
      "step": 1040
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.4375,
      "learning_rate": 5.9457637138304664e-05,
      "loss": 1.3066,
      "step": 1045
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.90625,
      "learning_rate": 5.945234908474523e-05,
      "loss": 1.3902,
      "step": 1050
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.84375,
      "learning_rate": 5.94470356142136e-05,
      "loss": 1.3194,
      "step": 1055
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.96875,
      "learning_rate": 5.944169673129522e-05,
      "loss": 1.3439,
      "step": 1060
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.125,
      "learning_rate": 5.943633244059749e-05,
      "loss": 1.366,
      "step": 1065
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.625,
      "learning_rate": 5.9430942746749696e-05,
      "loss": 1.3423,
      "step": 1070
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.65625,
      "learning_rate": 5.942552765440308e-05,
      "loss": 1.3562,
      "step": 1075
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.40625,
      "learning_rate": 5.9420087168230794e-05,
      "loss": 1.3145,
      "step": 1080
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.90625,
      "learning_rate": 5.94146212929279e-05,
      "loss": 1.3223,
      "step": 1085
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.65625,
      "learning_rate": 5.940913003321138e-05,
      "loss": 1.3113,
      "step": 1090
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.0625,
      "learning_rate": 5.9403613393820105e-05,
      "loss": 1.2579,
      "step": 1095
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.5,
      "learning_rate": 5.939807137951486e-05,
      "loss": 1.4243,
      "step": 1100
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.46875,
      "learning_rate": 5.939250399507833e-05,
      "loss": 1.3892,
      "step": 1105
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.25,
      "learning_rate": 5.938691124531509e-05,
      "loss": 1.3237,
      "step": 1110
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.78125,
      "learning_rate": 5.938129313505162e-05,
      "loss": 1.3506,
      "step": 1115
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.25,
      "learning_rate": 5.9375649669136244e-05,
      "loss": 1.2949,
      "step": 1120
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.59375,
      "learning_rate": 5.9369980852439214e-05,
      "loss": 1.2863,
      "step": 1125
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.71875,
      "learning_rate": 5.936428668985265e-05,
      "loss": 1.3694,
      "step": 1130
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.8125,
      "learning_rate": 5.935856718629051e-05,
      "loss": 1.3392,
      "step": 1135
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.15625,
      "learning_rate": 5.935282234668867e-05,
      "loss": 1.3389,
      "step": 1140
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.90625,
      "learning_rate": 5.9347052176004836e-05,
      "loss": 1.3579,
      "step": 1145
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.125,
      "learning_rate": 5.934125667921859e-05,
      "loss": 1.3148,
      "step": 1150
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.21875,
      "learning_rate": 5.9335435861331355e-05,
      "loss": 1.3288,
      "step": 1155
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.4375,
      "learning_rate": 5.9329589727366435e-05,
      "loss": 1.3332,
      "step": 1160
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.8125,
      "learning_rate": 5.932371828236895e-05,
      "loss": 1.3346,
      "step": 1165
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.9375,
      "learning_rate": 5.931782153140588e-05,
      "loss": 1.3085,
      "step": 1170
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.65625,
      "learning_rate": 5.931189947956605e-05,
      "loss": 1.2766,
      "step": 1175
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.03125,
      "learning_rate": 5.9305952131960096e-05,
      "loss": 1.3103,
      "step": 1180
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.21875,
      "learning_rate": 5.929997949372049e-05,
      "loss": 1.2864,
      "step": 1185
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.8125,
      "learning_rate": 5.929398157000156e-05,
      "loss": 1.3362,
      "step": 1190
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.75,
      "learning_rate": 5.9287958365979407e-05,
      "loss": 1.2611,
      "step": 1195
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.40625,
      "learning_rate": 5.9281909886852004e-05,
      "loss": 1.244,
      "step": 1200
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.25,
      "learning_rate": 5.927583613783907e-05,
      "loss": 1.3666,
      "step": 1205
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.09375,
      "learning_rate": 5.926973712418219e-05,
      "loss": 1.306,
      "step": 1210
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.875,
      "learning_rate": 5.926361285114473e-05,
      "loss": 1.4385,
      "step": 1215
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.53125,
      "learning_rate": 5.925746332401183e-05,
      "loss": 1.3202,
      "step": 1220
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.4375,
      "learning_rate": 5.9251288548090484e-05,
      "loss": 1.3279,
      "step": 1225
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.65625,
      "learning_rate": 5.92450885287094e-05,
      "loss": 1.2451,
      "step": 1230
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.1875,
      "learning_rate": 5.923886327121914e-05,
      "loss": 1.2847,
      "step": 1235
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.65625,
      "learning_rate": 5.9232612780991996e-05,
      "loss": 1.3613,
      "step": 1240
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.0625,
      "learning_rate": 5.9226337063422067e-05,
      "loss": 1.4253,
      "step": 1245
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.59375,
      "learning_rate": 5.922003612392521e-05,
      "loss": 1.3019,
      "step": 1250
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.5,
      "learning_rate": 5.9213709967939036e-05,
      "loss": 1.3341,
      "step": 1255
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.65625,
      "learning_rate": 5.920735860092294e-05,
      "loss": 1.3104,
      "step": 1260
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.65625,
      "learning_rate": 5.920098202835807e-05,
      "loss": 1.3679,
      "step": 1265
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.375,
      "learning_rate": 5.919458025574732e-05,
      "loss": 1.251,
      "step": 1270
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.65625,
      "learning_rate": 5.9188153288615304e-05,
      "loss": 1.3897,
      "step": 1275
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.03125,
      "learning_rate": 5.918170113250845e-05,
      "loss": 1.3073,
      "step": 1280
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.78125,
      "learning_rate": 5.9175223792994846e-05,
      "loss": 1.3152,
      "step": 1285
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.5625,
      "learning_rate": 5.916872127566437e-05,
      "loss": 1.3615,
      "step": 1290
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.8125,
      "learning_rate": 5.916219358612859e-05,
      "loss": 1.3319,
      "step": 1295
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.8125,
      "learning_rate": 5.9155640730020824e-05,
      "loss": 1.2954,
      "step": 1300
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.0,
      "learning_rate": 5.9149062712996084e-05,
      "loss": 1.3416,
      "step": 1305
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.5,
      "learning_rate": 5.914245954073111e-05,
      "loss": 1.3452,
      "step": 1310
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.0625,
      "learning_rate": 5.9135831218924354e-05,
      "loss": 1.3642,
      "step": 1315
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.84375,
      "learning_rate": 5.912917775329597e-05,
      "loss": 1.2947,
      "step": 1320
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.65625,
      "learning_rate": 5.91224991495878e-05,
      "loss": 1.4229,
      "step": 1325
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.71875,
      "learning_rate": 5.9115795413563396e-05,
      "loss": 1.3313,
      "step": 1330
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.78125,
      "learning_rate": 5.910906655100797e-05,
      "loss": 1.3698,
      "step": 1335
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.8125,
      "learning_rate": 5.9102312567728466e-05,
      "loss": 1.3391,
      "step": 1340
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.28125,
      "learning_rate": 5.9095533469553453e-05,
      "loss": 1.3375,
      "step": 1345
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.875,
      "learning_rate": 5.908872926233322e-05,
      "loss": 1.3414,
      "step": 1350
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.1875,
      "learning_rate": 5.908189995193969e-05,
      "loss": 1.389,
      "step": 1355
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.21875,
      "learning_rate": 5.907504554426647e-05,
      "loss": 1.3755,
      "step": 1360
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8125,
      "learning_rate": 5.906816604522882e-05,
      "loss": 1.3034,
      "step": 1365
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.0,
      "learning_rate": 5.906126146076365e-05,
      "loss": 1.2712,
      "step": 1370
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.0625,
      "learning_rate": 5.905433179682952e-05,
      "loss": 1.3502,
      "step": 1375
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.125,
      "learning_rate": 5.9047377059406646e-05,
      "loss": 1.3117,
      "step": 1380
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.40625,
      "learning_rate": 5.904039725449686e-05,
      "loss": 1.363,
      "step": 1385
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.125,
      "learning_rate": 5.903339238812364e-05,
      "loss": 1.3283,
      "step": 1390
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.5,
      "learning_rate": 5.902636246633208e-05,
      "loss": 1.38,
      "step": 1395
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.34375,
      "learning_rate": 5.9019307495188924e-05,
      "loss": 1.2899,
      "step": 1400
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.25,
      "learning_rate": 5.9012227480782505e-05,
      "loss": 1.3645,
      "step": 1405
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.21875,
      "learning_rate": 5.900512242922278e-05,
      "loss": 1.3104,
      "step": 1410
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.5625,
      "learning_rate": 5.899799234664131e-05,
      "loss": 1.3189,
      "step": 1415
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.78125,
      "learning_rate": 5.899083723919126e-05,
      "loss": 1.2704,
      "step": 1420
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.4375,
      "learning_rate": 5.898365711304738e-05,
      "loss": 1.3263,
      "step": 1425
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.28125,
      "learning_rate": 5.8976451974406034e-05,
      "loss": 1.3194,
      "step": 1430
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.4375,
      "learning_rate": 5.896922182948516e-05,
      "loss": 1.3556,
      "step": 1435
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.4375,
      "learning_rate": 5.8961966684524244e-05,
      "loss": 1.2764,
      "step": 1440
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.625,
      "learning_rate": 5.8954686545784414e-05,
      "loss": 1.3496,
      "step": 1445
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.625,
      "learning_rate": 5.894738141954831e-05,
      "loss": 1.3491,
      "step": 1450
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.09375,
      "learning_rate": 5.894005131212015e-05,
      "loss": 1.3982,
      "step": 1455
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.21875,
      "learning_rate": 5.8932696229825736e-05,
      "loss": 1.3221,
      "step": 1460
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.84375,
      "learning_rate": 5.8925316179012384e-05,
      "loss": 1.3247,
      "step": 1465
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.375,
      "learning_rate": 5.8917911166048995e-05,
      "loss": 1.2892,
      "step": 1470
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.15625,
      "learning_rate": 5.891048119732597e-05,
      "loss": 1.3018,
      "step": 1475
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.0,
      "learning_rate": 5.890302627925528e-05,
      "loss": 1.4015,
      "step": 1480
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.125,
      "learning_rate": 5.889554641827043e-05,
      "loss": 1.3275,
      "step": 1485
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.6875,
      "learning_rate": 5.888804162082642e-05,
      "loss": 1.3106,
      "step": 1490
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.0625,
      "learning_rate": 5.888051189339978e-05,
      "loss": 1.2854,
      "step": 1495
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.71875,
      "learning_rate": 5.8872957242488585e-05,
      "loss": 1.2805,
      "step": 1500
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.96875,
      "learning_rate": 5.886537767461238e-05,
      "loss": 1.3378,
      "step": 1505
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.09375,
      "learning_rate": 5.885777319631224e-05,
      "loss": 1.3369,
      "step": 1510
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.71875,
      "learning_rate": 5.885014381415071e-05,
      "loss": 1.3096,
      "step": 1515
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.625,
      "learning_rate": 5.8842489534711856e-05,
      "loss": 1.3195,
      "step": 1520
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.03125,
      "learning_rate": 5.88348103646012e-05,
      "loss": 1.2828,
      "step": 1525
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.46875,
      "learning_rate": 5.882710631044577e-05,
      "loss": 1.3882,
      "step": 1530
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.875,
      "learning_rate": 5.881937737889406e-05,
      "loss": 1.3717,
      "step": 1535
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.9375,
      "learning_rate": 5.8811623576616026e-05,
      "loss": 1.2351,
      "step": 1540
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.71875,
      "learning_rate": 5.8803844910303105e-05,
      "loss": 1.2742,
      "step": 1545
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.28125,
      "learning_rate": 5.879604138666816e-05,
      "loss": 1.304,
      "step": 1550
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.03125,
      "learning_rate": 5.878821301244554e-05,
      "loss": 1.3395,
      "step": 1555
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.28125,
      "learning_rate": 5.878035979439101e-05,
      "loss": 1.3145,
      "step": 1560
    },
    {
      "epoch": 0.28,
      "grad_norm": 3.8125,
      "learning_rate": 5.87724817392818e-05,
      "loss": 1.3266,
      "step": 1565
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.375,
      "learning_rate": 5.876457885391656e-05,
      "loss": 1.3551,
      "step": 1570
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.96875,
      "learning_rate": 5.8756651145115384e-05,
      "loss": 1.2578,
      "step": 1575
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.9375,
      "learning_rate": 5.874869861971976e-05,
      "loss": 1.4112,
      "step": 1580
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.59375,
      "learning_rate": 5.874072128459261e-05,
      "loss": 1.2546,
      "step": 1585
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.25,
      "learning_rate": 5.8732719146618265e-05,
      "loss": 1.2999,
      "step": 1590
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.8125,
      "learning_rate": 5.872469221270246e-05,
      "loss": 1.3047,
      "step": 1595
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.21875,
      "learning_rate": 5.8716640489772335e-05,
      "loss": 1.3674,
      "step": 1600
    },
    {
      "epoch": 0.28,
      "grad_norm": 4.5,
      "learning_rate": 5.870856398477641e-05,
      "loss": 1.2913,
      "step": 1605
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.4375,
      "learning_rate": 5.870046270468459e-05,
      "loss": 1.3916,
      "step": 1610
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.0625,
      "learning_rate": 5.8692336656488196e-05,
      "loss": 1.3267,
      "step": 1615
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.53125,
      "learning_rate": 5.868418584719986e-05,
      "loss": 1.3203,
      "step": 1620
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.40625,
      "learning_rate": 5.867601028385365e-05,
      "loss": 1.4038,
      "step": 1625
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.0625,
      "learning_rate": 5.866780997350494e-05,
      "loss": 1.3253,
      "step": 1630
    },
    {
      "epoch": 0.29,
      "grad_norm": 5.21875,
      "learning_rate": 5.86595849232305e-05,
      "loss": 1.3407,
      "step": 1635
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.90625,
      "learning_rate": 5.865133514012843e-05,
      "loss": 1.2964,
      "step": 1640
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.25,
      "learning_rate": 5.864306063131818e-05,
      "loss": 1.3601,
      "step": 1645
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.125,
      "learning_rate": 5.863476140394053e-05,
      "loss": 1.3301,
      "step": 1650
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.9375,
      "learning_rate": 5.862643746515761e-05,
      "loss": 1.3416,
      "step": 1655
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.09375,
      "learning_rate": 5.8618088822152866e-05,
      "loss": 1.3301,
      "step": 1660
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.40625,
      "learning_rate": 5.8609715482131055e-05,
      "loss": 1.3167,
      "step": 1665
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.15625,
      "learning_rate": 5.860131745231826e-05,
      "loss": 1.3531,
      "step": 1670
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.65625,
      "learning_rate": 5.859289473996186e-05,
      "loss": 1.3402,
      "step": 1675
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.75,
      "learning_rate": 5.8584447352330533e-05,
      "loss": 1.2596,
      "step": 1680
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.1875,
      "learning_rate": 5.8575975296714274e-05,
      "loss": 1.3238,
      "step": 1685
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.625,
      "learning_rate": 5.856747858042435e-05,
      "loss": 1.3372,
      "step": 1690
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.125,
      "learning_rate": 5.85589572107933e-05,
      "loss": 1.3193,
      "step": 1695
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.875,
      "learning_rate": 5.855041119517494e-05,
      "loss": 1.284,
      "step": 1700
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.5,
      "learning_rate": 5.854184054094439e-05,
      "loss": 1.2943,
      "step": 1705
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.6875,
      "learning_rate": 5.853324525549799e-05,
      "loss": 1.2727,
      "step": 1710
    },
    {
      "epoch": 0.3,
      "grad_norm": 3.6875,
      "learning_rate": 5.852462534625336e-05,
      "loss": 1.3013,
      "step": 1715
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.5625,
      "learning_rate": 5.8515980820649354e-05,
      "loss": 1.2731,
      "step": 1720
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.5625,
      "learning_rate": 5.850731168614608e-05,
      "loss": 1.3445,
      "step": 1725
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.1875,
      "learning_rate": 5.849861795022489e-05,
      "loss": 1.2977,
      "step": 1730
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.4375,
      "learning_rate": 5.848989962038836e-05,
      "loss": 1.3489,
      "step": 1735
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.03125,
      "learning_rate": 5.848115670416028e-05,
      "loss": 1.3006,
      "step": 1740
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.625,
      "learning_rate": 5.847238920908566e-05,
      "loss": 1.357,
      "step": 1745
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.25,
      "learning_rate": 5.846359714273075e-05,
      "loss": 1.37,
      "step": 1750
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.5,
      "learning_rate": 5.845478051268296e-05,
      "loss": 1.315,
      "step": 1755
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.15625,
      "learning_rate": 5.844593932655094e-05,
      "loss": 1.3086,
      "step": 1760
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.65625,
      "learning_rate": 5.84370735919645e-05,
      "loss": 1.2737,
      "step": 1765
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.40625,
      "learning_rate": 5.8428183316574645e-05,
      "loss": 1.2527,
      "step": 1770
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.8125,
      "learning_rate": 5.841926850805357e-05,
      "loss": 1.3043,
      "step": 1775
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.5625,
      "learning_rate": 5.8410329174094624e-05,
      "loss": 1.3782,
      "step": 1780
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.984375,
      "learning_rate": 5.8401365322412335e-05,
      "loss": 1.253,
      "step": 1785
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.03125,
      "learning_rate": 5.839237696074237e-05,
      "loss": 1.3019,
      "step": 1790
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.53125,
      "learning_rate": 5.8383364096841584e-05,
      "loss": 1.349,
      "step": 1795
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.78125,
      "learning_rate": 5.837432673848794e-05,
      "loss": 1.3437,
      "step": 1800
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.5,
      "learning_rate": 5.836526489348056e-05,
      "loss": 1.2328,
      "step": 1805
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.375,
      "learning_rate": 5.835617856963968e-05,
      "loss": 1.2784,
      "step": 1810
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.71875,
      "learning_rate": 5.834706777480669e-05,
      "loss": 1.3552,
      "step": 1815
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.03125,
      "learning_rate": 5.8337932516844065e-05,
      "loss": 1.4127,
      "step": 1820
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.46875,
      "learning_rate": 5.832877280363542e-05,
      "loss": 1.311,
      "step": 1825
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.8125,
      "learning_rate": 5.831958864308545e-05,
      "loss": 1.3119,
      "step": 1830
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.21875,
      "learning_rate": 5.831038004311996e-05,
      "loss": 1.3521,
      "step": 1835
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.65625,
      "learning_rate": 5.830114701168586e-05,
      "loss": 1.3077,
      "step": 1840
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.4375,
      "learning_rate": 5.829188955675111e-05,
      "loss": 1.3005,
      "step": 1845
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.8125,
      "learning_rate": 5.828260768630478e-05,
      "loss": 1.3011,
      "step": 1850
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.46875,
      "learning_rate": 5.827330140835699e-05,
      "loss": 1.3626,
      "step": 1855
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.15625,
      "learning_rate": 5.826397073093893e-05,
      "loss": 1.339,
      "step": 1860
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.71875,
      "learning_rate": 5.825461566210285e-05,
      "loss": 1.2936,
      "step": 1865
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.1875,
      "learning_rate": 5.824523620992205e-05,
      "loss": 1.3297,
      "step": 1870
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.84375,
      "learning_rate": 5.823583238249084e-05,
      "loss": 1.3285,
      "step": 1875
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.34375,
      "learning_rate": 5.8226404187924633e-05,
      "loss": 1.2949,
      "step": 1880
    },
    {
      "epoch": 0.33,
      "grad_norm": 4.1875,
      "learning_rate": 5.821695163435982e-05,
      "loss": 1.4273,
      "step": 1885
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.15625,
      "learning_rate": 5.820747472995381e-05,
      "loss": 1.2731,
      "step": 1890
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.0625,
      "learning_rate": 5.819797348288504e-05,
      "loss": 1.3678,
      "step": 1895
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.34375,
      "learning_rate": 5.8188447901352976e-05,
      "loss": 1.3279,
      "step": 1900
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.15625,
      "learning_rate": 5.817889799357805e-05,
      "loss": 1.2751,
      "step": 1905
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.65625,
      "learning_rate": 5.81693237678017e-05,
      "loss": 1.2946,
      "step": 1910
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.5,
      "learning_rate": 5.815972523228636e-05,
      "loss": 1.3346,
      "step": 1915
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.15625,
      "learning_rate": 5.815010239531543e-05,
      "loss": 1.3255,
      "step": 1920
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.40625,
      "learning_rate": 5.8140455265193284e-05,
      "loss": 1.3371,
      "step": 1925
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.625,
      "learning_rate": 5.8130783850245265e-05,
      "loss": 1.2516,
      "step": 1930
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.125,
      "learning_rate": 5.812108815881766e-05,
      "loss": 1.3151,
      "step": 1935
    },
    {
      "epoch": 0.34,
      "grad_norm": 4.6875,
      "learning_rate": 5.8111368199277735e-05,
      "loss": 1.3182,
      "step": 1940
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.5625,
      "learning_rate": 5.810162398001367e-05,
      "loss": 1.2896,
      "step": 1945
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.15625,
      "learning_rate": 5.809185550943461e-05,
      "loss": 1.3718,
      "step": 1950
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.1875,
      "learning_rate": 5.808206279597058e-05,
      "loss": 1.3822,
      "step": 1955
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.15625,
      "learning_rate": 5.8072245848072595e-05,
      "loss": 1.2806,
      "step": 1960
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.78125,
      "learning_rate": 5.80624046742125e-05,
      "loss": 1.2678,
      "step": 1965
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.375,
      "learning_rate": 5.805253928288313e-05,
      "loss": 1.3332,
      "step": 1970
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.53125,
      "learning_rate": 5.804264968259817e-05,
      "loss": 1.3929,
      "step": 1975
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.90625,
      "learning_rate": 5.803273588189221e-05,
      "loss": 1.346,
      "step": 1980
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.90625,
      "learning_rate": 5.8022797889320716e-05,
      "loss": 1.3509,
      "step": 1985
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.4375,
      "learning_rate": 5.801283571346004e-05,
      "loss": 1.3379,
      "step": 1990
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.34375,
      "learning_rate": 5.8002849362907425e-05,
      "loss": 1.3061,
      "step": 1995
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.8125,
      "learning_rate": 5.7992838846280926e-05,
      "loss": 1.2657,
      "step": 2000
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.5,
      "learning_rate": 5.798280417221949e-05,
      "loss": 1.3385,
      "step": 2005
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.9375,
      "learning_rate": 5.797274534938291e-05,
      "loss": 1.307,
      "step": 2010
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.53125,
      "learning_rate": 5.7962662386451805e-05,
      "loss": 1.3082,
      "step": 2015
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.1875,
      "learning_rate": 5.795255529212763e-05,
      "loss": 1.3263,
      "step": 2020
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.5,
      "learning_rate": 5.794242407513268e-05,
      "loss": 1.3369,
      "step": 2025
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.75,
      "learning_rate": 5.7932268744210046e-05,
      "loss": 1.3142,
      "step": 2030
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.46875,
      "learning_rate": 5.7922089308123646e-05,
      "loss": 1.2688,
      "step": 2035
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.25,
      "learning_rate": 5.7911885775658186e-05,
      "loss": 1.3064,
      "step": 2040
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.84375,
      "learning_rate": 5.790165815561918e-05,
      "loss": 1.2826,
      "step": 2045
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.96875,
      "learning_rate": 5.7891406456832925e-05,
      "loss": 1.2759,
      "step": 2050
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.90625,
      "learning_rate": 5.788113068814648e-05,
      "loss": 1.318,
      "step": 2055
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.46875,
      "learning_rate": 5.787083085842772e-05,
      "loss": 1.3142,
      "step": 2060
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.5,
      "learning_rate": 5.786050697656523e-05,
      "loss": 1.2876,
      "step": 2065
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.28125,
      "learning_rate": 5.78501590514684e-05,
      "loss": 1.3256,
      "step": 2070
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.703125,
      "learning_rate": 5.783978709206732e-05,
      "loss": 1.3258,
      "step": 2075
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.65625,
      "learning_rate": 5.782939110731287e-05,
      "loss": 1.2908,
      "step": 2080
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.28125,
      "learning_rate": 5.7818971106176634e-05,
      "loss": 1.2957,
      "step": 2085
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.5,
      "learning_rate": 5.780852709765094e-05,
      "loss": 1.355,
      "step": 2090
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.0,
      "learning_rate": 5.7798059090748805e-05,
      "loss": 1.2604,
      "step": 2095
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.09375,
      "learning_rate": 5.778756709450399e-05,
      "loss": 1.2796,
      "step": 2100
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.5625,
      "learning_rate": 5.7777051117970946e-05,
      "loss": 1.3614,
      "step": 2105
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.3125,
      "learning_rate": 5.77665111702248e-05,
      "loss": 1.3198,
      "step": 2110
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.09375,
      "learning_rate": 5.775594726036139e-05,
      "loss": 1.2494,
      "step": 2115
    },
    {
      "epoch": 0.38,
      "grad_norm": 3.984375,
      "learning_rate": 5.774535939749723e-05,
      "loss": 1.2933,
      "step": 2120
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.8125,
      "learning_rate": 5.77347475907695e-05,
      "loss": 1.3178,
      "step": 2125
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.75,
      "learning_rate": 5.772411184933603e-05,
      "loss": 1.3002,
      "step": 2130
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.53125,
      "learning_rate": 5.771345218237534e-05,
      "loss": 1.3109,
      "step": 2135
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.96875,
      "learning_rate": 5.770276859908655e-05,
      "loss": 1.3076,
      "step": 2140
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.5625,
      "learning_rate": 5.769206110868947e-05,
      "loss": 1.3253,
      "step": 2145
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.65625,
      "learning_rate": 5.76813297204245e-05,
      "loss": 1.3409,
      "step": 2150
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.53125,
      "learning_rate": 5.767057444355269e-05,
      "loss": 1.3253,
      "step": 2155
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.40625,
      "learning_rate": 5.765979528735569e-05,
      "loss": 1.2958,
      "step": 2160
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.5625,
      "learning_rate": 5.7648992261135766e-05,
      "loss": 1.2519,
      "step": 2165
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.84375,
      "learning_rate": 5.7638165374215786e-05,
      "loss": 1.2364,
      "step": 2170
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.8125,
      "learning_rate": 5.7627314635939185e-05,
      "loss": 1.2876,
      "step": 2175
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.46875,
      "learning_rate": 5.7616440055670037e-05,
      "loss": 1.2515,
      "step": 2180
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.71875,
      "learning_rate": 5.7605541642792915e-05,
      "loss": 1.3361,
      "step": 2185
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.125,
      "learning_rate": 5.7594619406713025e-05,
      "loss": 1.28,
      "step": 2190
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.34375,
      "learning_rate": 5.758367335685611e-05,
      "loss": 1.2606,
      "step": 2195
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.9375,
      "learning_rate": 5.7572703502668436e-05,
      "loss": 1.3897,
      "step": 2200
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.21875,
      "learning_rate": 5.756170985361686e-05,
      "loss": 1.2162,
      "step": 2205
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.28125,
      "learning_rate": 5.755069241918873e-05,
      "loss": 1.3689,
      "step": 2210
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.5625,
      "learning_rate": 5.753965120889196e-05,
      "loss": 1.3492,
      "step": 2215
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.4375,
      "learning_rate": 5.752858623225495e-05,
      "loss": 1.333,
      "step": 2220
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.78125,
      "learning_rate": 5.751749749882663e-05,
      "loss": 1.3392,
      "step": 2225
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.5625,
      "learning_rate": 5.750638501817642e-05,
      "loss": 1.3787,
      "step": 2230
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.53125,
      "learning_rate": 5.7495248799894225e-05,
      "loss": 1.2737,
      "step": 2235
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.875,
      "learning_rate": 5.7484088853590474e-05,
      "loss": 1.2746,
      "step": 2240
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.09375,
      "learning_rate": 5.747290518889604e-05,
      "loss": 1.2974,
      "step": 2245
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.1875,
      "learning_rate": 5.7461697815462265e-05,
      "loss": 1.3759,
      "step": 2250
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.65625,
      "learning_rate": 5.745046674296095e-05,
      "loss": 1.326,
      "step": 2255
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.84375,
      "learning_rate": 5.743921198108437e-05,
      "loss": 1.2786,
      "step": 2260
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.09375,
      "learning_rate": 5.7427933539545225e-05,
      "loss": 1.3348,
      "step": 2265
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.53125,
      "learning_rate": 5.7416631428076655e-05,
      "loss": 1.3611,
      "step": 2270
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.4375,
      "learning_rate": 5.740530565643223e-05,
      "loss": 1.3215,
      "step": 2275
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.5,
      "learning_rate": 5.739395623438594e-05,
      "loss": 1.337,
      "step": 2280
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.28125,
      "learning_rate": 5.738258317173217e-05,
      "loss": 1.3059,
      "step": 2285
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.5,
      "learning_rate": 5.737118647828571e-05,
      "loss": 1.3581,
      "step": 2290
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.09375,
      "learning_rate": 5.7359766163881766e-05,
      "loss": 1.2102,
      "step": 2295
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.78125,
      "learning_rate": 5.73483222383759e-05,
      "loss": 1.3794,
      "step": 2300
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.0625,
      "learning_rate": 5.733685471164407e-05,
      "loss": 1.3096,
      "step": 2305
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.6875,
      "learning_rate": 5.73253635935826e-05,
      "loss": 1.3031,
      "step": 2310
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.15625,
      "learning_rate": 5.731384889410814e-05,
      "loss": 1.2819,
      "step": 2315
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.75,
      "learning_rate": 5.730231062315775e-05,
      "loss": 1.2891,
      "step": 2320
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.0625,
      "learning_rate": 5.729074879068878e-05,
      "loss": 1.3334,
      "step": 2325
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.8125,
      "learning_rate": 5.727916340667895e-05,
      "loss": 1.3738,
      "step": 2330
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.21875,
      "learning_rate": 5.7267554481126255e-05,
      "loss": 1.3433,
      "step": 2335
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.625,
      "learning_rate": 5.725592202404907e-05,
      "loss": 1.2954,
      "step": 2340
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.4375,
      "learning_rate": 5.7244266045486025e-05,
      "loss": 1.323,
      "step": 2345
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.5625,
      "learning_rate": 5.723258655549608e-05,
      "loss": 1.2917,
      "step": 2350
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.1875,
      "learning_rate": 5.7220883564158474e-05,
      "loss": 1.3366,
      "step": 2355
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.03125,
      "learning_rate": 5.7209157081572725e-05,
      "loss": 1.3243,
      "step": 2360
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.40625,
      "learning_rate": 5.7197407117858625e-05,
      "loss": 1.3289,
      "step": 2365
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.125,
      "learning_rate": 5.718563368315623e-05,
      "loss": 1.3062,
      "step": 2370
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.6875,
      "learning_rate": 5.7173836787625854e-05,
      "loss": 1.3856,
      "step": 2375
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.5,
      "learning_rate": 5.7162016441448056e-05,
      "loss": 1.4069,
      "step": 2380
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.46875,
      "learning_rate": 5.715017265482363e-05,
      "loss": 1.3393,
      "step": 2385
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.6875,
      "learning_rate": 5.7138305437973613e-05,
      "loss": 1.3196,
      "step": 2390
    },
    {
      "epoch": 0.42,
      "grad_norm": 4.4375,
      "learning_rate": 5.712641480113923e-05,
      "loss": 1.2874,
      "step": 2395
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.8125,
      "learning_rate": 5.7114500754581954e-05,
      "loss": 1.3249,
      "step": 2400
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.6875,
      "learning_rate": 5.710256330858343e-05,
      "loss": 1.327,
      "step": 2405
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.0625,
      "learning_rate": 5.709060247344552e-05,
      "loss": 1.3701,
      "step": 2410
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.15625,
      "learning_rate": 5.707861825949026e-05,
      "loss": 1.3252,
      "step": 2415
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.40625,
      "learning_rate": 5.706661067705984e-05,
      "loss": 1.3281,
      "step": 2420
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.3125,
      "learning_rate": 5.705457973651668e-05,
      "loss": 1.3507,
      "step": 2425
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.5625,
      "learning_rate": 5.7042525448243286e-05,
      "loss": 1.3503,
      "step": 2430
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.0,
      "learning_rate": 5.7030447822642346e-05,
      "loss": 1.2939,
      "step": 2435
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.84375,
      "learning_rate": 5.701834687013669e-05,
      "loss": 1.2999,
      "step": 2440
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.65625,
      "learning_rate": 5.700622260116927e-05,
      "loss": 1.3386,
      "step": 2445
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.78125,
      "learning_rate": 5.699407502620317e-05,
      "loss": 1.2755,
      "step": 2450
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.28125,
      "learning_rate": 5.6981904155721574e-05,
      "loss": 1.335,
      "step": 2455
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.09375,
      "learning_rate": 5.696971000022778e-05,
      "loss": 1.2703,
      "step": 2460
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.28125,
      "learning_rate": 5.695749257024517e-05,
      "loss": 1.254,
      "step": 2465
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.09375,
      "learning_rate": 5.694525187631722e-05,
      "loss": 1.2533,
      "step": 2470
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.46875,
      "learning_rate": 5.6932987929007484e-05,
      "loss": 1.2611,
      "step": 2475
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.875,
      "learning_rate": 5.692070073889958e-05,
      "loss": 1.3172,
      "step": 2480
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.28125,
      "learning_rate": 5.690839031659718e-05,
      "loss": 1.2717,
      "step": 2485
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.625,
      "learning_rate": 5.6896056672724014e-05,
      "loss": 1.3477,
      "step": 2490
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.34375,
      "learning_rate": 5.6883699817923856e-05,
      "loss": 1.3107,
      "step": 2495
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.65625,
      "learning_rate": 5.6871319762860475e-05,
      "loss": 1.3247,
      "step": 2500
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.59375,
      "learning_rate": 5.6858916518217714e-05,
      "loss": 1.282,
      "step": 2505
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.9375,
      "learning_rate": 5.6846490094699386e-05,
      "loss": 1.2465,
      "step": 2510
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.90625,
      "learning_rate": 5.6834040503029334e-05,
      "loss": 1.2966,
      "step": 2515
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.96875,
      "learning_rate": 5.682156775395137e-05,
      "loss": 1.2818,
      "step": 2520
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.5625,
      "learning_rate": 5.6809071858229325e-05,
      "loss": 1.2737,
      "step": 2525
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.4375,
      "learning_rate": 5.679655282664698e-05,
      "loss": 1.3038,
      "step": 2530
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.125,
      "learning_rate": 5.6784010670008073e-05,
      "loss": 1.314,
      "step": 2535
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.0625,
      "learning_rate": 5.6771445399136324e-05,
      "loss": 1.3201,
      "step": 2540
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.90625,
      "learning_rate": 5.6758857024875384e-05,
      "loss": 1.3113,
      "step": 2545
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.96875,
      "learning_rate": 5.674624555808886e-05,
      "loss": 1.2482,
      "step": 2550
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.8125,
      "learning_rate": 5.673361100966025e-05,
      "loss": 1.3105,
      "step": 2555
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.28125,
      "learning_rate": 5.672095339049302e-05,
      "loss": 1.308,
      "step": 2560
    },
    {
      "epoch": 0.45,
      "grad_norm": 4.65625,
      "learning_rate": 5.670827271151051e-05,
      "loss": 1.3011,
      "step": 2565
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.46875,
      "learning_rate": 5.669556898365598e-05,
      "loss": 1.3295,
      "step": 2570
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.78125,
      "learning_rate": 5.668284221789255e-05,
      "loss": 1.3619,
      "step": 2575
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.625,
      "learning_rate": 5.667009242520327e-05,
      "loss": 1.2631,
      "step": 2580
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.40625,
      "learning_rate": 5.665731961659102e-05,
      "loss": 1.3128,
      "step": 2585
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.25,
      "learning_rate": 5.6644523803078564e-05,
      "loss": 1.2335,
      "step": 2590
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.1875,
      "learning_rate": 5.663170499570851e-05,
      "loss": 1.3206,
      "step": 2595
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.8125,
      "learning_rate": 5.661886320554331e-05,
      "loss": 1.3087,
      "step": 2600
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.1875,
      "learning_rate": 5.660599844366527e-05,
      "loss": 1.2921,
      "step": 2605
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.5,
      "learning_rate": 5.6593110721176475e-05,
      "loss": 1.23,
      "step": 2610
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.59375,
      "learning_rate": 5.6580200049198876e-05,
      "loss": 1.3199,
      "step": 2615
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.9375,
      "learning_rate": 5.656726643887419e-05,
      "loss": 1.2793,
      "step": 2620
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.21875,
      "learning_rate": 5.655430990136395e-05,
      "loss": 1.3256,
      "step": 2625
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.0625,
      "learning_rate": 5.654133044784948e-05,
      "loss": 1.2995,
      "step": 2630
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.4375,
      "learning_rate": 5.652832808953185e-05,
      "loss": 1.3027,
      "step": 2635
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.15625,
      "learning_rate": 5.6515302837631935e-05,
      "loss": 1.2884,
      "step": 2640
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.0625,
      "learning_rate": 5.650225470339034e-05,
      "loss": 1.335,
      "step": 2645
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.75,
      "learning_rate": 5.648918369806742e-05,
      "loss": 1.3514,
      "step": 2650
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.25,
      "learning_rate": 5.6476089832943275e-05,
      "loss": 1.2674,
      "step": 2655
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.40625,
      "learning_rate": 5.646297311931775e-05,
      "loss": 1.3246,
      "step": 2660
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.0625,
      "learning_rate": 5.644983356851035e-05,
      "loss": 1.2846,
      "step": 2665
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.96875,
      "learning_rate": 5.643667119186036e-05,
      "loss": 1.2634,
      "step": 2670
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.4375,
      "learning_rate": 5.642348600072672e-05,
      "loss": 1.3831,
      "step": 2675
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.09375,
      "learning_rate": 5.641027800648806e-05,
      "loss": 1.3404,
      "step": 2680
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.984375,
      "learning_rate": 5.6397047220542705e-05,
      "loss": 1.3466,
      "step": 2685
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.59375,
      "learning_rate": 5.6383793654308653e-05,
      "loss": 1.2297,
      "step": 2690
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.59375,
      "learning_rate": 5.637051731922353e-05,
      "loss": 1.3816,
      "step": 2695
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.65625,
      "learning_rate": 5.6357218226744634e-05,
      "loss": 1.3857,
      "step": 2700
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.25,
      "learning_rate": 5.63438963883489e-05,
      "loss": 1.3098,
      "step": 2705
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.4375,
      "learning_rate": 5.6330551815532915e-05,
      "loss": 1.2842,
      "step": 2710
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.3125,
      "learning_rate": 5.631718451981282e-05,
      "loss": 1.3243,
      "step": 2715
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.6875,
      "learning_rate": 5.6303794512724444e-05,
      "loss": 1.2961,
      "step": 2720
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.375,
      "learning_rate": 5.629038180582315e-05,
      "loss": 1.2975,
      "step": 2725
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.6875,
      "learning_rate": 5.627694641068394e-05,
      "loss": 1.295,
      "step": 2730
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.46875,
      "learning_rate": 5.626348833890136e-05,
      "loss": 1.3153,
      "step": 2735
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.78125,
      "learning_rate": 5.6250007602089556e-05,
      "loss": 1.2853,
      "step": 2740
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.09375,
      "learning_rate": 5.623650421188221e-05,
      "loss": 1.2951,
      "step": 2745
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.1875,
      "learning_rate": 5.6222978179932566e-05,
      "loss": 1.3122,
      "step": 2750
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.3125,
      "learning_rate": 5.62094295179134e-05,
      "loss": 1.3803,
      "step": 2755
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.5625,
      "learning_rate": 5.6195858237517024e-05,
      "loss": 1.3136,
      "step": 2760
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.65625,
      "learning_rate": 5.618226435045526e-05,
      "loss": 1.2738,
      "step": 2765
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.1875,
      "learning_rate": 5.616864786845945e-05,
      "loss": 1.3019,
      "step": 2770
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.4375,
      "learning_rate": 5.615500880328044e-05,
      "loss": 1.3231,
      "step": 2775
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.4375,
      "learning_rate": 5.6141347166688534e-05,
      "loss": 1.3113,
      "step": 2780
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.5625,
      "learning_rate": 5.6127662970473556e-05,
      "loss": 1.3466,
      "step": 2785
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.59375,
      "learning_rate": 5.611395622644477e-05,
      "loss": 1.293,
      "step": 2790
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.75,
      "learning_rate": 5.610022694643091e-05,
      "loss": 1.335,
      "step": 2795
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.40625,
      "learning_rate": 5.608647514228015e-05,
      "loss": 1.3079,
      "step": 2800
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.3125,
      "learning_rate": 5.607270082586013e-05,
      "loss": 1.3268,
      "step": 2805
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.1875,
      "learning_rate": 5.605890400905786e-05,
      "loss": 1.3984,
      "step": 2810
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.875,
      "learning_rate": 5.604508470377983e-05,
      "loss": 1.3175,
      "step": 2815
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.25,
      "learning_rate": 5.60312429219519e-05,
      "loss": 1.2461,
      "step": 2820
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.09375,
      "learning_rate": 5.601737867551934e-05,
      "loss": 1.2589,
      "step": 2825
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.8125,
      "learning_rate": 5.6003491976446804e-05,
      "loss": 1.2646,
      "step": 2830
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.875,
      "learning_rate": 5.598958283671832e-05,
      "loss": 1.2677,
      "step": 2835
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.0625,
      "learning_rate": 5.59756512683373e-05,
      "loss": 1.3203,
      "step": 2840
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.65625,
      "learning_rate": 5.5961697283326465e-05,
      "loss": 1.3257,
      "step": 2845
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.78125,
      "learning_rate": 5.5947720893727943e-05,
      "loss": 1.2212,
      "step": 2850
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.28125,
      "learning_rate": 5.593372211160314e-05,
      "loss": 1.3295,
      "step": 2855
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.90625,
      "learning_rate": 5.591970094903284e-05,
      "loss": 1.3295,
      "step": 2860
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.0625,
      "learning_rate": 5.5905657418117084e-05,
      "loss": 1.31,
      "step": 2865
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.0,
      "learning_rate": 5.589159153097526e-05,
      "loss": 1.3057,
      "step": 2870
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.5625,
      "learning_rate": 5.5877503299746044e-05,
      "loss": 1.3155,
      "step": 2875
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.5,
      "learning_rate": 5.586339273658737e-05,
      "loss": 1.3647,
      "step": 2880
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.3125,
      "learning_rate": 5.584925985367647e-05,
      "loss": 1.3519,
      "step": 2885
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.4375,
      "learning_rate": 5.583510466320983e-05,
      "loss": 1.3283,
      "step": 2890
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.9375,
      "learning_rate": 5.582092717740318e-05,
      "loss": 1.3263,
      "step": 2895
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.78125,
      "learning_rate": 5.5806727408491494e-05,
      "loss": 1.3202,
      "step": 2900
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.40625,
      "learning_rate": 5.579250536872897e-05,
      "loss": 1.3079,
      "step": 2905
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.75,
      "learning_rate": 5.577826107038904e-05,
      "loss": 1.2571,
      "step": 2910
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.375,
      "learning_rate": 5.576399452576435e-05,
      "loss": 1.275,
      "step": 2915
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.53125,
      "learning_rate": 5.574970574716672e-05,
      "loss": 1.3317,
      "step": 2920
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.875,
      "learning_rate": 5.5735394746927157e-05,
      "loss": 1.3375,
      "step": 2925
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.28125,
      "learning_rate": 5.572106153739588e-05,
      "loss": 1.3079,
      "step": 2930
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.8125,
      "learning_rate": 5.570670613094223e-05,
      "loss": 1.314,
      "step": 2935
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.96875,
      "learning_rate": 5.569232853995475e-05,
      "loss": 1.2896,
      "step": 2940
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.65625,
      "learning_rate": 5.567792877684109e-05,
      "loss": 1.2462,
      "step": 2945
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.71875,
      "learning_rate": 5.566350685402804e-05,
      "loss": 1.4189,
      "step": 2950
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.90625,
      "learning_rate": 5.564906278396153e-05,
      "loss": 1.2991,
      "step": 2955
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.96875,
      "learning_rate": 5.56345965791066e-05,
      "loss": 1.3341,
      "step": 2960
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.4375,
      "learning_rate": 5.562010825194737e-05,
      "loss": 1.2373,
      "step": 2965
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.28125,
      "learning_rate": 5.560559781498709e-05,
      "loss": 1.3048,
      "step": 2970
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.375,
      "learning_rate": 5.5591065280748035e-05,
      "loss": 1.3548,
      "step": 2975
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.625,
      "learning_rate": 5.557651066177161e-05,
      "loss": 1.3368,
      "step": 2980
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.46875,
      "learning_rate": 5.556193397061824e-05,
      "loss": 1.212,
      "step": 2985
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.25,
      "learning_rate": 5.55473352198674e-05,
      "loss": 1.2773,
      "step": 2990
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.6875,
      "learning_rate": 5.553271442211764e-05,
      "loss": 1.3321,
      "step": 2995
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.0,
      "learning_rate": 5.5518071589986475e-05,
      "loss": 1.2887,
      "step": 3000
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.9375,
      "learning_rate": 5.550340673611049e-05,
      "loss": 1.3299,
      "step": 3005
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.46875,
      "learning_rate": 5.548871987314524e-05,
      "loss": 1.3793,
      "step": 3010
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.0625,
      "learning_rate": 5.547401101376529e-05,
      "loss": 1.2792,
      "step": 3015
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.46875,
      "learning_rate": 5.5459280170664174e-05,
      "loss": 1.2922,
      "step": 3020
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.09375,
      "learning_rate": 5.544452735655443e-05,
      "loss": 1.2014,
      "step": 3025
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.84375,
      "learning_rate": 5.542975258416751e-05,
      "loss": 1.3246,
      "step": 3030
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.21875,
      "learning_rate": 5.5414955866253854e-05,
      "loss": 1.2227,
      "step": 3035
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.03125,
      "learning_rate": 5.540013721558283e-05,
      "loss": 1.2552,
      "step": 3040
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.65625,
      "learning_rate": 5.53852966449427e-05,
      "loss": 1.345,
      "step": 3045
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.53125,
      "learning_rate": 5.53704341671407e-05,
      "loss": 1.2734,
      "step": 3050
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.65625,
      "learning_rate": 5.535554979500295e-05,
      "loss": 1.2668,
      "step": 3055
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.71875,
      "learning_rate": 5.5340643541374435e-05,
      "loss": 1.291,
      "step": 3060
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.46875,
      "learning_rate": 5.532571541911906e-05,
      "loss": 1.2496,
      "step": 3065
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.1875,
      "learning_rate": 5.531076544111958e-05,
      "loss": 1.3392,
      "step": 3070
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.3125,
      "learning_rate": 5.5295793620277634e-05,
      "loss": 1.3181,
      "step": 3075
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.03125,
      "learning_rate": 5.5280799969513686e-05,
      "loss": 1.2371,
      "step": 3080
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.34375,
      "learning_rate": 5.5265784501767056e-05,
      "loss": 1.3203,
      "step": 3085
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.46875,
      "learning_rate": 5.5250747229995884e-05,
      "loss": 1.3282,
      "step": 3090
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.15625,
      "learning_rate": 5.5235688167177135e-05,
      "loss": 1.3521,
      "step": 3095
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.65625,
      "learning_rate": 5.522060732630656e-05,
      "loss": 1.2624,
      "step": 3100
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.40625,
      "learning_rate": 5.520550472039873e-05,
      "loss": 1.2576,
      "step": 3105
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.40625,
      "learning_rate": 5.519038036248698e-05,
      "loss": 1.3183,
      "step": 3110
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.34375,
      "learning_rate": 5.5175234265623426e-05,
      "loss": 1.3224,
      "step": 3115
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.75,
      "learning_rate": 5.5160066442878936e-05,
      "loss": 1.2699,
      "step": 3120
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.0,
      "learning_rate": 5.5144876907343145e-05,
      "loss": 1.4023,
      "step": 3125
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.1875,
      "learning_rate": 5.51296656721244e-05,
      "loss": 1.304,
      "step": 3130
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.65625,
      "learning_rate": 5.5114432750349784e-05,
      "loss": 1.2936,
      "step": 3135
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.5625,
      "learning_rate": 5.509917815516511e-05,
      "loss": 1.3187,
      "step": 3140
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.375,
      "learning_rate": 5.5083901899734875e-05,
      "loss": 1.27,
      "step": 3145
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.6875,
      "learning_rate": 5.50686039972423e-05,
      "loss": 1.3432,
      "step": 3150
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.84375,
      "learning_rate": 5.5053284460889234e-05,
      "loss": 1.2972,
      "step": 3155
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.40625,
      "learning_rate": 5.5037943303896246e-05,
      "loss": 1.2965,
      "step": 3160
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.125,
      "learning_rate": 5.502258053950253e-05,
      "loss": 1.2841,
      "step": 3165
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.34375,
      "learning_rate": 5.500719618096595e-05,
      "loss": 1.2324,
      "step": 3170
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.03125,
      "learning_rate": 5.4991790241563e-05,
      "loss": 1.2665,
      "step": 3175
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.5,
      "learning_rate": 5.497636273458878e-05,
      "loss": 1.2642,
      "step": 3180
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.46875,
      "learning_rate": 5.496091367335703e-05,
      "loss": 1.3122,
      "step": 3185
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.40625,
      "learning_rate": 5.494544307120007e-05,
      "loss": 1.343,
      "step": 3190
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.375,
      "learning_rate": 5.492995094146881e-05,
      "loss": 1.286,
      "step": 3195
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.859375,
      "learning_rate": 5.491443729753276e-05,
      "loss": 1.2499,
      "step": 3200
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.90625,
      "learning_rate": 5.489890215277997e-05,
      "loss": 1.3874,
      "step": 3205
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.21875,
      "learning_rate": 5.4883345520617064e-05,
      "loss": 1.2641,
      "step": 3210
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.71875,
      "learning_rate": 5.486776741446919e-05,
      "loss": 1.3424,
      "step": 3215
    },
    {
      "epoch": 0.57,
      "grad_norm": 3.625,
      "learning_rate": 5.485216784778006e-05,
      "loss": 1.3308,
      "step": 3220
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.96875,
      "learning_rate": 5.4836546834011864e-05,
      "loss": 1.2457,
      "step": 3225
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.78125,
      "learning_rate": 5.482090438664533e-05,
      "loss": 1.3051,
      "step": 3230
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.375,
      "learning_rate": 5.480524051917968e-05,
      "loss": 1.2924,
      "step": 3235
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.375,
      "learning_rate": 5.4789555245132614e-05,
      "loss": 1.3335,
      "step": 3240
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.90625,
      "learning_rate": 5.47738485780403e-05,
      "loss": 1.3339,
      "step": 3245
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.75,
      "learning_rate": 5.475812053145738e-05,
      "loss": 1.3429,
      "step": 3250
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.21875,
      "learning_rate": 5.474237111895695e-05,
      "loss": 1.3206,
      "step": 3255
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.8125,
      "learning_rate": 5.472660035413052e-05,
      "loss": 1.285,
      "step": 3260
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.59375,
      "learning_rate": 5.471080825058807e-05,
      "loss": 1.2688,
      "step": 3265
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.0,
      "learning_rate": 5.469499482195794e-05,
      "loss": 1.2592,
      "step": 3270
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.03125,
      "learning_rate": 5.467916008188692e-05,
      "loss": 1.3642,
      "step": 3275
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.84375,
      "learning_rate": 5.4663304044040157e-05,
      "loss": 1.2997,
      "step": 3280
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.96875,
      "learning_rate": 5.46474267221012e-05,
      "loss": 1.3506,
      "step": 3285
    },
    {
      "epoch": 0.58,
      "grad_norm": 3.828125,
      "learning_rate": 5.463152812977196e-05,
      "loss": 1.2474,
      "step": 3290
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.625,
      "learning_rate": 5.4615608280772714e-05,
      "loss": 1.2861,
      "step": 3295
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.4375,
      "learning_rate": 5.459966718884206e-05,
      "loss": 1.2688,
      "step": 3300
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.09375,
      "learning_rate": 5.458370486773694e-05,
      "loss": 1.2938,
      "step": 3305
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.3125,
      "learning_rate": 5.4567721331232634e-05,
      "loss": 1.321,
      "step": 3310
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.125,
      "learning_rate": 5.455171659312269e-05,
      "loss": 1.279,
      "step": 3315
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.375,
      "learning_rate": 5.4535690667219e-05,
      "loss": 1.2736,
      "step": 3320
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.5,
      "learning_rate": 5.4519643567351704e-05,
      "loss": 1.3178,
      "step": 3325
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.65625,
      "learning_rate": 5.4503575307369225e-05,
      "loss": 1.2609,
      "step": 3330
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.59375,
      "learning_rate": 5.448748590113826e-05,
      "loss": 1.3433,
      "step": 3335
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.78125,
      "learning_rate": 5.447137536254374e-05,
      "loss": 1.3147,
      "step": 3340
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.796875,
      "learning_rate": 5.445524370548885e-05,
      "loss": 1.3675,
      "step": 3345
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.0625,
      "learning_rate": 5.443909094389497e-05,
      "loss": 1.3018,
      "step": 3350
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.09375,
      "learning_rate": 5.442291709170172e-05,
      "loss": 1.302,
      "step": 3355
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.40625,
      "learning_rate": 5.4406722162866915e-05,
      "loss": 1.2886,
      "step": 3360
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.3125,
      "learning_rate": 5.439050617136656e-05,
      "loss": 1.2772,
      "step": 3365
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.0625,
      "learning_rate": 5.437426913119482e-05,
      "loss": 1.2609,
      "step": 3370
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.71875,
      "learning_rate": 5.435801105636403e-05,
      "loss": 1.2131,
      "step": 3375
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.125,
      "learning_rate": 5.434173196090471e-05,
      "loss": 1.2784,
      "step": 3380
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.71875,
      "learning_rate": 5.432543185886548e-05,
      "loss": 1.2992,
      "step": 3385
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.875,
      "learning_rate": 5.4309110764313104e-05,
      "loss": 1.1948,
      "step": 3390
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.59375,
      "learning_rate": 5.429276869133247e-05,
      "loss": 1.3929,
      "step": 3395
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.09375,
      "learning_rate": 5.4276405654026564e-05,
      "loss": 1.304,
      "step": 3400
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.9375,
      "learning_rate": 5.426002166651644e-05,
      "loss": 1.3069,
      "step": 3405
    },
    {
      "epoch": 0.6,
      "grad_norm": 4.25,
      "learning_rate": 5.424361674294128e-05,
      "loss": 1.2989,
      "step": 3410
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.40625,
      "learning_rate": 5.422719089745829e-05,
      "loss": 1.3277,
      "step": 3415
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.65625,
      "learning_rate": 5.421074414424276e-05,
      "loss": 1.3051,
      "step": 3420
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.03125,
      "learning_rate": 5.4194276497488015e-05,
      "loss": 1.2802,
      "step": 3425
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.1875,
      "learning_rate": 5.417778797140539e-05,
      "loss": 1.3996,
      "step": 3430
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.78125,
      "learning_rate": 5.416127858022426e-05,
      "loss": 1.2566,
      "step": 3435
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.65625,
      "learning_rate": 5.414474833819202e-05,
      "loss": 1.3191,
      "step": 3440
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.6875,
      "learning_rate": 5.412819725957403e-05,
      "loss": 1.2944,
      "step": 3445
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.59375,
      "learning_rate": 5.411162535865364e-05,
      "loss": 1.2609,
      "step": 3450
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.375,
      "learning_rate": 5.409503264973217e-05,
      "loss": 1.3204,
      "step": 3455
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.75,
      "learning_rate": 5.407841914712891e-05,
      "loss": 1.2741,
      "step": 3460
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.6875,
      "learning_rate": 5.406178486518108e-05,
      "loss": 1.2724,
      "step": 3465
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.28125,
      "learning_rate": 5.4045129818243844e-05,
      "loss": 1.3163,
      "step": 3470
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.53125,
      "learning_rate": 5.402845402069027e-05,
      "loss": 1.2788,
      "step": 3475
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.03125,
      "learning_rate": 5.401175748691133e-05,
      "loss": 1.2765,
      "step": 3480
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.53125,
      "learning_rate": 5.399504023131593e-05,
      "loss": 1.3242,
      "step": 3485
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.1875,
      "learning_rate": 5.397830226833081e-05,
      "loss": 1.3258,
      "step": 3490
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.375,
      "learning_rate": 5.396154361240061e-05,
      "loss": 1.2323,
      "step": 3495
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.90625,
      "learning_rate": 5.3944764277987825e-05,
      "loss": 1.3304,
      "step": 3500
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.25,
      "learning_rate": 5.3927964279572776e-05,
      "loss": 1.3493,
      "step": 3505
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.375,
      "learning_rate": 5.391114363165365e-05,
      "loss": 1.2971,
      "step": 3510
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.59375,
      "learning_rate": 5.3894302348746404e-05,
      "loss": 1.298,
      "step": 3515
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.4375,
      "learning_rate": 5.3877440445384875e-05,
      "loss": 1.3231,
      "step": 3520
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.84375,
      "learning_rate": 5.386055793612061e-05,
      "loss": 1.3774,
      "step": 3525
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.4375,
      "learning_rate": 5.3843654835523025e-05,
      "loss": 1.2998,
      "step": 3530
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.71875,
      "learning_rate": 5.382673115817923e-05,
      "loss": 1.2647,
      "step": 3535
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.875,
      "learning_rate": 5.380978691869414e-05,
      "loss": 1.2855,
      "step": 3540
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.6875,
      "learning_rate": 5.37928221316904e-05,
      "loss": 1.2911,
      "step": 3545
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.15625,
      "learning_rate": 5.3775836811808384e-05,
      "loss": 1.2909,
      "step": 3550
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.03125,
      "learning_rate": 5.375883097370619e-05,
      "loss": 1.3173,
      "step": 3555
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.8125,
      "learning_rate": 5.374180463205962e-05,
      "loss": 1.3441,
      "step": 3560
    },
    {
      "epoch": 0.63,
      "grad_norm": 3.90625,
      "learning_rate": 5.372475780156219e-05,
      "loss": 1.3172,
      "step": 3565
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.0625,
      "learning_rate": 5.370769049692506e-05,
      "loss": 1.3002,
      "step": 3570
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.34375,
      "learning_rate": 5.369060273287709e-05,
      "loss": 1.2938,
      "step": 3575
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.65625,
      "learning_rate": 5.3673494524164775e-05,
      "loss": 1.3679,
      "step": 3580
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.25,
      "learning_rate": 5.365636588555228e-05,
      "loss": 1.2639,
      "step": 3585
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.5625,
      "learning_rate": 5.3639216831821364e-05,
      "loss": 1.2428,
      "step": 3590
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.1875,
      "learning_rate": 5.362204737777144e-05,
      "loss": 1.3073,
      "step": 3595
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.53125,
      "learning_rate": 5.3604857538219515e-05,
      "loss": 1.2776,
      "step": 3600
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.984375,
      "learning_rate": 5.358764732800017e-05,
      "loss": 1.2903,
      "step": 3605
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.625,
      "learning_rate": 5.357041676196558e-05,
      "loss": 1.2906,
      "step": 3610
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.71875,
      "learning_rate": 5.35531658549855e-05,
      "loss": 1.212,
      "step": 3615
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.28125,
      "learning_rate": 5.353589462194721e-05,
      "loss": 1.4188,
      "step": 3620
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.625,
      "learning_rate": 5.351860307775556e-05,
      "loss": 1.3063,
      "step": 3625
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.4375,
      "learning_rate": 5.350129123733291e-05,
      "loss": 1.3076,
      "step": 3630
    },
    {
      "epoch": 0.64,
      "grad_norm": 4.53125,
      "learning_rate": 5.348395911561912e-05,
      "loss": 1.2788,
      "step": 3635
    },
    {
      "epoch": 0.65,
      "grad_norm": 3.875,
      "learning_rate": 5.346660672757161e-05,
      "loss": 1.2826,
      "step": 3640
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.9375,
      "learning_rate": 5.3449234088165236e-05,
      "loss": 1.3334,
      "step": 3645
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.625,
      "learning_rate": 5.343184121239233e-05,
      "loss": 1.2731,
      "step": 3650
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.5,
      "learning_rate": 5.341442811526273e-05,
      "loss": 1.3533,
      "step": 3655
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.125,
      "learning_rate": 5.339699481180369e-05,
      "loss": 1.3177,
      "step": 3660
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.4375,
      "learning_rate": 5.33795413170599e-05,
      "loss": 1.3122,
      "step": 3665
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.59375,
      "learning_rate": 5.33620676460935e-05,
      "loss": 1.3433,
      "step": 3670
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.46875,
      "learning_rate": 5.334457381398402e-05,
      "loss": 1.2886,
      "step": 3675
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.3125,
      "learning_rate": 5.332705983582841e-05,
      "loss": 1.2162,
      "step": 3680
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.59375,
      "learning_rate": 5.330952572674098e-05,
      "loss": 1.2925,
      "step": 3685
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.65625,
      "learning_rate": 5.329197150185342e-05,
      "loss": 1.3291,
      "step": 3690
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.1875,
      "learning_rate": 5.327439717631479e-05,
      "loss": 1.2377,
      "step": 3695
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.5625,
      "learning_rate": 5.32568027652915e-05,
      "loss": 1.3072,
      "step": 3700
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.59375,
      "learning_rate": 5.323918828396728e-05,
      "loss": 1.3183,
      "step": 3705
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.125,
      "learning_rate": 5.322155374754317e-05,
      "loss": 1.325,
      "step": 3710
    },
    {
      "epoch": 0.66,
      "grad_norm": 3.9375,
      "learning_rate": 5.320389917123756e-05,
      "loss": 1.2649,
      "step": 3715
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.0625,
      "learning_rate": 5.318622457028609e-05,
      "loss": 1.2945,
      "step": 3720
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.5,
      "learning_rate": 5.3168529959941684e-05,
      "loss": 1.3274,
      "step": 3725
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.75,
      "learning_rate": 5.315081535547458e-05,
      "loss": 1.3552,
      "step": 3730
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.8125,
      "learning_rate": 5.3133080772172216e-05,
      "loss": 1.288,
      "step": 3735
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.125,
      "learning_rate": 5.311532622533929e-05,
      "loss": 1.2358,
      "step": 3740
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.5,
      "learning_rate": 5.309755173029775e-05,
      "loss": 1.2855,
      "step": 3745
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.5,
      "learning_rate": 5.307975730238673e-05,
      "loss": 1.3186,
      "step": 3750
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.125,
      "learning_rate": 5.306194295696257e-05,
      "loss": 1.3483,
      "step": 3755
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.09375,
      "learning_rate": 5.304410870939882e-05,
      "loss": 1.2975,
      "step": 3760
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.15625,
      "learning_rate": 5.302625457508618e-05,
      "loss": 1.3327,
      "step": 3765
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.3125,
      "learning_rate": 5.3008380569432526e-05,
      "loss": 1.3524,
      "step": 3770
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.25,
      "learning_rate": 5.299048670786288e-05,
      "loss": 1.2576,
      "step": 3775
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.5,
      "learning_rate": 5.297257300581938e-05,
      "loss": 1.3048,
      "step": 3780
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.46875,
      "learning_rate": 5.295463947876134e-05,
      "loss": 1.2838,
      "step": 3785
    },
    {
      "epoch": 0.67,
      "grad_norm": 3.890625,
      "learning_rate": 5.293668614216513e-05,
      "loss": 1.2591,
      "step": 3790
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.03125,
      "learning_rate": 5.2918713011524216e-05,
      "loss": 1.271,
      "step": 3795
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.75,
      "learning_rate": 5.290072010234919e-05,
      "loss": 1.2851,
      "step": 3800
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.65625,
      "learning_rate": 5.288270743016767e-05,
      "loss": 1.3198,
      "step": 3805
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.71875,
      "learning_rate": 5.286467501052434e-05,
      "loss": 1.3168,
      "step": 3810
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.75,
      "learning_rate": 5.284662285898094e-05,
      "loss": 1.2694,
      "step": 3815
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.59375,
      "learning_rate": 5.2828550991116234e-05,
      "loss": 1.3057,
      "step": 3820
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.59375,
      "learning_rate": 5.2810459422525973e-05,
      "loss": 1.334,
      "step": 3825
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.15625,
      "learning_rate": 5.2792348168822954e-05,
      "loss": 1.2662,
      "step": 3830
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.9375,
      "learning_rate": 5.277421724563693e-05,
      "loss": 1.3281,
      "step": 3835
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.71875,
      "learning_rate": 5.275606666861464e-05,
      "loss": 1.2915,
      "step": 3840
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.84375,
      "learning_rate": 5.2737896453419784e-05,
      "loss": 1.2695,
      "step": 3845
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.875,
      "learning_rate": 5.2719706615733e-05,
      "loss": 1.2676,
      "step": 3850
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.5,
      "learning_rate": 5.270149717125189e-05,
      "loss": 1.3432,
      "step": 3855
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.875,
      "learning_rate": 5.268326813569092e-05,
      "loss": 1.3136,
      "step": 3860
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.5,
      "learning_rate": 5.266501952478153e-05,
      "loss": 1.3011,
      "step": 3865
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.03125,
      "learning_rate": 5.264675135427199e-05,
      "loss": 1.2212,
      "step": 3870
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.34375,
      "learning_rate": 5.26284636399275e-05,
      "loss": 1.3356,
      "step": 3875
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.625,
      "learning_rate": 5.261015639753011e-05,
      "loss": 1.2184,
      "step": 3880
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.625,
      "learning_rate": 5.25918296428787e-05,
      "loss": 1.2665,
      "step": 3885
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.5,
      "learning_rate": 5.2573483391789026e-05,
      "loss": 1.3721,
      "step": 3890
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.65625,
      "learning_rate": 5.255511766009364e-05,
      "loss": 1.2977,
      "step": 3895
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.5,
      "learning_rate": 5.253673246364191e-05,
      "loss": 1.3042,
      "step": 3900
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.59375,
      "learning_rate": 5.251832781830002e-05,
      "loss": 1.2613,
      "step": 3905
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.53125,
      "learning_rate": 5.249990373995092e-05,
      "loss": 1.3259,
      "step": 3910
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.09375,
      "learning_rate": 5.248146024449433e-05,
      "loss": 1.2898,
      "step": 3915
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.5,
      "learning_rate": 5.246299734784675e-05,
      "loss": 1.2752,
      "step": 3920
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.1875,
      "learning_rate": 5.244451506594138e-05,
      "loss": 1.2829,
      "step": 3925
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.6875,
      "learning_rate": 5.2426013414728213e-05,
      "loss": 1.3093,
      "step": 3930
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.25,
      "learning_rate": 5.240749241017389e-05,
      "loss": 1.2484,
      "step": 3935
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.6875,
      "learning_rate": 5.238895206826178e-05,
      "loss": 1.3355,
      "step": 3940
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.46875,
      "learning_rate": 5.2370392404991965e-05,
      "loss": 1.2665,
      "step": 3945
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.3125,
      "learning_rate": 5.235181343638116e-05,
      "loss": 1.3069,
      "step": 3950
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.8125,
      "learning_rate": 5.233321517846277e-05,
      "loss": 1.3489,
      "step": 3955
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.15625,
      "learning_rate": 5.231459764728684e-05,
      "loss": 1.28,
      "step": 3960
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.125,
      "learning_rate": 5.229596085892002e-05,
      "loss": 1.2844,
      "step": 3965
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.5625,
      "learning_rate": 5.227730482944563e-05,
      "loss": 1.2787,
      "step": 3970
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.375,
      "learning_rate": 5.225862957496355e-05,
      "loss": 1.2874,
      "step": 3975
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.03125,
      "learning_rate": 5.223993511159028e-05,
      "loss": 1.3417,
      "step": 3980
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.96875,
      "learning_rate": 5.222122145545888e-05,
      "loss": 1.3058,
      "step": 3985
    },
    {
      "epoch": 0.71,
      "grad_norm": 3.984375,
      "learning_rate": 5.220248862271896e-05,
      "loss": 1.2472,
      "step": 3990
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.09375,
      "learning_rate": 5.218373662953672e-05,
      "loss": 1.2801,
      "step": 3995
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.28125,
      "learning_rate": 5.2164965492094874e-05,
      "loss": 1.2347,
      "step": 4000
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.4375,
      "learning_rate": 5.2146175226592634e-05,
      "loss": 1.2724,
      "step": 4005
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.09375,
      "learning_rate": 5.212736584924576e-05,
      "loss": 1.3077,
      "step": 4010
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.21875,
      "learning_rate": 5.210853737628647e-05,
      "loss": 1.3172,
      "step": 4015
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.40625,
      "learning_rate": 5.20896898239635e-05,
      "loss": 1.272,
      "step": 4020
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.03125,
      "learning_rate": 5.207082320854201e-05,
      "loss": 1.2565,
      "step": 4025
    },
    {
      "epoch": 0.71,
      "grad_norm": 4.59375,
      "learning_rate": 5.205193754630362e-05,
      "loss": 1.3023,
      "step": 4030
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.59375,
      "learning_rate": 5.203303285354642e-05,
      "loss": 1.2946,
      "step": 4035
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.4375,
      "learning_rate": 5.2014109146584905e-05,
      "loss": 1.2827,
      "step": 4040
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.03125,
      "learning_rate": 5.199516644174995e-05,
      "loss": 1.3168,
      "step": 4045
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.46875,
      "learning_rate": 5.197620475538885e-05,
      "loss": 1.3188,
      "step": 4050
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.28125,
      "learning_rate": 5.195722410386529e-05,
      "loss": 1.2968,
      "step": 4055
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.5625,
      "learning_rate": 5.193822450355933e-05,
      "loss": 1.2972,
      "step": 4060
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.5625,
      "learning_rate": 5.191920597086732e-05,
      "loss": 1.2119,
      "step": 4065
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.21875,
      "learning_rate": 5.190016852220202e-05,
      "loss": 1.3148,
      "step": 4070
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.53125,
      "learning_rate": 5.188111217399249e-05,
      "loss": 1.2963,
      "step": 4075
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.5,
      "learning_rate": 5.186203694268408e-05,
      "loss": 1.3657,
      "step": 4080
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.0,
      "learning_rate": 5.1842942844738456e-05,
      "loss": 1.3618,
      "step": 4085
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.53125,
      "learning_rate": 5.182382989663357e-05,
      "loss": 1.3017,
      "step": 4090
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.25,
      "learning_rate": 5.1804698114863634e-05,
      "loss": 1.2431,
      "step": 4095
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.28125,
      "learning_rate": 5.17855475159391e-05,
      "loss": 1.2422,
      "step": 4100
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.15625,
      "learning_rate": 5.176637811638669e-05,
      "loss": 1.2612,
      "step": 4105
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.8125,
      "learning_rate": 5.1747189932749315e-05,
      "loss": 1.3168,
      "step": 4110
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.125,
      "learning_rate": 5.172798298158613e-05,
      "loss": 1.2903,
      "step": 4115
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.3125,
      "learning_rate": 5.170875727947245e-05,
      "loss": 1.3711,
      "step": 4120
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.34375,
      "learning_rate": 5.1689512842999796e-05,
      "loss": 1.2792,
      "step": 4125
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.5625,
      "learning_rate": 5.1670249688775865e-05,
      "loss": 1.3702,
      "step": 4130
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.8125,
      "learning_rate": 5.165096783342448e-05,
      "loss": 1.2846,
      "step": 4135
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.46875,
      "learning_rate": 5.1631667293585637e-05,
      "loss": 1.2773,
      "step": 4140
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.375,
      "learning_rate": 5.161234808591542e-05,
      "loss": 1.3789,
      "step": 4145
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.84375,
      "learning_rate": 5.159301022708603e-05,
      "loss": 1.3553,
      "step": 4150
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.125,
      "learning_rate": 5.15736537337858e-05,
      "loss": 1.2859,
      "step": 4155
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.5625,
      "learning_rate": 5.1554278622719096e-05,
      "loss": 1.3654,
      "step": 4160
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.96875,
      "learning_rate": 5.153488491060638e-05,
      "loss": 1.3205,
      "step": 4165
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.5625,
      "learning_rate": 5.151547261418416e-05,
      "loss": 1.2788,
      "step": 4170
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.3125,
      "learning_rate": 5.149604175020498e-05,
      "loss": 1.2954,
      "step": 4175
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.09375,
      "learning_rate": 5.1476592335437406e-05,
      "loss": 1.2905,
      "step": 4180
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.3125,
      "learning_rate": 5.1457124386666025e-05,
      "loss": 1.293,
      "step": 4185
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.375,
      "learning_rate": 5.143763792069139e-05,
      "loss": 1.2971,
      "step": 4190
    },
    {
      "epoch": 0.74,
      "grad_norm": 3.859375,
      "learning_rate": 5.141813295433007e-05,
      "loss": 1.2769,
      "step": 4195
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.28125,
      "learning_rate": 5.139860950441457e-05,
      "loss": 1.2683,
      "step": 4200
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.0625,
      "learning_rate": 5.137906758779337e-05,
      "loss": 1.2046,
      "step": 4205
    },
    {
      "epoch": 0.75,
      "grad_norm": 3.890625,
      "learning_rate": 5.1359507221330876e-05,
      "loss": 1.2322,
      "step": 4210
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.625,
      "learning_rate": 5.133992842190741e-05,
      "loss": 1.3289,
      "step": 4215
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.71875,
      "learning_rate": 5.132033120641921e-05,
      "loss": 1.332,
      "step": 4220
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.875,
      "learning_rate": 5.130071559177839e-05,
      "loss": 1.3364,
      "step": 4225
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.59375,
      "learning_rate": 5.128108159491297e-05,
      "loss": 1.3073,
      "step": 4230
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.65625,
      "learning_rate": 5.126142923276681e-05,
      "loss": 1.2453,
      "step": 4235
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.15625,
      "learning_rate": 5.1241758522299644e-05,
      "loss": 1.2716,
      "step": 4240
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.875,
      "learning_rate": 5.1222069480487016e-05,
      "loss": 1.356,
      "step": 4245
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.9375,
      "learning_rate": 5.1202362124320285e-05,
      "loss": 1.21,
      "step": 4250
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.90625,
      "learning_rate": 5.118263647080665e-05,
      "loss": 1.3601,
      "step": 4255
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.375,
      "learning_rate": 5.116289253696905e-05,
      "loss": 1.3043,
      "step": 4260
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.53125,
      "learning_rate": 5.114313033984627e-05,
      "loss": 1.279,
      "step": 4265
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.34375,
      "learning_rate": 5.112334989649278e-05,
      "loss": 1.2923,
      "step": 4270
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.03125,
      "learning_rate": 5.110355122397885e-05,
      "loss": 1.3014,
      "step": 4275
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.9375,
      "learning_rate": 5.108373433939045e-05,
      "loss": 1.4038,
      "step": 4280
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.21875,
      "learning_rate": 5.1063899259829276e-05,
      "loss": 1.3483,
      "step": 4285
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.84375,
      "learning_rate": 5.1044046002412736e-05,
      "loss": 1.3516,
      "step": 4290
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.65625,
      "learning_rate": 5.102417458427392e-05,
      "loss": 1.3442,
      "step": 4295
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.0625,
      "learning_rate": 5.1004285022561584e-05,
      "loss": 1.2925,
      "step": 4300
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.21875,
      "learning_rate": 5.0984377334440154e-05,
      "loss": 1.2403,
      "step": 4305
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.3125,
      "learning_rate": 5.096445153708969e-05,
      "loss": 1.3107,
      "step": 4310
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.4375,
      "learning_rate": 5.0944507647705873e-05,
      "loss": 1.297,
      "step": 4315
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.96875,
      "learning_rate": 5.09245456835e-05,
      "loss": 1.3165,
      "step": 4320
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.03125,
      "learning_rate": 5.090456566169899e-05,
      "loss": 1.3021,
      "step": 4325
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.4375,
      "learning_rate": 5.0884567599545316e-05,
      "loss": 1.325,
      "step": 4330
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.875,
      "learning_rate": 5.0864551514297027e-05,
      "loss": 1.2788,
      "step": 4335
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.15625,
      "learning_rate": 5.084451742322774e-05,
      "loss": 1.2565,
      "step": 4340
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.0,
      "learning_rate": 5.0824465343626595e-05,
      "loss": 1.3245,
      "step": 4345
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.65625,
      "learning_rate": 5.0804395292798254e-05,
      "loss": 1.2891,
      "step": 4350
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.3125,
      "learning_rate": 5.078430728806291e-05,
      "loss": 1.2448,
      "step": 4355
    },
    {
      "epoch": 0.77,
      "grad_norm": 3.828125,
      "learning_rate": 5.076420134675621e-05,
      "loss": 1.241,
      "step": 4360
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.40625,
      "learning_rate": 5.0744077486229334e-05,
      "loss": 1.3236,
      "step": 4365
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.875,
      "learning_rate": 5.0723935723848884e-05,
      "loss": 1.3193,
      "step": 4370
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.3125,
      "learning_rate": 5.0703776076996916e-05,
      "loss": 1.2602,
      "step": 4375
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.46875,
      "learning_rate": 5.0683598563070944e-05,
      "loss": 1.3601,
      "step": 4380
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.40625,
      "learning_rate": 5.066340319948387e-05,
      "loss": 1.2986,
      "step": 4385
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.09375,
      "learning_rate": 5.064319000366403e-05,
      "loss": 1.2844,
      "step": 4390
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.125,
      "learning_rate": 5.062295899305512e-05,
      "loss": 1.3314,
      "step": 4395
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.46875,
      "learning_rate": 5.060271018511622e-05,
      "loss": 1.2696,
      "step": 4400
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.1875,
      "learning_rate": 5.0582443597321796e-05,
      "loss": 1.2538,
      "step": 4405
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.8125,
      "learning_rate": 5.056215924716162e-05,
      "loss": 1.3236,
      "step": 4410
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.75,
      "learning_rate": 5.05418571521408e-05,
      "loss": 1.3391,
      "step": 4415
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.46875,
      "learning_rate": 5.052153732977978e-05,
      "loss": 1.2908,
      "step": 4420
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.6875,
      "learning_rate": 5.0501199797614274e-05,
      "loss": 1.2552,
      "step": 4425
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.4375,
      "learning_rate": 5.0480844573195307e-05,
      "loss": 1.2899,
      "step": 4430
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.5,
      "learning_rate": 5.0460471674089145e-05,
      "loss": 1.2976,
      "step": 4435
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.3125,
      "learning_rate": 5.0440081117877326e-05,
      "loss": 1.2849,
      "step": 4440
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.6875,
      "learning_rate": 5.041967292215661e-05,
      "loss": 1.2499,
      "step": 4445
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.3125,
      "learning_rate": 5.0399247104539e-05,
      "loss": 1.238,
      "step": 4450
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.25,
      "learning_rate": 5.037880368265169e-05,
      "loss": 1.296,
      "step": 4455
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.75,
      "learning_rate": 5.035834267413708e-05,
      "loss": 1.3124,
      "step": 4460
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.09375,
      "learning_rate": 5.0337864096652734e-05,
      "loss": 1.3708,
      "step": 4465
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.09375,
      "learning_rate": 5.031736796787138e-05,
      "loss": 1.3199,
      "step": 4470
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.34375,
      "learning_rate": 5.02968543054809e-05,
      "loss": 1.2688,
      "step": 4475
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.28125,
      "learning_rate": 5.0276323127184296e-05,
      "loss": 1.2993,
      "step": 4480
    },
    {
      "epoch": 0.79,
      "grad_norm": 4.84375,
      "learning_rate": 5.0255774450699695e-05,
      "loss": 1.2572,
      "step": 4485
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.40625,
      "learning_rate": 5.023520829376033e-05,
      "loss": 1.2632,
      "step": 4490
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.3125,
      "learning_rate": 5.02146246741145e-05,
      "loss": 1.2746,
      "step": 4495
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.15625,
      "learning_rate": 5.0194023609525604e-05,
      "loss": 1.3369,
      "step": 4500
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.28125,
      "learning_rate": 5.017340511777205e-05,
      "loss": 1.2647,
      "step": 4505
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.90625,
      "learning_rate": 5.0152769216647325e-05,
      "loss": 1.2443,
      "step": 4510
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.3125,
      "learning_rate": 5.013211592395993e-05,
      "loss": 1.3092,
      "step": 4515
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.6875,
      "learning_rate": 5.011144525753339e-05,
      "loss": 1.2733,
      "step": 4520
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.8125,
      "learning_rate": 5.009075723520617e-05,
      "loss": 1.3003,
      "step": 4525
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.6875,
      "learning_rate": 5.007005187483178e-05,
      "loss": 1.2904,
      "step": 4530
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.15625,
      "learning_rate": 5.0049329194278636e-05,
      "loss": 1.346,
      "step": 4535
    },
    {
      "epoch": 0.8,
      "grad_norm": 4.59375,
      "learning_rate": 5.0028589211430144e-05,
      "loss": 1.2387,
      "step": 4540
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.59375,
      "learning_rate": 5.0007831944184615e-05,
      "loss": 1.3197,
      "step": 4545
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.25,
      "learning_rate": 4.9987057410455304e-05,
      "loss": 1.2982,
      "step": 4550
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.21875,
      "learning_rate": 4.9966265628170314e-05,
      "loss": 1.3044,
      "step": 4555
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.0625,
      "learning_rate": 4.9945456615272696e-05,
      "loss": 1.3252,
      "step": 4560
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.5625,
      "learning_rate": 4.992463038972032e-05,
      "loss": 1.2837,
      "step": 4565
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.875,
      "learning_rate": 4.9903786969485956e-05,
      "loss": 1.3049,
      "step": 4570
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.03125,
      "learning_rate": 4.9882926372557183e-05,
      "loss": 1.3271,
      "step": 4575
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.34375,
      "learning_rate": 4.986204861693639e-05,
      "loss": 1.2451,
      "step": 4580
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.15625,
      "learning_rate": 4.984115372064081e-05,
      "loss": 1.3575,
      "step": 4585
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.53125,
      "learning_rate": 4.982024170170244e-05,
      "loss": 1.3739,
      "step": 4590
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.4375,
      "learning_rate": 4.9799312578168083e-05,
      "loss": 1.2905,
      "step": 4595
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.71875,
      "learning_rate": 4.977836636809927e-05,
      "loss": 1.2897,
      "step": 4600
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.46875,
      "learning_rate": 4.97574030895723e-05,
      "loss": 1.3045,
      "step": 4605
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.34375,
      "learning_rate": 4.973642276067818e-05,
      "loss": 1.2284,
      "step": 4610
    },
    {
      "epoch": 0.82,
      "grad_norm": 3.9375,
      "learning_rate": 4.971542539952267e-05,
      "loss": 1.2265,
      "step": 4615
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.09375,
      "learning_rate": 4.969441102422618e-05,
      "loss": 1.2963,
      "step": 4620
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.25,
      "learning_rate": 4.9673379652923835e-05,
      "loss": 1.294,
      "step": 4625
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.5625,
      "learning_rate": 4.9652331303765426e-05,
      "loss": 1.4051,
      "step": 4630
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.125,
      "learning_rate": 4.963126599491538e-05,
      "loss": 1.2579,
      "step": 4635
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.625,
      "learning_rate": 4.961018374455277e-05,
      "loss": 1.3055,
      "step": 4640
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.21875,
      "learning_rate": 4.95890845708713e-05,
      "loss": 1.2871,
      "step": 4645
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.03125,
      "learning_rate": 4.956796849207924e-05,
      "loss": 1.284,
      "step": 4650
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.25,
      "learning_rate": 4.95468355263995e-05,
      "loss": 1.282,
      "step": 4655
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.75,
      "learning_rate": 4.952568569206954e-05,
      "loss": 1.2939,
      "step": 4660
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.25,
      "learning_rate": 4.9504519007341355e-05,
      "loss": 1.2865,
      "step": 4665
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.9375,
      "learning_rate": 4.948333549048151e-05,
      "loss": 1.2585,
      "step": 4670
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.84375,
      "learning_rate": 4.946213515977111e-05,
      "loss": 1.2563,
      "step": 4675
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.03125,
      "learning_rate": 4.944091803350573e-05,
      "loss": 1.2754,
      "step": 4680
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.875,
      "learning_rate": 4.9419684129995454e-05,
      "loss": 1.3099,
      "step": 4685
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.3125,
      "learning_rate": 4.939843346756486e-05,
      "loss": 1.2882,
      "step": 4690
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.875,
      "learning_rate": 4.9377166064552976e-05,
      "loss": 1.2923,
      "step": 4695
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.0,
      "learning_rate": 4.9355881939313264e-05,
      "loss": 1.2603,
      "step": 4700
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.8125,
      "learning_rate": 4.933458111021366e-05,
      "loss": 1.2562,
      "step": 4705
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.6875,
      "learning_rate": 4.931326359563645e-05,
      "loss": 1.3307,
      "step": 4710
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.40625,
      "learning_rate": 4.929192941397839e-05,
      "loss": 1.3033,
      "step": 4715
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.84375,
      "learning_rate": 4.927057858365056e-05,
      "loss": 1.3993,
      "step": 4720
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.875,
      "learning_rate": 4.9249211123078454e-05,
      "loss": 1.3516,
      "step": 4725
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.625,
      "learning_rate": 4.922782705070187e-05,
      "loss": 1.3344,
      "step": 4730
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.1875,
      "learning_rate": 4.9206426384975e-05,
      "loss": 1.2776,
      "step": 4735
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.96875,
      "learning_rate": 4.9185009144366305e-05,
      "loss": 1.3762,
      "step": 4740
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.9375,
      "learning_rate": 4.9163575347358565e-05,
      "loss": 1.2513,
      "step": 4745
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.03125,
      "learning_rate": 4.9142125012448864e-05,
      "loss": 1.2112,
      "step": 4750
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.1875,
      "learning_rate": 4.912065815814854e-05,
      "loss": 1.29,
      "step": 4755
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.0625,
      "learning_rate": 4.90991748029832e-05,
      "loss": 1.3301,
      "step": 4760
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.25,
      "learning_rate": 4.9077674965492675e-05,
      "loss": 1.3531,
      "step": 4765
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.90625,
      "learning_rate": 4.905615866423103e-05,
      "loss": 1.32,
      "step": 4770
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.875,
      "learning_rate": 4.903462591776655e-05,
      "loss": 1.2463,
      "step": 4775
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.34375,
      "learning_rate": 4.901307674468168e-05,
      "loss": 1.2764,
      "step": 4780
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.84375,
      "learning_rate": 4.899151116357307e-05,
      "loss": 1.3328,
      "step": 4785
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.78125,
      "learning_rate": 4.896992919305153e-05,
      "loss": 1.3076,
      "step": 4790
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.984375,
      "learning_rate": 4.894833085174197e-05,
      "loss": 1.3817,
      "step": 4795
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.3125,
      "learning_rate": 4.89267161582835e-05,
      "loss": 1.2894,
      "step": 4800
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.59375,
      "learning_rate": 4.8905085131329287e-05,
      "loss": 1.2889,
      "step": 4805
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.96875,
      "learning_rate": 4.8883437789546605e-05,
      "loss": 1.3055,
      "step": 4810
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.40625,
      "learning_rate": 4.886177415161681e-05,
      "loss": 1.2963,
      "step": 4815
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.15625,
      "learning_rate": 4.884009423623534e-05,
      "loss": 1.3166,
      "step": 4820
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.78125,
      "learning_rate": 4.881839806211167e-05,
      "loss": 1.272,
      "step": 4825
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.6875,
      "learning_rate": 4.8796685647969267e-05,
      "loss": 1.253,
      "step": 4830
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.75,
      "learning_rate": 4.877495701254566e-05,
      "loss": 1.3501,
      "step": 4835
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.15625,
      "learning_rate": 4.875321217459237e-05,
      "loss": 1.239,
      "step": 4840
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.59375,
      "learning_rate": 4.87314511528749e-05,
      "loss": 1.3132,
      "step": 4845
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.28125,
      "learning_rate": 4.870967396617268e-05,
      "loss": 1.2862,
      "step": 4850
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.5,
      "learning_rate": 4.868788063327915e-05,
      "loss": 1.2742,
      "step": 4855
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.0625,
      "learning_rate": 4.866607117300166e-05,
      "loss": 1.3141,
      "step": 4860
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.4375,
      "learning_rate": 4.864424560416146e-05,
      "loss": 1.3136,
      "step": 4865
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.125,
      "learning_rate": 4.862240394559372e-05,
      "loss": 1.3039,
      "step": 4870
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.625,
      "learning_rate": 4.8600546216147496e-05,
      "loss": 1.1936,
      "step": 4875
    },
    {
      "epoch": 0.86,
      "grad_norm": 4.46875,
      "learning_rate": 4.85786724346857e-05,
      "loss": 1.2879,
      "step": 4880
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.5,
      "learning_rate": 4.8556782620085106e-05,
      "loss": 1.2209,
      "step": 4885
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.90625,
      "learning_rate": 4.8534876791236335e-05,
      "loss": 1.3453,
      "step": 4890
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.71875,
      "learning_rate": 4.8512954967043796e-05,
      "loss": 1.3152,
      "step": 4895
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.28125,
      "learning_rate": 4.849101716642573e-05,
      "loss": 1.3382,
      "step": 4900
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.0625,
      "learning_rate": 4.846906340831417e-05,
      "loss": 1.248,
      "step": 4905
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.3125,
      "learning_rate": 4.844709371165489e-05,
      "loss": 1.2956,
      "step": 4910
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.0,
      "learning_rate": 4.842510809540744e-05,
      "loss": 1.2005,
      "step": 4915
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.78125,
      "learning_rate": 4.8403106578545106e-05,
      "loss": 1.321,
      "step": 4920
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.71875,
      "learning_rate": 4.8381089180054905e-05,
      "loss": 1.3446,
      "step": 4925
    },
    {
      "epoch": 0.87,
      "grad_norm": 3.6875,
      "learning_rate": 4.835905591893753e-05,
      "loss": 1.247,
      "step": 4930
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.28125,
      "learning_rate": 4.8337006814207396e-05,
      "loss": 1.3416,
      "step": 4935
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.65625,
      "learning_rate": 4.831494188489257e-05,
      "loss": 1.3565,
      "step": 4940
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.40625,
      "learning_rate": 4.829286115003478e-05,
      "loss": 1.2465,
      "step": 4945
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.78125,
      "learning_rate": 4.827076462868941e-05,
      "loss": 1.2708,
      "step": 4950
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.90625,
      "learning_rate": 4.8248652339925455e-05,
      "loss": 1.2907,
      "step": 4955
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.890625,
      "learning_rate": 4.822652430282549e-05,
      "loss": 1.3552,
      "step": 4960
    },
    {
      "epoch": 0.88,
      "grad_norm": 3.96875,
      "learning_rate": 4.820438053648573e-05,
      "loss": 1.2673,
      "step": 4965
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.75,
      "learning_rate": 4.8182221060015935e-05,
      "loss": 1.2592,
      "step": 4970
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.40625,
      "learning_rate": 4.816004589253943e-05,
      "loss": 1.3148,
      "step": 4975
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.8125,
      "learning_rate": 4.813785505319307e-05,
      "loss": 1.2941,
      "step": 4980
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.03125,
      "learning_rate": 4.811564856112726e-05,
      "loss": 1.2849,
      "step": 4985
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.0625,
      "learning_rate": 4.809342643550588e-05,
      "loss": 1.214,
      "step": 4990
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.25,
      "learning_rate": 4.807118869550634e-05,
      "loss": 1.2822,
      "step": 4995
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.03125,
      "learning_rate": 4.804893536031948e-05,
      "loss": 1.2748,
      "step": 5000
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.5,
      "learning_rate": 4.802666644914964e-05,
      "loss": 1.2134,
      "step": 5005
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.75,
      "learning_rate": 4.8004381981214566e-05,
      "loss": 1.2885,
      "step": 5010
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.75,
      "learning_rate": 4.7982081975745476e-05,
      "loss": 1.226,
      "step": 5015
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.6875,
      "learning_rate": 4.7959766451986947e-05,
      "loss": 1.2971,
      "step": 5020
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.28125,
      "learning_rate": 4.793743542919697e-05,
      "loss": 1.2564,
      "step": 5025
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.0,
      "learning_rate": 4.791508892664691e-05,
      "loss": 1.2823,
      "step": 5030
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.03125,
      "learning_rate": 4.7892726963621504e-05,
      "loss": 1.306,
      "step": 5035
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.8125,
      "learning_rate": 4.78703495594188e-05,
      "loss": 1.3166,
      "step": 5040
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.78125,
      "learning_rate": 4.78479567333502e-05,
      "loss": 1.2067,
      "step": 5045
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.25,
      "learning_rate": 4.782554850474042e-05,
      "loss": 1.3576,
      "step": 5050
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.03125,
      "learning_rate": 4.780312489292742e-05,
      "loss": 1.2865,
      "step": 5055
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.0,
      "learning_rate": 4.778068591726249e-05,
      "loss": 1.3035,
      "step": 5060
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.4375,
      "learning_rate": 4.775823159711016e-05,
      "loss": 1.2351,
      "step": 5065
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.84375,
      "learning_rate": 4.773576195184819e-05,
      "loss": 1.2538,
      "step": 5070
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.15625,
      "learning_rate": 4.771327700086758e-05,
      "loss": 1.3922,
      "step": 5075
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.34375,
      "learning_rate": 4.769077676357253e-05,
      "loss": 1.2832,
      "step": 5080
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.9375,
      "learning_rate": 4.766826125938043e-05,
      "loss": 1.2454,
      "step": 5085
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.21875,
      "learning_rate": 4.764573050772187e-05,
      "loss": 1.2927,
      "step": 5090
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.6875,
      "learning_rate": 4.7623184528040554e-05,
      "loss": 1.3244,
      "step": 5095
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.953125,
      "learning_rate": 4.7600623339793375e-05,
      "loss": 1.2799,
      "step": 5100
    },
    {
      "epoch": 0.9,
      "grad_norm": 3.953125,
      "learning_rate": 4.757804696245031e-05,
      "loss": 1.3255,
      "step": 5105
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.84375,
      "learning_rate": 4.755545541549447e-05,
      "loss": 1.3351,
      "step": 5110
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.9375,
      "learning_rate": 4.753284871842205e-05,
      "loss": 1.3204,
      "step": 5115
    },
    {
      "epoch": 0.91,
      "grad_norm": 3.734375,
      "learning_rate": 4.751022689074232e-05,
      "loss": 1.2998,
      "step": 5120
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.40625,
      "learning_rate": 4.7487589951977604e-05,
      "loss": 1.2709,
      "step": 5125
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.1875,
      "learning_rate": 4.7464937921663267e-05,
      "loss": 1.3357,
      "step": 5130
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.4375,
      "learning_rate": 4.7442270819347705e-05,
      "loss": 1.3468,
      "step": 5135
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.1875,
      "learning_rate": 4.741958866459231e-05,
      "loss": 1.2886,
      "step": 5140
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.8125,
      "learning_rate": 4.739689147697147e-05,
      "loss": 1.348,
      "step": 5145
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.46875,
      "learning_rate": 4.7374179276072554e-05,
      "loss": 1.2766,
      "step": 5150
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.4375,
      "learning_rate": 4.735145208149587e-05,
      "loss": 1.2971,
      "step": 5155
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.96875,
      "learning_rate": 4.7328709912854676e-05,
      "loss": 1.301,
      "step": 5160
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.0625,
      "learning_rate": 4.730595278977514e-05,
      "loss": 1.2901,
      "step": 5165
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.28125,
      "learning_rate": 4.728318073189636e-05,
      "loss": 1.2202,
      "step": 5170
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.40625,
      "learning_rate": 4.726039375887032e-05,
      "loss": 1.2634,
      "step": 5175
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.40625,
      "learning_rate": 4.7237591890361846e-05,
      "loss": 1.3181,
      "step": 5180
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.125,
      "learning_rate": 4.721477514604864e-05,
      "loss": 1.287,
      "step": 5185
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.8125,
      "learning_rate": 4.719194354562124e-05,
      "loss": 1.2871,
      "step": 5190
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.6875,
      "learning_rate": 4.7169097108783e-05,
      "loss": 1.2688,
      "step": 5195
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.765625,
      "learning_rate": 4.714623585525009e-05,
      "loss": 1.2646,
      "step": 5200
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.09375,
      "learning_rate": 4.7123359804751446e-05,
      "loss": 1.2724,
      "step": 5205
    },
    {
      "epoch": 0.92,
      "grad_norm": 4.75,
      "learning_rate": 4.7100468977028794e-05,
      "loss": 1.2753,
      "step": 5210
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.953125,
      "learning_rate": 4.7077563391836605e-05,
      "loss": 1.2553,
      "step": 5215
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.90625,
      "learning_rate": 4.705464306894207e-05,
      "loss": 1.3248,
      "step": 5220
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.5,
      "learning_rate": 4.703170802812512e-05,
      "loss": 1.2768,
      "step": 5225
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.78125,
      "learning_rate": 4.700875828917839e-05,
      "loss": 1.2913,
      "step": 5230
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.65625,
      "learning_rate": 4.6985793871907175e-05,
      "loss": 1.2903,
      "step": 5235
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.3125,
      "learning_rate": 4.696281479612946e-05,
      "loss": 1.2847,
      "step": 5240
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.5,
      "learning_rate": 4.6939821081675855e-05,
      "loss": 1.3054,
      "step": 5245
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.375,
      "learning_rate": 4.6916812748389656e-05,
      "loss": 1.308,
      "step": 5250
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.46875,
      "learning_rate": 4.68937898161267e-05,
      "loss": 1.3282,
      "step": 5255
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.15625,
      "learning_rate": 4.6870752304755475e-05,
      "loss": 1.217,
      "step": 5260
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.625,
      "learning_rate": 4.684770023415704e-05,
      "loss": 1.2743,
      "step": 5265
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.5625,
      "learning_rate": 4.6824633624225006e-05,
      "loss": 1.298,
      "step": 5270
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.09375,
      "learning_rate": 4.680155249486555e-05,
      "loss": 1.3485,
      "step": 5275
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.1875,
      "learning_rate": 4.677845686599737e-05,
      "loss": 1.2938,
      "step": 5280
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.71875,
      "learning_rate": 4.675534675755167e-05,
      "loss": 1.3163,
      "step": 5285
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.875,
      "learning_rate": 4.673222218947215e-05,
      "loss": 1.2243,
      "step": 5290
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.78125,
      "learning_rate": 4.670908318171501e-05,
      "loss": 1.1996,
      "step": 5295
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.28125,
      "learning_rate": 4.668592975424888e-05,
      "loss": 1.256,
      "step": 5300
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.40625,
      "learning_rate": 4.666276192705486e-05,
      "loss": 1.2628,
      "step": 5305
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.1875,
      "learning_rate": 4.6639579720126455e-05,
      "loss": 1.3257,
      "step": 5310
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.375,
      "learning_rate": 4.661638315346961e-05,
      "loss": 1.2853,
      "step": 5315
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.03125,
      "learning_rate": 4.659317224710263e-05,
      "loss": 1.3387,
      "step": 5320
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.25,
      "learning_rate": 4.6569947021056194e-05,
      "loss": 1.3179,
      "step": 5325
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.78125,
      "learning_rate": 4.654670749537338e-05,
      "loss": 1.2654,
      "step": 5330
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.125,
      "learning_rate": 4.652345369010956e-05,
      "loss": 1.2965,
      "step": 5335
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.34375,
      "learning_rate": 4.650018562533246e-05,
      "loss": 1.3444,
      "step": 5340
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.8125,
      "learning_rate": 4.647690332112209e-05,
      "loss": 1.2672,
      "step": 5345
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.75,
      "learning_rate": 4.645360679757077e-05,
      "loss": 1.3395,
      "step": 5350
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.34375,
      "learning_rate": 4.6430296074783066e-05,
      "loss": 1.2981,
      "step": 5355
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.46875,
      "learning_rate": 4.6406971172875825e-05,
      "loss": 1.2583,
      "step": 5360
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.40625,
      "learning_rate": 4.638363211197811e-05,
      "loss": 1.3008,
      "step": 5365
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.09375,
      "learning_rate": 4.63602789122312e-05,
      "loss": 1.2887,
      "step": 5370
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.625,
      "learning_rate": 4.63369115937886e-05,
      "loss": 1.3329,
      "step": 5375
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.0625,
      "learning_rate": 4.631353017681597e-05,
      "loss": 1.2983,
      "step": 5380
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.15625,
      "learning_rate": 4.6290134681491165e-05,
      "loss": 1.2293,
      "step": 5385
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.0625,
      "learning_rate": 4.626672512800417e-05,
      "loss": 1.3206,
      "step": 5390
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.5,
      "learning_rate": 4.6243301536557114e-05,
      "loss": 1.2805,
      "step": 5395
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.09375,
      "learning_rate": 4.621986392736423e-05,
      "loss": 1.352,
      "step": 5400
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.875,
      "learning_rate": 4.619641232065184e-05,
      "loss": 1.2629,
      "step": 5405
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.75,
      "learning_rate": 4.6172946736658366e-05,
      "loss": 1.295,
      "step": 5410
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.28125,
      "learning_rate": 4.61494671956343e-05,
      "loss": 1.2634,
      "step": 5415
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.90625,
      "learning_rate": 4.6125973717842145e-05,
      "loss": 1.2701,
      "step": 5420
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.75,
      "learning_rate": 4.6102466323556464e-05,
      "loss": 1.2746,
      "step": 5425
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.15625,
      "learning_rate": 4.607894503306379e-05,
      "loss": 1.403,
      "step": 5430
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.09375,
      "learning_rate": 4.605540986666269e-05,
      "loss": 1.2862,
      "step": 5435
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.25,
      "learning_rate": 4.6031860844663694e-05,
      "loss": 1.3084,
      "step": 5440
    },
    {
      "epoch": 0.96,
      "grad_norm": 3.96875,
      "learning_rate": 4.600829798738927e-05,
      "loss": 1.3343,
      "step": 5445
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.4375,
      "learning_rate": 4.5984721315173856e-05,
      "loss": 1.3125,
      "step": 5450
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.03125,
      "learning_rate": 4.5961130848363776e-05,
      "loss": 1.2542,
      "step": 5455
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.09375,
      "learning_rate": 4.593752660731731e-05,
      "loss": 1.2577,
      "step": 5460
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.1875,
      "learning_rate": 4.591390861240455e-05,
      "loss": 1.2446,
      "step": 5465
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.40625,
      "learning_rate": 4.589027688400754e-05,
      "loss": 1.2402,
      "step": 5470
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.40625,
      "learning_rate": 4.586663144252011e-05,
      "loss": 1.2589,
      "step": 5475
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.125,
      "learning_rate": 4.584297230834797e-05,
      "loss": 1.2659,
      "step": 5480
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.09375,
      "learning_rate": 4.581929950190861e-05,
      "loss": 1.2953,
      "step": 5485
    },
    {
      "epoch": 0.97,
      "grad_norm": 3.796875,
      "learning_rate": 4.579561304363135e-05,
      "loss": 1.2839,
      "step": 5490
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.75,
      "learning_rate": 4.577191295395728e-05,
      "loss": 1.2788,
      "step": 5495
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.90625,
      "learning_rate": 4.574819925333923e-05,
      "loss": 1.2983,
      "step": 5500
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.4375,
      "learning_rate": 4.5724471962241815e-05,
      "loss": 1.3001,
      "step": 5505
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.0,
      "learning_rate": 4.5700731101141364e-05,
      "loss": 1.3513,
      "step": 5510
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.25,
      "learning_rate": 4.5676976690525895e-05,
      "loss": 1.2569,
      "step": 5515
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.5,
      "learning_rate": 4.565320875089516e-05,
      "loss": 1.3074,
      "step": 5520
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.75,
      "learning_rate": 4.562942730276055e-05,
      "loss": 1.2869,
      "step": 5525
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.21875,
      "learning_rate": 4.560563236664513e-05,
      "loss": 1.2873,
      "step": 5530
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.46875,
      "learning_rate": 4.558182396308361e-05,
      "loss": 1.2841,
      "step": 5535
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.78125,
      "learning_rate": 4.555800211262231e-05,
      "loss": 1.3453,
      "step": 5540
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.03125,
      "learning_rate": 4.5534166835819155e-05,
      "loss": 1.2259,
      "step": 5545
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.625,
      "learning_rate": 4.551031815324367e-05,
      "loss": 1.2489,
      "step": 5550
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.6875,
      "learning_rate": 4.548645608547692e-05,
      "loss": 1.3278,
      "step": 5555
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.96875,
      "learning_rate": 4.5462580653111563e-05,
      "loss": 1.3266,
      "step": 5560
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.34375,
      "learning_rate": 4.543869187675176e-05,
      "loss": 1.3331,
      "step": 5565
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.75,
      "learning_rate": 4.541478977701321e-05,
      "loss": 1.2837,
      "step": 5570
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.8125,
      "learning_rate": 4.5390874374523066e-05,
      "loss": 1.2806,
      "step": 5575
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.5,
      "learning_rate": 4.536694568992002e-05,
      "loss": 1.3459,
      "step": 5580
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.6875,
      "learning_rate": 4.534300374385418e-05,
      "loss": 1.2935,
      "step": 5585
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.5,
      "learning_rate": 4.531904855698714e-05,
      "loss": 1.2311,
      "step": 5590
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.15625,
      "learning_rate": 4.529508014999187e-05,
      "loss": 1.2909,
      "step": 5595
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.28125,
      "learning_rate": 4.5271098543552774e-05,
      "loss": 1.2652,
      "step": 5600
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.5625,
      "learning_rate": 4.524710375836568e-05,
      "loss": 1.2896,
      "step": 5605
    },
    {
      "epoch": 0.99,
      "grad_norm": 4.4375,
      "learning_rate": 4.522309581513773e-05,
      "loss": 1.2788,
      "step": 5610
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.0,
      "learning_rate": 4.5199074734587454e-05,
      "loss": 1.2915,
      "step": 5615
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.625,
      "learning_rate": 4.5175040537444726e-05,
      "loss": 1.2556,
      "step": 5620
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.09375,
      "learning_rate": 4.51509932444507e-05,
      "loss": 1.3896,
      "step": 5625
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.21875,
      "learning_rate": 4.5126932876357896e-05,
      "loss": 1.2715,
      "step": 5630
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.375,
      "learning_rate": 4.510285945393005e-05,
      "loss": 1.3126,
      "step": 5635
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.5,
      "learning_rate": 4.5078772997942224e-05,
      "loss": 1.3096,
      "step": 5640
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.890625,
      "learning_rate": 4.5054673529180676e-05,
      "loss": 1.2639,
      "step": 5645
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.5,
      "learning_rate": 4.503056106844291e-05,
      "loss": 1.1886,
      "step": 5650
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.953125,
      "learning_rate": 4.500643563653767e-05,
      "loss": 1.1993,
      "step": 5655
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.1875,
      "learning_rate": 4.498229725428486e-05,
      "loss": 1.2049,
      "step": 5660
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.65625,
      "learning_rate": 4.4958145942515565e-05,
      "loss": 1.2029,
      "step": 5665
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.84375,
      "learning_rate": 4.493398172207205e-05,
      "loss": 1.1845,
      "step": 5670
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.25,
      "learning_rate": 4.49098046138077e-05,
      "loss": 1.2204,
      "step": 5675
    },
    {
      "epoch": 1.01,
      "grad_norm": 3.9375,
      "learning_rate": 4.4885614638587024e-05,
      "loss": 1.2347,
      "step": 5680
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.53125,
      "learning_rate": 4.486141181728564e-05,
      "loss": 1.312,
      "step": 5685
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.65625,
      "learning_rate": 4.483719617079024e-05,
      "loss": 1.2256,
      "step": 5690
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.21875,
      "learning_rate": 4.481296771999861e-05,
      "loss": 1.2535,
      "step": 5695
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.78125,
      "learning_rate": 4.478872648581956e-05,
      "loss": 1.2195,
      "step": 5700
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.5625,
      "learning_rate": 4.476447248917295e-05,
      "loss": 1.2377,
      "step": 5705
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.21875,
      "learning_rate": 4.4740205750989626e-05,
      "loss": 1.1829,
      "step": 5710
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.21875,
      "learning_rate": 4.471592629221148e-05,
      "loss": 1.2236,
      "step": 5715
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.09375,
      "learning_rate": 4.469163413379132e-05,
      "loss": 1.2187,
      "step": 5720
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.875,
      "learning_rate": 4.4667329296692966e-05,
      "loss": 1.2311,
      "step": 5725
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.125,
      "learning_rate": 4.464301180189115e-05,
      "loss": 1.2426,
      "step": 5730
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.03125,
      "learning_rate": 4.461868167037154e-05,
      "loss": 1.1811,
      "step": 5735
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.625,
      "learning_rate": 4.459433892313069e-05,
      "loss": 1.2966,
      "step": 5740
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.375,
      "learning_rate": 4.456998358117608e-05,
      "loss": 1.2073,
      "step": 5745
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.78125,
      "learning_rate": 4.454561566552602e-05,
      "loss": 1.2092,
      "step": 5750
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.8125,
      "learning_rate": 4.4521235197209686e-05,
      "loss": 1.2631,
      "step": 5755
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.65625,
      "learning_rate": 4.449684219726709e-05,
      "loss": 1.2517,
      "step": 5760
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.5625,
      "learning_rate": 4.447243668674905e-05,
      "loss": 1.2079,
      "step": 5765
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.6875,
      "learning_rate": 4.44480186867172e-05,
      "loss": 1.2512,
      "step": 5770
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.1875,
      "learning_rate": 4.442358821824392e-05,
      "loss": 1.2197,
      "step": 5775
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.0625,
      "learning_rate": 4.439914530241238e-05,
      "loss": 1.2712,
      "step": 5780
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.09375,
      "learning_rate": 4.4374689960316486e-05,
      "loss": 1.2516,
      "step": 5785
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.75,
      "learning_rate": 4.435022221306084e-05,
      "loss": 1.2293,
      "step": 5790
    },
    {
      "epoch": 1.03,
      "grad_norm": 3.96875,
      "learning_rate": 4.4325742081760793e-05,
      "loss": 1.275,
      "step": 5795
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.0,
      "learning_rate": 4.430124958754235e-05,
      "loss": 1.2475,
      "step": 5800
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.4375,
      "learning_rate": 4.4276744751542196e-05,
      "loss": 1.1762,
      "step": 5805
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.65625,
      "learning_rate": 4.4252227594907685e-05,
      "loss": 1.2229,
      "step": 5810
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.59375,
      "learning_rate": 4.422769813879676e-05,
      "loss": 1.2236,
      "step": 5815
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.6875,
      "learning_rate": 4.4203156404378025e-05,
      "loss": 1.2489,
      "step": 5820
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.65625,
      "learning_rate": 4.417860241283064e-05,
      "loss": 1.2097,
      "step": 5825
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.78125,
      "learning_rate": 4.415403618534439e-05,
      "loss": 1.2719,
      "step": 5830
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.28125,
      "learning_rate": 4.412945774311956e-05,
      "loss": 1.276,
      "step": 5835
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.125,
      "learning_rate": 4.410486710736704e-05,
      "loss": 1.2583,
      "step": 5840
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.5,
      "learning_rate": 4.4080264299308166e-05,
      "loss": 1.2111,
      "step": 5845
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.6875,
      "learning_rate": 4.405564934017487e-05,
      "loss": 1.2251,
      "step": 5850
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.4375,
      "learning_rate": 4.4031022251209494e-05,
      "loss": 1.2126,
      "step": 5855
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.53125,
      "learning_rate": 4.400638305366489e-05,
      "loss": 1.1829,
      "step": 5860
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.15625,
      "learning_rate": 4.3981731768804336e-05,
      "loss": 1.2408,
      "step": 5865
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.4375,
      "learning_rate": 4.395706841790157e-05,
      "loss": 1.1975,
      "step": 5870
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.8125,
      "learning_rate": 4.393239302224071e-05,
      "loss": 1.3135,
      "step": 5875
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.375,
      "learning_rate": 4.390770560311628e-05,
      "loss": 1.1736,
      "step": 5880
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.21875,
      "learning_rate": 4.388300618183319e-05,
      "loss": 1.2355,
      "step": 5885
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.15625,
      "learning_rate": 4.38582947797067e-05,
      "loss": 1.2522,
      "step": 5890
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.875,
      "learning_rate": 4.3833571418062416e-05,
      "loss": 1.2311,
      "step": 5895
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.1875,
      "learning_rate": 4.380883611823624e-05,
      "loss": 1.2135,
      "step": 5900
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.21875,
      "learning_rate": 4.3784088901574415e-05,
      "loss": 1.2067,
      "step": 5905
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.8125,
      "learning_rate": 4.375932978943345e-05,
      "loss": 1.2137,
      "step": 5910
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.625,
      "learning_rate": 4.373455880318011e-05,
      "loss": 1.1993,
      "step": 5915
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.3125,
      "learning_rate": 4.370977596419141e-05,
      "loss": 1.2493,
      "step": 5920
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.5625,
      "learning_rate": 4.368498129385461e-05,
      "loss": 1.3239,
      "step": 5925
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.34375,
      "learning_rate": 4.366017481356716e-05,
      "loss": 1.253,
      "step": 5930
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.28125,
      "learning_rate": 4.363535654473672e-05,
      "loss": 1.2516,
      "step": 5935
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.1875,
      "learning_rate": 4.361052650878111e-05,
      "loss": 1.2904,
      "step": 5940
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.6875,
      "learning_rate": 4.358568472712831e-05,
      "loss": 1.2502,
      "step": 5945
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.46875,
      "learning_rate": 4.356083122121644e-05,
      "loss": 1.2033,
      "step": 5950
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.1875,
      "learning_rate": 4.353596601249372e-05,
      "loss": 1.205,
      "step": 5955
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.46875,
      "learning_rate": 4.35110891224185e-05,
      "loss": 1.1544,
      "step": 5960
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.96875,
      "learning_rate": 4.34862005724592e-05,
      "loss": 1.2133,
      "step": 5965
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.78125,
      "learning_rate": 4.3461300384094256e-05,
      "loss": 1.1757,
      "step": 5970
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.25,
      "learning_rate": 4.3436388578812236e-05,
      "loss": 1.1731,
      "step": 5975
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.40625,
      "learning_rate": 4.3411465178111656e-05,
      "loss": 1.2755,
      "step": 5980
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.03125,
      "learning_rate": 4.338653020350108e-05,
      "loss": 1.3168,
      "step": 5985
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.03125,
      "learning_rate": 4.336158367649905e-05,
      "loss": 1.2721,
      "step": 5990
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.78125,
      "learning_rate": 4.333662561863406e-05,
      "loss": 1.1861,
      "step": 5995
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.78125,
      "learning_rate": 4.3311656051444594e-05,
      "loss": 1.2178,
      "step": 6000
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.65625,
      "learning_rate": 4.328667499647903e-05,
      "loss": 1.2756,
      "step": 6005
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.40625,
      "learning_rate": 4.326168247529567e-05,
      "loss": 1.3336,
      "step": 6010
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.21875,
      "learning_rate": 4.323667850946273e-05,
      "loss": 1.3161,
      "step": 6015
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.21875,
      "learning_rate": 4.321166312055829e-05,
      "loss": 1.3181,
      "step": 6020
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.78125,
      "learning_rate": 4.318663633017026e-05,
      "loss": 1.2342,
      "step": 6025
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.65625,
      "learning_rate": 4.316159815989645e-05,
      "loss": 1.1907,
      "step": 6030
    },
    {
      "epoch": 1.07,
      "grad_norm": 3.921875,
      "learning_rate": 4.313654863134444e-05,
      "loss": 1.2357,
      "step": 6035
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.875,
      "learning_rate": 4.3111487766131624e-05,
      "loss": 1.1942,
      "step": 6040
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.84375,
      "learning_rate": 4.308641558588519e-05,
      "loss": 1.2796,
      "step": 6045
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.4375,
      "learning_rate": 4.3061332112242086e-05,
      "loss": 1.2781,
      "step": 6050
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.25,
      "learning_rate": 4.3036237366849e-05,
      "loss": 1.2587,
      "step": 6055
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.1875,
      "learning_rate": 4.301113137136236e-05,
      "loss": 1.198,
      "step": 6060
    },
    {
      "epoch": 1.07,
      "grad_norm": 3.71875,
      "learning_rate": 4.298601414744829e-05,
      "loss": 1.222,
      "step": 6065
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.28125,
      "learning_rate": 4.296088571678261e-05,
      "loss": 1.2194,
      "step": 6070
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.34375,
      "learning_rate": 4.293574610105081e-05,
      "loss": 1.2135,
      "step": 6075
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.65625,
      "learning_rate": 4.291059532194803e-05,
      "loss": 1.2257,
      "step": 6080
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.25,
      "learning_rate": 4.2885433401179054e-05,
      "loss": 1.2826,
      "step": 6085
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.15625,
      "learning_rate": 4.2860260360458256e-05,
      "loss": 1.1905,
      "step": 6090
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.25,
      "learning_rate": 4.2835076221509646e-05,
      "loss": 1.2797,
      "step": 6095
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.28125,
      "learning_rate": 4.2809881006066775e-05,
      "loss": 1.2085,
      "step": 6100
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.4375,
      "learning_rate": 4.278467473587278e-05,
      "loss": 1.2315,
      "step": 6105
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.375,
      "learning_rate": 4.275945743268031e-05,
      "loss": 1.2416,
      "step": 6110
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.03125,
      "learning_rate": 4.2734229118251553e-05,
      "loss": 1.1506,
      "step": 6115
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.65625,
      "learning_rate": 4.270898981435821e-05,
      "loss": 1.1416,
      "step": 6120
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.34375,
      "learning_rate": 4.268373954278145e-05,
      "loss": 1.2071,
      "step": 6125
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.59375,
      "learning_rate": 4.2658478325311905e-05,
      "loss": 1.1974,
      "step": 6130
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.21875,
      "learning_rate": 4.263320618374964e-05,
      "loss": 1.2206,
      "step": 6135
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.1875,
      "learning_rate": 4.260792313990421e-05,
      "loss": 1.213,
      "step": 6140
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.46875,
      "learning_rate": 4.2582629215594504e-05,
      "loss": 1.2882,
      "step": 6145
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.96875,
      "learning_rate": 4.255732443264882e-05,
      "loss": 1.2809,
      "step": 6150
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.25,
      "learning_rate": 4.2532008812904866e-05,
      "loss": 1.3132,
      "step": 6155
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.625,
      "learning_rate": 4.250668237820965e-05,
      "loss": 1.1401,
      "step": 6160
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.5,
      "learning_rate": 4.2481345150419566e-05,
      "loss": 1.265,
      "step": 6165
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.71875,
      "learning_rate": 4.245599715140027e-05,
      "loss": 1.3294,
      "step": 6170
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.46875,
      "learning_rate": 4.243063840302673e-05,
      "loss": 1.2237,
      "step": 6175
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.5,
      "learning_rate": 4.2405268927183214e-05,
      "loss": 1.2142,
      "step": 6180
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.0625,
      "learning_rate": 4.2379888745763224e-05,
      "loss": 1.2268,
      "step": 6185
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.15625,
      "learning_rate": 4.2354497880669514e-05,
      "loss": 1.2154,
      "step": 6190
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.1875,
      "learning_rate": 4.232909635381404e-05,
      "loss": 1.185,
      "step": 6195
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.46875,
      "learning_rate": 4.230368418711798e-05,
      "loss": 1.3333,
      "step": 6200
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.40625,
      "learning_rate": 4.227826140251169e-05,
      "loss": 1.2565,
      "step": 6205
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.28125,
      "learning_rate": 4.225282802193466e-05,
      "loss": 1.2973,
      "step": 6210
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.65625,
      "learning_rate": 4.2227384067335566e-05,
      "loss": 1.2033,
      "step": 6215
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.0625,
      "learning_rate": 4.220192956067219e-05,
      "loss": 1.3041,
      "step": 6220
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.15625,
      "learning_rate": 4.2176464523911406e-05,
      "loss": 1.2824,
      "step": 6225
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.34375,
      "learning_rate": 4.215098897902922e-05,
      "loss": 1.1925,
      "step": 6230
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.84375,
      "learning_rate": 4.2125502948010646e-05,
      "loss": 1.2424,
      "step": 6235
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.0,
      "learning_rate": 4.2100006452849804e-05,
      "loss": 1.1535,
      "step": 6240
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.4375,
      "learning_rate": 4.207449951554979e-05,
      "loss": 1.2146,
      "step": 6245
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.46875,
      "learning_rate": 4.204898215812277e-05,
      "loss": 1.2141,
      "step": 6250
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.3125,
      "learning_rate": 4.2023454402589856e-05,
      "loss": 1.2369,
      "step": 6255
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.84375,
      "learning_rate": 4.1997916270981144e-05,
      "loss": 1.2443,
      "step": 6260
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.1875,
      "learning_rate": 4.197236778533571e-05,
      "loss": 1.1905,
      "step": 6265
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.8125,
      "learning_rate": 4.194680896770154e-05,
      "loss": 1.2564,
      "step": 6270
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.65625,
      "learning_rate": 4.1921239840135525e-05,
      "loss": 1.2885,
      "step": 6275
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.3125,
      "learning_rate": 4.189566042470349e-05,
      "loss": 1.1706,
      "step": 6280
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.65625,
      "learning_rate": 4.18700707434801e-05,
      "loss": 1.3361,
      "step": 6285
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.3125,
      "learning_rate": 4.1844470818548924e-05,
      "loss": 1.2021,
      "step": 6290
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.53125,
      "learning_rate": 4.181886067200233e-05,
      "loss": 1.2032,
      "step": 6295
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.3125,
      "learning_rate": 4.1793240325941505e-05,
      "loss": 1.2259,
      "step": 6300
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.625,
      "learning_rate": 4.176760980247649e-05,
      "loss": 1.215,
      "step": 6305
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.25,
      "learning_rate": 4.174196912372605e-05,
      "loss": 1.2053,
      "step": 6310
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.5625,
      "learning_rate": 4.1716318311817746e-05,
      "loss": 1.195,
      "step": 6315
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.53125,
      "learning_rate": 4.169065738888787e-05,
      "loss": 1.2607,
      "step": 6320
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.0625,
      "learning_rate": 4.166498637708145e-05,
      "loss": 1.2782,
      "step": 6325
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.59375,
      "learning_rate": 4.163930529855223e-05,
      "loss": 1.231,
      "step": 6330
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.84375,
      "learning_rate": 4.1613614175462606e-05,
      "loss": 1.2344,
      "step": 6335
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.53125,
      "learning_rate": 4.158791302998367e-05,
      "loss": 1.2016,
      "step": 6340
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.125,
      "learning_rate": 4.156220188429517e-05,
      "loss": 1.2371,
      "step": 6345
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.46875,
      "learning_rate": 4.153648076058545e-05,
      "loss": 1.2503,
      "step": 6350
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.46875,
      "learning_rate": 4.1510749681051504e-05,
      "loss": 1.2649,
      "step": 6355
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.9375,
      "learning_rate": 4.148500866789888e-05,
      "loss": 1.2615,
      "step": 6360
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.03125,
      "learning_rate": 4.145925774334174e-05,
      "loss": 1.2473,
      "step": 6365
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.5625,
      "learning_rate": 4.143349692960277e-05,
      "loss": 1.2527,
      "step": 6370
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.15625,
      "learning_rate": 4.140772624891319e-05,
      "loss": 1.2399,
      "step": 6375
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.40625,
      "learning_rate": 4.1381945723512756e-05,
      "loss": 1.2365,
      "step": 6380
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.0625,
      "learning_rate": 4.135615537564969e-05,
      "loss": 1.3182,
      "step": 6385
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.34375,
      "learning_rate": 4.133035522758073e-05,
      "loss": 1.2551,
      "step": 6390
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.84375,
      "learning_rate": 4.1304545301571024e-05,
      "loss": 1.2442,
      "step": 6395
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.1875,
      "learning_rate": 4.12787256198942e-05,
      "loss": 1.2467,
      "step": 6400
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.6875,
      "learning_rate": 4.1252896204832294e-05,
      "loss": 1.2771,
      "step": 6405
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.09375,
      "learning_rate": 4.122705707867571e-05,
      "loss": 1.2577,
      "step": 6410
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.09375,
      "learning_rate": 4.120120826372329e-05,
      "loss": 1.2464,
      "step": 6415
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.375,
      "learning_rate": 4.1175349782282184e-05,
      "loss": 1.204,
      "step": 6420
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.375,
      "learning_rate": 4.114948165666791e-05,
      "loss": 1.2572,
      "step": 6425
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.1875,
      "learning_rate": 4.112360390920431e-05,
      "loss": 1.2345,
      "step": 6430
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.34375,
      "learning_rate": 4.1097716562223514e-05,
      "loss": 1.2233,
      "step": 6435
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.4375,
      "learning_rate": 4.107181963806597e-05,
      "loss": 1.2604,
      "step": 6440
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.25,
      "learning_rate": 4.104591315908033e-05,
      "loss": 1.1759,
      "step": 6445
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.75,
      "learning_rate": 4.101999714762356e-05,
      "loss": 1.1818,
      "step": 6450
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.15625,
      "learning_rate": 4.09940716260608e-05,
      "loss": 1.1738,
      "step": 6455
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.625,
      "learning_rate": 4.096813661676545e-05,
      "loss": 1.2228,
      "step": 6460
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.25,
      "learning_rate": 4.0942192142119024e-05,
      "loss": 1.2425,
      "step": 6465
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.125,
      "learning_rate": 4.091623822451129e-05,
      "loss": 1.2344,
      "step": 6470
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.0625,
      "learning_rate": 4.089027488634009e-05,
      "loss": 1.1715,
      "step": 6475
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.03125,
      "learning_rate": 4.086430215001144e-05,
      "loss": 1.3268,
      "step": 6480
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.4375,
      "learning_rate": 4.083832003793945e-05,
      "loss": 1.2293,
      "step": 6485
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.3125,
      "learning_rate": 4.0812328572546336e-05,
      "loss": 1.2874,
      "step": 6490
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.71875,
      "learning_rate": 4.0786327776262355e-05,
      "loss": 1.2482,
      "step": 6495
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.375,
      "learning_rate": 4.076031767152586e-05,
      "loss": 1.3249,
      "step": 6500
    },
    {
      "epoch": 1.15,
      "grad_norm": 3.984375,
      "learning_rate": 4.073429828078319e-05,
      "loss": 1.3375,
      "step": 6505
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.75,
      "learning_rate": 4.070826962648875e-05,
      "loss": 1.2449,
      "step": 6510
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.625,
      "learning_rate": 4.068223173110489e-05,
      "loss": 1.2831,
      "step": 6515
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.96875,
      "learning_rate": 4.065618461710196e-05,
      "loss": 1.2977,
      "step": 6520
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.1875,
      "learning_rate": 4.0630128306958254e-05,
      "loss": 1.2328,
      "step": 6525
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.21875,
      "learning_rate": 4.0604062823160024e-05,
      "loss": 1.2707,
      "step": 6530
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.4375,
      "learning_rate": 4.0577988188201414e-05,
      "loss": 1.279,
      "step": 6535
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.375,
      "learning_rate": 4.0551904424584485e-05,
      "loss": 1.2234,
      "step": 6540
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.46875,
      "learning_rate": 4.0525811554819145e-05,
      "loss": 1.1966,
      "step": 6545
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.28125,
      "learning_rate": 4.049970960142321e-05,
      "loss": 1.1876,
      "step": 6550
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.1875,
      "learning_rate": 4.047359858692229e-05,
      "loss": 1.2647,
      "step": 6555
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.3125,
      "learning_rate": 4.044747853384983e-05,
      "loss": 1.2892,
      "step": 6560
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.3125,
      "learning_rate": 4.042134946474708e-05,
      "loss": 1.1884,
      "step": 6565
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.625,
      "learning_rate": 4.039521140216308e-05,
      "loss": 1.2284,
      "step": 6570
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.6875,
      "learning_rate": 4.036906436865459e-05,
      "loss": 1.1783,
      "step": 6575
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.375,
      "learning_rate": 4.0342908386786175e-05,
      "loss": 1.2079,
      "step": 6580
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.75,
      "learning_rate": 4.031674347913005e-05,
      "loss": 1.2053,
      "step": 6585
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.5625,
      "learning_rate": 4.02905696682662e-05,
      "loss": 1.1107,
      "step": 6590
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.03125,
      "learning_rate": 4.0264386976782256e-05,
      "loss": 1.2973,
      "step": 6595
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.71875,
      "learning_rate": 4.0238195427273515e-05,
      "loss": 1.2198,
      "step": 6600
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.875,
      "learning_rate": 4.021199504234294e-05,
      "loss": 1.2684,
      "step": 6605
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.34375,
      "learning_rate": 4.018578584460107e-05,
      "loss": 1.2408,
      "step": 6610
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.96875,
      "learning_rate": 4.015956785666611e-05,
      "loss": 1.2367,
      "step": 6615
    },
    {
      "epoch": 1.17,
      "grad_norm": 3.953125,
      "learning_rate": 4.0133341101163824e-05,
      "loss": 1.1958,
      "step": 6620
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.0,
      "learning_rate": 4.010710560072753e-05,
      "loss": 1.2547,
      "step": 6625
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.46875,
      "learning_rate": 4.00808613779981e-05,
      "loss": 1.2206,
      "step": 6630
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.9375,
      "learning_rate": 4.005460845562395e-05,
      "loss": 1.267,
      "step": 6635
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.3125,
      "learning_rate": 4.002834685626098e-05,
      "loss": 1.2417,
      "step": 6640
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.4375,
      "learning_rate": 4.000207660257261e-05,
      "loss": 1.2821,
      "step": 6645
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.6875,
      "learning_rate": 3.997579771722967e-05,
      "loss": 1.2522,
      "step": 6650
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.28125,
      "learning_rate": 3.994951022291051e-05,
      "loss": 1.1806,
      "step": 6655
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.46875,
      "learning_rate": 3.992321414230086e-05,
      "loss": 1.1642,
      "step": 6660
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.875,
      "learning_rate": 3.9896909498093895e-05,
      "loss": 1.2136,
      "step": 6665
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.0,
      "learning_rate": 3.9870596312990126e-05,
      "loss": 1.1851,
      "step": 6670
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.25,
      "learning_rate": 3.9844274609697496e-05,
      "loss": 1.1859,
      "step": 6675
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.25,
      "learning_rate": 3.9817944410931276e-05,
      "loss": 1.2189,
      "step": 6680
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.8125,
      "learning_rate": 3.979160573941406e-05,
      "loss": 1.2484,
      "step": 6685
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.125,
      "learning_rate": 3.976525861787574e-05,
      "loss": 1.2842,
      "step": 6690
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.15625,
      "learning_rate": 3.973890306905355e-05,
      "loss": 1.148,
      "step": 6695
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.5,
      "learning_rate": 3.9712539115691944e-05,
      "loss": 1.2832,
      "step": 6700
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.3125,
      "learning_rate": 3.968616678054267e-05,
      "loss": 1.2673,
      "step": 6705
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.46875,
      "learning_rate": 3.965978608636466e-05,
      "loss": 1.2478,
      "step": 6710
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.0625,
      "learning_rate": 3.963339705592413e-05,
      "loss": 1.2625,
      "step": 6715
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.03125,
      "learning_rate": 3.960699971199442e-05,
      "loss": 1.1947,
      "step": 6720
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.4375,
      "learning_rate": 3.9580594077356086e-05,
      "loss": 1.259,
      "step": 6725
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.96875,
      "learning_rate": 3.955418017479682e-05,
      "loss": 1.2197,
      "step": 6730
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.1875,
      "learning_rate": 3.952775802711148e-05,
      "loss": 1.2944,
      "step": 6735
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.9375,
      "learning_rate": 3.9501327657101985e-05,
      "loss": 1.2542,
      "step": 6740
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.15625,
      "learning_rate": 3.947488908757741e-05,
      "loss": 1.2235,
      "step": 6745
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.8125,
      "learning_rate": 3.9448442341353846e-05,
      "loss": 1.3071,
      "step": 6750
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.3125,
      "learning_rate": 3.942198744125449e-05,
      "loss": 1.1837,
      "step": 6755
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.375,
      "learning_rate": 3.939552441010956e-05,
      "loss": 1.2894,
      "step": 6760
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.75,
      "learning_rate": 3.936905327075628e-05,
      "loss": 1.2089,
      "step": 6765
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.5625,
      "learning_rate": 3.934257404603888e-05,
      "loss": 1.2504,
      "step": 6770
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.8125,
      "learning_rate": 3.931608675880857e-05,
      "loss": 1.2669,
      "step": 6775
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.09375,
      "learning_rate": 3.9289591431923505e-05,
      "loss": 1.2545,
      "step": 6780
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.78125,
      "learning_rate": 3.92630880882488e-05,
      "loss": 1.2244,
      "step": 6785
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.4375,
      "learning_rate": 3.9236576750656486e-05,
      "loss": 1.2,
      "step": 6790
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.125,
      "learning_rate": 3.921005744202544e-05,
      "loss": 1.1967,
      "step": 6795
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.875,
      "learning_rate": 3.918353018524151e-05,
      "loss": 1.2683,
      "step": 6800
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.25,
      "learning_rate": 3.915699500319733e-05,
      "loss": 1.1731,
      "step": 6805
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.71875,
      "learning_rate": 3.913045191879239e-05,
      "loss": 1.1733,
      "step": 6810
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.03125,
      "learning_rate": 3.910390095493302e-05,
      "loss": 1.2795,
      "step": 6815
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.71875,
      "learning_rate": 3.907734213453234e-05,
      "loss": 1.223,
      "step": 6820
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.4375,
      "learning_rate": 3.905077548051024e-05,
      "loss": 1.2375,
      "step": 6825
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.28125,
      "learning_rate": 3.902420101579337e-05,
      "loss": 1.2068,
      "step": 6830
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.84375,
      "learning_rate": 3.8997618763315135e-05,
      "loss": 1.2004,
      "step": 6835
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.03125,
      "learning_rate": 3.8971028746015656e-05,
      "loss": 1.2176,
      "step": 6840
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.46875,
      "learning_rate": 3.894443098684175e-05,
      "loss": 1.21,
      "step": 6845
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.75,
      "learning_rate": 3.891782550874691e-05,
      "loss": 1.1994,
      "step": 6850
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.46875,
      "learning_rate": 3.889121233469131e-05,
      "loss": 1.2446,
      "step": 6855
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.5,
      "learning_rate": 3.886459148764174e-05,
      "loss": 1.1825,
      "step": 6860
    },
    {
      "epoch": 1.22,
      "grad_norm": 3.96875,
      "learning_rate": 3.883796299057164e-05,
      "loss": 1.1753,
      "step": 6865
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.0,
      "learning_rate": 3.8811326866461034e-05,
      "loss": 1.2289,
      "step": 6870
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.46875,
      "learning_rate": 3.878468313829651e-05,
      "loss": 1.2698,
      "step": 6875
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.34375,
      "learning_rate": 3.875803182907127e-05,
      "loss": 1.1907,
      "step": 6880
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.46875,
      "learning_rate": 3.8731372961785e-05,
      "loss": 1.2376,
      "step": 6885
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.09375,
      "learning_rate": 3.8704706559443956e-05,
      "loss": 1.3123,
      "step": 6890
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.0,
      "learning_rate": 3.867803264506086e-05,
      "loss": 1.1943,
      "step": 6895
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.4375,
      "learning_rate": 3.865135124165495e-05,
      "loss": 1.2202,
      "step": 6900
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.5,
      "learning_rate": 3.86246623722519e-05,
      "loss": 1.2567,
      "step": 6905
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.09375,
      "learning_rate": 3.859796605988383e-05,
      "loss": 1.2359,
      "step": 6910
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.90625,
      "learning_rate": 3.857126232758931e-05,
      "loss": 1.2316,
      "step": 6915
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.34375,
      "learning_rate": 3.8544551198413276e-05,
      "loss": 1.2515,
      "step": 6920
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.34375,
      "learning_rate": 3.8517832695407084e-05,
      "loss": 1.2577,
      "step": 6925
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.1875,
      "learning_rate": 3.8491106841628414e-05,
      "loss": 1.2484,
      "step": 6930
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.46875,
      "learning_rate": 3.846437366014132e-05,
      "loss": 1.2581,
      "step": 6935
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.96875,
      "learning_rate": 3.8437633174016166e-05,
      "loss": 1.1939,
      "step": 6940
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.34375,
      "learning_rate": 3.841088540632963e-05,
      "loss": 1.158,
      "step": 6945
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.9375,
      "learning_rate": 3.838413038016467e-05,
      "loss": 1.2439,
      "step": 6950
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.28125,
      "learning_rate": 3.83573681186105e-05,
      "loss": 1.327,
      "step": 6955
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.3125,
      "learning_rate": 3.833059864476257e-05,
      "loss": 1.2533,
      "step": 6960
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.03125,
      "learning_rate": 3.8303821981722585e-05,
      "loss": 1.2454,
      "step": 6965
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.21875,
      "learning_rate": 3.8277038152598444e-05,
      "loss": 1.1471,
      "step": 6970
    },
    {
      "epoch": 1.24,
      "grad_norm": 3.984375,
      "learning_rate": 3.82502471805042e-05,
      "loss": 1.163,
      "step": 6975
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.4375,
      "learning_rate": 3.822344908856011e-05,
      "loss": 1.2945,
      "step": 6980
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.09375,
      "learning_rate": 3.819664389989257e-05,
      "loss": 1.1877,
      "step": 6985
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.25,
      "learning_rate": 3.816983163763405e-05,
      "loss": 1.2883,
      "step": 6990
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.03125,
      "learning_rate": 3.8143012324923204e-05,
      "loss": 1.3264,
      "step": 6995
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.0,
      "learning_rate": 3.811618598490471e-05,
      "loss": 1.2202,
      "step": 7000
    },
    {
      "epoch": 1.24,
      "grad_norm": 3.890625,
      "learning_rate": 3.8089352640729335e-05,
      "loss": 1.1479,
      "step": 7005
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.34375,
      "learning_rate": 3.806251231555388e-05,
      "loss": 1.2507,
      "step": 7010
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.09375,
      "learning_rate": 3.8035665032541186e-05,
      "loss": 1.1932,
      "step": 7015
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.375,
      "learning_rate": 3.8008810814860076e-05,
      "loss": 1.2855,
      "step": 7020
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.28125,
      "learning_rate": 3.798194968568539e-05,
      "loss": 1.2523,
      "step": 7025
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.9375,
      "learning_rate": 3.79550816681979e-05,
      "loss": 1.2133,
      "step": 7030
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.34375,
      "learning_rate": 3.792820678558434e-05,
      "loss": 1.2357,
      "step": 7035
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.75,
      "learning_rate": 3.7901325061037354e-05,
      "loss": 1.2459,
      "step": 7040
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.34375,
      "learning_rate": 3.7874436517755506e-05,
      "loss": 1.242,
      "step": 7045
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.0625,
      "learning_rate": 3.784754117894326e-05,
      "loss": 1.198,
      "step": 7050
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.84375,
      "learning_rate": 3.782063906781089e-05,
      "loss": 1.2639,
      "step": 7055
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.625,
      "learning_rate": 3.779373020757456e-05,
      "loss": 1.2507,
      "step": 7060
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.03125,
      "learning_rate": 3.776681462145625e-05,
      "loss": 1.2248,
      "step": 7065
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.625,
      "learning_rate": 3.773989233268375e-05,
      "loss": 1.2225,
      "step": 7070
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.375,
      "learning_rate": 3.771296336449061e-05,
      "loss": 1.1804,
      "step": 7075
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.15625,
      "learning_rate": 3.7686027740116165e-05,
      "loss": 1.2351,
      "step": 7080
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.5625,
      "learning_rate": 3.765908548280548e-05,
      "loss": 1.214,
      "step": 7085
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.375,
      "learning_rate": 3.763213661580937e-05,
      "loss": 1.2337,
      "step": 7090
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.0,
      "learning_rate": 3.7605181162384326e-05,
      "loss": 1.2682,
      "step": 7095
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.4375,
      "learning_rate": 3.757821914579252e-05,
      "loss": 1.2228,
      "step": 7100
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.40625,
      "learning_rate": 3.755125058930183e-05,
      "loss": 1.2619,
      "step": 7105
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.5,
      "learning_rate": 3.752427551618573e-05,
      "loss": 1.1934,
      "step": 7110
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.0625,
      "learning_rate": 3.7497293949723344e-05,
      "loss": 1.2883,
      "step": 7115
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.46875,
      "learning_rate": 3.7470305913199386e-05,
      "loss": 1.285,
      "step": 7120
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.53125,
      "learning_rate": 3.7443311429904164e-05,
      "loss": 1.2227,
      "step": 7125
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.46875,
      "learning_rate": 3.7416310523133556e-05,
      "loss": 1.1527,
      "step": 7130
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.375,
      "learning_rate": 3.7389303216188964e-05,
      "loss": 1.3002,
      "step": 7135
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.40625,
      "learning_rate": 3.736228953237731e-05,
      "loss": 1.1666,
      "step": 7140
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.5,
      "learning_rate": 3.733526949501105e-05,
      "loss": 1.2881,
      "step": 7145
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.3125,
      "learning_rate": 3.730824312740811e-05,
      "loss": 1.2361,
      "step": 7150
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.3125,
      "learning_rate": 3.7281210452891846e-05,
      "loss": 1.2538,
      "step": 7155
    },
    {
      "epoch": 1.27,
      "grad_norm": 3.8125,
      "learning_rate": 3.725417149479111e-05,
      "loss": 1.2228,
      "step": 7160
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.25,
      "learning_rate": 3.722712627644014e-05,
      "loss": 1.2783,
      "step": 7165
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.03125,
      "learning_rate": 3.7200074821178587e-05,
      "loss": 1.2249,
      "step": 7170
    },
    {
      "epoch": 1.27,
      "grad_norm": 3.40625,
      "learning_rate": 3.717301715235149e-05,
      "loss": 1.2403,
      "step": 7175
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.6875,
      "learning_rate": 3.714595329330924e-05,
      "loss": 1.2009,
      "step": 7180
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.4375,
      "learning_rate": 3.711888326740757e-05,
      "loss": 1.2153,
      "step": 7185
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.75,
      "learning_rate": 3.709180709800756e-05,
      "loss": 1.2137,
      "step": 7190
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.25,
      "learning_rate": 3.706472480847556e-05,
      "loss": 1.3122,
      "step": 7195
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.0625,
      "learning_rate": 3.703763642218321e-05,
      "loss": 1.2654,
      "step": 7200
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.46875,
      "learning_rate": 3.701054196250741e-05,
      "loss": 1.1606,
      "step": 7205
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.40625,
      "learning_rate": 3.6983441452830306e-05,
      "loss": 1.3224,
      "step": 7210
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.40625,
      "learning_rate": 3.695633491653929e-05,
      "loss": 1.229,
      "step": 7215
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.25,
      "learning_rate": 3.6929222377026907e-05,
      "loss": 1.1734,
      "step": 7220
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.0625,
      "learning_rate": 3.69021038576909e-05,
      "loss": 1.3508,
      "step": 7225
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.25,
      "learning_rate": 3.687497938193421e-05,
      "loss": 1.1976,
      "step": 7230
    },
    {
      "epoch": 1.28,
      "grad_norm": 3.921875,
      "learning_rate": 3.684784897316485e-05,
      "loss": 1.2657,
      "step": 7235
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.125,
      "learning_rate": 3.682071265479602e-05,
      "loss": 1.2413,
      "step": 7240
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.71875,
      "learning_rate": 3.679357045024596e-05,
      "loss": 1.2066,
      "step": 7245
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.28125,
      "learning_rate": 3.676642238293804e-05,
      "loss": 1.2192,
      "step": 7250
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.375,
      "learning_rate": 3.673926847630067e-05,
      "loss": 1.2644,
      "step": 7255
    },
    {
      "epoch": 1.29,
      "grad_norm": 3.78125,
      "learning_rate": 3.671210875376729e-05,
      "loss": 1.2022,
      "step": 7260
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.34375,
      "learning_rate": 3.668494323877635e-05,
      "loss": 1.3457,
      "step": 7265
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.125,
      "learning_rate": 3.665777195477135e-05,
      "loss": 1.2585,
      "step": 7270
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.8125,
      "learning_rate": 3.663059492520072e-05,
      "loss": 1.2622,
      "step": 7275
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.4375,
      "learning_rate": 3.660341217351785e-05,
      "loss": 1.2034,
      "step": 7280
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.53125,
      "learning_rate": 3.6576223723181104e-05,
      "loss": 1.1774,
      "step": 7285
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.25,
      "learning_rate": 3.654902959765372e-05,
      "loss": 1.2666,
      "step": 7290
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.34375,
      "learning_rate": 3.6521829820403864e-05,
      "loss": 1.2841,
      "step": 7295
    },
    {
      "epoch": 1.29,
      "grad_norm": 3.75,
      "learning_rate": 3.6494624414904575e-05,
      "loss": 1.2286,
      "step": 7300
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.15625,
      "learning_rate": 3.646741340463373e-05,
      "loss": 1.2205,
      "step": 7305
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.28125,
      "learning_rate": 3.644019681307406e-05,
      "loss": 1.2346,
      "step": 7310
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.125,
      "learning_rate": 3.641297466371313e-05,
      "loss": 1.1978,
      "step": 7315
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.28125,
      "learning_rate": 3.638574698004325e-05,
      "loss": 1.2718,
      "step": 7320
    },
    {
      "epoch": 1.3,
      "grad_norm": 3.984375,
      "learning_rate": 3.6358513785561554e-05,
      "loss": 1.1844,
      "step": 7325
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.34375,
      "learning_rate": 3.633127510376991e-05,
      "loss": 1.2515,
      "step": 7330
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.46875,
      "learning_rate": 3.630403095817493e-05,
      "loss": 1.2636,
      "step": 7335
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.1875,
      "learning_rate": 3.6276781372287934e-05,
      "loss": 1.2779,
      "step": 7340
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.75,
      "learning_rate": 3.6249526369624934e-05,
      "loss": 1.2245,
      "step": 7345
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.90625,
      "learning_rate": 3.622226597370663e-05,
      "loss": 1.2662,
      "step": 7350
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.65625,
      "learning_rate": 3.619500020805837e-05,
      "loss": 1.2206,
      "step": 7355
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.5625,
      "learning_rate": 3.616772909621012e-05,
      "loss": 1.3105,
      "step": 7360
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.5,
      "learning_rate": 3.61404526616965e-05,
      "loss": 1.2933,
      "step": 7365
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.78125,
      "learning_rate": 3.6113170928056666e-05,
      "loss": 1.2003,
      "step": 7370
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.625,
      "learning_rate": 3.608588391883441e-05,
      "loss": 1.2041,
      "step": 7375
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.9375,
      "learning_rate": 3.605859165757803e-05,
      "loss": 1.1733,
      "step": 7380
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.46875,
      "learning_rate": 3.603129416784038e-05,
      "loss": 1.2631,
      "step": 7385
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.78125,
      "learning_rate": 3.60039914731788e-05,
      "loss": 1.1757,
      "step": 7390
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.1875,
      "learning_rate": 3.5976683597155156e-05,
      "loss": 1.2808,
      "step": 7395
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.5625,
      "learning_rate": 3.594937056333577e-05,
      "loss": 1.2256,
      "step": 7400
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.125,
      "learning_rate": 3.5922052395291406e-05,
      "loss": 1.2752,
      "step": 7405
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.5,
      "learning_rate": 3.5894729116597277e-05,
      "loss": 1.3149,
      "step": 7410
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.5,
      "learning_rate": 3.586740075083298e-05,
      "loss": 1.2811,
      "step": 7415
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.71875,
      "learning_rate": 3.584006732158255e-05,
      "loss": 1.229,
      "step": 7420
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.78125,
      "learning_rate": 3.581272885243433e-05,
      "loss": 1.2844,
      "step": 7425
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.40625,
      "learning_rate": 3.578538536698104e-05,
      "loss": 1.1966,
      "step": 7430
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.40625,
      "learning_rate": 3.575803688881976e-05,
      "loss": 1.2922,
      "step": 7435
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.84375,
      "learning_rate": 3.573068344155184e-05,
      "loss": 1.2633,
      "step": 7440
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.75,
      "learning_rate": 3.570332504878292e-05,
      "loss": 1.2893,
      "step": 7445
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.8125,
      "learning_rate": 3.567596173412292e-05,
      "loss": 1.2356,
      "step": 7450
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.71875,
      "learning_rate": 3.564859352118601e-05,
      "loss": 1.1968,
      "step": 7455
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.8125,
      "learning_rate": 3.562122043359057e-05,
      "loss": 1.2618,
      "step": 7460
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.40625,
      "learning_rate": 3.5593842494959216e-05,
      "loss": 1.2348,
      "step": 7465
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.1875,
      "learning_rate": 3.556645972891871e-05,
      "loss": 1.2285,
      "step": 7470
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.6875,
      "learning_rate": 3.553907215910001e-05,
      "loss": 1.2448,
      "step": 7475
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.375,
      "learning_rate": 3.5511679809138235e-05,
      "loss": 1.1997,
      "step": 7480
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.21875,
      "learning_rate": 3.548428270267257e-05,
      "loss": 1.2353,
      "step": 7485
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.46875,
      "learning_rate": 3.5456880863346376e-05,
      "loss": 1.218,
      "step": 7490
    },
    {
      "epoch": 1.33,
      "grad_norm": 3.921875,
      "learning_rate": 3.542947431480703e-05,
      "loss": 1.1787,
      "step": 7495
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.21875,
      "learning_rate": 3.540206308070603e-05,
      "loss": 1.2362,
      "step": 7500
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.34375,
      "learning_rate": 3.5374647184698896e-05,
      "loss": 1.239,
      "step": 7505
    },
    {
      "epoch": 1.33,
      "grad_norm": 3.875,
      "learning_rate": 3.534722665044514e-05,
      "loss": 1.1829,
      "step": 7510
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.0625,
      "learning_rate": 3.531980150160834e-05,
      "loss": 1.2116,
      "step": 7515
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.0,
      "learning_rate": 3.529237176185601e-05,
      "loss": 1.2746,
      "step": 7520
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.71875,
      "learning_rate": 3.526493745485962e-05,
      "loss": 1.2576,
      "step": 7525
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.1875,
      "learning_rate": 3.523749860429463e-05,
      "loss": 1.2221,
      "step": 7530
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.0625,
      "learning_rate": 3.521005523384037e-05,
      "loss": 1.1911,
      "step": 7535
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.78125,
      "learning_rate": 3.51826073671801e-05,
      "loss": 1.2157,
      "step": 7540
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.1875,
      "learning_rate": 3.515515502800095e-05,
      "loss": 1.2778,
      "step": 7545
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.4375,
      "learning_rate": 3.512769823999392e-05,
      "loss": 1.2832,
      "step": 7550
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.6875,
      "learning_rate": 3.510023702685381e-05,
      "loss": 1.2897,
      "step": 7555
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.375,
      "learning_rate": 3.507277141227931e-05,
      "loss": 1.2216,
      "step": 7560
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.625,
      "learning_rate": 3.504530141997283e-05,
      "loss": 1.2385,
      "step": 7565
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.9375,
      "learning_rate": 3.501782707364063e-05,
      "loss": 1.2352,
      "step": 7570
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.625,
      "learning_rate": 3.499034839699267e-05,
      "loss": 1.1935,
      "step": 7575
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.3125,
      "learning_rate": 3.496286541374267e-05,
      "loss": 1.2684,
      "step": 7580
    },
    {
      "epoch": 1.34,
      "grad_norm": 3.796875,
      "learning_rate": 3.493537814760808e-05,
      "loss": 1.2503,
      "step": 7585
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.15625,
      "learning_rate": 3.490788662231003e-05,
      "loss": 1.2168,
      "step": 7590
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.75,
      "learning_rate": 3.48803908615733e-05,
      "loss": 1.2524,
      "step": 7595
    },
    {
      "epoch": 1.35,
      "grad_norm": 3.921875,
      "learning_rate": 3.4852890889126404e-05,
      "loss": 1.2563,
      "step": 7600
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.09375,
      "learning_rate": 3.482538672870141e-05,
      "loss": 1.2519,
      "step": 7605
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.0,
      "learning_rate": 3.4797878404034025e-05,
      "loss": 1.2164,
      "step": 7610
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.9375,
      "learning_rate": 3.477036593886357e-05,
      "loss": 1.3107,
      "step": 7615
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.28125,
      "learning_rate": 3.474284935693291e-05,
      "loss": 1.2211,
      "step": 7620
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.75,
      "learning_rate": 3.471532868198848e-05,
      "loss": 1.2178,
      "step": 7625
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.34375,
      "learning_rate": 3.468780393778027e-05,
      "loss": 1.1606,
      "step": 7630
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.71875,
      "learning_rate": 3.466027514806171e-05,
      "loss": 1.28,
      "step": 7635
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.375,
      "learning_rate": 3.46327423365898e-05,
      "loss": 1.3125,
      "step": 7640
    },
    {
      "epoch": 1.35,
      "grad_norm": 3.984375,
      "learning_rate": 3.460520552712497e-05,
      "loss": 1.2196,
      "step": 7645
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.40625,
      "learning_rate": 3.4577664743431095e-05,
      "loss": 1.2336,
      "step": 7650
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.34375,
      "learning_rate": 3.455012000927551e-05,
      "loss": 1.1643,
      "step": 7655
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.71875,
      "learning_rate": 3.452257134842894e-05,
      "loss": 1.2263,
      "step": 7660
    },
    {
      "epoch": 1.36,
      "grad_norm": 3.9375,
      "learning_rate": 3.4495018784665495e-05,
      "loss": 1.2134,
      "step": 7665
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.34375,
      "learning_rate": 3.446746234176266e-05,
      "loss": 1.2095,
      "step": 7670
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.375,
      "learning_rate": 3.443990204350127e-05,
      "loss": 1.2502,
      "step": 7675
    },
    {
      "epoch": 1.36,
      "grad_norm": 3.78125,
      "learning_rate": 3.441233791366546e-05,
      "loss": 1.1875,
      "step": 7680
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.71875,
      "learning_rate": 3.438476997604274e-05,
      "loss": 1.2597,
      "step": 7685
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.8125,
      "learning_rate": 3.435719825442381e-05,
      "loss": 1.2673,
      "step": 7690
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.46875,
      "learning_rate": 3.432962277260273e-05,
      "loss": 1.2411,
      "step": 7695
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.40625,
      "learning_rate": 3.430204355437674e-05,
      "loss": 1.2837,
      "step": 7700
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.6875,
      "learning_rate": 3.427446062354632e-05,
      "loss": 1.2769,
      "step": 7705
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.71875,
      "learning_rate": 3.424687400391518e-05,
      "loss": 1.283,
      "step": 7710
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.375,
      "learning_rate": 3.421928371929018e-05,
      "loss": 1.2279,
      "step": 7715
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.84375,
      "learning_rate": 3.4191689793481346e-05,
      "loss": 1.2468,
      "step": 7720
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.21875,
      "learning_rate": 3.416409225030189e-05,
      "loss": 1.2283,
      "step": 7725
    },
    {
      "epoch": 1.37,
      "grad_norm": 3.953125,
      "learning_rate": 3.413649111356808e-05,
      "loss": 1.1719,
      "step": 7730
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.78125,
      "learning_rate": 3.410888640709933e-05,
      "loss": 1.2873,
      "step": 7735
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.25,
      "learning_rate": 3.408127815471812e-05,
      "loss": 1.18,
      "step": 7740
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.96875,
      "learning_rate": 3.4053666380249986e-05,
      "loss": 1.2085,
      "step": 7745
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.4375,
      "learning_rate": 3.402605110752353e-05,
      "loss": 1.1924,
      "step": 7750
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.71875,
      "learning_rate": 3.3998432360370334e-05,
      "loss": 1.2658,
      "step": 7755
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.0,
      "learning_rate": 3.397081016262499e-05,
      "loss": 1.1559,
      "step": 7760
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.875,
      "learning_rate": 3.39431845381251e-05,
      "loss": 1.1612,
      "step": 7765
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.65625,
      "learning_rate": 3.3915555510711185e-05,
      "loss": 1.2686,
      "step": 7770
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.59375,
      "learning_rate": 3.3887923104226714e-05,
      "loss": 1.2432,
      "step": 7775
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.3125,
      "learning_rate": 3.386028734251808e-05,
      "loss": 1.0922,
      "step": 7780
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.34375,
      "learning_rate": 3.383264824943455e-05,
      "loss": 1.2121,
      "step": 7785
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.65625,
      "learning_rate": 3.38050058488283e-05,
      "loss": 1.2373,
      "step": 7790
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.5,
      "learning_rate": 3.377736016455433e-05,
      "loss": 1.1687,
      "step": 7795
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.0,
      "learning_rate": 3.374971122047048e-05,
      "loss": 1.3151,
      "step": 7800
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.5625,
      "learning_rate": 3.372205904043742e-05,
      "loss": 1.259,
      "step": 7805
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.78125,
      "learning_rate": 3.369440364831861e-05,
      "loss": 1.1974,
      "step": 7810
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.5,
      "learning_rate": 3.366674506798025e-05,
      "loss": 1.2669,
      "step": 7815
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.46875,
      "learning_rate": 3.363908332329134e-05,
      "loss": 1.1812,
      "step": 7820
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.125,
      "learning_rate": 3.361141843812356e-05,
      "loss": 1.2599,
      "step": 7825
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.03125,
      "learning_rate": 3.358375043635133e-05,
      "loss": 1.216,
      "step": 7830
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.90625,
      "learning_rate": 3.3556079341851775e-05,
      "loss": 1.1921,
      "step": 7835
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.125,
      "learning_rate": 3.352840517850465e-05,
      "loss": 1.1924,
      "step": 7840
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.53125,
      "learning_rate": 3.3500727970192374e-05,
      "loss": 1.1985,
      "step": 7845
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.6875,
      "learning_rate": 3.347304774080002e-05,
      "loss": 1.201,
      "step": 7850
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.5625,
      "learning_rate": 3.344536451421522e-05,
      "loss": 1.2272,
      "step": 7855
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.40625,
      "learning_rate": 3.341767831432824e-05,
      "loss": 1.2578,
      "step": 7860
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.9375,
      "learning_rate": 3.338998916503186e-05,
      "loss": 1.2344,
      "step": 7865
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.375,
      "learning_rate": 3.3362297090221466e-05,
      "loss": 1.28,
      "step": 7870
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.40625,
      "learning_rate": 3.333460211379493e-05,
      "loss": 1.2487,
      "step": 7875
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.125,
      "learning_rate": 3.3306904259652626e-05,
      "loss": 1.274,
      "step": 7880
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.6875,
      "learning_rate": 3.3279203551697414e-05,
      "loss": 1.2464,
      "step": 7885
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.34375,
      "learning_rate": 3.3251500013834655e-05,
      "loss": 1.3006,
      "step": 7890
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.53125,
      "learning_rate": 3.32237936699721e-05,
      "loss": 1.2043,
      "step": 7895
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.25,
      "learning_rate": 3.319608454401994e-05,
      "loss": 1.2738,
      "step": 7900
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.1875,
      "learning_rate": 3.316837265989077e-05,
      "loss": 1.1612,
      "step": 7905
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.21875,
      "learning_rate": 3.314065804149958e-05,
      "loss": 1.2169,
      "step": 7910
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.1875,
      "learning_rate": 3.3112940712763685e-05,
      "loss": 1.2207,
      "step": 7915
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.6875,
      "learning_rate": 3.308522069760278e-05,
      "loss": 1.2492,
      "step": 7920
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.1875,
      "learning_rate": 3.305749801993882e-05,
      "loss": 1.1049,
      "step": 7925
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.03125,
      "learning_rate": 3.3029772703696145e-05,
      "loss": 1.1853,
      "step": 7930
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.03125,
      "learning_rate": 3.3002044772801284e-05,
      "loss": 1.3082,
      "step": 7935
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.59375,
      "learning_rate": 3.2974314251183066e-05,
      "loss": 1.2403,
      "step": 7940
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.46875,
      "learning_rate": 3.294658116277257e-05,
      "loss": 1.2429,
      "step": 7945
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.6875,
      "learning_rate": 3.291884553150305e-05,
      "loss": 1.2617,
      "step": 7950
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.40625,
      "learning_rate": 3.2891107381309974e-05,
      "loss": 1.2781,
      "step": 7955
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.40625,
      "learning_rate": 3.286336673613099e-05,
      "loss": 1.2413,
      "step": 7960
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.34375,
      "learning_rate": 3.28356236199059e-05,
      "loss": 1.2386,
      "step": 7965
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.375,
      "learning_rate": 3.280787805657662e-05,
      "loss": 1.2417,
      "step": 7970
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.8125,
      "learning_rate": 3.2780130070087196e-05,
      "loss": 1.1677,
      "step": 7975
    },
    {
      "epoch": 1.41,
      "grad_norm": 3.96875,
      "learning_rate": 3.2752379684383754e-05,
      "loss": 1.2378,
      "step": 7980
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.71875,
      "learning_rate": 3.27246269234145e-05,
      "loss": 1.2372,
      "step": 7985
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.34375,
      "learning_rate": 3.269687181112968e-05,
      "loss": 1.2368,
      "step": 7990
    },
    {
      "epoch": 1.42,
      "grad_norm": 3.671875,
      "learning_rate": 3.2669114371481587e-05,
      "loss": 1.2243,
      "step": 7995
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.875,
      "learning_rate": 3.2641354628424494e-05,
      "loss": 1.2218,
      "step": 8000
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.1875,
      "learning_rate": 3.261359260591468e-05,
      "loss": 1.2455,
      "step": 8005
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.625,
      "learning_rate": 3.25858283279104e-05,
      "loss": 1.2707,
      "step": 8010
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.65625,
      "learning_rate": 3.2558061818371835e-05,
      "loss": 1.2104,
      "step": 8015
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.5625,
      "learning_rate": 3.25302931012611e-05,
      "loss": 1.3039,
      "step": 8020
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.78125,
      "learning_rate": 3.2502522200542216e-05,
      "loss": 1.2611,
      "step": 8025
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.625,
      "learning_rate": 3.247474914018109e-05,
      "loss": 1.1853,
      "step": 8030
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.34375,
      "learning_rate": 3.244697394414549e-05,
      "loss": 1.3064,
      "step": 8035
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.65625,
      "learning_rate": 3.241919663640504e-05,
      "loss": 1.248,
      "step": 8040
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.71875,
      "learning_rate": 3.2391417240931166e-05,
      "loss": 1.2026,
      "step": 8045
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.28125,
      "learning_rate": 3.236363578169707e-05,
      "loss": 1.2549,
      "step": 8050
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.9375,
      "learning_rate": 3.2335852282677844e-05,
      "loss": 1.2203,
      "step": 8055
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.125,
      "learning_rate": 3.230806676785021e-05,
      "loss": 1.1889,
      "step": 8060
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.3125,
      "learning_rate": 3.2280279261192685e-05,
      "loss": 1.1957,
      "step": 8065
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.25,
      "learning_rate": 3.225248978668553e-05,
      "loss": 1.2039,
      "step": 8070
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.84375,
      "learning_rate": 3.222469836831067e-05,
      "loss": 1.2068,
      "step": 8075
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.0,
      "learning_rate": 3.219690503005171e-05,
      "loss": 1.2231,
      "step": 8080
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.21875,
      "learning_rate": 3.2169109795893924e-05,
      "loss": 1.1968,
      "step": 8085
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.375,
      "learning_rate": 3.2141312689824204e-05,
      "loss": 1.2683,
      "step": 8090
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.8125,
      "learning_rate": 3.21135137358311e-05,
      "loss": 1.2401,
      "step": 8095
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.28125,
      "learning_rate": 3.2085712957904704e-05,
      "loss": 1.2254,
      "step": 8100
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.09375,
      "learning_rate": 3.205791038003671e-05,
      "loss": 1.2193,
      "step": 8105
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.84375,
      "learning_rate": 3.203010602622036e-05,
      "loss": 1.2358,
      "step": 8110
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.71875,
      "learning_rate": 3.200229992045044e-05,
      "loss": 1.2211,
      "step": 8115
    },
    {
      "epoch": 1.44,
      "grad_norm": 3.765625,
      "learning_rate": 3.197449208672321e-05,
      "loss": 1.1589,
      "step": 8120
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.75,
      "learning_rate": 3.194668254903648e-05,
      "loss": 1.2716,
      "step": 8125
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.21875,
      "learning_rate": 3.191887133138947e-05,
      "loss": 1.2663,
      "step": 8130
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.46875,
      "learning_rate": 3.1891058457782895e-05,
      "loss": 1.2231,
      "step": 8135
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.1875,
      "learning_rate": 3.186324395221888e-05,
      "loss": 1.2072,
      "step": 8140
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.0,
      "learning_rate": 3.183542783870096e-05,
      "loss": 1.2664,
      "step": 8145
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.15625,
      "learning_rate": 3.180761014123407e-05,
      "loss": 1.2123,
      "step": 8150
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.75,
      "learning_rate": 3.177979088382447e-05,
      "loss": 1.1911,
      "step": 8155
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.59375,
      "learning_rate": 3.175197009047982e-05,
      "loss": 1.265,
      "step": 8160
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.59375,
      "learning_rate": 3.172414778520908e-05,
      "loss": 1.175,
      "step": 8165
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.71875,
      "learning_rate": 3.169632399202252e-05,
      "loss": 1.3004,
      "step": 8170
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.8125,
      "learning_rate": 3.1668498734931665e-05,
      "loss": 1.2015,
      "step": 8175
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.34375,
      "learning_rate": 3.164067203794936e-05,
      "loss": 1.2129,
      "step": 8180
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.25,
      "learning_rate": 3.161284392508966e-05,
      "loss": 1.2226,
      "step": 8185
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.0625,
      "learning_rate": 3.1585014420367823e-05,
      "loss": 1.338,
      "step": 8190
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.5625,
      "learning_rate": 3.155718354780034e-05,
      "loss": 1.2668,
      "step": 8195
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.5,
      "learning_rate": 3.1529351331404876e-05,
      "loss": 1.3103,
      "step": 8200
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.21875,
      "learning_rate": 3.1501517795200246e-05,
      "loss": 1.2984,
      "step": 8205
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.09375,
      "learning_rate": 3.147368296320642e-05,
      "loss": 1.2456,
      "step": 8210
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.59375,
      "learning_rate": 3.1445846859444435e-05,
      "loss": 1.1962,
      "step": 8215
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.375,
      "learning_rate": 3.141800950793652e-05,
      "loss": 1.2575,
      "step": 8220
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.21875,
      "learning_rate": 3.13901709327059e-05,
      "loss": 1.2426,
      "step": 8225
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.125,
      "learning_rate": 3.1362331157776876e-05,
      "loss": 1.2526,
      "step": 8230
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.78125,
      "learning_rate": 3.13344902071748e-05,
      "loss": 1.2345,
      "step": 8235
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.71875,
      "learning_rate": 3.130664810492604e-05,
      "loss": 1.2419,
      "step": 8240
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.28125,
      "learning_rate": 3.127880487505792e-05,
      "loss": 1.2911,
      "step": 8245
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.65625,
      "learning_rate": 3.125096054159879e-05,
      "loss": 1.2828,
      "step": 8250
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.1875,
      "learning_rate": 3.122311512857791e-05,
      "loss": 1.3087,
      "step": 8255
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.8125,
      "learning_rate": 3.11952686600255e-05,
      "loss": 1.2053,
      "step": 8260
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.875,
      "learning_rate": 3.116742115997268e-05,
      "loss": 1.193,
      "step": 8265
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.90625,
      "learning_rate": 3.113957265245145e-05,
      "loss": 1.2133,
      "step": 8270
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.84375,
      "learning_rate": 3.111172316149469e-05,
      "loss": 1.2015,
      "step": 8275
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.28125,
      "learning_rate": 3.108387271113614e-05,
      "loss": 1.233,
      "step": 8280
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.21875,
      "learning_rate": 3.105602132541033e-05,
      "loss": 1.24,
      "step": 8285
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.34375,
      "learning_rate": 3.1028169028352663e-05,
      "loss": 1.1698,
      "step": 8290
    },
    {
      "epoch": 1.47,
      "grad_norm": 3.921875,
      "learning_rate": 3.100031584399925e-05,
      "loss": 1.3059,
      "step": 8295
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.21875,
      "learning_rate": 3.0972461796387027e-05,
      "loss": 1.2142,
      "step": 8300
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.0625,
      "learning_rate": 3.094460690955365e-05,
      "loss": 1.2578,
      "step": 8305
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.21875,
      "learning_rate": 3.091675120753749e-05,
      "loss": 1.2503,
      "step": 8310
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.09375,
      "learning_rate": 3.088889471437766e-05,
      "loss": 1.288,
      "step": 8315
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.0625,
      "learning_rate": 3.086103745411391e-05,
      "loss": 1.2587,
      "step": 8320
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.375,
      "learning_rate": 3.083317945078667e-05,
      "loss": 1.2464,
      "step": 8325
    },
    {
      "epoch": 1.48,
      "grad_norm": 3.890625,
      "learning_rate": 3.080532072843704e-05,
      "loss": 1.2546,
      "step": 8330
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.5,
      "learning_rate": 3.07774613111067e-05,
      "loss": 1.1686,
      "step": 8335
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.53125,
      "learning_rate": 3.074960122283792e-05,
      "loss": 1.2485,
      "step": 8340
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.65625,
      "learning_rate": 3.0721740487673606e-05,
      "loss": 1.2651,
      "step": 8345
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.21875,
      "learning_rate": 3.069387912965717e-05,
      "loss": 1.2785,
      "step": 8350
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.09375,
      "learning_rate": 3.0666017172832594e-05,
      "loss": 1.3134,
      "step": 8355
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.15625,
      "learning_rate": 3.063815464124436e-05,
      "loss": 1.2162,
      "step": 8360
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.3125,
      "learning_rate": 3.061029155893744e-05,
      "loss": 1.1924,
      "step": 8365
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.40625,
      "learning_rate": 3.058242794995731e-05,
      "loss": 1.2103,
      "step": 8370
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.28125,
      "learning_rate": 3.055456383834987e-05,
      "loss": 1.1494,
      "step": 8375
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.34375,
      "learning_rate": 3.052669924816144e-05,
      "loss": 1.2265,
      "step": 8380
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.03125,
      "learning_rate": 3.0498834203438825e-05,
      "loss": 1.2099,
      "step": 8385
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.6875,
      "learning_rate": 3.0470968728229156e-05,
      "loss": 1.1725,
      "step": 8390
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.0,
      "learning_rate": 3.044310284657994e-05,
      "loss": 1.2158,
      "step": 8395
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.78125,
      "learning_rate": 3.041523658253906e-05,
      "loss": 1.2203,
      "step": 8400
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.375,
      "learning_rate": 3.038736996015473e-05,
      "loss": 1.2895,
      "step": 8405
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.25,
      "learning_rate": 3.035950300347544e-05,
      "loss": 1.284,
      "step": 8410
    },
    {
      "epoch": 1.49,
      "grad_norm": 3.96875,
      "learning_rate": 3.0331635736550014e-05,
      "loss": 1.2822,
      "step": 8415
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.65625,
      "learning_rate": 3.0303768183427504e-05,
      "loss": 1.2586,
      "step": 8420
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.09375,
      "learning_rate": 3.0275900368157222e-05,
      "loss": 1.2682,
      "step": 8425
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.71875,
      "learning_rate": 3.0248032314788733e-05,
      "loss": 1.1847,
      "step": 8430
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.625,
      "learning_rate": 3.022016404737175e-05,
      "loss": 1.2741,
      "step": 8435
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.40625,
      "learning_rate": 3.019229558995623e-05,
      "loss": 1.2583,
      "step": 8440
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.28125,
      "learning_rate": 3.0164426966592264e-05,
      "loss": 1.1743,
      "step": 8445
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.375,
      "learning_rate": 3.0136558201330086e-05,
      "loss": 1.3288,
      "step": 8450
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.28125,
      "learning_rate": 3.0108689318220058e-05,
      "loss": 1.1603,
      "step": 8455
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.8125,
      "learning_rate": 3.0080820341312637e-05,
      "loss": 1.1689,
      "step": 8460
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.15625,
      "learning_rate": 3.0052951294658376e-05,
      "loss": 1.2477,
      "step": 8465
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.90625,
      "learning_rate": 3.002508220230788e-05,
      "loss": 1.2148,
      "step": 8470
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.4375,
      "learning_rate": 2.9997213088311784e-05,
      "loss": 1.2865,
      "step": 8475
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.03125,
      "learning_rate": 2.996934397672076e-05,
      "loss": 1.2427,
      "step": 8480
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.03125,
      "learning_rate": 2.9941474891585464e-05,
      "loss": 1.2736,
      "step": 8485
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.53125,
      "learning_rate": 2.9913605856956522e-05,
      "loss": 1.2205,
      "step": 8490
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.15625,
      "learning_rate": 2.9885736896884532e-05,
      "loss": 1.3078,
      "step": 8495
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.15625,
      "learning_rate": 2.985786803542004e-05,
      "loss": 1.2615,
      "step": 8500
    },
    {
      "epoch": 1.51,
      "grad_norm": 3.984375,
      "learning_rate": 2.9829999296613477e-05,
      "loss": 1.2039,
      "step": 8505
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.0,
      "learning_rate": 2.9802130704515188e-05,
      "loss": 1.2577,
      "step": 8510
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.71875,
      "learning_rate": 2.9774262283175376e-05,
      "loss": 1.2148,
      "step": 8515
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.21875,
      "learning_rate": 2.9746394056644106e-05,
      "loss": 1.293,
      "step": 8520
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.0625,
      "learning_rate": 2.9718526048971282e-05,
      "loss": 1.1719,
      "step": 8525
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.5625,
      "learning_rate": 2.9690658284206603e-05,
      "loss": 1.2483,
      "step": 8530
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.46875,
      "learning_rate": 2.966279078639956e-05,
      "loss": 1.2634,
      "step": 8535
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.8125,
      "learning_rate": 2.9634923579599448e-05,
      "loss": 1.2112,
      "step": 8540
    },
    {
      "epoch": 1.51,
      "grad_norm": 3.984375,
      "learning_rate": 2.9607056687855252e-05,
      "loss": 1.3329,
      "step": 8545
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.0625,
      "learning_rate": 2.957919013521574e-05,
      "loss": 1.274,
      "step": 8550
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.4375,
      "learning_rate": 2.9551323945729357e-05,
      "loss": 1.2245,
      "step": 8555
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.25,
      "learning_rate": 2.952345814344423e-05,
      "loss": 1.2301,
      "step": 8560
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.625,
      "learning_rate": 2.949559275240818e-05,
      "loss": 1.1579,
      "step": 8565
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.4375,
      "learning_rate": 2.946772779666864e-05,
      "loss": 1.2767,
      "step": 8570
    },
    {
      "epoch": 1.52,
      "grad_norm": 3.890625,
      "learning_rate": 2.9439863300272683e-05,
      "loss": 1.2106,
      "step": 8575
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.0625,
      "learning_rate": 2.9411999287267002e-05,
      "loss": 1.2151,
      "step": 8580
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.71875,
      "learning_rate": 2.9384135781697858e-05,
      "loss": 1.2881,
      "step": 8585
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.90625,
      "learning_rate": 2.9356272807611054e-05,
      "loss": 1.2159,
      "step": 8590
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.6875,
      "learning_rate": 2.932841038905197e-05,
      "loss": 1.2455,
      "step": 8595
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.53125,
      "learning_rate": 2.930054855006548e-05,
      "loss": 1.2668,
      "step": 8600
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.34375,
      "learning_rate": 2.9272687314695975e-05,
      "loss": 1.2085,
      "step": 8605
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.75,
      "learning_rate": 2.9244826706987314e-05,
      "loss": 1.1821,
      "step": 8610
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.90625,
      "learning_rate": 2.9216966750982808e-05,
      "loss": 1.2171,
      "step": 8615
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.1875,
      "learning_rate": 2.9189107470725232e-05,
      "loss": 1.225,
      "step": 8620
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.28125,
      "learning_rate": 2.9161248890256763e-05,
      "loss": 1.2498,
      "step": 8625
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.40625,
      "learning_rate": 2.913339103361896e-05,
      "loss": 1.2482,
      "step": 8630
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.28125,
      "learning_rate": 2.9105533924852786e-05,
      "loss": 1.1949,
      "step": 8635
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.40625,
      "learning_rate": 2.9077677587998524e-05,
      "loss": 1.2296,
      "step": 8640
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.5625,
      "learning_rate": 2.9049822047095822e-05,
      "loss": 1.2159,
      "step": 8645
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.9375,
      "learning_rate": 2.902196732618363e-05,
      "loss": 1.2127,
      "step": 8650
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.46875,
      "learning_rate": 2.8994113449300177e-05,
      "loss": 1.2132,
      "step": 8655
    },
    {
      "epoch": 1.53,
      "grad_norm": 3.984375,
      "learning_rate": 2.8966260440482975e-05,
      "loss": 1.2137,
      "step": 8660
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.9375,
      "learning_rate": 2.8938408323768812e-05,
      "loss": 1.2337,
      "step": 8665
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.96875,
      "learning_rate": 2.8910557123193657e-05,
      "loss": 1.1794,
      "step": 8670
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.71875,
      "learning_rate": 2.8882706862792727e-05,
      "loss": 1.2256,
      "step": 8675
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.25,
      "learning_rate": 2.885485756660041e-05,
      "loss": 1.2319,
      "step": 8680
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.28125,
      "learning_rate": 2.8827009258650253e-05,
      "loss": 1.1926,
      "step": 8685
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.0,
      "learning_rate": 2.8799161962974983e-05,
      "loss": 1.2584,
      "step": 8690
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.71875,
      "learning_rate": 2.8771315703606418e-05,
      "loss": 1.2559,
      "step": 8695
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.46875,
      "learning_rate": 2.8743470504575492e-05,
      "loss": 1.2884,
      "step": 8700
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.0,
      "learning_rate": 2.871562638991226e-05,
      "loss": 1.2673,
      "step": 8705
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.0,
      "learning_rate": 2.868778338364578e-05,
      "loss": 1.2837,
      "step": 8710
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.21875,
      "learning_rate": 2.8659941509804196e-05,
      "loss": 1.1962,
      "step": 8715
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.34375,
      "learning_rate": 2.863210079241466e-05,
      "loss": 1.3414,
      "step": 8720
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.78125,
      "learning_rate": 2.8604261255503318e-05,
      "loss": 1.3023,
      "step": 8725
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.53125,
      "learning_rate": 2.857642292309533e-05,
      "loss": 1.2483,
      "step": 8730
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.96875,
      "learning_rate": 2.8548585819214768e-05,
      "loss": 1.2163,
      "step": 8735
    },
    {
      "epoch": 1.55,
      "grad_norm": 3.6875,
      "learning_rate": 2.8520749967884678e-05,
      "loss": 1.1771,
      "step": 8740
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.03125,
      "learning_rate": 2.8492915393127036e-05,
      "loss": 1.2,
      "step": 8745
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.1875,
      "learning_rate": 2.8465082118962684e-05,
      "loss": 1.2173,
      "step": 8750
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.84375,
      "learning_rate": 2.8437250169411343e-05,
      "loss": 1.2842,
      "step": 8755
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.53125,
      "learning_rate": 2.8409419568491622e-05,
      "loss": 1.2555,
      "step": 8760
    },
    {
      "epoch": 1.55,
      "grad_norm": 3.984375,
      "learning_rate": 2.838159034022093e-05,
      "loss": 1.2138,
      "step": 8765
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.84375,
      "learning_rate": 2.8353762508615525e-05,
      "loss": 1.177,
      "step": 8770
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.8125,
      "learning_rate": 2.8325936097690432e-05,
      "loss": 1.1827,
      "step": 8775
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.4375,
      "learning_rate": 2.829811113145945e-05,
      "loss": 1.2602,
      "step": 8780
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.21875,
      "learning_rate": 2.8270287633935162e-05,
      "loss": 1.2557,
      "step": 8785
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.46875,
      "learning_rate": 2.8242465629128863e-05,
      "loss": 1.2532,
      "step": 8790
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.1875,
      "learning_rate": 2.8214645141050548e-05,
      "loss": 1.1997,
      "step": 8795
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.46875,
      "learning_rate": 2.818682619370893e-05,
      "loss": 1.2626,
      "step": 8800
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.84375,
      "learning_rate": 2.8159008811111362e-05,
      "loss": 1.1736,
      "step": 8805
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.03125,
      "learning_rate": 2.813119301726388e-05,
      "loss": 1.1969,
      "step": 8810
    },
    {
      "epoch": 1.56,
      "grad_norm": 3.890625,
      "learning_rate": 2.8103378836171118e-05,
      "loss": 1.1917,
      "step": 8815
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.875,
      "learning_rate": 2.807556629183633e-05,
      "loss": 1.3117,
      "step": 8820
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.125,
      "learning_rate": 2.804775540826136e-05,
      "loss": 1.1878,
      "step": 8825
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.0625,
      "learning_rate": 2.8019946209446638e-05,
      "loss": 1.2779,
      "step": 8830
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.5625,
      "learning_rate": 2.7992138719391096e-05,
      "loss": 1.1921,
      "step": 8835
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.03125,
      "learning_rate": 2.796433296209223e-05,
      "loss": 1.2221,
      "step": 8840
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.71875,
      "learning_rate": 2.7936528961546024e-05,
      "loss": 1.2652,
      "step": 8845
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.34375,
      "learning_rate": 2.7908726741746935e-05,
      "loss": 1.2786,
      "step": 8850
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.09375,
      "learning_rate": 2.7880926326687916e-05,
      "loss": 1.246,
      "step": 8855
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.71875,
      "learning_rate": 2.7853127740360328e-05,
      "loss": 1.2047,
      "step": 8860
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.25,
      "learning_rate": 2.7825331006753964e-05,
      "loss": 1.285,
      "step": 8865
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.40625,
      "learning_rate": 2.7797536149857047e-05,
      "loss": 1.2147,
      "step": 8870
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.46875,
      "learning_rate": 2.7769743193656144e-05,
      "loss": 1.2411,
      "step": 8875
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.875,
      "learning_rate": 2.774195216213618e-05,
      "loss": 1.21,
      "step": 8880
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.03125,
      "learning_rate": 2.7714163079280456e-05,
      "loss": 1.2104,
      "step": 8885
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.4375,
      "learning_rate": 2.768637596907055e-05,
      "loss": 1.2257,
      "step": 8890
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.53125,
      "learning_rate": 2.7658590855486366e-05,
      "loss": 1.2227,
      "step": 8895
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.78125,
      "learning_rate": 2.7630807762506068e-05,
      "loss": 1.2063,
      "step": 8900
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.625,
      "learning_rate": 2.7603026714106082e-05,
      "loss": 1.2168,
      "step": 8905
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.4375,
      "learning_rate": 2.7575247734261086e-05,
      "loss": 1.271,
      "step": 8910
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.5625,
      "learning_rate": 2.754747084694395e-05,
      "loss": 1.2211,
      "step": 8915
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.0,
      "learning_rate": 2.7519696076125733e-05,
      "loss": 1.2754,
      "step": 8920
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.5,
      "learning_rate": 2.74919234457757e-05,
      "loss": 1.198,
      "step": 8925
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.34375,
      "learning_rate": 2.746415297986123e-05,
      "loss": 1.2156,
      "step": 8930
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.4375,
      "learning_rate": 2.7436384702347875e-05,
      "loss": 1.2143,
      "step": 8935
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.71875,
      "learning_rate": 2.740861863719926e-05,
      "loss": 1.2102,
      "step": 8940
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.3125,
      "learning_rate": 2.738085480837711e-05,
      "loss": 1.1612,
      "step": 8945
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.4375,
      "learning_rate": 2.735309323984124e-05,
      "loss": 1.2098,
      "step": 8950
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.5625,
      "learning_rate": 2.7325333955549514e-05,
      "loss": 1.2207,
      "step": 8955
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.46875,
      "learning_rate": 2.729757697945779e-05,
      "loss": 1.288,
      "step": 8960
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.40625,
      "learning_rate": 2.726982233551997e-05,
      "loss": 1.2354,
      "step": 8965
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.8125,
      "learning_rate": 2.7242070047687924e-05,
      "loss": 1.2786,
      "step": 8970
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.53125,
      "learning_rate": 2.7214320139911502e-05,
      "loss": 1.2516,
      "step": 8975
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.1875,
      "learning_rate": 2.7186572636138486e-05,
      "loss": 1.3034,
      "step": 8980
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.25,
      "learning_rate": 2.7158827560314584e-05,
      "loss": 1.2262,
      "step": 8985
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.65625,
      "learning_rate": 2.7131084936383415e-05,
      "loss": 1.2658,
      "step": 8990
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.375,
      "learning_rate": 2.710334478828651e-05,
      "loss": 1.2092,
      "step": 8995
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.59375,
      "learning_rate": 2.70756071399632e-05,
      "loss": 1.3058,
      "step": 9000
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.3125,
      "learning_rate": 2.7047872015350716e-05,
      "loss": 1.2714,
      "step": 9005
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.40625,
      "learning_rate": 2.7020139438384086e-05,
      "loss": 1.1851,
      "step": 9010
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.78125,
      "learning_rate": 2.6992409432996128e-05,
      "loss": 1.1893,
      "step": 9015
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.4375,
      "learning_rate": 2.6964682023117474e-05,
      "loss": 1.2345,
      "step": 9020
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.53125,
      "learning_rate": 2.6936957232676483e-05,
      "loss": 1.2383,
      "step": 9025
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.8125,
      "learning_rate": 2.6909235085599265e-05,
      "loss": 1.2523,
      "step": 9030
    },
    {
      "epoch": 1.6,
      "grad_norm": 3.953125,
      "learning_rate": 2.688151560580967e-05,
      "loss": 1.22,
      "step": 9035
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.53125,
      "learning_rate": 2.6853798817229224e-05,
      "loss": 1.1901,
      "step": 9040
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.5625,
      "learning_rate": 2.6826084743777124e-05,
      "loss": 1.1808,
      "step": 9045
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.625,
      "learning_rate": 2.679837340937024e-05,
      "loss": 1.2417,
      "step": 9050
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.25,
      "learning_rate": 2.6770664837923066e-05,
      "loss": 1.2738,
      "step": 9055
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.21875,
      "learning_rate": 2.674295905334773e-05,
      "loss": 1.19,
      "step": 9060
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.21875,
      "learning_rate": 2.6715256079553932e-05,
      "loss": 1.1987,
      "step": 9065
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.5625,
      "learning_rate": 2.6687555940448957e-05,
      "loss": 1.1559,
      "step": 9070
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.3125,
      "learning_rate": 2.6659858659937654e-05,
      "loss": 1.245,
      "step": 9075
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.09375,
      "learning_rate": 2.6632164261922395e-05,
      "loss": 1.2034,
      "step": 9080
    },
    {
      "epoch": 1.61,
      "grad_norm": 3.890625,
      "learning_rate": 2.6604472770303054e-05,
      "loss": 1.2308,
      "step": 9085
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.40625,
      "learning_rate": 2.657678420897702e-05,
      "loss": 1.3012,
      "step": 9090
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.03125,
      "learning_rate": 2.6549098601839126e-05,
      "loss": 1.2385,
      "step": 9095
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.375,
      "learning_rate": 2.6521415972781684e-05,
      "loss": 1.2172,
      "step": 9100
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.1875,
      "learning_rate": 2.6493736345694416e-05,
      "loss": 1.2625,
      "step": 9105
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.5625,
      "learning_rate": 2.6466059744464445e-05,
      "loss": 1.1966,
      "step": 9110
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.59375,
      "learning_rate": 2.6438386192976317e-05,
      "loss": 1.2159,
      "step": 9115
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.875,
      "learning_rate": 2.6410715715111932e-05,
      "loss": 1.2293,
      "step": 9120
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.65625,
      "learning_rate": 2.6383048334750513e-05,
      "loss": 1.2052,
      "step": 9125
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.78125,
      "learning_rate": 2.6355384075768643e-05,
      "loss": 1.2416,
      "step": 9130
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.25,
      "learning_rate": 2.632772296204019e-05,
      "loss": 1.1974,
      "step": 9135
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.4375,
      "learning_rate": 2.630006501743631e-05,
      "loss": 1.1715,
      "step": 9140
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.8125,
      "learning_rate": 2.627241026582544e-05,
      "loss": 1.2225,
      "step": 9145
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.21875,
      "learning_rate": 2.6244758731073237e-05,
      "loss": 1.2184,
      "step": 9150
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.6875,
      "learning_rate": 2.6217110437042598e-05,
      "loss": 1.2246,
      "step": 9155
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.28125,
      "learning_rate": 2.6189465407593638e-05,
      "loss": 1.2882,
      "step": 9160
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.3125,
      "learning_rate": 2.616182366658362e-05,
      "loss": 1.209,
      "step": 9165
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.1875,
      "learning_rate": 2.613418523786699e-05,
      "loss": 1.2612,
      "step": 9170
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.6875,
      "learning_rate": 2.610655014529534e-05,
      "loss": 1.2252,
      "step": 9175
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.5,
      "learning_rate": 2.6078918412717357e-05,
      "loss": 1.2659,
      "step": 9180
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.28125,
      "learning_rate": 2.6051290063978867e-05,
      "loss": 1.216,
      "step": 9185
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.34375,
      "learning_rate": 2.6023665122922733e-05,
      "loss": 1.2545,
      "step": 9190
    },
    {
      "epoch": 1.63,
      "grad_norm": 3.984375,
      "learning_rate": 2.5996043613388908e-05,
      "loss": 1.2076,
      "step": 9195
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.53125,
      "learning_rate": 2.596842555921439e-05,
      "loss": 1.2028,
      "step": 9200
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.75,
      "learning_rate": 2.5940810984233167e-05,
      "loss": 1.284,
      "step": 9205
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.375,
      "learning_rate": 2.591319991227623e-05,
      "loss": 1.262,
      "step": 9210
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.84375,
      "learning_rate": 2.588559236717157e-05,
      "loss": 1.3166,
      "step": 9215
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.0625,
      "learning_rate": 2.5857988372744105e-05,
      "loss": 1.1672,
      "step": 9220
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.25,
      "learning_rate": 2.5830387952815718e-05,
      "loss": 1.2457,
      "step": 9225
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.625,
      "learning_rate": 2.580279113120517e-05,
      "loss": 1.2493,
      "step": 9230
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.5,
      "learning_rate": 2.5775197931728153e-05,
      "loss": 1.2051,
      "step": 9235
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.375,
      "learning_rate": 2.574760837819722e-05,
      "loss": 1.1574,
      "step": 9240
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.40625,
      "learning_rate": 2.572002249442178e-05,
      "loss": 1.2136,
      "step": 9245
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.15625,
      "learning_rate": 2.5692440304208052e-05,
      "loss": 1.2651,
      "step": 9250
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.40625,
      "learning_rate": 2.566486183135911e-05,
      "loss": 1.2261,
      "step": 9255
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.71875,
      "learning_rate": 2.5637287099674775e-05,
      "loss": 1.2706,
      "step": 9260
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.5625,
      "learning_rate": 2.560971613295168e-05,
      "loss": 1.2091,
      "step": 9265
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.25,
      "learning_rate": 2.558214895498318e-05,
      "loss": 1.283,
      "step": 9270
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.28125,
      "learning_rate": 2.5554585589559366e-05,
      "loss": 1.2048,
      "step": 9275
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.125,
      "learning_rate": 2.5527026060467052e-05,
      "loss": 1.2221,
      "step": 9280
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.40625,
      "learning_rate": 2.549947039148974e-05,
      "loss": 1.3335,
      "step": 9285
    },
    {
      "epoch": 1.65,
      "grad_norm": 3.96875,
      "learning_rate": 2.5471918606407575e-05,
      "loss": 1.2139,
      "step": 9290
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.15625,
      "learning_rate": 2.5444370728997383e-05,
      "loss": 1.2791,
      "step": 9295
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.28125,
      "learning_rate": 2.5416826783032596e-05,
      "loss": 1.2211,
      "step": 9300
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.09375,
      "learning_rate": 2.538928679228326e-05,
      "loss": 1.234,
      "step": 9305
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.96875,
      "learning_rate": 2.5361750780516014e-05,
      "loss": 1.2598,
      "step": 9310
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.8125,
      "learning_rate": 2.533421877149405e-05,
      "loss": 1.2217,
      "step": 9315
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.71875,
      "learning_rate": 2.5306690788977108e-05,
      "loss": 1.2758,
      "step": 9320
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.84375,
      "learning_rate": 2.527916685672148e-05,
      "loss": 1.1914,
      "step": 9325
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.59375,
      "learning_rate": 2.5251646998479924e-05,
      "loss": 1.246,
      "step": 9330
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.84375,
      "learning_rate": 2.522413123800171e-05,
      "loss": 1.2388,
      "step": 9335
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.90625,
      "learning_rate": 2.519661959903255e-05,
      "loss": 1.2098,
      "step": 9340
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.25,
      "learning_rate": 2.5169112105314614e-05,
      "loss": 1.2659,
      "step": 9345
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.65625,
      "learning_rate": 2.5141608780586494e-05,
      "loss": 1.2059,
      "step": 9350
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.46875,
      "learning_rate": 2.5114109648583172e-05,
      "loss": 1.2435,
      "step": 9355
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.125,
      "learning_rate": 2.5086614733036024e-05,
      "loss": 1.207,
      "step": 9360
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.8125,
      "learning_rate": 2.5059124057672797e-05,
      "loss": 1.2742,
      "step": 9365
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.4375,
      "learning_rate": 2.5031637646217557e-05,
      "loss": 1.2421,
      "step": 9370
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.0,
      "learning_rate": 2.500415552239069e-05,
      "loss": 1.2334,
      "step": 9375
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.625,
      "learning_rate": 2.4976677709908902e-05,
      "loss": 1.1658,
      "step": 9380
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.59375,
      "learning_rate": 2.4949204232485163e-05,
      "loss": 1.3034,
      "step": 9385
    },
    {
      "epoch": 1.66,
      "grad_norm": 3.9375,
      "learning_rate": 2.492173511382871e-05,
      "loss": 1.2718,
      "step": 9390
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.4375,
      "learning_rate": 2.4894270377645012e-05,
      "loss": 1.2526,
      "step": 9395
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.875,
      "learning_rate": 2.486681004763575e-05,
      "loss": 1.3093,
      "step": 9400
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.03125,
      "learning_rate": 2.483935414749882e-05,
      "loss": 1.2795,
      "step": 9405
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.6875,
      "learning_rate": 2.4811902700928293e-05,
      "loss": 1.254,
      "step": 9410
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.34375,
      "learning_rate": 2.4784455731614378e-05,
      "loss": 1.2126,
      "step": 9415
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.5625,
      "learning_rate": 2.4757013263243443e-05,
      "loss": 1.2063,
      "step": 9420
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.375,
      "learning_rate": 2.4729575319497944e-05,
      "loss": 1.1925,
      "step": 9425
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.21875,
      "learning_rate": 2.470214192405647e-05,
      "loss": 1.202,
      "step": 9430
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.125,
      "learning_rate": 2.467471310059365e-05,
      "loss": 1.2715,
      "step": 9435
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.21875,
      "learning_rate": 2.4647288872780175e-05,
      "loss": 1.2031,
      "step": 9440
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.5625,
      "learning_rate": 2.461986926428279e-05,
      "loss": 1.2453,
      "step": 9445
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.0,
      "learning_rate": 2.4592454298764244e-05,
      "loss": 1.184,
      "step": 9450
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.59375,
      "learning_rate": 2.4565043999883262e-05,
      "loss": 1.3203,
      "step": 9455
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.625,
      "learning_rate": 2.4537638391294565e-05,
      "loss": 1.182,
      "step": 9460
    },
    {
      "epoch": 1.68,
      "grad_norm": 3.96875,
      "learning_rate": 2.4510237496648813e-05,
      "loss": 1.2617,
      "step": 9465
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.78125,
      "learning_rate": 2.4482841339592592e-05,
      "loss": 1.2159,
      "step": 9470
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.15625,
      "learning_rate": 2.4455449943768427e-05,
      "loss": 1.2835,
      "step": 9475
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.125,
      "learning_rate": 2.4428063332814694e-05,
      "loss": 1.2629,
      "step": 9480
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.46875,
      "learning_rate": 2.440068153036567e-05,
      "loss": 1.1939,
      "step": 9485
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.34375,
      "learning_rate": 2.437330456005149e-05,
      "loss": 1.2551,
      "step": 9490
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.3125,
      "learning_rate": 2.4345932445498088e-05,
      "loss": 1.2447,
      "step": 9495
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.15625,
      "learning_rate": 2.4318565210327217e-05,
      "loss": 1.23,
      "step": 9500
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.34375,
      "learning_rate": 2.4291202878156442e-05,
      "loss": 1.2258,
      "step": 9505
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.53125,
      "learning_rate": 2.4263845472599058e-05,
      "loss": 1.3003,
      "step": 9510
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.5625,
      "learning_rate": 2.423649301726415e-05,
      "loss": 1.2588,
      "step": 9515
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.5625,
      "learning_rate": 2.42091455357565e-05,
      "loss": 1.1817,
      "step": 9520
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.78125,
      "learning_rate": 2.4181803051676597e-05,
      "loss": 1.2335,
      "step": 9525
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.25,
      "learning_rate": 2.4154465588620656e-05,
      "loss": 1.2761,
      "step": 9530
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.65625,
      "learning_rate": 2.412713317018052e-05,
      "loss": 1.2143,
      "step": 9535
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.375,
      "learning_rate": 2.4099805819943678e-05,
      "loss": 1.2038,
      "step": 9540
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.75,
      "learning_rate": 2.4072483561493276e-05,
      "loss": 1.2474,
      "step": 9545
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.0625,
      "learning_rate": 2.404516641840803e-05,
      "loss": 1.2524,
      "step": 9550
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.84375,
      "learning_rate": 2.4017854414262276e-05,
      "loss": 1.2824,
      "step": 9555
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.0,
      "learning_rate": 2.399054757262589e-05,
      "loss": 1.1828,
      "step": 9560
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.625,
      "learning_rate": 2.3963245917064284e-05,
      "loss": 1.1856,
      "step": 9565
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.625,
      "learning_rate": 2.3935949471138434e-05,
      "loss": 1.2389,
      "step": 9570
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.375,
      "learning_rate": 2.39086582584048e-05,
      "loss": 1.251,
      "step": 9575
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.6875,
      "learning_rate": 2.3881372302415304e-05,
      "loss": 1.2674,
      "step": 9580
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.78125,
      "learning_rate": 2.385409162671737e-05,
      "loss": 1.2149,
      "step": 9585
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.28125,
      "learning_rate": 2.382681625485383e-05,
      "loss": 1.2698,
      "step": 9590
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.625,
      "learning_rate": 2.3799546210362964e-05,
      "loss": 1.1687,
      "step": 9595
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.25,
      "learning_rate": 2.3772281516778444e-05,
      "loss": 1.2271,
      "step": 9600
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.125,
      "learning_rate": 2.3745022197629315e-05,
      "loss": 1.2536,
      "step": 9605
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.25,
      "learning_rate": 2.3717768276440008e-05,
      "loss": 1.2312,
      "step": 9610
    },
    {
      "epoch": 1.7,
      "grad_norm": 3.734375,
      "learning_rate": 2.3690519776730285e-05,
      "loss": 1.2273,
      "step": 9615
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.25,
      "learning_rate": 2.3663276722015216e-05,
      "loss": 1.3296,
      "step": 9620
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.71875,
      "learning_rate": 2.3636039135805184e-05,
      "loss": 1.2372,
      "step": 9625
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.375,
      "learning_rate": 2.360880704160586e-05,
      "loss": 1.1788,
      "step": 9630
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.0625,
      "learning_rate": 2.3581580462918153e-05,
      "loss": 1.1941,
      "step": 9635
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.21875,
      "learning_rate": 2.3554359423238237e-05,
      "loss": 1.234,
      "step": 9640
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.0625,
      "learning_rate": 2.3527143946057487e-05,
      "loss": 1.2099,
      "step": 9645
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.71875,
      "learning_rate": 2.3499934054862485e-05,
      "loss": 1.2352,
      "step": 9650
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.5625,
      "learning_rate": 2.3472729773135007e-05,
      "loss": 1.2427,
      "step": 9655
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.125,
      "learning_rate": 2.344553112435196e-05,
      "loss": 1.2594,
      "step": 9660
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.375,
      "learning_rate": 2.341833813198541e-05,
      "loss": 1.2753,
      "step": 9665
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.21875,
      "learning_rate": 2.3391150819502537e-05,
      "loss": 1.2217,
      "step": 9670
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.6875,
      "learning_rate": 2.3363969210365603e-05,
      "loss": 1.2343,
      "step": 9675
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.75,
      "learning_rate": 2.3336793328031985e-05,
      "loss": 1.3,
      "step": 9680
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.34375,
      "learning_rate": 2.330962319595408e-05,
      "loss": 1.1879,
      "step": 9685
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.625,
      "learning_rate": 2.3282458837579332e-05,
      "loss": 1.308,
      "step": 9690
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.4375,
      "learning_rate": 2.325530027635024e-05,
      "loss": 1.2088,
      "step": 9695
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.34375,
      "learning_rate": 2.322814753570424e-05,
      "loss": 1.2339,
      "step": 9700
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.4375,
      "learning_rate": 2.3201000639073782e-05,
      "loss": 1.2529,
      "step": 9705
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.84375,
      "learning_rate": 2.317385960988627e-05,
      "loss": 1.1942,
      "step": 9710
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.625,
      "learning_rate": 2.3146724471564027e-05,
      "loss": 1.219,
      "step": 9715
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.28125,
      "learning_rate": 2.3119595247524316e-05,
      "loss": 1.2757,
      "step": 9720
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.375,
      "learning_rate": 2.309247196117928e-05,
      "loss": 1.2365,
      "step": 9725
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.4375,
      "learning_rate": 2.306535463593593e-05,
      "loss": 1.2427,
      "step": 9730
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.3125,
      "learning_rate": 2.303824329519616e-05,
      "loss": 1.2575,
      "step": 9735
    },
    {
      "epoch": 1.73,
      "grad_norm": 3.875,
      "learning_rate": 2.3011137962356688e-05,
      "loss": 1.2286,
      "step": 9740
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.9375,
      "learning_rate": 2.298403866080903e-05,
      "loss": 1.1721,
      "step": 9745
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.21875,
      "learning_rate": 2.2956945413939523e-05,
      "loss": 1.1961,
      "step": 9750
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.46875,
      "learning_rate": 2.292985824512926e-05,
      "loss": 1.2077,
      "step": 9755
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.78125,
      "learning_rate": 2.2902777177754086e-05,
      "loss": 1.2,
      "step": 9760
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.21875,
      "learning_rate": 2.2875702235184604e-05,
      "loss": 1.1728,
      "step": 9765
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.90625,
      "learning_rate": 2.28486334407861e-05,
      "loss": 1.23,
      "step": 9770
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.59375,
      "learning_rate": 2.282157081791859e-05,
      "loss": 1.2252,
      "step": 9775
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.53125,
      "learning_rate": 2.279451438993675e-05,
      "loss": 1.2351,
      "step": 9780
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.40625,
      "learning_rate": 2.2767464180189878e-05,
      "loss": 1.2357,
      "step": 9785
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.28125,
      "learning_rate": 2.2740420212021957e-05,
      "loss": 1.1844,
      "step": 9790
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.28125,
      "learning_rate": 2.2713382508771552e-05,
      "loss": 1.2376,
      "step": 9795
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.5,
      "learning_rate": 2.2686351093771822e-05,
      "loss": 1.207,
      "step": 9800
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.09375,
      "learning_rate": 2.2659325990350515e-05,
      "loss": 1.1888,
      "step": 9805
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.3125,
      "learning_rate": 2.2632307221829922e-05,
      "loss": 1.2653,
      "step": 9810
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.125,
      "learning_rate": 2.2605294811526856e-05,
      "loss": 1.2371,
      "step": 9815
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.90625,
      "learning_rate": 2.257828878275268e-05,
      "loss": 1.1258,
      "step": 9820
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.28125,
      "learning_rate": 2.2551289158813215e-05,
      "loss": 1.2575,
      "step": 9825
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.28125,
      "learning_rate": 2.2524295963008756e-05,
      "loss": 1.2713,
      "step": 9830
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.21875,
      "learning_rate": 2.249730921863407e-05,
      "loss": 1.2667,
      "step": 9835
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.46875,
      "learning_rate": 2.247032894897834e-05,
      "loss": 1.3168,
      "step": 9840
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.25,
      "learning_rate": 2.2443355177325178e-05,
      "loss": 1.2969,
      "step": 9845
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.34375,
      "learning_rate": 2.2416387926952558e-05,
      "loss": 1.2525,
      "step": 9850
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.78125,
      "learning_rate": 2.2389427221132862e-05,
      "loss": 1.252,
      "step": 9855
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.1875,
      "learning_rate": 2.2362473083132805e-05,
      "loss": 1.2446,
      "step": 9860
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.375,
      "learning_rate": 2.2335525536213447e-05,
      "loss": 1.2247,
      "step": 9865
    },
    {
      "epoch": 1.75,
      "grad_norm": 3.984375,
      "learning_rate": 2.2308584603630127e-05,
      "loss": 1.2229,
      "step": 9870
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.25,
      "learning_rate": 2.228165030863252e-05,
      "loss": 1.2547,
      "step": 9875
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.84375,
      "learning_rate": 2.2254722674464545e-05,
      "loss": 1.2873,
      "step": 9880
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.40625,
      "learning_rate": 2.222780172436438e-05,
      "loss": 1.1998,
      "step": 9885
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.59375,
      "learning_rate": 2.220088748156444e-05,
      "loss": 1.2603,
      "step": 9890
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.75,
      "learning_rate": 2.2173979969291325e-05,
      "loss": 1.2554,
      "step": 9895
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.75,
      "learning_rate": 2.2147079210765873e-05,
      "loss": 1.165,
      "step": 9900
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.46875,
      "learning_rate": 2.2120185229203064e-05,
      "loss": 1.1839,
      "step": 9905
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.28125,
      "learning_rate": 2.2093298047812023e-05,
      "loss": 1.3033,
      "step": 9910
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.34375,
      "learning_rate": 2.206641768979603e-05,
      "loss": 1.2337,
      "step": 9915
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.78125,
      "learning_rate": 2.2039544178352465e-05,
      "loss": 1.2301,
      "step": 9920
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.1875,
      "learning_rate": 2.201267753667278e-05,
      "loss": 1.2458,
      "step": 9925
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.28125,
      "learning_rate": 2.1985817787942542e-05,
      "loss": 1.2615,
      "step": 9930
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.78125,
      "learning_rate": 2.1958964955341318e-05,
      "loss": 1.216,
      "step": 9935
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.28125,
      "learning_rate": 2.1932119062042753e-05,
      "loss": 1.2266,
      "step": 9940
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.1875,
      "learning_rate": 2.1905280131214484e-05,
      "loss": 1.2197,
      "step": 9945
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.96875,
      "learning_rate": 2.187844818601813e-05,
      "loss": 1.1808,
      "step": 9950
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.375,
      "learning_rate": 2.18516232496093e-05,
      "loss": 1.2173,
      "step": 9955
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.40625,
      "learning_rate": 2.182480534513754e-05,
      "loss": 1.222,
      "step": 9960
    },
    {
      "epoch": 1.77,
      "grad_norm": 3.84375,
      "learning_rate": 2.1797994495746325e-05,
      "loss": 1.2228,
      "step": 9965
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.6875,
      "learning_rate": 2.1771190724573066e-05,
      "loss": 1.1286,
      "step": 9970
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.40625,
      "learning_rate": 2.1744394054749027e-05,
      "loss": 1.1704,
      "step": 9975
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.59375,
      "learning_rate": 2.171760450939938e-05,
      "loss": 1.2409,
      "step": 9980
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.875,
      "learning_rate": 2.1690822111643146e-05,
      "loss": 1.2584,
      "step": 9985
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.125,
      "learning_rate": 2.1664046884593147e-05,
      "loss": 1.2551,
      "step": 9990
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.375,
      "learning_rate": 2.1637278851356034e-05,
      "loss": 1.2601,
      "step": 9995
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.9375,
      "learning_rate": 2.1610518035032265e-05,
      "loss": 1.2196,
      "step": 10000
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.6875,
      "learning_rate": 2.1583764458716046e-05,
      "loss": 1.1912,
      "step": 10005
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.09375,
      "learning_rate": 2.1557018145495357e-05,
      "loss": 1.2636,
      "step": 10010
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.28125,
      "learning_rate": 2.1530279118451885e-05,
      "loss": 1.2286,
      "step": 10015
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.09375,
      "learning_rate": 2.1503547400661045e-05,
      "loss": 1.1407,
      "step": 10020
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.03125,
      "learning_rate": 2.1476823015191943e-05,
      "loss": 1.3042,
      "step": 10025
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.21875,
      "learning_rate": 2.1450105985107368e-05,
      "loss": 1.2152,
      "step": 10030
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.03125,
      "learning_rate": 2.1423396333463733e-05,
      "loss": 1.2356,
      "step": 10035
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.46875,
      "learning_rate": 2.1396694083311116e-05,
      "loss": 1.282,
      "step": 10040
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.125,
      "learning_rate": 2.1369999257693177e-05,
      "loss": 1.1886,
      "step": 10045
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.375,
      "learning_rate": 2.1343311879647188e-05,
      "loss": 1.166,
      "step": 10050
    },
    {
      "epoch": 1.78,
      "grad_norm": 3.90625,
      "learning_rate": 2.1316631972203997e-05,
      "loss": 1.3303,
      "step": 10055
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.6875,
      "learning_rate": 2.1289959558387974e-05,
      "loss": 1.2369,
      "step": 10060
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.375,
      "learning_rate": 2.126329466121707e-05,
      "loss": 1.2935,
      "step": 10065
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.21875,
      "learning_rate": 2.1236637303702725e-05,
      "loss": 1.2315,
      "step": 10070
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.3125,
      "learning_rate": 2.120998750884985e-05,
      "loss": 1.2752,
      "step": 10075
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.4375,
      "learning_rate": 2.1183345299656873e-05,
      "loss": 1.21,
      "step": 10080
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.5,
      "learning_rate": 2.115671069911565e-05,
      "loss": 1.2175,
      "step": 10085
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.34375,
      "learning_rate": 2.1130083730211464e-05,
      "loss": 1.2338,
      "step": 10090
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.5,
      "learning_rate": 2.1103464415923034e-05,
      "loss": 1.2426,
      "step": 10095
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.34375,
      "learning_rate": 2.107685277922244e-05,
      "loss": 1.1281,
      "step": 10100
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.40625,
      "learning_rate": 2.105024884307519e-05,
      "loss": 1.2764,
      "step": 10105
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.90625,
      "learning_rate": 2.1023652630440103e-05,
      "loss": 1.2374,
      "step": 10110
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.03125,
      "learning_rate": 2.0997064164269346e-05,
      "loss": 1.3095,
      "step": 10115
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.59375,
      "learning_rate": 2.097048346750839e-05,
      "loss": 1.1059,
      "step": 10120
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.71875,
      "learning_rate": 2.0943910563096023e-05,
      "loss": 1.2832,
      "step": 10125
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.5625,
      "learning_rate": 2.0917345473964282e-05,
      "loss": 1.3174,
      "step": 10130
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.0625,
      "learning_rate": 2.089078822303849e-05,
      "loss": 1.1703,
      "step": 10135
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.4375,
      "learning_rate": 2.086423883323717e-05,
      "loss": 1.1923,
      "step": 10140
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.4375,
      "learning_rate": 2.0837697327472093e-05,
      "loss": 1.2809,
      "step": 10145
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.625,
      "learning_rate": 2.081116372864822e-05,
      "loss": 1.1973,
      "step": 10150
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.5625,
      "learning_rate": 2.0784638059663678e-05,
      "loss": 1.1783,
      "step": 10155
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.75,
      "learning_rate": 2.075812034340974e-05,
      "loss": 1.2367,
      "step": 10160
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.625,
      "learning_rate": 2.0731610602770848e-05,
      "loss": 1.2496,
      "step": 10165
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.4375,
      "learning_rate": 2.0705108860624534e-05,
      "loss": 1.237,
      "step": 10170
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.21875,
      "learning_rate": 2.067861513984144e-05,
      "loss": 1.1803,
      "step": 10175
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.34375,
      "learning_rate": 2.065212946328528e-05,
      "loss": 1.2181,
      "step": 10180
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.15625,
      "learning_rate": 2.0625651853812823e-05,
      "loss": 1.2144,
      "step": 10185
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.96875,
      "learning_rate": 2.0599182334273897e-05,
      "loss": 1.2775,
      "step": 10190
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.9375,
      "learning_rate": 2.0572720927511328e-05,
      "loss": 1.2114,
      "step": 10195
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.28125,
      "learning_rate": 2.0546267656360938e-05,
      "loss": 1.2475,
      "step": 10200
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.0,
      "learning_rate": 2.051982254365155e-05,
      "loss": 1.196,
      "step": 10205
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.0,
      "learning_rate": 2.049338561220492e-05,
      "loss": 1.2303,
      "step": 10210
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.0,
      "learning_rate": 2.0466956884835765e-05,
      "loss": 1.2322,
      "step": 10215
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.46875,
      "learning_rate": 2.044053638435171e-05,
      "loss": 1.3535,
      "step": 10220
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.78125,
      "learning_rate": 2.0414124133553268e-05,
      "loss": 1.3262,
      "step": 10225
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.46875,
      "learning_rate": 2.0387720155233867e-05,
      "loss": 1.2842,
      "step": 10230
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.96875,
      "learning_rate": 2.0361324472179784e-05,
      "loss": 1.2659,
      "step": 10235
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.15625,
      "learning_rate": 2.033493710717011e-05,
      "loss": 1.2254,
      "step": 10240
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.5,
      "learning_rate": 2.0308558082976797e-05,
      "loss": 1.2179,
      "step": 10245
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.625,
      "learning_rate": 2.0282187422364568e-05,
      "loss": 1.26,
      "step": 10250
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.59375,
      "learning_rate": 2.0255825148090936e-05,
      "loss": 1.2971,
      "step": 10255
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.25,
      "learning_rate": 2.0229471282906196e-05,
      "loss": 1.1954,
      "step": 10260
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.1875,
      "learning_rate": 2.0203125849553353e-05,
      "loss": 1.2139,
      "step": 10265
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.96875,
      "learning_rate": 2.0176788870768166e-05,
      "loss": 1.215,
      "step": 10270
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.375,
      "learning_rate": 2.015046036927909e-05,
      "loss": 1.2333,
      "step": 10275
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.8125,
      "learning_rate": 2.012414036780726e-05,
      "loss": 1.2548,
      "step": 10280
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.5,
      "learning_rate": 2.0097828889066455e-05,
      "loss": 1.236,
      "step": 10285
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.46875,
      "learning_rate": 2.0071525955763142e-05,
      "loss": 1.2928,
      "step": 10290
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.3125,
      "learning_rate": 2.0045231590596376e-05,
      "loss": 1.214,
      "step": 10295
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.40625,
      "learning_rate": 2.001894581625784e-05,
      "loss": 1.1807,
      "step": 10300
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.25,
      "learning_rate": 1.9992668655431786e-05,
      "loss": 1.1957,
      "step": 10305
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.4375,
      "learning_rate": 1.9966400130795052e-05,
      "loss": 1.1615,
      "step": 10310
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.0625,
      "learning_rate": 1.9940140265017024e-05,
      "loss": 1.2096,
      "step": 10315
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.28125,
      "learning_rate": 1.9913889080759586e-05,
      "loss": 1.2963,
      "step": 10320
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.65625,
      "learning_rate": 1.988764660067715e-05,
      "loss": 1.2614,
      "step": 10325
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.53125,
      "learning_rate": 1.986141284741663e-05,
      "loss": 1.2647,
      "step": 10330
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.90625,
      "learning_rate": 1.9835187843617382e-05,
      "loss": 1.2544,
      "step": 10335
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.6875,
      "learning_rate": 1.9808971611911228e-05,
      "loss": 1.226,
      "step": 10340
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.84375,
      "learning_rate": 1.9782764174922415e-05,
      "loss": 1.1379,
      "step": 10345
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.78125,
      "learning_rate": 1.9756565555267587e-05,
      "loss": 1.2248,
      "step": 10350
    },
    {
      "epoch": 1.83,
      "grad_norm": 3.984375,
      "learning_rate": 1.9730375775555813e-05,
      "loss": 1.1641,
      "step": 10355
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.4375,
      "learning_rate": 1.970419485838851e-05,
      "loss": 1.2268,
      "step": 10360
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.0,
      "learning_rate": 1.9678022826359442e-05,
      "loss": 1.2274,
      "step": 10365
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.0625,
      "learning_rate": 1.9651859702054715e-05,
      "loss": 1.2171,
      "step": 10370
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.4375,
      "learning_rate": 1.9625705508052748e-05,
      "loss": 1.1871,
      "step": 10375
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.09375,
      "learning_rate": 1.9599560266924248e-05,
      "loss": 1.2849,
      "step": 10380
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.9375,
      "learning_rate": 1.95734240012322e-05,
      "loss": 1.2191,
      "step": 10385
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.4375,
      "learning_rate": 1.9547296733531824e-05,
      "loss": 1.2566,
      "step": 10390
    },
    {
      "epoch": 1.84,
      "grad_norm": 3.953125,
      "learning_rate": 1.9521178486370627e-05,
      "loss": 1.2119,
      "step": 10395
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.71875,
      "learning_rate": 1.949506928228828e-05,
      "loss": 1.2778,
      "step": 10400
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.625,
      "learning_rate": 1.9468969143816655e-05,
      "loss": 1.282,
      "step": 10405
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.59375,
      "learning_rate": 1.9442878093479836e-05,
      "loss": 1.2939,
      "step": 10410
    },
    {
      "epoch": 1.85,
      "grad_norm": 3.953125,
      "learning_rate": 1.9416796153794027e-05,
      "loss": 1.2547,
      "step": 10415
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.8125,
      "learning_rate": 1.939072334726758e-05,
      "loss": 1.2143,
      "step": 10420
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.0,
      "learning_rate": 1.9364659696400978e-05,
      "loss": 1.2312,
      "step": 10425
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.78125,
      "learning_rate": 1.9338605223686777e-05,
      "loss": 1.2821,
      "step": 10430
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.46875,
      "learning_rate": 1.9312559951609644e-05,
      "loss": 1.124,
      "step": 10435
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.5625,
      "learning_rate": 1.92865239026463e-05,
      "loss": 1.2027,
      "step": 10440
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.0,
      "learning_rate": 1.926049709926548e-05,
      "loss": 1.2098,
      "step": 10445
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.1875,
      "learning_rate": 1.9234479563927956e-05,
      "loss": 1.281,
      "step": 10450
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.0625,
      "learning_rate": 1.9208471319086508e-05,
      "loss": 1.2019,
      "step": 10455
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.21875,
      "learning_rate": 1.9182472387185897e-05,
      "loss": 1.1232,
      "step": 10460
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.625,
      "learning_rate": 1.9156482790662843e-05,
      "loss": 1.329,
      "step": 10465
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.90625,
      "learning_rate": 1.9130502551945996e-05,
      "loss": 1.2313,
      "step": 10470
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.125,
      "learning_rate": 1.9104531693455965e-05,
      "loss": 1.2072,
      "step": 10475
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.09375,
      "learning_rate": 1.907857023760524e-05,
      "loss": 1.2114,
      "step": 10480
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.65625,
      "learning_rate": 1.9052618206798195e-05,
      "loss": 1.2417,
      "step": 10485
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.40625,
      "learning_rate": 1.9026675623431073e-05,
      "loss": 1.258,
      "step": 10490
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.40625,
      "learning_rate": 1.9000742509891978e-05,
      "loss": 1.1946,
      "step": 10495
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.53125,
      "learning_rate": 1.897481888856081e-05,
      "loss": 1.1545,
      "step": 10500
    },
    {
      "epoch": 1.86,
      "grad_norm": 3.78125,
      "learning_rate": 1.8948904781809318e-05,
      "loss": 1.2631,
      "step": 10505
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.625,
      "learning_rate": 1.892300021200102e-05,
      "loss": 1.1224,
      "step": 10510
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.15625,
      "learning_rate": 1.8897105201491177e-05,
      "loss": 1.2031,
      "step": 10515
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.6875,
      "learning_rate": 1.887121977262685e-05,
      "loss": 1.1983,
      "step": 10520
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.21875,
      "learning_rate": 1.8845343947746813e-05,
      "loss": 1.2446,
      "step": 10525
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.3125,
      "learning_rate": 1.8819477749181528e-05,
      "loss": 1.24,
      "step": 10530
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.40625,
      "learning_rate": 1.8793621199253184e-05,
      "loss": 1.2421,
      "step": 10535
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.375,
      "learning_rate": 1.8767774320275628e-05,
      "loss": 1.2379,
      "step": 10540
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.21875,
      "learning_rate": 1.8741937134554347e-05,
      "loss": 1.2196,
      "step": 10545
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.5625,
      "learning_rate": 1.871610966438649e-05,
      "loss": 1.2877,
      "step": 10550
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.625,
      "learning_rate": 1.8690291932060786e-05,
      "loss": 1.1435,
      "step": 10555
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.6875,
      "learning_rate": 1.866448395985761e-05,
      "loss": 1.3104,
      "step": 10560
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.0625,
      "learning_rate": 1.863868577004889e-05,
      "loss": 1.2981,
      "step": 10565
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.53125,
      "learning_rate": 1.8612897384898088e-05,
      "loss": 1.2738,
      "step": 10570
    },
    {
      "epoch": 1.87,
      "grad_norm": 3.9375,
      "learning_rate": 1.8587118826660236e-05,
      "loss": 1.1909,
      "step": 10575
    },
    {
      "epoch": 1.87,
      "grad_norm": 3.84375,
      "learning_rate": 1.8561350117581876e-05,
      "loss": 1.2391,
      "step": 10580
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.15625,
      "learning_rate": 1.8535591279901043e-05,
      "loss": 1.2303,
      "step": 10585
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.59375,
      "learning_rate": 1.8509842335847268e-05,
      "loss": 1.2008,
      "step": 10590
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.5,
      "learning_rate": 1.848410330764152e-05,
      "loss": 1.2388,
      "step": 10595
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.25,
      "learning_rate": 1.845837421749624e-05,
      "loss": 1.2994,
      "step": 10600
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.71875,
      "learning_rate": 1.8432655087615286e-05,
      "loss": 1.2355,
      "step": 10605
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.5,
      "learning_rate": 1.8406945940193905e-05,
      "loss": 1.2061,
      "step": 10610
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.28125,
      "learning_rate": 1.838124679741873e-05,
      "loss": 1.354,
      "step": 10615
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.5,
      "learning_rate": 1.8355557681467778e-05,
      "loss": 1.2445,
      "step": 10620
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.40625,
      "learning_rate": 1.8329878614510394e-05,
      "loss": 1.2624,
      "step": 10625
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.3125,
      "learning_rate": 1.8304209618707276e-05,
      "loss": 1.1915,
      "step": 10630
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.03125,
      "learning_rate": 1.8278550716210384e-05,
      "loss": 1.2452,
      "step": 10635
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.375,
      "learning_rate": 1.825290192916303e-05,
      "loss": 1.2235,
      "step": 10640
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.84375,
      "learning_rate": 1.822726327969975e-05,
      "loss": 1.2426,
      "step": 10645
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.03125,
      "learning_rate": 1.820163478994635e-05,
      "loss": 1.2017,
      "step": 10650
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.78125,
      "learning_rate": 1.817601648201986e-05,
      "loss": 1.2144,
      "step": 10655
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.9375,
      "learning_rate": 1.815040837802853e-05,
      "loss": 1.306,
      "step": 10660
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.625,
      "learning_rate": 1.8124810500071796e-05,
      "loss": 1.189,
      "step": 10665
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.25,
      "learning_rate": 1.809922287024029e-05,
      "loss": 1.1786,
      "step": 10670
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.6875,
      "learning_rate": 1.807364551061577e-05,
      "loss": 1.183,
      "step": 10675
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.78125,
      "learning_rate": 1.8048078443271136e-05,
      "loss": 1.2631,
      "step": 10680
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.21875,
      "learning_rate": 1.8022521690270435e-05,
      "loss": 1.2272,
      "step": 10685
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.84375,
      "learning_rate": 1.7996975273668795e-05,
      "loss": 1.2642,
      "step": 10690
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.59375,
      "learning_rate": 1.7971439215512408e-05,
      "loss": 1.2779,
      "step": 10695
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.0,
      "learning_rate": 1.794591353783855e-05,
      "loss": 1.1558,
      "step": 10700
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.6875,
      "learning_rate": 1.7920398262675535e-05,
      "loss": 1.1882,
      "step": 10705
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.1875,
      "learning_rate": 1.7894893412042675e-05,
      "loss": 1.2556,
      "step": 10710
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.8125,
      "learning_rate": 1.7869399007950325e-05,
      "loss": 1.1912,
      "step": 10715
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.15625,
      "learning_rate": 1.7843915072399783e-05,
      "loss": 1.2509,
      "step": 10720
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.8125,
      "learning_rate": 1.781844162738336e-05,
      "loss": 1.2804,
      "step": 10725
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.125,
      "learning_rate": 1.7792978694884282e-05,
      "loss": 1.2376,
      "step": 10730
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.15625,
      "learning_rate": 1.7767526296876696e-05,
      "loss": 1.2882,
      "step": 10735
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.625,
      "learning_rate": 1.7742084455325692e-05,
      "loss": 1.2569,
      "step": 10740
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.0625,
      "learning_rate": 1.7716653192187213e-05,
      "loss": 1.1303,
      "step": 10745
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.71875,
      "learning_rate": 1.7691232529408093e-05,
      "loss": 1.1813,
      "step": 10750
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.1875,
      "learning_rate": 1.7665822488926018e-05,
      "loss": 1.2501,
      "step": 10755
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.71875,
      "learning_rate": 1.7640423092669488e-05,
      "loss": 1.2762,
      "step": 10760
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.3125,
      "learning_rate": 1.7615034362557857e-05,
      "loss": 1.2352,
      "step": 10765
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.84375,
      "learning_rate": 1.7589656320501243e-05,
      "loss": 1.2967,
      "step": 10770
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.6875,
      "learning_rate": 1.756428898840054e-05,
      "loss": 1.3101,
      "step": 10775
    },
    {
      "epoch": 1.91,
      "grad_norm": 3.78125,
      "learning_rate": 1.753893238814741e-05,
      "loss": 1.2059,
      "step": 10780
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.03125,
      "learning_rate": 1.7513586541624254e-05,
      "loss": 1.2329,
      "step": 10785
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.96875,
      "learning_rate": 1.748825147070418e-05,
      "loss": 1.2249,
      "step": 10790
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.875,
      "learning_rate": 1.7462927197251008e-05,
      "loss": 1.2879,
      "step": 10795
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.4375,
      "learning_rate": 1.743761374311924e-05,
      "loss": 1.2719,
      "step": 10800
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.5625,
      "learning_rate": 1.7412311130154042e-05,
      "loss": 1.2503,
      "step": 10805
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.15625,
      "learning_rate": 1.738701938019123e-05,
      "loss": 1.2268,
      "step": 10810
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.6875,
      "learning_rate": 1.7361738515057212e-05,
      "loss": 1.2435,
      "step": 10815
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.625,
      "learning_rate": 1.733646855656904e-05,
      "loss": 1.2645,
      "step": 10820
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.125,
      "learning_rate": 1.731120952653434e-05,
      "loss": 1.1833,
      "step": 10825
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.53125,
      "learning_rate": 1.7285961446751323e-05,
      "loss": 1.2146,
      "step": 10830
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.5,
      "learning_rate": 1.7260724339008706e-05,
      "loss": 1.246,
      "step": 10835
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.1875,
      "learning_rate": 1.7235498225085774e-05,
      "loss": 1.2344,
      "step": 10840
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.875,
      "learning_rate": 1.7210283126752328e-05,
      "loss": 1.2346,
      "step": 10845
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.40625,
      "learning_rate": 1.7185079065768637e-05,
      "loss": 1.2486,
      "step": 10850
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.125,
      "learning_rate": 1.7159886063885474e-05,
      "loss": 1.2362,
      "step": 10855
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.1875,
      "learning_rate": 1.7134704142844058e-05,
      "loss": 1.211,
      "step": 10860
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.46875,
      "learning_rate": 1.7109533324376017e-05,
      "loss": 1.3006,
      "step": 10865
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.65625,
      "learning_rate": 1.7084373630203427e-05,
      "loss": 1.2022,
      "step": 10870
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.53125,
      "learning_rate": 1.705922508203878e-05,
      "loss": 1.2281,
      "step": 10875
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.0625,
      "learning_rate": 1.7034087701584896e-05,
      "loss": 1.2491,
      "step": 10880
    },
    {
      "epoch": 1.93,
      "grad_norm": 3.953125,
      "learning_rate": 1.7008961510534997e-05,
      "loss": 1.2503,
      "step": 10885
    },
    {
      "epoch": 1.93,
      "grad_norm": 3.546875,
      "learning_rate": 1.6983846530572664e-05,
      "loss": 1.1476,
      "step": 10890
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.6875,
      "learning_rate": 1.6958742783371756e-05,
      "loss": 1.1949,
      "step": 10895
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.90625,
      "learning_rate": 1.6933650290596465e-05,
      "loss": 1.1943,
      "step": 10900
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.6875,
      "learning_rate": 1.6908569073901278e-05,
      "loss": 1.2673,
      "step": 10905
    },
    {
      "epoch": 1.93,
      "grad_norm": 3.8125,
      "learning_rate": 1.6883499154930928e-05,
      "loss": 1.2509,
      "step": 10910
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.0,
      "learning_rate": 1.6858440555320408e-05,
      "loss": 1.2243,
      "step": 10915
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.84375,
      "learning_rate": 1.683339329669495e-05,
      "loss": 1.2138,
      "step": 10920
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.96875,
      "learning_rate": 1.6808357400670002e-05,
      "loss": 1.1986,
      "step": 10925
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.15625,
      "learning_rate": 1.678333288885118e-05,
      "loss": 1.277,
      "step": 10930
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.53125,
      "learning_rate": 1.6758319782834325e-05,
      "loss": 1.2099,
      "step": 10935
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.1875,
      "learning_rate": 1.6733318104205368e-05,
      "loss": 1.2928,
      "step": 10940
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.8125,
      "learning_rate": 1.670832787454043e-05,
      "loss": 1.276,
      "step": 10945
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.9375,
      "learning_rate": 1.6683349115405733e-05,
      "loss": 1.2391,
      "step": 10950
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.5,
      "learning_rate": 1.6658381848357613e-05,
      "loss": 1.2587,
      "step": 10955
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.9375,
      "learning_rate": 1.663342609494246e-05,
      "loss": 1.2599,
      "step": 10960
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.65625,
      "learning_rate": 1.6608481876696742e-05,
      "loss": 1.2571,
      "step": 10965
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.875,
      "learning_rate": 1.6583549215147014e-05,
      "loss": 1.3088,
      "step": 10970
    },
    {
      "epoch": 1.94,
      "grad_norm": 3.75,
      "learning_rate": 1.655862813180978e-05,
      "loss": 1.2296,
      "step": 10975
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.625,
      "learning_rate": 1.6533718648191607e-05,
      "loss": 1.2204,
      "step": 10980
    },
    {
      "epoch": 1.95,
      "grad_norm": 3.96875,
      "learning_rate": 1.6508820785789048e-05,
      "loss": 1.3175,
      "step": 10985
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.25,
      "learning_rate": 1.6483934566088598e-05,
      "loss": 1.2415,
      "step": 10990
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.65625,
      "learning_rate": 1.645906001056673e-05,
      "loss": 1.2501,
      "step": 10995
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.1875,
      "learning_rate": 1.643419714068985e-05,
      "loss": 1.2471,
      "step": 11000
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.3125,
      "learning_rate": 1.640934597791425e-05,
      "loss": 1.2796,
      "step": 11005
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.875,
      "learning_rate": 1.638450654368615e-05,
      "loss": 1.2118,
      "step": 11010
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.28125,
      "learning_rate": 1.6359678859441667e-05,
      "loss": 1.2012,
      "step": 11015
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.25,
      "learning_rate": 1.6334862946606726e-05,
      "loss": 1.2075,
      "step": 11020
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.40625,
      "learning_rate": 1.6310058826597115e-05,
      "loss": 1.2756,
      "step": 11025
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.8125,
      "learning_rate": 1.6285266520818474e-05,
      "loss": 1.2687,
      "step": 11030
    },
    {
      "epoch": 1.96,
      "grad_norm": 3.921875,
      "learning_rate": 1.626048605066619e-05,
      "loss": 1.1698,
      "step": 11035
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.84375,
      "learning_rate": 1.6235717437525485e-05,
      "loss": 1.2954,
      "step": 11040
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.21875,
      "learning_rate": 1.621096070277133e-05,
      "loss": 1.2346,
      "step": 11045
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.28125,
      "learning_rate": 1.6186215867768445e-05,
      "loss": 1.2863,
      "step": 11050
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.375,
      "learning_rate": 1.616148295387128e-05,
      "loss": 1.308,
      "step": 11055
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.5625,
      "learning_rate": 1.6136761982424006e-05,
      "loss": 1.2754,
      "step": 11060
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.4375,
      "learning_rate": 1.611205297476049e-05,
      "loss": 1.2071,
      "step": 11065
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.53125,
      "learning_rate": 1.6087355952204242e-05,
      "loss": 1.2192,
      "step": 11070
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.53125,
      "learning_rate": 1.6062670936068466e-05,
      "loss": 1.2669,
      "step": 11075
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.6875,
      "learning_rate": 1.6037997947656e-05,
      "loss": 1.2244,
      "step": 11080
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.875,
      "learning_rate": 1.6013337008259272e-05,
      "loss": 1.2916,
      "step": 11085
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.65625,
      "learning_rate": 1.5988688139160333e-05,
      "loss": 1.2581,
      "step": 11090
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.875,
      "learning_rate": 1.5964051361630855e-05,
      "loss": 1.279,
      "step": 11095
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.4375,
      "learning_rate": 1.5939426696931997e-05,
      "loss": 1.2235,
      "step": 11100
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.28125,
      "learning_rate": 1.591481416631453e-05,
      "loss": 1.2298,
      "step": 11105
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.3125,
      "learning_rate": 1.589021379101873e-05,
      "loss": 1.2104,
      "step": 11110
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.28125,
      "learning_rate": 1.5865625592274362e-05,
      "loss": 1.2318,
      "step": 11115
    },
    {
      "epoch": 1.97,
      "grad_norm": 3.75,
      "learning_rate": 1.5841049591300717e-05,
      "loss": 1.1771,
      "step": 11120
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.4375,
      "learning_rate": 1.5816485809306543e-05,
      "loss": 1.224,
      "step": 11125
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.40625,
      "learning_rate": 1.579193426749005e-05,
      "loss": 1.1915,
      "step": 11130
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.09375,
      "learning_rate": 1.576739498703888e-05,
      "loss": 1.2214,
      "step": 11135
    },
    {
      "epoch": 1.97,
      "grad_norm": 3.984375,
      "learning_rate": 1.57428679891301e-05,
      "loss": 1.2232,
      "step": 11140
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.34375,
      "learning_rate": 1.5718353294930157e-05,
      "loss": 1.2187,
      "step": 11145
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.34375,
      "learning_rate": 1.56938509255949e-05,
      "loss": 1.2003,
      "step": 11150
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.03125,
      "learning_rate": 1.566936090226954e-05,
      "loss": 1.227,
      "step": 11155
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.78125,
      "learning_rate": 1.5644883246088645e-05,
      "loss": 1.254,
      "step": 11160
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.71875,
      "learning_rate": 1.5620417978176067e-05,
      "loss": 1.2335,
      "step": 11165
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.3125,
      "learning_rate": 1.559596511964501e-05,
      "loss": 1.234,
      "step": 11170
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.0625,
      "learning_rate": 1.557152469159795e-05,
      "loss": 1.289,
      "step": 11175
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.09375,
      "learning_rate": 1.554709671512665e-05,
      "loss": 1.2494,
      "step": 11180
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.28125,
      "learning_rate": 1.552268121131211e-05,
      "loss": 1.2839,
      "step": 11185
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.40625,
      "learning_rate": 1.5498278201224583e-05,
      "loss": 1.2122,
      "step": 11190
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.03125,
      "learning_rate": 1.5473887705923518e-05,
      "loss": 1.2544,
      "step": 11195
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.125,
      "learning_rate": 1.5449509746457583e-05,
      "loss": 1.2481,
      "step": 11200
    },
    {
      "epoch": 1.99,
      "grad_norm": 3.890625,
      "learning_rate": 1.5425144343864632e-05,
      "loss": 1.2549,
      "step": 11205
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.28125,
      "learning_rate": 1.540079151917165e-05,
      "loss": 1.2285,
      "step": 11210
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.125,
      "learning_rate": 1.5376451293394795e-05,
      "loss": 1.1655,
      "step": 11215
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.03125,
      "learning_rate": 1.535212368753937e-05,
      "loss": 1.2322,
      "step": 11220
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.34375,
      "learning_rate": 1.532780872259974e-05,
      "loss": 1.2208,
      "step": 11225
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.15625,
      "learning_rate": 1.5303506419559393e-05,
      "loss": 1.136,
      "step": 11230
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.28125,
      "learning_rate": 1.5279216799390896e-05,
      "loss": 1.2763,
      "step": 11235
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.84375,
      "learning_rate": 1.5254939883055829e-05,
      "loss": 1.2371,
      "step": 11240
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.5625,
      "learning_rate": 1.5230675691504857e-05,
      "loss": 1.2051,
      "step": 11245
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.0625,
      "learning_rate": 1.5206424245677632e-05,
      "loss": 1.2487,
      "step": 11250
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.53125,
      "learning_rate": 1.518218556650283e-05,
      "loss": 1.1531,
      "step": 11255
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.71875,
      "learning_rate": 1.5157959674898093e-05,
      "loss": 1.2918,
      "step": 11260
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.5625,
      "learning_rate": 1.5133746591770042e-05,
      "loss": 1.1509,
      "step": 11265
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.21875,
      "learning_rate": 1.510954633801421e-05,
      "loss": 1.2398,
      "step": 11270
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.15625,
      "learning_rate": 1.5085358934515104e-05,
      "loss": 1.2054,
      "step": 11275
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.59375,
      "learning_rate": 1.5061184402146114e-05,
      "loss": 1.1741,
      "step": 11280
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.96875,
      "learning_rate": 1.5037022761769532e-05,
      "loss": 1.2437,
      "step": 11285
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.9375,
      "learning_rate": 1.5012874034236515e-05,
      "loss": 1.2112,
      "step": 11290
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.75,
      "learning_rate": 1.4988738240387083e-05,
      "loss": 1.2235,
      "step": 11295
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.53125,
      "learning_rate": 1.49646154010501e-05,
      "loss": 1.215,
      "step": 11300
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.15625,
      "learning_rate": 1.4940505537043241e-05,
      "loss": 1.1498,
      "step": 11305
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.78125,
      "learning_rate": 1.4916408669172988e-05,
      "loss": 1.2098,
      "step": 11310
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.71875,
      "learning_rate": 1.4892324818234619e-05,
      "loss": 1.1875,
      "step": 11315
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.625,
      "learning_rate": 1.4868254005012142e-05,
      "loss": 1.18,
      "step": 11320
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.75,
      "learning_rate": 1.484419625027835e-05,
      "loss": 1.1851,
      "step": 11325
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.375,
      "learning_rate": 1.4820151574794762e-05,
      "loss": 1.2454,
      "step": 11330
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.8125,
      "learning_rate": 1.4796119999311588e-05,
      "loss": 1.1825,
      "step": 11335
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.96875,
      "learning_rate": 1.4772101544567736e-05,
      "loss": 1.1918,
      "step": 11340
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.921875,
      "learning_rate": 1.4748096231290848e-05,
      "loss": 1.1976,
      "step": 11345
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.4375,
      "learning_rate": 1.4724104080197138e-05,
      "loss": 1.2413,
      "step": 11350
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.28125,
      "learning_rate": 1.4700125111991513e-05,
      "loss": 1.183,
      "step": 11355
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.15625,
      "learning_rate": 1.4676159347367507e-05,
      "loss": 1.2035,
      "step": 11360
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.09375,
      "learning_rate": 1.4652206807007225e-05,
      "loss": 1.2674,
      "step": 11365
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.75,
      "learning_rate": 1.4628267511581385e-05,
      "loss": 1.2273,
      "step": 11370
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.796875,
      "learning_rate": 1.4604341481749274e-05,
      "loss": 1.2421,
      "step": 11375
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.84375,
      "learning_rate": 1.4580428738158726e-05,
      "loss": 1.2685,
      "step": 11380
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.78125,
      "learning_rate": 1.4556529301446108e-05,
      "loss": 1.223,
      "step": 11385
    },
    {
      "epoch": 2.02,
      "grad_norm": 5.3125,
      "learning_rate": 1.4532643192236318e-05,
      "loss": 1.1947,
      "step": 11390
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.96875,
      "learning_rate": 1.4508770431142716e-05,
      "loss": 1.2627,
      "step": 11395
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.625,
      "learning_rate": 1.4484911038767183e-05,
      "loss": 1.231,
      "step": 11400
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.625,
      "learning_rate": 1.4461065035700039e-05,
      "loss": 1.1734,
      "step": 11405
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.0,
      "learning_rate": 1.4437232442520072e-05,
      "loss": 1.163,
      "step": 11410
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.40625,
      "learning_rate": 1.4413413279794458e-05,
      "loss": 1.2159,
      "step": 11415
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.375,
      "learning_rate": 1.4389607568078809e-05,
      "loss": 1.2378,
      "step": 11420
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.1875,
      "learning_rate": 1.4365815327917157e-05,
      "loss": 1.1857,
      "step": 11425
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.4375,
      "learning_rate": 1.4342036579841848e-05,
      "loss": 1.176,
      "step": 11430
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.53125,
      "learning_rate": 1.4318271344373625e-05,
      "loss": 1.2452,
      "step": 11435
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.5625,
      "learning_rate": 1.429451964202157e-05,
      "loss": 1.2421,
      "step": 11440
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.03125,
      "learning_rate": 1.4270781493283054e-05,
      "loss": 1.247,
      "step": 11445
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.25,
      "learning_rate": 1.4247056918643782e-05,
      "loss": 1.2056,
      "step": 11450
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.15625,
      "learning_rate": 1.4223345938577747e-05,
      "loss": 1.2407,
      "step": 11455
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.625,
      "learning_rate": 1.4199648573547167e-05,
      "loss": 1.3027,
      "step": 11460
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.84375,
      "learning_rate": 1.4175964844002574e-05,
      "loss": 1.1677,
      "step": 11465
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.84375,
      "learning_rate": 1.4152294770382703e-05,
      "loss": 1.2503,
      "step": 11470
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.34375,
      "learning_rate": 1.4128638373114476e-05,
      "loss": 1.1914,
      "step": 11475
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.21875,
      "learning_rate": 1.4104995672613054e-05,
      "loss": 1.2181,
      "step": 11480
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.65625,
      "learning_rate": 1.408136668928176e-05,
      "loss": 1.2696,
      "step": 11485
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.5,
      "learning_rate": 1.4057751443512093e-05,
      "loss": 1.1018,
      "step": 11490
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.875,
      "learning_rate": 1.403414995568366e-05,
      "loss": 1.2709,
      "step": 11495
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.75,
      "learning_rate": 1.401056224616424e-05,
      "loss": 1.2611,
      "step": 11500
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.1875,
      "learning_rate": 1.3986988335309692e-05,
      "loss": 1.3003,
      "step": 11505
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.6875,
      "learning_rate": 1.3963428243463982e-05,
      "loss": 1.1981,
      "step": 11510
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.90625,
      "learning_rate": 1.3939881990959147e-05,
      "loss": 1.192,
      "step": 11515
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.4375,
      "learning_rate": 1.3916349598115287e-05,
      "loss": 1.2897,
      "step": 11520
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.28125,
      "learning_rate": 1.389283108524051e-05,
      "loss": 1.2152,
      "step": 11525
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.0,
      "learning_rate": 1.3869326472630983e-05,
      "loss": 1.2433,
      "step": 11530
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.15625,
      "learning_rate": 1.3845835780570876e-05,
      "loss": 1.2339,
      "step": 11535
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.34375,
      "learning_rate": 1.3822359029332308e-05,
      "loss": 1.2686,
      "step": 11540
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.21875,
      "learning_rate": 1.379889623917539e-05,
      "loss": 1.2561,
      "step": 11545
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.78125,
      "learning_rate": 1.3775447430348227e-05,
      "loss": 1.2147,
      "step": 11550
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.625,
      "learning_rate": 1.375201262308678e-05,
      "loss": 1.2291,
      "step": 11555
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.625,
      "learning_rate": 1.372859183761498e-05,
      "loss": 1.203,
      "step": 11560
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.21875,
      "learning_rate": 1.370518509414465e-05,
      "loss": 1.2442,
      "step": 11565
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.40625,
      "learning_rate": 1.368179241287547e-05,
      "loss": 1.1387,
      "step": 11570
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.34375,
      "learning_rate": 1.3658413813995008e-05,
      "loss": 1.2217,
      "step": 11575
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.8125,
      "learning_rate": 1.3635049317678682e-05,
      "loss": 1.2324,
      "step": 11580
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.90625,
      "learning_rate": 1.3611698944089723e-05,
      "loss": 1.2554,
      "step": 11585
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.125,
      "learning_rate": 1.3588362713379181e-05,
      "loss": 1.2289,
      "step": 11590
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.78125,
      "learning_rate": 1.3565040645685918e-05,
      "loss": 1.1574,
      "step": 11595
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.96875,
      "learning_rate": 1.3541732761136536e-05,
      "loss": 1.2241,
      "step": 11600
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.9375,
      "learning_rate": 1.351843907984543e-05,
      "loss": 1.3095,
      "step": 11605
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.6875,
      "learning_rate": 1.3495159621914723e-05,
      "loss": 1.2143,
      "step": 11610
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.59375,
      "learning_rate": 1.3471894407434278e-05,
      "loss": 1.2177,
      "step": 11615
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.875,
      "learning_rate": 1.3448643456481635e-05,
      "loss": 1.1781,
      "step": 11620
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.0625,
      "learning_rate": 1.3425406789122058e-05,
      "loss": 1.1749,
      "step": 11625
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.65625,
      "learning_rate": 1.3402184425408467e-05,
      "loss": 1.2989,
      "step": 11630
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.9375,
      "learning_rate": 1.3378976385381449e-05,
      "loss": 1.2312,
      "step": 11635
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.09375,
      "learning_rate": 1.3355782689069215e-05,
      "loss": 1.2329,
      "step": 11640
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.59375,
      "learning_rate": 1.3332603356487622e-05,
      "loss": 1.2531,
      "step": 11645
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.28125,
      "learning_rate": 1.3309438407640094e-05,
      "loss": 1.2564,
      "step": 11650
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.78125,
      "learning_rate": 1.3286287862517671e-05,
      "loss": 1.2358,
      "step": 11655
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.375,
      "learning_rate": 1.3263151741098968e-05,
      "loss": 1.1978,
      "step": 11660
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.75,
      "learning_rate": 1.3240030063350117e-05,
      "loss": 1.2038,
      "step": 11665
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.3125,
      "learning_rate": 1.3216922849224808e-05,
      "loss": 1.2005,
      "step": 11670
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.59375,
      "learning_rate": 1.319383011866428e-05,
      "loss": 1.3006,
      "step": 11675
    },
    {
      "epoch": 2.07,
      "grad_norm": 3.90625,
      "learning_rate": 1.3170751891597216e-05,
      "loss": 1.211,
      "step": 11680
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.875,
      "learning_rate": 1.314768818793981e-05,
      "loss": 1.1973,
      "step": 11685
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.0625,
      "learning_rate": 1.3124639027595737e-05,
      "loss": 1.2116,
      "step": 11690
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.1875,
      "learning_rate": 1.3101604430456085e-05,
      "loss": 1.1875,
      "step": 11695
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.125,
      "learning_rate": 1.3078584416399403e-05,
      "loss": 1.2534,
      "step": 11700
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.25,
      "learning_rate": 1.3055579005291644e-05,
      "loss": 1.2312,
      "step": 11705
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.75,
      "learning_rate": 1.3032588216986167e-05,
      "loss": 1.1792,
      "step": 11710
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.15625,
      "learning_rate": 1.3009612071323704e-05,
      "loss": 1.1538,
      "step": 11715
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.71875,
      "learning_rate": 1.2986650588132362e-05,
      "loss": 1.2282,
      "step": 11720
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.15625,
      "learning_rate": 1.2963703787227565e-05,
      "loss": 1.2437,
      "step": 11725
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.5625,
      "learning_rate": 1.2940771688412095e-05,
      "loss": 1.2169,
      "step": 11730
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.75,
      "learning_rate": 1.2917854311476042e-05,
      "loss": 1.2458,
      "step": 11735
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.53125,
      "learning_rate": 1.2894951676196797e-05,
      "loss": 1.1886,
      "step": 11740
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.5625,
      "learning_rate": 1.2872063802338993e-05,
      "loss": 1.2291,
      "step": 11745
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.3125,
      "learning_rate": 1.2849190709654569e-05,
      "loss": 1.0974,
      "step": 11750
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.6875,
      "learning_rate": 1.282633241788268e-05,
      "loss": 1.2064,
      "step": 11755
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.5625,
      "learning_rate": 1.2803488946749727e-05,
      "loss": 1.2217,
      "step": 11760
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.1875,
      "learning_rate": 1.2780660315969305e-05,
      "loss": 1.1908,
      "step": 11765
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.9375,
      "learning_rate": 1.275784654524222e-05,
      "loss": 1.1512,
      "step": 11770
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.09375,
      "learning_rate": 1.2735047654256423e-05,
      "loss": 1.2046,
      "step": 11775
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.59375,
      "learning_rate": 1.2712263662687052e-05,
      "loss": 1.2452,
      "step": 11780
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.78125,
      "learning_rate": 1.2689494590196393e-05,
      "loss": 1.232,
      "step": 11785
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.1875,
      "learning_rate": 1.2666740456433805e-05,
      "loss": 1.1803,
      "step": 11790
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.0625,
      "learning_rate": 1.2644001281035827e-05,
      "loss": 1.254,
      "step": 11795
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.90625,
      "learning_rate": 1.2621277083626053e-05,
      "loss": 1.2314,
      "step": 11800
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.921875,
      "learning_rate": 1.2598567883815138e-05,
      "loss": 1.2098,
      "step": 11805
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.375,
      "learning_rate": 1.2575873701200808e-05,
      "loss": 1.2614,
      "step": 11810
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.375,
      "learning_rate": 1.2553194555367852e-05,
      "loss": 1.2503,
      "step": 11815
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.375,
      "learning_rate": 1.2530530465888038e-05,
      "loss": 1.1943,
      "step": 11820
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.4375,
      "learning_rate": 1.2507881452320168e-05,
      "loss": 1.2694,
      "step": 11825
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.8125,
      "learning_rate": 1.2485247534210033e-05,
      "loss": 1.235,
      "step": 11830
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.34375,
      "learning_rate": 1.2462628731090398e-05,
      "loss": 1.1934,
      "step": 11835
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.875,
      "learning_rate": 1.244002506248097e-05,
      "loss": 1.2025,
      "step": 11840
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.59375,
      "learning_rate": 1.2417436547888406e-05,
      "loss": 1.2334,
      "step": 11845
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.9375,
      "learning_rate": 1.2394863206806299e-05,
      "loss": 1.2341,
      "step": 11850
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.46875,
      "learning_rate": 1.2372305058715108e-05,
      "loss": 1.1738,
      "step": 11855
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.96875,
      "learning_rate": 1.2349762123082216e-05,
      "loss": 1.1949,
      "step": 11860
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.40625,
      "learning_rate": 1.2327234419361873e-05,
      "loss": 1.2212,
      "step": 11865
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.25,
      "learning_rate": 1.2304721966995162e-05,
      "loss": 1.2751,
      "step": 11870
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.6875,
      "learning_rate": 1.228222478541002e-05,
      "loss": 1.2377,
      "step": 11875
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.21875,
      "learning_rate": 1.2259742894021228e-05,
      "loss": 1.242,
      "step": 11880
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.34375,
      "learning_rate": 1.2237276312230327e-05,
      "loss": 1.1998,
      "step": 11885
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.59375,
      "learning_rate": 1.2214825059425674e-05,
      "loss": 1.1997,
      "step": 11890
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.46875,
      "learning_rate": 1.2192389154982405e-05,
      "loss": 1.2709,
      "step": 11895
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.25,
      "learning_rate": 1.2169968618262373e-05,
      "loss": 1.2318,
      "step": 11900
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.25,
      "learning_rate": 1.2147563468614206e-05,
      "loss": 1.1543,
      "step": 11905
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.96875,
      "learning_rate": 1.2125173725373247e-05,
      "loss": 1.22,
      "step": 11910
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.15625,
      "learning_rate": 1.2102799407861507e-05,
      "loss": 1.2233,
      "step": 11915
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.5,
      "learning_rate": 1.2080440535387745e-05,
      "loss": 1.2098,
      "step": 11920
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.34375,
      "learning_rate": 1.2058097127247363e-05,
      "loss": 1.2235,
      "step": 11925
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.0625,
      "learning_rate": 1.203576920272239e-05,
      "loss": 1.171,
      "step": 11930
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.65625,
      "learning_rate": 1.2013456781081533e-05,
      "loss": 1.3391,
      "step": 11935
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.46875,
      "learning_rate": 1.19911598815801e-05,
      "loss": 1.2118,
      "step": 11940
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.6875,
      "learning_rate": 1.1968878523460024e-05,
      "loss": 1.2064,
      "step": 11945
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.859375,
      "learning_rate": 1.1946612725949785e-05,
      "loss": 1.1682,
      "step": 11950
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.4375,
      "learning_rate": 1.1924362508264476e-05,
      "loss": 1.2179,
      "step": 11955
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.125,
      "learning_rate": 1.190212788960572e-05,
      "loss": 1.2331,
      "step": 11960
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.46875,
      "learning_rate": 1.1879908889161698e-05,
      "loss": 1.2432,
      "step": 11965
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.75,
      "learning_rate": 1.1857705526107094e-05,
      "loss": 1.2453,
      "step": 11970
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.03125,
      "learning_rate": 1.1835517819603119e-05,
      "loss": 1.1749,
      "step": 11975
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.375,
      "learning_rate": 1.1813345788797434e-05,
      "loss": 1.2463,
      "step": 11980
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.0625,
      "learning_rate": 1.1791189452824213e-05,
      "loss": 1.2137,
      "step": 11985
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.375,
      "learning_rate": 1.1769048830804074e-05,
      "loss": 1.2506,
      "step": 11990
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.15625,
      "learning_rate": 1.174692394184404e-05,
      "loss": 1.2114,
      "step": 11995
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.40625,
      "learning_rate": 1.172481480503762e-05,
      "loss": 1.2779,
      "step": 12000
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.0625,
      "learning_rate": 1.170272143946469e-05,
      "loss": 1.3237,
      "step": 12005
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.0,
      "learning_rate": 1.16806438641915e-05,
      "loss": 1.2011,
      "step": 12010
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.4375,
      "learning_rate": 1.165858209827071e-05,
      "loss": 1.2506,
      "step": 12015
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.875,
      "learning_rate": 1.1636536160741325e-05,
      "loss": 1.2015,
      "step": 12020
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.96875,
      "learning_rate": 1.161450607062867e-05,
      "loss": 1.2242,
      "step": 12025
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.875,
      "learning_rate": 1.1592491846944419e-05,
      "loss": 1.1798,
      "step": 12030
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.21875,
      "learning_rate": 1.1570493508686539e-05,
      "loss": 1.1721,
      "step": 12035
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.25,
      "learning_rate": 1.15485110748393e-05,
      "loss": 1.2821,
      "step": 12040
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.125,
      "learning_rate": 1.1526544564373234e-05,
      "loss": 1.2259,
      "step": 12045
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.0,
      "learning_rate": 1.1504593996245156e-05,
      "loss": 1.1761,
      "step": 12050
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.9375,
      "learning_rate": 1.1482659389398079e-05,
      "loss": 1.2192,
      "step": 12055
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.90625,
      "learning_rate": 1.1460740762761275e-05,
      "loss": 1.2518,
      "step": 12060
    },
    {
      "epoch": 2.14,
      "grad_norm": 5.40625,
      "learning_rate": 1.143883813525022e-05,
      "loss": 1.2474,
      "step": 12065
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.21875,
      "learning_rate": 1.1416951525766588e-05,
      "loss": 1.1971,
      "step": 12070
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.59375,
      "learning_rate": 1.1395080953198206e-05,
      "loss": 1.2473,
      "step": 12075
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.90625,
      "learning_rate": 1.1373226436419085e-05,
      "loss": 1.2219,
      "step": 12080
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.28125,
      "learning_rate": 1.1351387994289373e-05,
      "loss": 1.2296,
      "step": 12085
    },
    {
      "epoch": 2.14,
      "grad_norm": 5.09375,
      "learning_rate": 1.1329565645655343e-05,
      "loss": 1.2139,
      "step": 12090
    },
    {
      "epoch": 2.14,
      "grad_norm": 5.8125,
      "learning_rate": 1.1307759409349385e-05,
      "loss": 1.2684,
      "step": 12095
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.53125,
      "learning_rate": 1.1285969304189989e-05,
      "loss": 1.2209,
      "step": 12100
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.5,
      "learning_rate": 1.12641953489817e-05,
      "loss": 1.2325,
      "step": 12105
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.3125,
      "learning_rate": 1.1242437562515145e-05,
      "loss": 1.1674,
      "step": 12110
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.0,
      "learning_rate": 1.122069596356701e-05,
      "loss": 1.2634,
      "step": 12115
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.4375,
      "learning_rate": 1.1198970570899963e-05,
      "loss": 1.2642,
      "step": 12120
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.4375,
      "learning_rate": 1.1177261403262751e-05,
      "loss": 1.2223,
      "step": 12125
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.3125,
      "learning_rate": 1.1155568479390083e-05,
      "loss": 1.2342,
      "step": 12130
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.3125,
      "learning_rate": 1.113389181800264e-05,
      "loss": 1.258,
      "step": 12135
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.125,
      "learning_rate": 1.1112231437807092e-05,
      "loss": 1.2185,
      "step": 12140
    },
    {
      "epoch": 2.15,
      "grad_norm": 3.984375,
      "learning_rate": 1.1090587357496052e-05,
      "loss": 1.1993,
      "step": 12145
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.71875,
      "learning_rate": 1.106895959574805e-05,
      "loss": 1.2767,
      "step": 12150
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.15625,
      "learning_rate": 1.1047348171227558e-05,
      "loss": 1.229,
      "step": 12155
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.875,
      "learning_rate": 1.1025753102584933e-05,
      "loss": 1.15,
      "step": 12160
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.5,
      "learning_rate": 1.1004174408456427e-05,
      "loss": 1.1801,
      "step": 12165
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.21875,
      "learning_rate": 1.098261210746415e-05,
      "loss": 1.1574,
      "step": 12170
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.8125,
      "learning_rate": 1.0961066218216093e-05,
      "loss": 1.181,
      "step": 12175
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.28125,
      "learning_rate": 1.0939536759306027e-05,
      "loss": 1.1649,
      "step": 12180
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.25,
      "learning_rate": 1.09180237493136e-05,
      "loss": 1.2138,
      "step": 12185
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.46875,
      "learning_rate": 1.089652720680424e-05,
      "loss": 1.2191,
      "step": 12190
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.9375,
      "learning_rate": 1.087504715032918e-05,
      "loss": 1.1963,
      "step": 12195
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.09375,
      "learning_rate": 1.0853583598425386e-05,
      "loss": 1.2847,
      "step": 12200
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.15625,
      "learning_rate": 1.0832136569615616e-05,
      "loss": 1.2158,
      "step": 12205
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.3125,
      "learning_rate": 1.0810706082408387e-05,
      "loss": 1.1894,
      "step": 12210
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.09375,
      "learning_rate": 1.0789292155297889e-05,
      "loss": 1.1982,
      "step": 12215
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.375,
      "learning_rate": 1.0767894806764048e-05,
      "loss": 1.2524,
      "step": 12220
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.28125,
      "learning_rate": 1.0746514055272501e-05,
      "loss": 1.1822,
      "step": 12225
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.4375,
      "learning_rate": 1.0725149919274516e-05,
      "loss": 1.199,
      "step": 12230
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.375,
      "learning_rate": 1.070380241720706e-05,
      "loss": 1.2155,
      "step": 12235
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.25,
      "learning_rate": 1.0682471567492748e-05,
      "loss": 1.2446,
      "step": 12240
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.6875,
      "learning_rate": 1.0661157388539776e-05,
      "loss": 1.1959,
      "step": 12245
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.34375,
      "learning_rate": 1.0639859898742016e-05,
      "loss": 1.1996,
      "step": 12250
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.5,
      "learning_rate": 1.0618579116478912e-05,
      "loss": 1.2542,
      "step": 12255
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.09375,
      "learning_rate": 1.0597315060115462e-05,
      "loss": 1.2137,
      "step": 12260
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.96875,
      "learning_rate": 1.057606774800227e-05,
      "loss": 1.3289,
      "step": 12265
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.34375,
      "learning_rate": 1.0554837198475482e-05,
      "loss": 1.1656,
      "step": 12270
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.28125,
      "learning_rate": 1.0533623429856747e-05,
      "loss": 1.2656,
      "step": 12275
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.3125,
      "learning_rate": 1.0512426460453269e-05,
      "loss": 1.1964,
      "step": 12280
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.40625,
      "learning_rate": 1.0491246308557744e-05,
      "loss": 1.2557,
      "step": 12285
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.21875,
      "learning_rate": 1.0470082992448344e-05,
      "loss": 1.2265,
      "step": 12290
    },
    {
      "epoch": 2.18,
      "grad_norm": 5.15625,
      "learning_rate": 1.0448936530388726e-05,
      "loss": 1.2316,
      "step": 12295
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.96875,
      "learning_rate": 1.0427806940627992e-05,
      "loss": 1.21,
      "step": 12300
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.9375,
      "learning_rate": 1.0406694241400697e-05,
      "loss": 1.2146,
      "step": 12305
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.75,
      "learning_rate": 1.0385598450926796e-05,
      "loss": 1.2386,
      "step": 12310
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.25,
      "learning_rate": 1.0364519587411671e-05,
      "loss": 1.2426,
      "step": 12315
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.21875,
      "learning_rate": 1.0343457669046101e-05,
      "loss": 1.2029,
      "step": 12320
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.5,
      "learning_rate": 1.0322412714006203e-05,
      "loss": 1.207,
      "step": 12325
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.375,
      "learning_rate": 1.0301384740453518e-05,
      "loss": 1.1426,
      "step": 12330
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.40625,
      "learning_rate": 1.028037376653489e-05,
      "loss": 1.1714,
      "step": 12335
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.28125,
      "learning_rate": 1.0259379810382487e-05,
      "loss": 1.169,
      "step": 12340
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.09375,
      "learning_rate": 1.0238402890113813e-05,
      "loss": 1.3475,
      "step": 12345
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.46875,
      "learning_rate": 1.0217443023831667e-05,
      "loss": 1.1872,
      "step": 12350
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.96875,
      "learning_rate": 1.0196500229624113e-05,
      "loss": 1.2164,
      "step": 12355
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.125,
      "learning_rate": 1.0175574525564498e-05,
      "loss": 1.2419,
      "step": 12360
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.625,
      "learning_rate": 1.0154665929711419e-05,
      "loss": 1.2395,
      "step": 12365
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.0625,
      "learning_rate": 1.013377446010871e-05,
      "loss": 1.1668,
      "step": 12370
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.96875,
      "learning_rate": 1.0112900134785415e-05,
      "loss": 1.207,
      "step": 12375
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.4375,
      "learning_rate": 1.0092042971755805e-05,
      "loss": 1.3247,
      "step": 12380
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.46875,
      "learning_rate": 1.0071202989019303e-05,
      "loss": 1.2503,
      "step": 12385
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.890625,
      "learning_rate": 1.005038020456054e-05,
      "loss": 1.1916,
      "step": 12390
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.84375,
      "learning_rate": 1.0029574636349289e-05,
      "loss": 1.1929,
      "step": 12395
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.34375,
      "learning_rate": 1.0008786302340479e-05,
      "loss": 1.2172,
      "step": 12400
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.59375,
      "learning_rate": 9.98801522047414e-06,
      "loss": 1.2727,
      "step": 12405
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.0625,
      "learning_rate": 9.967261408675439e-06,
      "loss": 1.1781,
      "step": 12410
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.5625,
      "learning_rate": 9.946524884854628e-06,
      "loss": 1.2183,
      "step": 12415
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.71875,
      "learning_rate": 9.925805666907042e-06,
      "loss": 1.2361,
      "step": 12420
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.53125,
      "learning_rate": 9.905103772713085e-06,
      "loss": 1.1819,
      "step": 12425
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.15625,
      "learning_rate": 9.884419220138212e-06,
      "loss": 1.2409,
      "step": 12430
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.03125,
      "learning_rate": 9.86375202703289e-06,
      "loss": 1.2427,
      "step": 12435
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.90625,
      "learning_rate": 9.84310221123263e-06,
      "loss": 1.2343,
      "step": 12440
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.984375,
      "learning_rate": 9.822469790557954e-06,
      "loss": 1.205,
      "step": 12445
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.15625,
      "learning_rate": 9.801854782814324e-06,
      "loss": 1.2038,
      "step": 12450
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.0625,
      "learning_rate": 9.781257205792236e-06,
      "loss": 1.2113,
      "step": 12455
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.21875,
      "learning_rate": 9.760677077267121e-06,
      "loss": 1.1965,
      "step": 12460
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.96875,
      "learning_rate": 9.740114414999322e-06,
      "loss": 1.2775,
      "step": 12465
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.3125,
      "learning_rate": 9.719569236734145e-06,
      "loss": 1.2827,
      "step": 12470
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.46875,
      "learning_rate": 9.699041560201808e-06,
      "loss": 1.2849,
      "step": 12475
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.84375,
      "learning_rate": 9.678531403117393e-06,
      "loss": 1.2862,
      "step": 12480
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.3125,
      "learning_rate": 9.658038783180895e-06,
      "loss": 1.1669,
      "step": 12485
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.15625,
      "learning_rate": 9.63756371807716e-06,
      "loss": 1.196,
      "step": 12490
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.59375,
      "learning_rate": 9.617106225475886e-06,
      "loss": 1.1788,
      "step": 12495
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.28125,
      "learning_rate": 9.596666323031613e-06,
      "loss": 1.2413,
      "step": 12500
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.53125,
      "learning_rate": 9.576244028383702e-06,
      "loss": 1.2391,
      "step": 12505
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.59375,
      "learning_rate": 9.555839359156297e-06,
      "loss": 1.2009,
      "step": 12510
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.125,
      "learning_rate": 9.535452332958349e-06,
      "loss": 1.2493,
      "step": 12515
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.5,
      "learning_rate": 9.515082967383589e-06,
      "loss": 1.2127,
      "step": 12520
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.0,
      "learning_rate": 9.494731280010504e-06,
      "loss": 1.2991,
      "step": 12525
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.34375,
      "learning_rate": 9.474397288402301e-06,
      "loss": 1.2015,
      "step": 12530
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.1875,
      "learning_rate": 9.45408101010695e-06,
      "loss": 1.2464,
      "step": 12535
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.15625,
      "learning_rate": 9.433782462657117e-06,
      "loss": 1.2177,
      "step": 12540
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.03125,
      "learning_rate": 9.413501663570168e-06,
      "loss": 1.2475,
      "step": 12545
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.03125,
      "learning_rate": 9.39323863034815e-06,
      "loss": 1.1492,
      "step": 12550
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.5,
      "learning_rate": 9.3729933804778e-06,
      "loss": 1.2035,
      "step": 12555
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.34375,
      "learning_rate": 9.352765931430465e-06,
      "loss": 1.2241,
      "step": 12560
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.5625,
      "learning_rate": 9.332556300662167e-06,
      "loss": 1.3138,
      "step": 12565
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.25,
      "learning_rate": 9.31236450561355e-06,
      "loss": 1.1703,
      "step": 12570
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.3125,
      "learning_rate": 9.292190563709825e-06,
      "loss": 1.1405,
      "step": 12575
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.28125,
      "learning_rate": 9.272034492360855e-06,
      "loss": 1.2765,
      "step": 12580
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.46875,
      "learning_rate": 9.251896308961055e-06,
      "loss": 1.2769,
      "step": 12585
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.25,
      "learning_rate": 9.231776030889379e-06,
      "loss": 1.2174,
      "step": 12590
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.5,
      "learning_rate": 9.21167367550936e-06,
      "loss": 1.2512,
      "step": 12595
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.125,
      "learning_rate": 9.19158926016906e-06,
      "loss": 1.301,
      "step": 12600
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.03125,
      "learning_rate": 9.17152280220104e-06,
      "loss": 1.277,
      "step": 12605
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.9375,
      "learning_rate": 9.151474318922383e-06,
      "loss": 1.1845,
      "step": 12610
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.40625,
      "learning_rate": 9.13144382763466e-06,
      "loss": 1.1766,
      "step": 12615
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.90625,
      "learning_rate": 9.111431345623902e-06,
      "loss": 1.1864,
      "step": 12620
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.5625,
      "learning_rate": 9.09143689016061e-06,
      "loss": 1.2049,
      "step": 12625
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.4375,
      "learning_rate": 9.071460478499725e-06,
      "loss": 1.1791,
      "step": 12630
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.78125,
      "learning_rate": 9.051502127880625e-06,
      "loss": 1.1526,
      "step": 12635
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.59375,
      "learning_rate": 9.031561855527065e-06,
      "loss": 1.2475,
      "step": 12640
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.03125,
      "learning_rate": 9.011639678647246e-06,
      "loss": 1.166,
      "step": 12645
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.78125,
      "learning_rate": 8.991735614433739e-06,
      "loss": 1.259,
      "step": 12650
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.5625,
      "learning_rate": 8.971849680063446e-06,
      "loss": 1.2286,
      "step": 12655
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.46875,
      "learning_rate": 8.951981892697686e-06,
      "loss": 1.1004,
      "step": 12660
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.40625,
      "learning_rate": 8.932132269482087e-06,
      "loss": 1.2629,
      "step": 12665
    },
    {
      "epoch": 2.25,
      "grad_norm": 5.46875,
      "learning_rate": 8.912300827546578e-06,
      "loss": 1.237,
      "step": 12670
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.03125,
      "learning_rate": 8.892487584005434e-06,
      "loss": 1.2382,
      "step": 12675
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.40625,
      "learning_rate": 8.872692555957218e-06,
      "loss": 1.2339,
      "step": 12680
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.5625,
      "learning_rate": 8.852915760484751e-06,
      "loss": 1.2276,
      "step": 12685
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.6875,
      "learning_rate": 8.833157214655145e-06,
      "loss": 1.2687,
      "step": 12690
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.5625,
      "learning_rate": 8.813416935519762e-06,
      "loss": 1.2516,
      "step": 12695
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.6875,
      "learning_rate": 8.793694940114159e-06,
      "loss": 1.2138,
      "step": 12700
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.484375,
      "learning_rate": 8.773991245458181e-06,
      "loss": 1.251,
      "step": 12705
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.40625,
      "learning_rate": 8.754305868555845e-06,
      "loss": 1.2466,
      "step": 12710
    },
    {
      "epoch": 2.25,
      "grad_norm": 5.21875,
      "learning_rate": 8.734638826395337e-06,
      "loss": 1.2669,
      "step": 12715
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.625,
      "learning_rate": 8.71499013594906e-06,
      "loss": 1.2774,
      "step": 12720
    },
    {
      "epoch": 2.25,
      "grad_norm": 5.65625,
      "learning_rate": 8.695359814173555e-06,
      "loss": 1.3163,
      "step": 12725
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.5,
      "learning_rate": 8.67574787800954e-06,
      "loss": 1.157,
      "step": 12730
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.53125,
      "learning_rate": 8.656154344381815e-06,
      "loss": 1.3017,
      "step": 12735
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.1875,
      "learning_rate": 8.636579230199349e-06,
      "loss": 1.2112,
      "step": 12740
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.1875,
      "learning_rate": 8.617022552355193e-06,
      "loss": 1.2632,
      "step": 12745
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.3125,
      "learning_rate": 8.59748432772649e-06,
      "loss": 1.2442,
      "step": 12750
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.59375,
      "learning_rate": 8.57796457317446e-06,
      "loss": 1.2142,
      "step": 12755
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.21875,
      "learning_rate": 8.558463305544396e-06,
      "loss": 1.2106,
      "step": 12760
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.75,
      "learning_rate": 8.538980541665599e-06,
      "loss": 1.2024,
      "step": 12765
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.90625,
      "learning_rate": 8.519516298351436e-06,
      "loss": 1.2204,
      "step": 12770
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.84375,
      "learning_rate": 8.500070592399297e-06,
      "loss": 1.2423,
      "step": 12775
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.3125,
      "learning_rate": 8.480643440590524e-06,
      "loss": 1.1918,
      "step": 12780
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.8125,
      "learning_rate": 8.461234859690514e-06,
      "loss": 1.2232,
      "step": 12785
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.78125,
      "learning_rate": 8.4418448664486e-06,
      "loss": 1.2385,
      "step": 12790
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.875,
      "learning_rate": 8.422473477598068e-06,
      "loss": 1.2354,
      "step": 12795
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.125,
      "learning_rate": 8.403120709856165e-06,
      "loss": 1.1435,
      "step": 12800
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.71875,
      "learning_rate": 8.383786579924075e-06,
      "loss": 1.2146,
      "step": 12805
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.25,
      "learning_rate": 8.364471104486862e-06,
      "loss": 1.2101,
      "step": 12810
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.96875,
      "learning_rate": 8.345174300213533e-06,
      "loss": 1.2591,
      "step": 12815
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.8125,
      "learning_rate": 8.32589618375696e-06,
      "loss": 1.2192,
      "step": 12820
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.28125,
      "learning_rate": 8.306636771753893e-06,
      "loss": 1.2827,
      "step": 12825
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.5,
      "learning_rate": 8.287396080824941e-06,
      "loss": 1.1651,
      "step": 12830
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.25,
      "learning_rate": 8.268174127574562e-06,
      "loss": 1.1642,
      "step": 12835
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.03125,
      "learning_rate": 8.248970928591024e-06,
      "loss": 1.2455,
      "step": 12840
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.75,
      "learning_rate": 8.229786500446427e-06,
      "loss": 1.1773,
      "step": 12845
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.96875,
      "learning_rate": 8.210620859696673e-06,
      "loss": 1.1888,
      "step": 12850
    },
    {
      "epoch": 2.28,
      "grad_norm": 5.5625,
      "learning_rate": 8.191474022881452e-06,
      "loss": 1.2173,
      "step": 12855
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.5,
      "learning_rate": 8.172346006524206e-06,
      "loss": 1.2199,
      "step": 12860
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.9375,
      "learning_rate": 8.153236827132157e-06,
      "loss": 1.2301,
      "step": 12865
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.4375,
      "learning_rate": 8.13414650119626e-06,
      "loss": 1.186,
      "step": 12870
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.21875,
      "learning_rate": 8.115075045191213e-06,
      "loss": 1.2115,
      "step": 12875
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.84375,
      "learning_rate": 8.096022475575415e-06,
      "loss": 1.3112,
      "step": 12880
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.09375,
      "learning_rate": 8.076988808790983e-06,
      "loss": 1.2512,
      "step": 12885
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.15625,
      "learning_rate": 8.057974061263686e-06,
      "loss": 1.2413,
      "step": 12890
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.0,
      "learning_rate": 8.038978249403004e-06,
      "loss": 1.1906,
      "step": 12895
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.96875,
      "learning_rate": 8.02000138960207e-06,
      "loss": 1.1623,
      "step": 12900
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.625,
      "learning_rate": 8.001043498237623e-06,
      "loss": 1.1897,
      "step": 12905
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.84375,
      "learning_rate": 7.982104591670092e-06,
      "loss": 1.2242,
      "step": 12910
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.25,
      "learning_rate": 7.963184686243491e-06,
      "loss": 1.122,
      "step": 12915
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.0625,
      "learning_rate": 7.94428379828542e-06,
      "loss": 1.2384,
      "step": 12920
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.8125,
      "learning_rate": 7.925401944107099e-06,
      "loss": 1.222,
      "step": 12925
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.46875,
      "learning_rate": 7.906539140003314e-06,
      "loss": 1.1871,
      "step": 12930
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.890625,
      "learning_rate": 7.887695402252386e-06,
      "loss": 1.2027,
      "step": 12935
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.375,
      "learning_rate": 7.86887074711621e-06,
      "loss": 1.1551,
      "step": 12940
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.8125,
      "learning_rate": 7.850065190840212e-06,
      "loss": 1.1454,
      "step": 12945
    },
    {
      "epoch": 2.29,
      "grad_norm": 5.03125,
      "learning_rate": 7.831278749653317e-06,
      "loss": 1.3206,
      "step": 12950
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.25,
      "learning_rate": 7.812511439767975e-06,
      "loss": 1.239,
      "step": 12955
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.53125,
      "learning_rate": 7.79376327738012e-06,
      "loss": 1.2231,
      "step": 12960
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.34375,
      "learning_rate": 7.775034278669146e-06,
      "loss": 1.2539,
      "step": 12965
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.5625,
      "learning_rate": 7.756324459797925e-06,
      "loss": 1.1297,
      "step": 12970
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.5,
      "learning_rate": 7.73763383691277e-06,
      "loss": 1.2602,
      "step": 12975
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.25,
      "learning_rate": 7.71896242614345e-06,
      "loss": 1.2281,
      "step": 12980
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.09375,
      "learning_rate": 7.700310243603103e-06,
      "loss": 1.2094,
      "step": 12985
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.40625,
      "learning_rate": 7.681677305388332e-06,
      "loss": 1.2602,
      "step": 12990
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.46875,
      "learning_rate": 7.663063627579111e-06,
      "loss": 1.2418,
      "step": 12995
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.3125,
      "learning_rate": 7.644469226238764e-06,
      "loss": 1.2019,
      "step": 13000
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.53125,
      "learning_rate": 7.62589411741402e-06,
      "loss": 1.3093,
      "step": 13005
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.625,
      "learning_rate": 7.607338317134945e-06,
      "loss": 1.2237,
      "step": 13010
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.15625,
      "learning_rate": 7.588801841414926e-06,
      "loss": 1.2198,
      "step": 13015
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.78125,
      "learning_rate": 7.570284706250691e-06,
      "loss": 1.2645,
      "step": 13020
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.375,
      "learning_rate": 7.551786927622283e-06,
      "loss": 1.2441,
      "step": 13025
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.6875,
      "learning_rate": 7.53330852149301e-06,
      "loss": 1.1804,
      "step": 13030
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.3125,
      "learning_rate": 7.5148495038094985e-06,
      "loss": 1.2938,
      "step": 13035
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.4375,
      "learning_rate": 7.4964098905016375e-06,
      "loss": 1.2402,
      "step": 13040
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.3125,
      "learning_rate": 7.477989697482533e-06,
      "loss": 1.2788,
      "step": 13045
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.46875,
      "learning_rate": 7.459588940648574e-06,
      "loss": 1.3012,
      "step": 13050
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.40625,
      "learning_rate": 7.441207635879367e-06,
      "loss": 1.2432,
      "step": 13055
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.34375,
      "learning_rate": 7.422845799037707e-06,
      "loss": 1.1939,
      "step": 13060
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.375,
      "learning_rate": 7.4045034459696135e-06,
      "loss": 1.2259,
      "step": 13065
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.40625,
      "learning_rate": 7.386180592504287e-06,
      "loss": 1.1989,
      "step": 13070
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.28125,
      "learning_rate": 7.367877254454097e-06,
      "loss": 1.1798,
      "step": 13075
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.28125,
      "learning_rate": 7.349593447614576e-06,
      "loss": 1.2313,
      "step": 13080
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.15625,
      "learning_rate": 7.331329187764396e-06,
      "loss": 1.232,
      "step": 13085
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.53125,
      "learning_rate": 7.313084490665368e-06,
      "loss": 1.2528,
      "step": 13090
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.375,
      "learning_rate": 7.294859372062402e-06,
      "loss": 1.2367,
      "step": 13095
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.1875,
      "learning_rate": 7.276653847683528e-06,
      "loss": 1.1921,
      "step": 13100
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.875,
      "learning_rate": 7.258467933239876e-06,
      "loss": 1.2442,
      "step": 13105
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.71875,
      "learning_rate": 7.240301644425613e-06,
      "loss": 1.1734,
      "step": 13110
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.75,
      "learning_rate": 7.222154996918021e-06,
      "loss": 1.2215,
      "step": 13115
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.40625,
      "learning_rate": 7.20402800637741e-06,
      "loss": 1.2926,
      "step": 13120
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.0625,
      "learning_rate": 7.185920688447106e-06,
      "loss": 1.1718,
      "step": 13125
    },
    {
      "epoch": 2.33,
      "grad_norm": 5.0,
      "learning_rate": 7.16783305875348e-06,
      "loss": 1.2646,
      "step": 13130
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.3125,
      "learning_rate": 7.1497651329059185e-06,
      "loss": 1.2017,
      "step": 13135
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.53125,
      "learning_rate": 7.131716926496776e-06,
      "loss": 1.2526,
      "step": 13140
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.40625,
      "learning_rate": 7.113688455101417e-06,
      "loss": 1.2169,
      "step": 13145
    },
    {
      "epoch": 2.33,
      "grad_norm": 5.1875,
      "learning_rate": 7.095679734278162e-06,
      "loss": 1.2157,
      "step": 13150
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.40625,
      "learning_rate": 7.07769077956829e-06,
      "loss": 1.2365,
      "step": 13155
    },
    {
      "epoch": 2.33,
      "grad_norm": 5.03125,
      "learning_rate": 7.059721606496019e-06,
      "loss": 1.274,
      "step": 13160
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.21875,
      "learning_rate": 7.041772230568512e-06,
      "loss": 1.2208,
      "step": 13165
    },
    {
      "epoch": 2.33,
      "grad_norm": 6.0625,
      "learning_rate": 7.02384266727581e-06,
      "loss": 1.1886,
      "step": 13170
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.40625,
      "learning_rate": 7.005932932090895e-06,
      "loss": 1.1618,
      "step": 13175
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.03125,
      "learning_rate": 6.988043040469618e-06,
      "loss": 1.211,
      "step": 13180
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.8125,
      "learning_rate": 6.970173007850721e-06,
      "loss": 1.2251,
      "step": 13185
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.125,
      "learning_rate": 6.952322849655777e-06,
      "loss": 1.2558,
      "step": 13190
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.65625,
      "learning_rate": 6.934492581289233e-06,
      "loss": 1.1988,
      "step": 13195
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.4375,
      "learning_rate": 6.9166822181383695e-06,
      "loss": 1.2061,
      "step": 13200
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.59375,
      "learning_rate": 6.8988917755732825e-06,
      "loss": 1.2491,
      "step": 13205
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.3125,
      "learning_rate": 6.881121268946879e-06,
      "loss": 1.2104,
      "step": 13210
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.4375,
      "learning_rate": 6.863370713594876e-06,
      "loss": 1.2402,
      "step": 13215
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.65625,
      "learning_rate": 6.845640124835734e-06,
      "loss": 1.2133,
      "step": 13220
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.25,
      "learning_rate": 6.8279295179707215e-06,
      "loss": 1.2277,
      "step": 13225
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.625,
      "learning_rate": 6.810238908283852e-06,
      "loss": 1.2826,
      "step": 13230
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.25,
      "learning_rate": 6.792568311041858e-06,
      "loss": 1.2167,
      "step": 13235
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.09375,
      "learning_rate": 6.774917741494243e-06,
      "loss": 1.2804,
      "step": 13240
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.28125,
      "learning_rate": 6.757287214873207e-06,
      "loss": 1.1219,
      "step": 13245
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.75,
      "learning_rate": 6.739676746393638e-06,
      "loss": 1.2941,
      "step": 13250
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.40625,
      "learning_rate": 6.722086351253134e-06,
      "loss": 1.1911,
      "step": 13255
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.6875,
      "learning_rate": 6.7045160446319705e-06,
      "loss": 1.2473,
      "step": 13260
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.9375,
      "learning_rate": 6.68696584169307e-06,
      "loss": 1.2949,
      "step": 13265
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.78125,
      "learning_rate": 6.669435757582017e-06,
      "loss": 1.2424,
      "step": 13270
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.875,
      "learning_rate": 6.651925807427038e-06,
      "loss": 1.1743,
      "step": 13275
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.5625,
      "learning_rate": 6.634436006338978e-06,
      "loss": 1.2165,
      "step": 13280
    },
    {
      "epoch": 2.35,
      "grad_norm": 5.125,
      "learning_rate": 6.616966369411294e-06,
      "loss": 1.276,
      "step": 13285
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.65625,
      "learning_rate": 6.59951691172005e-06,
      "loss": 1.242,
      "step": 13290
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.40625,
      "learning_rate": 6.582087648323873e-06,
      "loss": 1.2212,
      "step": 13295
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.59375,
      "learning_rate": 6.564678594263987e-06,
      "loss": 1.1428,
      "step": 13300
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.5,
      "learning_rate": 6.547289764564165e-06,
      "loss": 1.1883,
      "step": 13305
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.46875,
      "learning_rate": 6.529921174230735e-06,
      "loss": 1.2039,
      "step": 13310
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.25,
      "learning_rate": 6.5125728382525285e-06,
      "loss": 1.2108,
      "step": 13315
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.46875,
      "learning_rate": 6.4952447716009554e-06,
      "loss": 1.1812,
      "step": 13320
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.53125,
      "learning_rate": 6.4779369892298675e-06,
      "loss": 1.2652,
      "step": 13325
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.46875,
      "learning_rate": 6.460649506075657e-06,
      "loss": 1.2179,
      "step": 13330
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.5,
      "learning_rate": 6.443382337057185e-06,
      "loss": 1.2241,
      "step": 13335
    },
    {
      "epoch": 2.36,
      "grad_norm": 5.34375,
      "learning_rate": 6.4261354970757846e-06,
      "loss": 1.23,
      "step": 13340
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.921875,
      "learning_rate": 6.408909001015226e-06,
      "loss": 1.2766,
      "step": 13345
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.65625,
      "learning_rate": 6.391702863741751e-06,
      "loss": 1.2437,
      "step": 13350
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.21875,
      "learning_rate": 6.374517100104024e-06,
      "loss": 1.2141,
      "step": 13355
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.9375,
      "learning_rate": 6.357351724933095e-06,
      "loss": 1.3076,
      "step": 13360
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.34375,
      "learning_rate": 6.340206753042478e-06,
      "loss": 1.2218,
      "step": 13365
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.84375,
      "learning_rate": 6.323082199228044e-06,
      "loss": 1.1901,
      "step": 13370
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.734375,
      "learning_rate": 6.305978078268032e-06,
      "loss": 1.2251,
      "step": 13375
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.15625,
      "learning_rate": 6.288894404923072e-06,
      "loss": 1.2667,
      "step": 13380
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.125,
      "learning_rate": 6.271831193936145e-06,
      "loss": 1.226,
      "step": 13385
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.34375,
      "learning_rate": 6.254788460032551e-06,
      "loss": 1.2075,
      "step": 13390
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.1875,
      "learning_rate": 6.237766217919948e-06,
      "loss": 1.1999,
      "step": 13395
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.875,
      "learning_rate": 6.220764482288291e-06,
      "loss": 1.1984,
      "step": 13400
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.6875,
      "learning_rate": 6.203783267809848e-06,
      "loss": 1.2679,
      "step": 13405
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.375,
      "learning_rate": 6.186822589139173e-06,
      "loss": 1.2431,
      "step": 13410
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.75,
      "learning_rate": 6.1698824609131035e-06,
      "loss": 1.2197,
      "step": 13415
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.53125,
      "learning_rate": 6.152962897750726e-06,
      "loss": 1.2248,
      "step": 13420
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.125,
      "learning_rate": 6.136063914253394e-06,
      "loss": 1.1751,
      "step": 13425
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.75,
      "learning_rate": 6.119185525004698e-06,
      "loss": 1.2829,
      "step": 13430
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.65625,
      "learning_rate": 6.102327744570467e-06,
      "loss": 1.2259,
      "step": 13435
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.53125,
      "learning_rate": 6.085490587498711e-06,
      "loss": 1.249,
      "step": 13440
    },
    {
      "epoch": 2.38,
      "grad_norm": 5.09375,
      "learning_rate": 6.0686740683196825e-06,
      "loss": 1.298,
      "step": 13445
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.3125,
      "learning_rate": 6.051878201545814e-06,
      "loss": 1.1801,
      "step": 13450
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.5,
      "learning_rate": 6.035103001671684e-06,
      "loss": 1.1638,
      "step": 13455
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.9375,
      "learning_rate": 6.018348483174074e-06,
      "loss": 1.2189,
      "step": 13460
    },
    {
      "epoch": 2.39,
      "grad_norm": 5.0625,
      "learning_rate": 6.001614660511907e-06,
      "loss": 1.1509,
      "step": 13465
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.28125,
      "learning_rate": 5.984901548126223e-06,
      "loss": 1.2472,
      "step": 13470
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.625,
      "learning_rate": 5.9682091604402234e-06,
      "loss": 1.2166,
      "step": 13475
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.5625,
      "learning_rate": 5.951537511859206e-06,
      "loss": 1.22,
      "step": 13480
    },
    {
      "epoch": 2.39,
      "grad_norm": 5.0,
      "learning_rate": 5.934886616770574e-06,
      "loss": 1.1693,
      "step": 13485
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.59375,
      "learning_rate": 5.918256489543819e-06,
      "loss": 1.2806,
      "step": 13490
    },
    {
      "epoch": 2.39,
      "grad_norm": 6.15625,
      "learning_rate": 5.901647144530523e-06,
      "loss": 1.2205,
      "step": 13495
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.8125,
      "learning_rate": 5.885058596064306e-06,
      "loss": 1.2368,
      "step": 13500
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.640625,
      "learning_rate": 5.868490858460861e-06,
      "loss": 1.1197,
      "step": 13505
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.9375,
      "learning_rate": 5.851943946017922e-06,
      "loss": 1.2083,
      "step": 13510
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.0,
      "learning_rate": 5.835417873015258e-06,
      "loss": 1.2314,
      "step": 13515
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.3125,
      "learning_rate": 5.818912653714611e-06,
      "loss": 1.229,
      "step": 13520
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.28125,
      "learning_rate": 5.802428302359793e-06,
      "loss": 1.2604,
      "step": 13525
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.90625,
      "learning_rate": 5.785964833176552e-06,
      "loss": 1.1857,
      "step": 13530
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.9375,
      "learning_rate": 5.76952226037264e-06,
      "loss": 1.2315,
      "step": 13535
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.1875,
      "learning_rate": 5.753100598137772e-06,
      "loss": 1.1852,
      "step": 13540
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.40625,
      "learning_rate": 5.736699860643623e-06,
      "loss": 1.2071,
      "step": 13545
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.90625,
      "learning_rate": 5.720320062043791e-06,
      "loss": 1.2514,
      "step": 13550
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.75,
      "learning_rate": 5.703961216473824e-06,
      "loss": 1.226,
      "step": 13555
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.15625,
      "learning_rate": 5.687623338051189e-06,
      "loss": 1.23,
      "step": 13560
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.75,
      "learning_rate": 5.671306440875224e-06,
      "loss": 1.279,
      "step": 13565
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.65625,
      "learning_rate": 5.655010539027215e-06,
      "loss": 1.2161,
      "step": 13570
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.0625,
      "learning_rate": 5.638735646570295e-06,
      "loss": 1.1989,
      "step": 13575
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.96875,
      "learning_rate": 5.622481777549462e-06,
      "loss": 1.3063,
      "step": 13580
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.21875,
      "learning_rate": 5.6062489459915865e-06,
      "loss": 1.2894,
      "step": 13585
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.890625,
      "learning_rate": 5.590037165905391e-06,
      "loss": 1.2357,
      "step": 13590
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.0,
      "learning_rate": 5.573846451281396e-06,
      "loss": 1.1786,
      "step": 13595
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.78125,
      "learning_rate": 5.557676816091979e-06,
      "loss": 1.2385,
      "step": 13600
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.34375,
      "learning_rate": 5.541528274291311e-06,
      "loss": 1.16,
      "step": 13605
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.09375,
      "learning_rate": 5.525400839815361e-06,
      "loss": 1.231,
      "step": 13610
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.8125,
      "learning_rate": 5.509294526581883e-06,
      "loss": 1.1719,
      "step": 13615
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.09375,
      "learning_rate": 5.493209348490411e-06,
      "loss": 1.2882,
      "step": 13620
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.6875,
      "learning_rate": 5.477145319422219e-06,
      "loss": 1.1774,
      "step": 13625
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.890625,
      "learning_rate": 5.461102453240349e-06,
      "loss": 1.2344,
      "step": 13630
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.28125,
      "learning_rate": 5.445080763789579e-06,
      "loss": 1.2527,
      "step": 13635
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.28125,
      "learning_rate": 5.429080264896408e-06,
      "loss": 1.2034,
      "step": 13640
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.1875,
      "learning_rate": 5.413100970369026e-06,
      "loss": 1.2204,
      "step": 13645
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.625,
      "learning_rate": 5.397142893997373e-06,
      "loss": 1.2548,
      "step": 13650
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.0625,
      "learning_rate": 5.38120604955303e-06,
      "loss": 1.2007,
      "step": 13655
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.3125,
      "learning_rate": 5.365290450789279e-06,
      "loss": 1.1728,
      "step": 13660
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.53125,
      "learning_rate": 5.349396111441065e-06,
      "loss": 1.245,
      "step": 13665
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.71875,
      "learning_rate": 5.333523045224991e-06,
      "loss": 1.3296,
      "step": 13670
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.21875,
      "learning_rate": 5.317671265839278e-06,
      "loss": 1.1998,
      "step": 13675
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.78125,
      "learning_rate": 5.301840786963804e-06,
      "loss": 1.2105,
      "step": 13680
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.5,
      "learning_rate": 5.286031622260055e-06,
      "loss": 1.1911,
      "step": 13685
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.46875,
      "learning_rate": 5.270243785371123e-06,
      "loss": 1.2265,
      "step": 13690
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.125,
      "learning_rate": 5.25447728992169e-06,
      "loss": 1.1662,
      "step": 13695
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.0,
      "learning_rate": 5.238732149518032e-06,
      "loss": 1.2045,
      "step": 13700
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.4375,
      "learning_rate": 5.22300837774798e-06,
      "loss": 1.18,
      "step": 13705
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.4375,
      "learning_rate": 5.207305988180936e-06,
      "loss": 1.1698,
      "step": 13710
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.25,
      "learning_rate": 5.1916249943678505e-06,
      "loss": 1.1973,
      "step": 13715
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.6875,
      "learning_rate": 5.175965409841199e-06,
      "loss": 1.2249,
      "step": 13720
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.53125,
      "learning_rate": 5.160327248114987e-06,
      "loss": 1.272,
      "step": 13725
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.625,
      "learning_rate": 5.144710522684738e-06,
      "loss": 1.2849,
      "step": 13730
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.40625,
      "learning_rate": 5.129115247027469e-06,
      "loss": 1.1437,
      "step": 13735
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.6875,
      "learning_rate": 5.1135414346016894e-06,
      "loss": 1.2092,
      "step": 13740
    },
    {
      "epoch": 2.44,
      "grad_norm": 5.0625,
      "learning_rate": 5.097989098847391e-06,
      "loss": 1.2192,
      "step": 13745
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.90625,
      "learning_rate": 5.0824582531860145e-06,
      "loss": 1.2569,
      "step": 13750
    },
    {
      "epoch": 2.44,
      "grad_norm": 5.34375,
      "learning_rate": 5.066948911020468e-06,
      "loss": 1.2137,
      "step": 13755
    },
    {
      "epoch": 2.44,
      "grad_norm": 5.25,
      "learning_rate": 5.051461085735106e-06,
      "loss": 1.1968,
      "step": 13760
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.65625,
      "learning_rate": 5.03599479069571e-06,
      "loss": 1.1551,
      "step": 13765
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.0,
      "learning_rate": 5.020550039249464e-06,
      "loss": 1.1821,
      "step": 13770
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.4375,
      "learning_rate": 5.005126844725002e-06,
      "loss": 1.2117,
      "step": 13775
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.875,
      "learning_rate": 4.989725220432311e-06,
      "loss": 1.219,
      "step": 13780
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.625,
      "learning_rate": 4.974345179662782e-06,
      "loss": 1.168,
      "step": 13785
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.84375,
      "learning_rate": 4.958986735689181e-06,
      "loss": 1.3122,
      "step": 13790
    },
    {
      "epoch": 2.44,
      "grad_norm": 5.1875,
      "learning_rate": 4.9436499017656424e-06,
      "loss": 1.2454,
      "step": 13795
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.5,
      "learning_rate": 4.928334691127626e-06,
      "loss": 1.1958,
      "step": 13800
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.8125,
      "learning_rate": 4.91304111699196e-06,
      "loss": 1.1721,
      "step": 13805
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.21875,
      "learning_rate": 4.897769192556783e-06,
      "loss": 1.1879,
      "step": 13810
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.09375,
      "learning_rate": 4.882518931001558e-06,
      "loss": 1.238,
      "step": 13815
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.40625,
      "learning_rate": 4.867290345487053e-06,
      "loss": 1.1882,
      "step": 13820
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.8125,
      "learning_rate": 4.8520834491553315e-06,
      "loss": 1.2289,
      "step": 13825
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.34375,
      "learning_rate": 4.836898255129725e-06,
      "loss": 1.2288,
      "step": 13830
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.96875,
      "learning_rate": 4.821734776514851e-06,
      "loss": 1.2269,
      "step": 13835
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.9375,
      "learning_rate": 4.806593026396593e-06,
      "loss": 1.2926,
      "step": 13840
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.34375,
      "learning_rate": 4.791473017842057e-06,
      "loss": 1.2239,
      "step": 13845
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.6875,
      "learning_rate": 4.776374763899604e-06,
      "loss": 1.2458,
      "step": 13850
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.53125,
      "learning_rate": 4.761298277598839e-06,
      "loss": 1.22,
      "step": 13855
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.65625,
      "learning_rate": 4.746243571950542e-06,
      "loss": 1.2482,
      "step": 13860
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.1875,
      "learning_rate": 4.731210659946726e-06,
      "loss": 1.1763,
      "step": 13865
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.25,
      "learning_rate": 4.71619955456058e-06,
      "loss": 1.2507,
      "step": 13870
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.21875,
      "learning_rate": 4.7012102687464965e-06,
      "loss": 1.2358,
      "step": 13875
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.125,
      "learning_rate": 4.686242815440008e-06,
      "loss": 1.2011,
      "step": 13880
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.90625,
      "learning_rate": 4.671297207557823e-06,
      "loss": 1.3662,
      "step": 13885
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.125,
      "learning_rate": 4.656373457997807e-06,
      "loss": 1.2752,
      "step": 13890
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.34375,
      "learning_rate": 4.64147157963893e-06,
      "loss": 1.2324,
      "step": 13895
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.78125,
      "learning_rate": 4.626591585341324e-06,
      "loss": 1.1464,
      "step": 13900
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.0,
      "learning_rate": 4.611733487946226e-06,
      "loss": 1.2287,
      "step": 13905
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.65625,
      "learning_rate": 4.596897300275949e-06,
      "loss": 1.2273,
      "step": 13910
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.5625,
      "learning_rate": 4.5820830351339304e-06,
      "loss": 1.2192,
      "step": 13915
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.3125,
      "learning_rate": 4.567290705304685e-06,
      "loss": 1.2464,
      "step": 13920
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.53125,
      "learning_rate": 4.552520323553772e-06,
      "loss": 1.2819,
      "step": 13925
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.40625,
      "learning_rate": 4.537771902627839e-06,
      "loss": 1.211,
      "step": 13930
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.5,
      "learning_rate": 4.5230454552545656e-06,
      "loss": 1.1863,
      "step": 13935
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.5625,
      "learning_rate": 4.508340994142679e-06,
      "loss": 1.198,
      "step": 13940
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.46875,
      "learning_rate": 4.493658531981923e-06,
      "loss": 1.1726,
      "step": 13945
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.8125,
      "learning_rate": 4.478998081443067e-06,
      "loss": 1.2258,
      "step": 13950
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.1875,
      "learning_rate": 4.4643596551778674e-06,
      "loss": 1.2225,
      "step": 13955
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.625,
      "learning_rate": 4.449743265819088e-06,
      "loss": 1.1837,
      "step": 13960
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.0625,
      "learning_rate": 4.435148925980476e-06,
      "loss": 1.2069,
      "step": 13965
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.65625,
      "learning_rate": 4.4205766482567514e-06,
      "loss": 1.2637,
      "step": 13970
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.0,
      "learning_rate": 4.406026445223566e-06,
      "loss": 1.2356,
      "step": 13975
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.75,
      "learning_rate": 4.391498329437574e-06,
      "loss": 1.2047,
      "step": 13980
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.3125,
      "learning_rate": 4.376992313436322e-06,
      "loss": 1.1981,
      "step": 13985
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.125,
      "learning_rate": 4.362508409738305e-06,
      "loss": 1.2055,
      "step": 13990
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.53125,
      "learning_rate": 4.348046630842935e-06,
      "loss": 1.203,
      "step": 13995
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.34375,
      "learning_rate": 4.333606989230534e-06,
      "loss": 1.1335,
      "step": 14000
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.3125,
      "learning_rate": 4.319189497362303e-06,
      "loss": 1.1772,
      "step": 14005
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.65625,
      "learning_rate": 4.304794167680341e-06,
      "loss": 1.2322,
      "step": 14010
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.59375,
      "learning_rate": 4.290421012607623e-06,
      "loss": 1.1897,
      "step": 14015
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.90625,
      "learning_rate": 4.2760700445479835e-06,
      "loss": 1.258,
      "step": 14020
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.59375,
      "learning_rate": 4.261741275886111e-06,
      "loss": 1.1733,
      "step": 14025
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.0625,
      "learning_rate": 4.247434718987543e-06,
      "loss": 1.1627,
      "step": 14030
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.625,
      "learning_rate": 4.233150386198624e-06,
      "loss": 1.2205,
      "step": 14035
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.5625,
      "learning_rate": 4.218888289846551e-06,
      "loss": 1.2929,
      "step": 14040
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.75,
      "learning_rate": 4.204648442239317e-06,
      "loss": 1.2437,
      "step": 14045
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.40625,
      "learning_rate": 4.190430855665703e-06,
      "loss": 1.2375,
      "step": 14050
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.0625,
      "learning_rate": 4.176235542395298e-06,
      "loss": 1.2609,
      "step": 14055
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.40625,
      "learning_rate": 4.162062514678462e-06,
      "loss": 1.2145,
      "step": 14060
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.84375,
      "learning_rate": 4.147911784746324e-06,
      "loss": 1.2177,
      "step": 14065
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.8125,
      "learning_rate": 4.133783364810767e-06,
      "loss": 1.2332,
      "step": 14070
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.3125,
      "learning_rate": 4.119677267064433e-06,
      "loss": 1.1548,
      "step": 14075
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.4375,
      "learning_rate": 4.105593503680672e-06,
      "loss": 1.181,
      "step": 14080
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.75,
      "learning_rate": 4.091532086813589e-06,
      "loss": 1.1565,
      "step": 14085
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.125,
      "learning_rate": 4.077493028597995e-06,
      "loss": 1.2753,
      "step": 14090
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.34375,
      "learning_rate": 4.063476341149404e-06,
      "loss": 1.1614,
      "step": 14095
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.375,
      "learning_rate": 4.049482036564012e-06,
      "loss": 1.213,
      "step": 14100
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.03125,
      "learning_rate": 4.035510126918732e-06,
      "loss": 1.2405,
      "step": 14105
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.71875,
      "learning_rate": 4.021560624271113e-06,
      "loss": 1.2894,
      "step": 14110
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.1875,
      "learning_rate": 4.0076335406593886e-06,
      "loss": 1.1662,
      "step": 14115
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.09375,
      "learning_rate": 3.9937288881024446e-06,
      "loss": 1.245,
      "step": 14120
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.875,
      "learning_rate": 3.979846678599809e-06,
      "loss": 1.2372,
      "step": 14125
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.3125,
      "learning_rate": 3.9659869241316215e-06,
      "loss": 1.2155,
      "step": 14130
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.71875,
      "learning_rate": 3.952149636658672e-06,
      "loss": 1.1334,
      "step": 14135
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.84375,
      "learning_rate": 3.938334828122344e-06,
      "loss": 1.2097,
      "step": 14140
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.9375,
      "learning_rate": 3.9245425104446285e-06,
      "loss": 1.1807,
      "step": 14145
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.875,
      "learning_rate": 3.910772695528111e-06,
      "loss": 1.2905,
      "step": 14150
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.59375,
      "learning_rate": 3.897025395255952e-06,
      "loss": 1.1643,
      "step": 14155
    },
    {
      "epoch": 2.51,
      "grad_norm": 5.53125,
      "learning_rate": 3.883300621491872e-06,
      "loss": 1.1922,
      "step": 14160
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.65625,
      "learning_rate": 3.86959838608017e-06,
      "loss": 1.2677,
      "step": 14165
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.5625,
      "learning_rate": 3.855918700845693e-06,
      "loss": 1.2631,
      "step": 14170
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.28125,
      "learning_rate": 3.842261577593805e-06,
      "loss": 1.2279,
      "step": 14175
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.0625,
      "learning_rate": 3.828627028110419e-06,
      "loss": 1.2304,
      "step": 14180
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.4375,
      "learning_rate": 3.815015064161984e-06,
      "loss": 1.2528,
      "step": 14185
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.53125,
      "learning_rate": 3.801425697495412e-06,
      "loss": 1.2509,
      "step": 14190
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.46875,
      "learning_rate": 3.7878589398381548e-06,
      "loss": 1.2176,
      "step": 14195
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.65625,
      "learning_rate": 3.7743148028981367e-06,
      "loss": 1.1542,
      "step": 14200
    },
    {
      "epoch": 2.52,
      "grad_norm": 5.0625,
      "learning_rate": 3.760793298363755e-06,
      "loss": 1.2211,
      "step": 14205
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.78125,
      "learning_rate": 3.747294437903882e-06,
      "loss": 1.1891,
      "step": 14210
    },
    {
      "epoch": 2.52,
      "grad_norm": 5.21875,
      "learning_rate": 3.733818233167853e-06,
      "loss": 1.1773,
      "step": 14215
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.25,
      "learning_rate": 3.720364695785451e-06,
      "loss": 1.2188,
      "step": 14220
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.84375,
      "learning_rate": 3.7069338373668795e-06,
      "loss": 1.2067,
      "step": 14225
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.953125,
      "learning_rate": 3.6935256695027963e-06,
      "loss": 1.193,
      "step": 14230
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.15625,
      "learning_rate": 3.6801402037642727e-06,
      "loss": 1.247,
      "step": 14235
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.6875,
      "learning_rate": 3.6667774517027653e-06,
      "loss": 1.1969,
      "step": 14240
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.34375,
      "learning_rate": 3.6534374248501524e-06,
      "loss": 1.1234,
      "step": 14245
    },
    {
      "epoch": 2.53,
      "grad_norm": 6.46875,
      "learning_rate": 3.640120134718702e-06,
      "loss": 1.2498,
      "step": 14250
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.25,
      "learning_rate": 3.6268255928010342e-06,
      "loss": 1.1945,
      "step": 14255
    },
    {
      "epoch": 2.53,
      "grad_norm": 5.09375,
      "learning_rate": 3.6135538105701683e-06,
      "loss": 1.2459,
      "step": 14260
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.125,
      "learning_rate": 3.6003047994794658e-06,
      "loss": 1.2033,
      "step": 14265
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.8125,
      "learning_rate": 3.587078570962644e-06,
      "loss": 1.1887,
      "step": 14270
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.28125,
      "learning_rate": 3.5738751364337565e-06,
      "loss": 1.2342,
      "step": 14275
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.71875,
      "learning_rate": 3.5606945072871875e-06,
      "loss": 1.2288,
      "step": 14280
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.5625,
      "learning_rate": 3.5475366948976294e-06,
      "loss": 1.2596,
      "step": 14285
    },
    {
      "epoch": 2.53,
      "grad_norm": 5.3125,
      "learning_rate": 3.534401710620099e-06,
      "loss": 1.1797,
      "step": 14290
    },
    {
      "epoch": 2.53,
      "grad_norm": 5.25,
      "learning_rate": 3.5212895657899114e-06,
      "loss": 1.2309,
      "step": 14295
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.40625,
      "learning_rate": 3.5082002717226592e-06,
      "loss": 1.297,
      "step": 14300
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.5,
      "learning_rate": 3.4951338397142197e-06,
      "loss": 1.2751,
      "step": 14305
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.875,
      "learning_rate": 3.4820902810407663e-06,
      "loss": 1.2834,
      "step": 14310
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.375,
      "learning_rate": 3.469069606958688e-06,
      "loss": 1.2813,
      "step": 14315
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.0625,
      "learning_rate": 3.456071828704653e-06,
      "loss": 1.1941,
      "step": 14320
    },
    {
      "epoch": 2.54,
      "grad_norm": 5.0625,
      "learning_rate": 3.4430969574955693e-06,
      "loss": 1.2214,
      "step": 14325
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.8125,
      "learning_rate": 3.4301450045285767e-06,
      "loss": 1.2623,
      "step": 14330
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.5,
      "learning_rate": 3.41721598098102e-06,
      "loss": 1.2812,
      "step": 14335
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.0,
      "learning_rate": 3.4043098980104747e-06,
      "loss": 1.1913,
      "step": 14340
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.4375,
      "learning_rate": 3.3914267667547116e-06,
      "loss": 1.2005,
      "step": 14345
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.53125,
      "learning_rate": 3.3785665983317005e-06,
      "loss": 1.1436,
      "step": 14350
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.1875,
      "learning_rate": 3.3657294038395836e-06,
      "loss": 1.1792,
      "step": 14355
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.40625,
      "learning_rate": 3.3529151943566927e-06,
      "loss": 1.197,
      "step": 14360
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.125,
      "learning_rate": 3.3401239809415052e-06,
      "loss": 1.2112,
      "step": 14365
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.15625,
      "learning_rate": 3.327355774632661e-06,
      "loss": 1.2661,
      "step": 14370
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.71875,
      "learning_rate": 3.3146105864489597e-06,
      "loss": 1.1639,
      "step": 14375
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.5625,
      "learning_rate": 3.301888427389306e-06,
      "loss": 1.2085,
      "step": 14380
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.15625,
      "learning_rate": 3.289189308432757e-06,
      "loss": 1.1612,
      "step": 14385
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.90625,
      "learning_rate": 3.276513240538478e-06,
      "loss": 1.2035,
      "step": 14390
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.625,
      "learning_rate": 3.263860234645737e-06,
      "loss": 1.1527,
      "step": 14395
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.15625,
      "learning_rate": 3.251230301673913e-06,
      "loss": 1.2506,
      "step": 14400
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.625,
      "learning_rate": 3.238623452522461e-06,
      "loss": 1.1533,
      "step": 14405
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.4375,
      "learning_rate": 3.226039698070912e-06,
      "loss": 1.2889,
      "step": 14410
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.875,
      "learning_rate": 3.213479049178878e-06,
      "loss": 1.2468,
      "step": 14415
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.375,
      "learning_rate": 3.200941516686022e-06,
      "loss": 1.2032,
      "step": 14420
    },
    {
      "epoch": 2.56,
      "grad_norm": 5.625,
      "learning_rate": 3.1884271114120765e-06,
      "loss": 1.2408,
      "step": 14425
    },
    {
      "epoch": 2.56,
      "grad_norm": 5.5,
      "learning_rate": 3.1759358441567756e-06,
      "loss": 1.2995,
      "step": 14430
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.375,
      "learning_rate": 3.163467725699939e-06,
      "loss": 1.2982,
      "step": 14435
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.59375,
      "learning_rate": 3.1510227668013637e-06,
      "loss": 1.2657,
      "step": 14440
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.625,
      "learning_rate": 3.1386009782008818e-06,
      "loss": 1.1995,
      "step": 14445
    },
    {
      "epoch": 2.56,
      "grad_norm": 5.0,
      "learning_rate": 3.1262023706183253e-06,
      "loss": 1.1654,
      "step": 14450
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.96875,
      "learning_rate": 3.1138269547535314e-06,
      "loss": 1.2838,
      "step": 14455
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.125,
      "learning_rate": 3.101474741286301e-06,
      "loss": 1.2064,
      "step": 14460
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.03125,
      "learning_rate": 3.089145740876427e-06,
      "loss": 1.2008,
      "step": 14465
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.9375,
      "learning_rate": 3.0768399641636714e-06,
      "loss": 1.3106,
      "step": 14470
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.6875,
      "learning_rate": 3.0645574217677497e-06,
      "loss": 1.2492,
      "step": 14475
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.65625,
      "learning_rate": 3.052298124288323e-06,
      "loss": 1.2166,
      "step": 14480
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.40625,
      "learning_rate": 3.0400620823050083e-06,
      "loss": 1.2448,
      "step": 14485
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.125,
      "learning_rate": 3.027849306377326e-06,
      "loss": 1.2154,
      "step": 14490
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.21875,
      "learning_rate": 3.0156598070447384e-06,
      "loss": 1.2059,
      "step": 14495
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.25,
      "learning_rate": 3.0034935948266275e-06,
      "loss": 1.2992,
      "step": 14500
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.34375,
      "learning_rate": 2.991350680222248e-06,
      "loss": 1.2047,
      "step": 14505
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.40625,
      "learning_rate": 2.9792310737107743e-06,
      "loss": 1.1961,
      "step": 14510
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.625,
      "learning_rate": 2.967134785751273e-06,
      "loss": 1.1616,
      "step": 14515
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.21875,
      "learning_rate": 2.955061826782661e-06,
      "loss": 1.1633,
      "step": 14520
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.15625,
      "learning_rate": 2.9430122072237373e-06,
      "loss": 1.2265,
      "step": 14525
    },
    {
      "epoch": 2.57,
      "grad_norm": 5.4375,
      "learning_rate": 2.9309859374731675e-06,
      "loss": 1.1996,
      "step": 14530
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.59375,
      "learning_rate": 2.918983027909442e-06,
      "loss": 1.3076,
      "step": 14535
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.09375,
      "learning_rate": 2.907003488890916e-06,
      "loss": 1.1145,
      "step": 14540
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.5625,
      "learning_rate": 2.8950473307557658e-06,
      "loss": 1.1708,
      "step": 14545
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.4375,
      "learning_rate": 2.8831145638219946e-06,
      "loss": 1.2462,
      "step": 14550
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.59375,
      "learning_rate": 2.871205198387402e-06,
      "loss": 1.2302,
      "step": 14555
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.125,
      "learning_rate": 2.85931924472963e-06,
      "loss": 1.2205,
      "step": 14560
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.34375,
      "learning_rate": 2.8474567131060738e-06,
      "loss": 1.2079,
      "step": 14565
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.25,
      "learning_rate": 2.83561761375394e-06,
      "loss": 1.2664,
      "step": 14570
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.25,
      "learning_rate": 2.8238019568902107e-06,
      "loss": 1.2239,
      "step": 14575
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.875,
      "learning_rate": 2.8120097527116383e-06,
      "loss": 1.2249,
      "step": 14580
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.53125,
      "learning_rate": 2.8002410113947264e-06,
      "loss": 1.2104,
      "step": 14585
    },
    {
      "epoch": 2.59,
      "grad_norm": 5.0,
      "learning_rate": 2.788495743095737e-06,
      "loss": 1.2833,
      "step": 14590
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.8125,
      "learning_rate": 2.776773957950679e-06,
      "loss": 1.1381,
      "step": 14595
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.75,
      "learning_rate": 2.7650756660752873e-06,
      "loss": 1.1904,
      "step": 14600
    },
    {
      "epoch": 2.59,
      "grad_norm": 5.40625,
      "learning_rate": 2.753400877565033e-06,
      "loss": 1.3166,
      "step": 14605
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.09375,
      "learning_rate": 2.7417496024951006e-06,
      "loss": 1.2379,
      "step": 14610
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.40625,
      "learning_rate": 2.7301218509203697e-06,
      "loss": 1.1392,
      "step": 14615
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.78125,
      "learning_rate": 2.7185176328754334e-06,
      "loss": 1.2676,
      "step": 14620
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.78125,
      "learning_rate": 2.7069369583745786e-06,
      "loss": 1.304,
      "step": 14625
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.34375,
      "learning_rate": 2.69537983741176e-06,
      "loss": 1.1828,
      "step": 14630
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.5625,
      "learning_rate": 2.68384627996061e-06,
      "loss": 1.2505,
      "step": 14635
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.5,
      "learning_rate": 2.672336295974447e-06,
      "loss": 1.2539,
      "step": 14640
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.84375,
      "learning_rate": 2.660849895386218e-06,
      "loss": 1.1943,
      "step": 14645
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.125,
      "learning_rate": 2.649387088108528e-06,
      "loss": 1.2522,
      "step": 14650
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.125,
      "learning_rate": 2.6379478840336236e-06,
      "loss": 1.2709,
      "step": 14655
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.1875,
      "learning_rate": 2.6265322930333844e-06,
      "loss": 1.2667,
      "step": 14660
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.71875,
      "learning_rate": 2.615140324959304e-06,
      "loss": 1.2275,
      "step": 14665
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.75,
      "learning_rate": 2.6037719896424917e-06,
      "loss": 1.2404,
      "step": 14670
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.46875,
      "learning_rate": 2.592427296893669e-06,
      "loss": 1.2323,
      "step": 14675
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.53125,
      "learning_rate": 2.581106256503152e-06,
      "loss": 1.2422,
      "step": 14680
    },
    {
      "epoch": 2.6,
      "grad_norm": 5.03125,
      "learning_rate": 2.5698088782408345e-06,
      "loss": 1.1812,
      "step": 14685
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.46875,
      "learning_rate": 2.558535171856212e-06,
      "loss": 1.1797,
      "step": 14690
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.6875,
      "learning_rate": 2.5472851470783253e-06,
      "loss": 1.2708,
      "step": 14695
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.28125,
      "learning_rate": 2.536058813615795e-06,
      "loss": 1.2745,
      "step": 14700
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.625,
      "learning_rate": 2.524856181156797e-06,
      "loss": 1.1406,
      "step": 14705
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.34375,
      "learning_rate": 2.5136772593690395e-06,
      "loss": 1.2081,
      "step": 14710
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.75,
      "learning_rate": 2.502522057899784e-06,
      "loss": 1.2162,
      "step": 14715
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.46875,
      "learning_rate": 2.491390586375818e-06,
      "loss": 1.2293,
      "step": 14720
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.4375,
      "learning_rate": 2.4802828544034427e-06,
      "loss": 1.208,
      "step": 14725
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.6875,
      "learning_rate": 2.4691988715684843e-06,
      "loss": 1.2732,
      "step": 14730
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.5625,
      "learning_rate": 2.4581386474362676e-06,
      "loss": 1.254,
      "step": 14735
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.96875,
      "learning_rate": 2.447102191551608e-06,
      "loss": 1.2072,
      "step": 14740
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.09375,
      "learning_rate": 2.4360895134388162e-06,
      "loss": 1.21,
      "step": 14745
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.96875,
      "learning_rate": 2.425100622601687e-06,
      "loss": 1.2603,
      "step": 14750
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.78125,
      "learning_rate": 2.414135528523481e-06,
      "loss": 1.2553,
      "step": 14755
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.40625,
      "learning_rate": 2.4031942406669127e-06,
      "loss": 1.2011,
      "step": 14760
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.625,
      "learning_rate": 2.392276768474182e-06,
      "loss": 1.267,
      "step": 14765
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.40625,
      "learning_rate": 2.381383121366899e-06,
      "loss": 1.2457,
      "step": 14770
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.1875,
      "learning_rate": 2.370513308746144e-06,
      "loss": 1.2169,
      "step": 14775
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.875,
      "learning_rate": 2.3596673399924095e-06,
      "loss": 1.2065,
      "step": 14780
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.1875,
      "learning_rate": 2.348845224465622e-06,
      "loss": 1.2515,
      "step": 14785
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.78125,
      "learning_rate": 2.3380469715051147e-06,
      "loss": 1.2639,
      "step": 14790
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.375,
      "learning_rate": 2.3272725904296287e-06,
      "loss": 1.2149,
      "step": 14795
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.34375,
      "learning_rate": 2.3165220905373107e-06,
      "loss": 1.253,
      "step": 14800
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.125,
      "learning_rate": 2.3057954811056958e-06,
      "loss": 1.227,
      "step": 14805
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.46875,
      "learning_rate": 2.295092771391696e-06,
      "loss": 1.2242,
      "step": 14810
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.5,
      "learning_rate": 2.2844139706316093e-06,
      "loss": 1.2294,
      "step": 14815
    },
    {
      "epoch": 2.63,
      "grad_norm": 5.125,
      "learning_rate": 2.2737590880410874e-06,
      "loss": 1.2813,
      "step": 14820
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.5625,
      "learning_rate": 2.263128132815149e-06,
      "loss": 1.233,
      "step": 14825
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.28125,
      "learning_rate": 2.2525211141281643e-06,
      "loss": 1.2906,
      "step": 14830
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.96875,
      "learning_rate": 2.241938041133842e-06,
      "loss": 1.2401,
      "step": 14835
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.5,
      "learning_rate": 2.2313789229652228e-06,
      "loss": 1.1943,
      "step": 14840
    },
    {
      "epoch": 2.63,
      "grad_norm": 5.21875,
      "learning_rate": 2.2208437687346973e-06,
      "loss": 1.2146,
      "step": 14845
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.375,
      "learning_rate": 2.2103325875339463e-06,
      "loss": 1.2441,
      "step": 14850
    },
    {
      "epoch": 2.63,
      "grad_norm": 5.5625,
      "learning_rate": 2.1998453884339744e-06,
      "loss": 1.2039,
      "step": 14855
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.859375,
      "learning_rate": 2.1893821804850978e-06,
      "loss": 1.1928,
      "step": 14860
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.28125,
      "learning_rate": 2.178942972716911e-06,
      "loss": 1.2162,
      "step": 14865
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.90625,
      "learning_rate": 2.16852777413831e-06,
      "loss": 1.23,
      "step": 14870
    },
    {
      "epoch": 2.64,
      "grad_norm": 5.1875,
      "learning_rate": 2.1581365937374733e-06,
      "loss": 1.1935,
      "step": 14875
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.96875,
      "learning_rate": 2.147769440481846e-06,
      "loss": 1.3079,
      "step": 14880
    },
    {
      "epoch": 2.64,
      "grad_norm": 5.21875,
      "learning_rate": 2.1374263233181256e-06,
      "loss": 1.1898,
      "step": 14885
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.5,
      "learning_rate": 2.1271072511723023e-06,
      "loss": 1.2012,
      "step": 14890
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.90625,
      "learning_rate": 2.1168122329495753e-06,
      "loss": 1.2945,
      "step": 14895
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.625,
      "learning_rate": 2.1065412775344085e-06,
      "loss": 1.2,
      "step": 14900
    },
    {
      "epoch": 2.64,
      "grad_norm": 5.5,
      "learning_rate": 2.0962943937904978e-06,
      "loss": 1.2661,
      "step": 14905
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.09375,
      "learning_rate": 2.086071590560766e-06,
      "loss": 1.2286,
      "step": 14910
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.65625,
      "learning_rate": 2.075872876667344e-06,
      "loss": 1.2287,
      "step": 14915
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.625,
      "learning_rate": 2.0656982609115803e-06,
      "loss": 1.2689,
      "step": 14920
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.15625,
      "learning_rate": 2.0555477520740384e-06,
      "loss": 1.2351,
      "step": 14925
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.4375,
      "learning_rate": 2.0454213589144577e-06,
      "loss": 1.246,
      "step": 14930
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.3125,
      "learning_rate": 2.0353190901717844e-06,
      "loss": 1.204,
      "step": 14935
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.28125,
      "learning_rate": 2.025240954564138e-06,
      "loss": 1.2059,
      "step": 14940
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.6875,
      "learning_rate": 2.015186960788802e-06,
      "loss": 1.2623,
      "step": 14945
    },
    {
      "epoch": 2.65,
      "grad_norm": 5.78125,
      "learning_rate": 2.0051571175222382e-06,
      "loss": 1.2947,
      "step": 14950
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.953125,
      "learning_rate": 1.9951514334200706e-06,
      "loss": 1.2189,
      "step": 14955
    },
    {
      "epoch": 2.65,
      "grad_norm": 5.34375,
      "learning_rate": 1.985169917117058e-06,
      "loss": 1.2386,
      "step": 14960
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.6875,
      "learning_rate": 1.9752125772271102e-06,
      "loss": 1.2289,
      "step": 14965
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.15625,
      "learning_rate": 1.965279422343287e-06,
      "loss": 1.2092,
      "step": 14970
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.875,
      "learning_rate": 1.955370461037751e-06,
      "loss": 1.2083,
      "step": 14975
    },
    {
      "epoch": 2.65,
      "grad_norm": 5.0625,
      "learning_rate": 1.945485701861809e-06,
      "loss": 1.2536,
      "step": 14980
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.25,
      "learning_rate": 1.93562515334587e-06,
      "loss": 1.2724,
      "step": 14985
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.15625,
      "learning_rate": 1.925788823999448e-06,
      "loss": 1.187,
      "step": 14990
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.53125,
      "learning_rate": 1.9159767223111592e-06,
      "loss": 1.2194,
      "step": 14995
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.53125,
      "learning_rate": 1.9061888567487162e-06,
      "loss": 1.2056,
      "step": 15000
    },
    {
      "epoch": 2.66,
      "grad_norm": 5.4375,
      "learning_rate": 1.8964252357589128e-06,
      "loss": 1.2409,
      "step": 15005
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.25,
      "learning_rate": 1.8866858677676191e-06,
      "loss": 1.2587,
      "step": 15010
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.875,
      "learning_rate": 1.8769707611797748e-06,
      "loss": 1.2194,
      "step": 15015
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.21875,
      "learning_rate": 1.867279924379387e-06,
      "loss": 1.2043,
      "step": 15020
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.90625,
      "learning_rate": 1.8576133657295091e-06,
      "loss": 1.2677,
      "step": 15025
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.5625,
      "learning_rate": 1.8479710935722483e-06,
      "loss": 1.2728,
      "step": 15030
    },
    {
      "epoch": 2.66,
      "grad_norm": 5.25,
      "learning_rate": 1.8383531162287582e-06,
      "loss": 1.2767,
      "step": 15035
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.25,
      "learning_rate": 1.8287594419992127e-06,
      "loss": 1.1646,
      "step": 15040
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.25,
      "learning_rate": 1.8191900791628292e-06,
      "loss": 1.2444,
      "step": 15045
    },
    {
      "epoch": 2.67,
      "grad_norm": 5.28125,
      "learning_rate": 1.8096450359778316e-06,
      "loss": 1.1962,
      "step": 15050
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.8125,
      "learning_rate": 1.8001243206814578e-06,
      "loss": 1.2632,
      "step": 15055
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.375,
      "learning_rate": 1.7906279414899619e-06,
      "loss": 1.3105,
      "step": 15060
    },
    {
      "epoch": 2.67,
      "grad_norm": 5.0625,
      "learning_rate": 1.7811559065985915e-06,
      "loss": 1.2764,
      "step": 15065
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.96875,
      "learning_rate": 1.7717082241815686e-06,
      "loss": 1.1784,
      "step": 15070
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.5,
      "learning_rate": 1.7622849023921207e-06,
      "loss": 1.2218,
      "step": 15075
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.1875,
      "learning_rate": 1.7528859493624494e-06,
      "loss": 1.2025,
      "step": 15080
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.125,
      "learning_rate": 1.74351137320371e-06,
      "loss": 1.2513,
      "step": 15085
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.0,
      "learning_rate": 1.734161182006041e-06,
      "loss": 1.1973,
      "step": 15090
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.625,
      "learning_rate": 1.7248353838385312e-06,
      "loss": 1.1656,
      "step": 15095
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.84375,
      "learning_rate": 1.7155339867492127e-06,
      "loss": 1.2393,
      "step": 15100
    },
    {
      "epoch": 2.68,
      "grad_norm": 5.125,
      "learning_rate": 1.706256998765061e-06,
      "loss": 1.2502,
      "step": 15105
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.84375,
      "learning_rate": 1.6970044278919927e-06,
      "loss": 1.1912,
      "step": 15110
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.125,
      "learning_rate": 1.6877762821148502e-06,
      "loss": 1.2212,
      "step": 15115
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.65625,
      "learning_rate": 1.678572569397393e-06,
      "loss": 1.2053,
      "step": 15120
    },
    {
      "epoch": 2.68,
      "grad_norm": 5.34375,
      "learning_rate": 1.669393297682298e-06,
      "loss": 1.1483,
      "step": 15125
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.6875,
      "learning_rate": 1.6602384748911515e-06,
      "loss": 1.25,
      "step": 15130
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.5625,
      "learning_rate": 1.6511081089244407e-06,
      "loss": 1.2511,
      "step": 15135
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.9375,
      "learning_rate": 1.642002207661546e-06,
      "loss": 1.1327,
      "step": 15140
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.28125,
      "learning_rate": 1.632920778960738e-06,
      "loss": 1.2231,
      "step": 15145
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.828125,
      "learning_rate": 1.623863830659158e-06,
      "loss": 1.2155,
      "step": 15150
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.125,
      "learning_rate": 1.6148313705728302e-06,
      "loss": 1.2029,
      "step": 15155
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.125,
      "learning_rate": 1.6058234064966492e-06,
      "loss": 1.2027,
      "step": 15160
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.59375,
      "learning_rate": 1.5968399462043536e-06,
      "loss": 1.2603,
      "step": 15165
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.3125,
      "learning_rate": 1.5878809974485487e-06,
      "loss": 1.2531,
      "step": 15170
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.9375,
      "learning_rate": 1.5789465679606962e-06,
      "loss": 1.2575,
      "step": 15175
    },
    {
      "epoch": 2.69,
      "grad_norm": 5.1875,
      "learning_rate": 1.5700366654510723e-06,
      "loss": 1.2439,
      "step": 15180
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.375,
      "learning_rate": 1.5611512976088026e-06,
      "loss": 1.2729,
      "step": 15185
    },
    {
      "epoch": 2.69,
      "grad_norm": 5.75,
      "learning_rate": 1.5522904721018438e-06,
      "loss": 1.223,
      "step": 15190
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.34375,
      "learning_rate": 1.5434541965769622e-06,
      "loss": 1.2872,
      "step": 15195
    },
    {
      "epoch": 2.69,
      "grad_norm": 5.5625,
      "learning_rate": 1.5346424786597379e-06,
      "loss": 1.2887,
      "step": 15200
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.125,
      "learning_rate": 1.5258553259545681e-06,
      "loss": 1.2042,
      "step": 15205
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.4375,
      "learning_rate": 1.517092746044647e-06,
      "loss": 1.2175,
      "step": 15210
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.3125,
      "learning_rate": 1.5083547464919455e-06,
      "loss": 1.2704,
      "step": 15215
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.53125,
      "learning_rate": 1.4996413348372584e-06,
      "loss": 1.2679,
      "step": 15220
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.4375,
      "learning_rate": 1.4909525186001272e-06,
      "loss": 1.2228,
      "step": 15225
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.3125,
      "learning_rate": 1.482288305278884e-06,
      "loss": 1.1883,
      "step": 15230
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.34375,
      "learning_rate": 1.4736487023506206e-06,
      "loss": 1.2245,
      "step": 15235
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.625,
      "learning_rate": 1.465033717271207e-06,
      "loss": 1.1921,
      "step": 15240
    },
    {
      "epoch": 2.7,
      "grad_norm": 5.71875,
      "learning_rate": 1.4564433574752456e-06,
      "loss": 1.2503,
      "step": 15245
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.875,
      "learning_rate": 1.4478776303761033e-06,
      "loss": 1.1965,
      "step": 15250
    },
    {
      "epoch": 2.7,
      "grad_norm": 5.0625,
      "learning_rate": 1.4393365433658867e-06,
      "loss": 1.2536,
      "step": 15255
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.5,
      "learning_rate": 1.43082010381543e-06,
      "loss": 1.2525,
      "step": 15260
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.890625,
      "learning_rate": 1.4223283190743109e-06,
      "loss": 1.1967,
      "step": 15265
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.53125,
      "learning_rate": 1.4138611964708203e-06,
      "loss": 1.2691,
      "step": 15270
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.3125,
      "learning_rate": 1.405418743311967e-06,
      "loss": 1.2037,
      "step": 15275
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.96875,
      "learning_rate": 1.3970009668834728e-06,
      "loss": 1.2393,
      "step": 15280
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.5625,
      "learning_rate": 1.3886078744497666e-06,
      "loss": 1.2493,
      "step": 15285
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.125,
      "learning_rate": 1.3802394732539613e-06,
      "loss": 1.228,
      "step": 15290
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.28125,
      "learning_rate": 1.371895770517877e-06,
      "loss": 1.2395,
      "step": 15295
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.0,
      "learning_rate": 1.3635767734420268e-06,
      "loss": 1.206,
      "step": 15300
    },
    {
      "epoch": 2.71,
      "grad_norm": 5.21875,
      "learning_rate": 1.3552824892055716e-06,
      "loss": 1.2609,
      "step": 15305
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.8125,
      "learning_rate": 1.3470129249663755e-06,
      "loss": 1.2489,
      "step": 15310
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.4375,
      "learning_rate": 1.3387680878609632e-06,
      "loss": 1.2318,
      "step": 15315
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.0625,
      "learning_rate": 1.3305479850045e-06,
      "loss": 1.2424,
      "step": 15320
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.9375,
      "learning_rate": 1.3223526234908344e-06,
      "loss": 1.2121,
      "step": 15325
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.5,
      "learning_rate": 1.3141820103924463e-06,
      "loss": 1.2529,
      "step": 15330
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.96875,
      "learning_rate": 1.3060361527604614e-06,
      "loss": 1.153,
      "step": 15335
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.3125,
      "learning_rate": 1.2979150576246468e-06,
      "loss": 1.2536,
      "step": 15340
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.28125,
      "learning_rate": 1.2898187319933962e-06,
      "loss": 1.2562,
      "step": 15345
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.71875,
      "learning_rate": 1.281747182853721e-06,
      "loss": 1.1804,
      "step": 15350
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.25,
      "learning_rate": 1.2737004171712585e-06,
      "loss": 1.2051,
      "step": 15355
    },
    {
      "epoch": 2.72,
      "grad_norm": 5.09375,
      "learning_rate": 1.2656784418902611e-06,
      "loss": 1.2003,
      "step": 15360
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.65625,
      "learning_rate": 1.2576812639335844e-06,
      "loss": 1.1739,
      "step": 15365
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.53125,
      "learning_rate": 1.2497088902026775e-06,
      "loss": 1.2231,
      "step": 15370
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.96875,
      "learning_rate": 1.2417613275775874e-06,
      "loss": 1.2153,
      "step": 15375
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.96875,
      "learning_rate": 1.2338385829169607e-06,
      "loss": 1.1916,
      "step": 15380
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.8125,
      "learning_rate": 1.2259406630580084e-06,
      "loss": 1.209,
      "step": 15385
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.53125,
      "learning_rate": 1.2180675748165314e-06,
      "loss": 1.2562,
      "step": 15390
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.3125,
      "learning_rate": 1.2102193249868987e-06,
      "loss": 1.177,
      "step": 15395
    },
    {
      "epoch": 2.73,
      "grad_norm": 5.46875,
      "learning_rate": 1.2023959203420354e-06,
      "loss": 1.2022,
      "step": 15400
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.4375,
      "learning_rate": 1.1945973676334344e-06,
      "loss": 1.2274,
      "step": 15405
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.40625,
      "learning_rate": 1.1868236735911454e-06,
      "loss": 1.1985,
      "step": 15410
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.53125,
      "learning_rate": 1.1790748449237488e-06,
      "loss": 1.1587,
      "step": 15415
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.875,
      "learning_rate": 1.1713508883183788e-06,
      "loss": 1.2371,
      "step": 15420
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.953125,
      "learning_rate": 1.1636518104407168e-06,
      "loss": 1.2591,
      "step": 15425
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.5,
      "learning_rate": 1.155977617934948e-06,
      "loss": 1.2539,
      "step": 15430
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.5625,
      "learning_rate": 1.148328317423798e-06,
      "loss": 1.279,
      "step": 15435
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.34375,
      "learning_rate": 1.1407039155085131e-06,
      "loss": 1.1732,
      "step": 15440
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.9375,
      "learning_rate": 1.1331044187688399e-06,
      "loss": 1.2721,
      "step": 15445
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.5,
      "learning_rate": 1.1255298337630392e-06,
      "loss": 1.2155,
      "step": 15450
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.4375,
      "learning_rate": 1.1179801670278823e-06,
      "loss": 1.2153,
      "step": 15455
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.9375,
      "learning_rate": 1.1104554250786203e-06,
      "loss": 1.29,
      "step": 15460
    },
    {
      "epoch": 2.74,
      "grad_norm": 5.125,
      "learning_rate": 1.1029556144090026e-06,
      "loss": 1.2462,
      "step": 15465
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.21875,
      "learning_rate": 1.0954807414912682e-06,
      "loss": 1.182,
      "step": 15470
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.28125,
      "learning_rate": 1.0880308127761274e-06,
      "loss": 1.1277,
      "step": 15475
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.25,
      "learning_rate": 1.0806058346927606e-06,
      "loss": 1.1808,
      "step": 15480
    },
    {
      "epoch": 2.74,
      "grad_norm": 5.21875,
      "learning_rate": 1.0732058136488255e-06,
      "loss": 1.2594,
      "step": 15485
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.4375,
      "learning_rate": 1.0658307560304436e-06,
      "loss": 1.2275,
      "step": 15490
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.0,
      "learning_rate": 1.058480668202181e-06,
      "loss": 1.23,
      "step": 15495
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.53125,
      "learning_rate": 1.0511555565070597e-06,
      "loss": 1.1641,
      "step": 15500
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.84375,
      "learning_rate": 1.0438554272665635e-06,
      "loss": 1.1708,
      "step": 15505
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.3125,
      "learning_rate": 1.036580286780593e-06,
      "loss": 1.2656,
      "step": 15510
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.84375,
      "learning_rate": 1.0293301413274958e-06,
      "loss": 1.221,
      "step": 15515
    },
    {
      "epoch": 2.75,
      "grad_norm": 5.28125,
      "learning_rate": 1.0221049971640505e-06,
      "loss": 1.2158,
      "step": 15520
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.3125,
      "learning_rate": 1.014904860525453e-06,
      "loss": 1.2776,
      "step": 15525
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.25,
      "learning_rate": 1.0077297376253258e-06,
      "loss": 1.2624,
      "step": 15530
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.46875,
      "learning_rate": 1.0005796346556995e-06,
      "loss": 1.212,
      "step": 15535
    },
    {
      "epoch": 2.75,
      "grad_norm": 5.4375,
      "learning_rate": 9.934545577870146e-07,
      "loss": 1.2415,
      "step": 15540
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.0,
      "learning_rate": 9.863545131681162e-07,
      "loss": 1.2322,
      "step": 15545
    },
    {
      "epoch": 2.76,
      "grad_norm": 5.125,
      "learning_rate": 9.792795069262495e-07,
      "loss": 1.2535,
      "step": 15550
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.40625,
      "learning_rate": 9.722295451670337e-07,
      "loss": 1.2222,
      "step": 15555
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.34375,
      "learning_rate": 9.652046339745024e-07,
      "loss": 1.1908,
      "step": 15560
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.375,
      "learning_rate": 9.582047794110493e-07,
      "loss": 1.1571,
      "step": 15565
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.5625,
      "learning_rate": 9.51229987517459e-07,
      "loss": 1.1873,
      "step": 15570
    },
    {
      "epoch": 2.76,
      "grad_norm": 5.25,
      "learning_rate": 9.442802643128767e-07,
      "loss": 1.2215,
      "step": 15575
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.34375,
      "learning_rate": 9.373556157948149e-07,
      "loss": 1.3061,
      "step": 15580
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.4375,
      "learning_rate": 9.304560479391566e-07,
      "loss": 1.2797,
      "step": 15585
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.875,
      "learning_rate": 9.235815667001324e-07,
      "loss": 1.2405,
      "step": 15590
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.875,
      "learning_rate": 9.167321780103199e-07,
      "loss": 1.197,
      "step": 15595
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.5625,
      "learning_rate": 9.099078877806577e-07,
      "loss": 1.2398,
      "step": 15600
    },
    {
      "epoch": 2.77,
      "grad_norm": 5.78125,
      "learning_rate": 9.031087019004047e-07,
      "loss": 1.2159,
      "step": 15605
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.25,
      "learning_rate": 8.963346262371675e-07,
      "loss": 1.1645,
      "step": 15610
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.125,
      "learning_rate": 8.895856666368862e-07,
      "loss": 1.1884,
      "step": 15615
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.40625,
      "learning_rate": 8.828618289238088e-07,
      "loss": 1.1647,
      "step": 15620
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.71875,
      "learning_rate": 8.761631189005171e-07,
      "loss": 1.2412,
      "step": 15625
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.25,
      "learning_rate": 8.694895423479199e-07,
      "loss": 1.1995,
      "step": 15630
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.34375,
      "learning_rate": 8.628411050252038e-07,
      "loss": 1.255,
      "step": 15635
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.53125,
      "learning_rate": 8.562178126698894e-07,
      "loss": 1.2921,
      "step": 15640
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.96875,
      "learning_rate": 8.496196709977877e-07,
      "loss": 1.2723,
      "step": 15645
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.09375,
      "learning_rate": 8.430466857029973e-07,
      "loss": 1.1927,
      "step": 15650
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.8125,
      "learning_rate": 8.36498862457914e-07,
      "loss": 1.2692,
      "step": 15655
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.4375,
      "learning_rate": 8.299762069132244e-07,
      "loss": 1.264,
      "step": 15660
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.0625,
      "learning_rate": 8.234787246978892e-07,
      "loss": 1.2398,
      "step": 15665
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.5625,
      "learning_rate": 8.170064214191464e-07,
      "loss": 1.258,
      "step": 15670
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.40625,
      "learning_rate": 8.10559302662508e-07,
      "loss": 1.2311,
      "step": 15675
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.5625,
      "learning_rate": 8.041373739917402e-07,
      "loss": 1.2606,
      "step": 15680
    },
    {
      "epoch": 2.78,
      "grad_norm": 5.1875,
      "learning_rate": 7.97740640948883e-07,
      "loss": 1.1358,
      "step": 15685
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.8125,
      "learning_rate": 7.913691090542308e-07,
      "loss": 1.2167,
      "step": 15690
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.125,
      "learning_rate": 7.850227838063284e-07,
      "loss": 1.194,
      "step": 15695
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.375,
      "learning_rate": 7.787016706819649e-07,
      "loss": 1.238,
      "step": 15700
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.34375,
      "learning_rate": 7.724057751361735e-07,
      "loss": 1.2405,
      "step": 15705
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.125,
      "learning_rate": 7.661351026022278e-07,
      "loss": 1.2148,
      "step": 15710
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.21875,
      "learning_rate": 7.598896584916326e-07,
      "loss": 1.2898,
      "step": 15715
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.34375,
      "learning_rate": 7.536694481941165e-07,
      "loss": 1.1764,
      "step": 15720
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.921875,
      "learning_rate": 7.474744770776387e-07,
      "loss": 1.2397,
      "step": 15725
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.75,
      "learning_rate": 7.41304750488373e-07,
      "loss": 1.2566,
      "step": 15730
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.65625,
      "learning_rate": 7.351602737507068e-07,
      "loss": 1.2023,
      "step": 15735
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.125,
      "learning_rate": 7.290410521672453e-07,
      "loss": 1.24,
      "step": 15740
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.25,
      "learning_rate": 7.22947091018784e-07,
      "loss": 1.2243,
      "step": 15745
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.59375,
      "learning_rate": 7.168783955643299e-07,
      "loss": 1.203,
      "step": 15750
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.53125,
      "learning_rate": 7.108349710410899e-07,
      "loss": 1.242,
      "step": 15755
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.25,
      "learning_rate": 7.048168226644491e-07,
      "loss": 1.1896,
      "step": 15760
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.75,
      "learning_rate": 6.988239556279929e-07,
      "loss": 1.2296,
      "step": 15765
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.6875,
      "learning_rate": 6.928563751034778e-07,
      "loss": 1.2388,
      "step": 15770
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.53125,
      "learning_rate": 6.869140862408441e-07,
      "loss": 1.1976,
      "step": 15775
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.46875,
      "learning_rate": 6.809970941682098e-07,
      "loss": 1.2042,
      "step": 15780
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.71875,
      "learning_rate": 6.751054039918503e-07,
      "loss": 1.2093,
      "step": 15785
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.78125,
      "learning_rate": 6.692390207962217e-07,
      "loss": 1.2112,
      "step": 15790
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.40625,
      "learning_rate": 6.633979496439247e-07,
      "loss": 1.1161,
      "step": 15795
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.1875,
      "learning_rate": 6.575821955757299e-07,
      "loss": 1.1945,
      "step": 15800
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.5,
      "learning_rate": 6.517917636105497e-07,
      "loss": 1.1423,
      "step": 15805
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.9375,
      "learning_rate": 6.460266587454433e-07,
      "loss": 1.208,
      "step": 15810
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.75,
      "learning_rate": 6.402868859556244e-07,
      "loss": 1.1694,
      "step": 15815
    },
    {
      "epoch": 2.8,
      "grad_norm": 5.0625,
      "learning_rate": 6.345724501944372e-07,
      "loss": 1.1911,
      "step": 15820
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.1875,
      "learning_rate": 6.288833563933538e-07,
      "loss": 1.2979,
      "step": 15825
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.5625,
      "learning_rate": 6.232196094619869e-07,
      "loss": 1.2027,
      "step": 15830
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.625,
      "learning_rate": 6.175812142880799e-07,
      "loss": 1.2395,
      "step": 15835
    },
    {
      "epoch": 2.81,
      "grad_norm": 5.28125,
      "learning_rate": 6.119681757374807e-07,
      "loss": 1.2291,
      "step": 15840
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.96875,
      "learning_rate": 6.063804986541677e-07,
      "loss": 1.153,
      "step": 15845
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.40625,
      "learning_rate": 6.008181878602337e-07,
      "loss": 1.2317,
      "step": 15850
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.375,
      "learning_rate": 5.952812481558689e-07,
      "loss": 1.2297,
      "step": 15855
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.03125,
      "learning_rate": 5.897696843193812e-07,
      "loss": 1.2349,
      "step": 15860
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.8125,
      "learning_rate": 5.842835011071724e-07,
      "loss": 1.1738,
      "step": 15865
    },
    {
      "epoch": 2.81,
      "grad_norm": 5.09375,
      "learning_rate": 5.788227032537419e-07,
      "loss": 1.2393,
      "step": 15870
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.75,
      "learning_rate": 5.733872954716835e-07,
      "loss": 1.2061,
      "step": 15875
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.1875,
      "learning_rate": 5.679772824516882e-07,
      "loss": 1.2361,
      "step": 15880
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.75,
      "learning_rate": 5.625926688625082e-07,
      "loss": 1.2011,
      "step": 15885
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.71875,
      "learning_rate": 5.572334593509998e-07,
      "loss": 1.239,
      "step": 15890
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.09375,
      "learning_rate": 5.518996585420866e-07,
      "loss": 1.2198,
      "step": 15895
    },
    {
      "epoch": 2.82,
      "grad_norm": 5.53125,
      "learning_rate": 5.465912710387633e-07,
      "loss": 1.2382,
      "step": 15900
    },
    {
      "epoch": 2.82,
      "grad_norm": 5.21875,
      "learning_rate": 5.413083014220954e-07,
      "loss": 1.2613,
      "step": 15905
    },
    {
      "epoch": 2.82,
      "grad_norm": 6.0,
      "learning_rate": 5.360507542512161e-07,
      "loss": 1.2447,
      "step": 15910
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.65625,
      "learning_rate": 5.308186340633093e-07,
      "loss": 1.2287,
      "step": 15915
    },
    {
      "epoch": 2.82,
      "grad_norm": 5.53125,
      "learning_rate": 5.25611945373633e-07,
      "loss": 1.1844,
      "step": 15920
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.5625,
      "learning_rate": 5.204306926754765e-07,
      "loss": 1.2053,
      "step": 15925
    },
    {
      "epoch": 2.82,
      "grad_norm": 5.03125,
      "learning_rate": 5.15274880440203e-07,
      "loss": 1.2335,
      "step": 15930
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.4375,
      "learning_rate": 5.101445131171934e-07,
      "loss": 1.1645,
      "step": 15935
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.84375,
      "learning_rate": 5.050395951338926e-07,
      "loss": 1.1898,
      "step": 15940
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.25,
      "learning_rate": 4.999601308957768e-07,
      "loss": 1.1604,
      "step": 15945
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.5625,
      "learning_rate": 4.949061247863462e-07,
      "loss": 1.2201,
      "step": 15950
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.8125,
      "learning_rate": 4.898775811671385e-07,
      "loss": 1.2203,
      "step": 15955
    },
    {
      "epoch": 2.83,
      "grad_norm": 5.15625,
      "learning_rate": 4.848745043777258e-07,
      "loss": 1.2214,
      "step": 15960
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.6875,
      "learning_rate": 4.798968987356911e-07,
      "loss": 1.2455,
      "step": 15965
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.71875,
      "learning_rate": 4.7494476853663484e-07,
      "loss": 1.2371,
      "step": 15970
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.75,
      "learning_rate": 4.7001811805418515e-07,
      "loss": 1.1929,
      "step": 15975
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.34375,
      "learning_rate": 4.651169515399678e-07,
      "loss": 1.26,
      "step": 15980
    },
    {
      "epoch": 2.83,
      "grad_norm": 5.09375,
      "learning_rate": 4.6024127322362274e-07,
      "loss": 1.1944,
      "step": 15985
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.375,
      "learning_rate": 4.553910873127942e-07,
      "loss": 1.183,
      "step": 15990
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.21875,
      "learning_rate": 4.5056639799312405e-07,
      "loss": 1.2059,
      "step": 15995
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.3125,
      "learning_rate": 4.457672094282583e-07,
      "loss": 1.2161,
      "step": 16000
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.96875,
      "learning_rate": 4.409935257598274e-07,
      "loss": 1.2133,
      "step": 16005
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.375,
      "learning_rate": 4.3624535110745267e-07,
      "loss": 1.2375,
      "step": 16010
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.625,
      "learning_rate": 4.315226895687463e-07,
      "loss": 1.2277,
      "step": 16015
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.8125,
      "learning_rate": 4.2682554521929816e-07,
      "loss": 1.1579,
      "step": 16020
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.1875,
      "learning_rate": 4.221539221126824e-07,
      "loss": 1.2338,
      "step": 16025
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.6875,
      "learning_rate": 4.175078242804442e-07,
      "loss": 1.2099,
      "step": 16030
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.96875,
      "learning_rate": 4.128872557320995e-07,
      "loss": 1.2116,
      "step": 16035
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.21875,
      "learning_rate": 4.082922204551354e-07,
      "loss": 1.2764,
      "step": 16040
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.78125,
      "learning_rate": 4.037227224150097e-07,
      "loss": 1.2124,
      "step": 16045
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.40625,
      "learning_rate": 3.991787655551282e-07,
      "loss": 1.2693,
      "step": 16050
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.0,
      "learning_rate": 3.9466035379687406e-07,
      "loss": 1.1926,
      "step": 16055
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.46875,
      "learning_rate": 3.901674910395647e-07,
      "loss": 1.2205,
      "step": 16060
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.78125,
      "learning_rate": 3.857001811604788e-07,
      "loss": 1.2319,
      "step": 16065
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.4375,
      "learning_rate": 3.8125842801485233e-07,
      "loss": 1.1998,
      "step": 16070
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.375,
      "learning_rate": 3.7684223543584915e-07,
      "loss": 1.2529,
      "step": 16075
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.0625,
      "learning_rate": 3.7245160723458406e-07,
      "loss": 1.2422,
      "step": 16080
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.25,
      "learning_rate": 3.680865472001127e-07,
      "loss": 1.2361,
      "step": 16085
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.78125,
      "learning_rate": 3.637470590994185e-07,
      "loss": 1.2167,
      "step": 16090
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.53125,
      "learning_rate": 3.594331466774192e-07,
      "loss": 1.1572,
      "step": 16095
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.03125,
      "learning_rate": 3.5514481365696683e-07,
      "loss": 1.2401,
      "step": 16100
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.5625,
      "learning_rate": 3.508820637388277e-07,
      "loss": 1.2385,
      "step": 16105
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.40625,
      "learning_rate": 3.4664490060169586e-07,
      "loss": 1.2307,
      "step": 16110
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.5,
      "learning_rate": 3.424333279021929e-07,
      "loss": 1.2208,
      "step": 16115
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.3125,
      "learning_rate": 3.3824734927484167e-07,
      "loss": 1.1774,
      "step": 16120
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.4375,
      "learning_rate": 3.340869683320824e-07,
      "loss": 1.2198,
      "step": 16125
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.375,
      "learning_rate": 3.2995218866427644e-07,
      "loss": 1.2034,
      "step": 16130
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.21875,
      "learning_rate": 3.258430138396695e-07,
      "loss": 1.2151,
      "step": 16135
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.34375,
      "learning_rate": 3.2175944740443165e-07,
      "loss": 1.2524,
      "step": 16140
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.71875,
      "learning_rate": 3.177014928826205e-07,
      "loss": 1.272,
      "step": 16145
    },
    {
      "epoch": 2.86,
      "grad_norm": 5.03125,
      "learning_rate": 3.136691537761982e-07,
      "loss": 1.2611,
      "step": 16150
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.75,
      "learning_rate": 3.0966243356501446e-07,
      "loss": 1.2921,
      "step": 16155
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.34375,
      "learning_rate": 3.0568133570681e-07,
      "loss": 1.1796,
      "step": 16160
    },
    {
      "epoch": 2.86,
      "grad_norm": 5.03125,
      "learning_rate": 3.0172586363723e-07,
      "loss": 1.2796,
      "step": 16165
    },
    {
      "epoch": 2.87,
      "grad_norm": 5.0625,
      "learning_rate": 2.977960207697805e-07,
      "loss": 1.2552,
      "step": 16170
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.1875,
      "learning_rate": 2.9389181049586875e-07,
      "loss": 1.1891,
      "step": 16175
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.21875,
      "learning_rate": 2.9001323618477296e-07,
      "loss": 1.2664,
      "step": 16180
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.1875,
      "learning_rate": 2.8616030118364886e-07,
      "loss": 1.2855,
      "step": 16185
    },
    {
      "epoch": 2.87,
      "grad_norm": 5.15625,
      "learning_rate": 2.8233300881752356e-07,
      "loss": 1.2714,
      "step": 16190
    },
    {
      "epoch": 2.87,
      "grad_norm": 5.6875,
      "learning_rate": 2.7853136238930154e-07,
      "loss": 1.1917,
      "step": 16195
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.3125,
      "learning_rate": 2.747553651797485e-07,
      "loss": 1.2613,
      "step": 16200
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.625,
      "learning_rate": 2.710050204475012e-07,
      "loss": 1.2138,
      "step": 16205
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.96875,
      "learning_rate": 2.6728033142905393e-07,
      "loss": 1.2404,
      "step": 16210
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.90625,
      "learning_rate": 2.6358130133876224e-07,
      "loss": 1.2025,
      "step": 16215
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.8125,
      "learning_rate": 2.5990793336883925e-07,
      "loss": 1.2518,
      "step": 16220
    },
    {
      "epoch": 2.88,
      "grad_norm": 5.15625,
      "learning_rate": 2.562602306893491e-07,
      "loss": 1.2982,
      "step": 16225
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.625,
      "learning_rate": 2.526381964482105e-07,
      "loss": 1.2893,
      "step": 16230
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.4375,
      "learning_rate": 2.490418337711864e-07,
      "loss": 1.215,
      "step": 16235
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.65625,
      "learning_rate": 2.4547114576189413e-07,
      "loss": 1.2429,
      "step": 16240
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.21875,
      "learning_rate": 2.419261355017821e-07,
      "loss": 1.1793,
      "step": 16245
    },
    {
      "epoch": 2.88,
      "grad_norm": 5.46875,
      "learning_rate": 2.3840680605014653e-07,
      "loss": 1.1375,
      "step": 16250
    },
    {
      "epoch": 2.88,
      "grad_norm": 5.09375,
      "learning_rate": 2.3491316044412458e-07,
      "loss": 1.2798,
      "step": 16255
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.03125,
      "learning_rate": 2.3144520169868122e-07,
      "loss": 1.1822,
      "step": 16260
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.59375,
      "learning_rate": 2.280029328066191e-07,
      "loss": 1.1957,
      "step": 16265
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.828125,
      "learning_rate": 2.2458635673856531e-07,
      "loss": 1.1985,
      "step": 16270
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.09375,
      "learning_rate": 2.211954764429813e-07,
      "loss": 1.268,
      "step": 16275
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.25,
      "learning_rate": 2.178302948461497e-07,
      "loss": 1.2402,
      "step": 16280
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.5625,
      "learning_rate": 2.144908148521707e-07,
      "loss": 1.2392,
      "step": 16285
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.40625,
      "learning_rate": 2.1117703934297904e-07,
      "loss": 1.2404,
      "step": 16290
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.15625,
      "learning_rate": 2.078889711783105e-07,
      "loss": 1.2165,
      "step": 16295
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.875,
      "learning_rate": 2.0462661319572197e-07,
      "loss": 1.2774,
      "step": 16300
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.40625,
      "learning_rate": 2.013899682105913e-07,
      "loss": 1.2393,
      "step": 16305
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.75,
      "learning_rate": 1.981790390160909e-07,
      "loss": 1.2066,
      "step": 16310
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.15625,
      "learning_rate": 1.949938283832109e-07,
      "loss": 1.1683,
      "step": 16315
    },
    {
      "epoch": 2.89,
      "grad_norm": 5.34375,
      "learning_rate": 1.9183433906074577e-07,
      "loss": 1.2378,
      "step": 16320
    },
    {
      "epoch": 2.89,
      "grad_norm": 5.1875,
      "learning_rate": 1.8870057377528783e-07,
      "loss": 1.2328,
      "step": 16325
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.5,
      "learning_rate": 1.855925352312371e-07,
      "loss": 1.1982,
      "step": 16330
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.28125,
      "learning_rate": 1.8251022611078806e-07,
      "loss": 1.2686,
      "step": 16335
    },
    {
      "epoch": 2.9,
      "grad_norm": 5.09375,
      "learning_rate": 1.7945364907392624e-07,
      "loss": 1.2487,
      "step": 16340
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.875,
      "learning_rate": 1.764228067584417e-07,
      "loss": 1.2633,
      "step": 16345
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.96875,
      "learning_rate": 1.7341770177990545e-07,
      "loss": 1.1945,
      "step": 16350
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.4375,
      "learning_rate": 1.7043833673168642e-07,
      "loss": 1.269,
      "step": 16355
    },
    {
      "epoch": 2.9,
      "grad_norm": 5.25,
      "learning_rate": 1.6748471418493116e-07,
      "loss": 1.3258,
      "step": 16360
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.375,
      "learning_rate": 1.6455683668857412e-07,
      "loss": 1.2335,
      "step": 16365
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.96875,
      "learning_rate": 1.6165470676933747e-07,
      "loss": 1.228,
      "step": 16370
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.65625,
      "learning_rate": 1.5877832693171778e-07,
      "loss": 1.2766,
      "step": 16375
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.125,
      "learning_rate": 1.5592769965798947e-07,
      "loss": 1.1978,
      "step": 16380
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.75,
      "learning_rate": 1.5310282740821135e-07,
      "loss": 1.2357,
      "step": 16385
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.0625,
      "learning_rate": 1.5030371262020005e-07,
      "loss": 1.1732,
      "step": 16390
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.75,
      "learning_rate": 1.4753035770955992e-07,
      "loss": 1.2298,
      "step": 16395
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.6875,
      "learning_rate": 1.447827650696565e-07,
      "loss": 1.2754,
      "step": 16400
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.59375,
      "learning_rate": 1.420609370716197e-07,
      "loss": 1.1565,
      "step": 16405
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.03125,
      "learning_rate": 1.3936487606435733e-07,
      "loss": 1.245,
      "step": 16410
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.46875,
      "learning_rate": 1.3669458437452818e-07,
      "loss": 1.141,
      "step": 16415
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.90625,
      "learning_rate": 1.3405006430655896e-07,
      "loss": 1.25,
      "step": 16420
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.75,
      "learning_rate": 1.3143131814263076e-07,
      "loss": 1.1836,
      "step": 16425
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.03125,
      "learning_rate": 1.2883834814269246e-07,
      "loss": 1.2555,
      "step": 16430
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.1875,
      "learning_rate": 1.262711565444341e-07,
      "loss": 1.1709,
      "step": 16435
    },
    {
      "epoch": 2.91,
      "grad_norm": 5.1875,
      "learning_rate": 1.237297455633135e-07,
      "loss": 1.4025,
      "step": 16440
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.6875,
      "learning_rate": 1.212141173925263e-07,
      "loss": 1.2073,
      "step": 16445
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.3125,
      "learning_rate": 1.1872427420302923e-07,
      "loss": 1.2578,
      "step": 16450
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.4375,
      "learning_rate": 1.162602181435235e-07,
      "loss": 1.1682,
      "step": 16455
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.59375,
      "learning_rate": 1.1382195134045148e-07,
      "loss": 1.2097,
      "step": 16460
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.75,
      "learning_rate": 1.1140947589800999e-07,
      "loss": 1.1716,
      "step": 16465
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.46875,
      "learning_rate": 1.0902279389812697e-07,
      "loss": 1.2083,
      "step": 16470
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.46875,
      "learning_rate": 1.0666190740047821e-07,
      "loss": 1.2104,
      "step": 16475
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.78125,
      "learning_rate": 1.0432681844247727e-07,
      "loss": 1.2306,
      "step": 16480
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.84375,
      "learning_rate": 1.0201752903926886e-07,
      "loss": 1.2515,
      "step": 16485
    },
    {
      "epoch": 2.92,
      "grad_norm": 5.09375,
      "learning_rate": 9.973404118374552e-08,
      "loss": 1.2126,
      "step": 16490
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.625,
      "learning_rate": 9.747635684651757e-08,
      "loss": 1.1876,
      "step": 16495
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.28125,
      "learning_rate": 9.524447797594316e-08,
      "loss": 1.2459,
      "step": 16500
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.5,
      "learning_rate": 9.303840649809825e-08,
      "loss": 1.2333,
      "step": 16505
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.84375,
      "learning_rate": 9.085814431679662e-08,
      "loss": 1.2255,
      "step": 16510
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.78125,
      "learning_rate": 8.870369331356987e-08,
      "loss": 1.2172,
      "step": 16515
    },
    {
      "epoch": 2.93,
      "grad_norm": 5.34375,
      "learning_rate": 8.657505534768073e-08,
      "loss": 1.259,
      "step": 16520
    },
    {
      "epoch": 2.93,
      "grad_norm": 5.03125,
      "learning_rate": 8.447223225611645e-08,
      "loss": 1.2263,
      "step": 16525
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.6875,
      "learning_rate": 8.239522585358206e-08,
      "loss": 1.1481,
      "step": 16530
    },
    {
      "epoch": 2.93,
      "grad_norm": 5.09375,
      "learning_rate": 8.03440379325071e-08,
      "loss": 1.2165,
      "step": 16535
    },
    {
      "epoch": 2.93,
      "grad_norm": 5.34375,
      "learning_rate": 7.831867026303896e-08,
      "loss": 1.1734,
      "step": 16540
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.75,
      "learning_rate": 7.63191245930428e-08,
      "loss": 1.2312,
      "step": 16545
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.1875,
      "learning_rate": 7.434540264809498e-08,
      "loss": 1.3157,
      "step": 16550
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.953125,
      "learning_rate": 7.239750613149299e-08,
      "loss": 1.2029,
      "step": 16555
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.46875,
      "learning_rate": 7.04754367242455e-08,
      "loss": 1.2632,
      "step": 16560
    },
    {
      "epoch": 2.94,
      "grad_norm": 5.28125,
      "learning_rate": 6.857919608506902e-08,
      "loss": 1.2953,
      "step": 16565
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.53125,
      "learning_rate": 6.670878585039452e-08,
      "loss": 1.1709,
      "step": 16570
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.15625,
      "learning_rate": 6.486420763436418e-08,
      "loss": 1.1543,
      "step": 16575
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.90625,
      "learning_rate": 6.304546302881797e-08,
      "loss": 1.2359,
      "step": 16580
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.625,
      "learning_rate": 6.125255360331373e-08,
      "loss": 1.2346,
      "step": 16585
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.4375,
      "learning_rate": 5.948548090510042e-08,
      "loss": 1.3088,
      "step": 16590
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.59375,
      "learning_rate": 5.774424645914156e-08,
      "loss": 1.188,
      "step": 16595
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.96875,
      "learning_rate": 5.6028851768095135e-08,
      "loss": 1.3111,
      "step": 16600
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.4375,
      "learning_rate": 5.433929831233031e-08,
      "loss": 1.1926,
      "step": 16605
    },
    {
      "epoch": 2.94,
      "grad_norm": 5.0,
      "learning_rate": 5.267558754990409e-08,
      "loss": 1.1958,
      "step": 16610
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.90625,
      "learning_rate": 5.103772091657133e-08,
      "loss": 1.2281,
      "step": 16615
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.78125,
      "learning_rate": 4.942569982579803e-08,
      "loss": 1.2216,
      "step": 16620
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.59375,
      "learning_rate": 4.7839525668728066e-08,
      "loss": 1.2009,
      "step": 16625
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.5,
      "learning_rate": 4.6279199814213135e-08,
      "loss": 1.2543,
      "step": 16630
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.25,
      "learning_rate": 4.474472360878945e-08,
      "loss": 1.2124,
      "step": 16635
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.21875,
      "learning_rate": 4.323609837669107e-08,
      "loss": 1.2459,
      "step": 16640
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.5625,
      "learning_rate": 4.175332541983989e-08,
      "loss": 1.2493,
      "step": 16645
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.625,
      "learning_rate": 4.0296406017849005e-08,
      "loss": 1.3077,
      "step": 16650
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.5,
      "learning_rate": 3.886534142801934e-08,
      "loss": 1.3042,
      "step": 16655
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.46875,
      "learning_rate": 3.746013288533967e-08,
      "loss": 1.2705,
      "step": 16660
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.59375,
      "learning_rate": 3.6080781602486625e-08,
      "loss": 1.2616,
      "step": 16665
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.34375,
      "learning_rate": 3.4727288769821344e-08,
      "loss": 1.1836,
      "step": 16670
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.9375,
      "learning_rate": 3.3399655555389485e-08,
      "loss": 1.1835,
      "step": 16675
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.1875,
      "learning_rate": 3.209788310492123e-08,
      "loss": 1.2594,
      "step": 16680
    },
    {
      "epoch": 2.96,
      "grad_norm": 5.0625,
      "learning_rate": 3.0821972541824617e-08,
      "loss": 1.2392,
      "step": 16685
    },
    {
      "epoch": 2.96,
      "grad_norm": 5.0,
      "learning_rate": 2.9571924967195517e-08,
      "loss": 1.2233,
      "step": 16690
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.625,
      "learning_rate": 2.8347741459807673e-08,
      "loss": 1.2079,
      "step": 16695
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.5,
      "learning_rate": 2.7149423076116008e-08,
      "loss": 1.2368,
      "step": 16700
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.28125,
      "learning_rate": 2.597697085024997e-08,
      "loss": 1.1561,
      "step": 16705
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.03125,
      "learning_rate": 2.4830385794020194e-08,
      "loss": 1.1925,
      "step": 16710
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.3125,
      "learning_rate": 2.370966889691517e-08,
      "loss": 1.2199,
      "step": 16715
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.375,
      "learning_rate": 2.2614821126097917e-08,
      "loss": 1.288,
      "step": 16720
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.78125,
      "learning_rate": 2.1545843426405977e-08,
      "loss": 1.3031,
      "step": 16725
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.59375,
      "learning_rate": 2.0502736720354743e-08,
      "loss": 1.2298,
      "step": 16730
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.96875,
      "learning_rate": 1.948550190813081e-08,
      "loss": 1.2347,
      "step": 16735
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.5,
      "learning_rate": 1.849413986758863e-08,
      "loss": 1.2037,
      "step": 16740
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.53125,
      "learning_rate": 1.7528651454267186e-08,
      "loss": 1.2405,
      "step": 16745
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.6875,
      "learning_rate": 1.6589037501363314e-08,
      "loss": 1.2105,
      "step": 16750
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.53125,
      "learning_rate": 1.5675298819755047e-08,
      "loss": 1.1745,
      "step": 16755
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.4375,
      "learning_rate": 1.4787436197984949e-08,
      "loss": 1.2091,
      "step": 16760
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.796875,
      "learning_rate": 1.3925450402266781e-08,
      "loss": 1.213,
      "step": 16765
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.5625,
      "learning_rate": 1.3089342176482167e-08,
      "loss": 1.227,
      "step": 16770
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.21875,
      "learning_rate": 1.227911224218059e-08,
      "loss": 1.288,
      "step": 16775
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.0625,
      "learning_rate": 1.1494761298582735e-08,
      "loss": 1.228,
      "step": 16780
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.5625,
      "learning_rate": 1.0736290022563822e-08,
      "loss": 1.3013,
      "step": 16785
    },
    {
      "epoch": 2.98,
      "grad_norm": 5.8125,
      "learning_rate": 1.0003699068683592e-08,
      "loss": 1.2895,
      "step": 16790
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.375,
      "learning_rate": 9.296989069156325e-09,
      "loss": 1.2222,
      "step": 16795
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.0,
      "learning_rate": 8.616160633857506e-09,
      "loss": 1.1691,
      "step": 16800
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.15625,
      "learning_rate": 7.961214350337143e-09,
      "loss": 1.2044,
      "step": 16805
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.875,
      "learning_rate": 7.332150783803115e-09,
      "loss": 1.1979,
      "step": 16810
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.21875,
      "learning_rate": 6.728970477131169e-09,
      "loss": 1.2296,
      "step": 16815
    },
    {
      "epoch": 2.98,
      "grad_norm": 5.125,
      "learning_rate": 6.151673950854919e-09,
      "loss": 1.2141,
      "step": 16820
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.96875,
      "learning_rate": 5.600261703175846e-09,
      "loss": 1.2413,
      "step": 16825
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.71875,
      "learning_rate": 5.074734209953303e-09,
      "loss": 1.2685,
      "step": 16830
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.65625,
      "learning_rate": 4.575091924707842e-09,
      "loss": 1.2457,
      "step": 16835
    },
    {
      "epoch": 2.98,
      "grad_norm": 5.28125,
      "learning_rate": 4.101335278631213e-09,
      "loss": 1.2232,
      "step": 16840
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.71875,
      "learning_rate": 3.6534646805597147e-09,
      "loss": 1.205,
      "step": 16845
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.625,
      "learning_rate": 3.231480517007501e-09,
      "loss": 1.2043,
      "step": 16850
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.28125,
      "learning_rate": 2.835383152136606e-09,
      "loss": 1.288,
      "step": 16855
    },
    {
      "epoch": 2.99,
      "grad_norm": 5.375,
      "learning_rate": 2.4651729277735955e-09,
      "loss": 1.1982,
      "step": 16860
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.921875,
      "learning_rate": 2.1208501634062405e-09,
      "loss": 1.1977,
      "step": 16865
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.46875,
      "learning_rate": 1.8024151561835123e-09,
      "loss": 1.2298,
      "step": 16870
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.09375,
      "learning_rate": 1.509868180902263e-09,
      "loss": 1.1935,
      "step": 16875
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.78125,
      "learning_rate": 1.2432094900338698e-09,
      "loss": 1.2461,
      "step": 16880
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.953125,
      "learning_rate": 1.0024393136975895e-09,
      "loss": 1.2549,
      "step": 16885
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.125,
      "learning_rate": 7.875578596772126e-10,
      "loss": 1.2171,
      "step": 16890
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.71875,
      "learning_rate": 5.985653134077396e-10,
      "loss": 1.2405,
      "step": 16895
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.40625,
      "learning_rate": 4.35461837992035e-10,
      "loss": 1.2141,
      "step": 16900
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.15625,
      "learning_rate": 2.982475741841739e-10,
      "loss": 1.2289,
      "step": 16905
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.09375,
      "learning_rate": 1.8692264039943397e-10,
      "loss": 1.2016,
      "step": 16910
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.1875,
      "learning_rate": 1.0148713270763387e-10,
      "loss": 1.2719,
      "step": 16915
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.09375,
      "learning_rate": 4.194112483979495e-11,
      "loss": 1.2198,
      "step": 16920
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.59375,
      "learning_rate": 8.284668184810328e-12,
      "loss": 1.2718,
      "step": 16925
    },
    {
      "epoch": 3.0,
      "step": 16929,
      "total_flos": 3.449475439306998e+18,
      "train_loss": 1.257252904005387,
      "train_runtime": 50222.2429,
      "train_samples_per_second": 21.574,
      "train_steps_per_second": 0.337
    }
  ],
  "logging_steps": 5,
  "max_steps": 16929,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 30000,
  "total_flos": 3.449475439306998e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}