{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2076465855114595,
  "eval_steps": 500,
  "global_step": 8000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 28.28061294555664,
      "learning_rate": 5e-06,
      "loss": 13.9306,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 31.21528434753418,
      "learning_rate": 1e-05,
      "loss": 13.7346,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.489715576171875,
      "learning_rate": 1.5e-05,
      "loss": 13.1528,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.441423416137695,
      "learning_rate": 2e-05,
      "loss": 12.0999,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 13.402457237243652,
      "learning_rate": 2.5e-05,
      "loss": 10.8301,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.77131462097168,
      "learning_rate": 3e-05,
      "loss": 9.726,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.1222710609436035,
      "learning_rate": 3.5e-05,
      "loss": 8.9503,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9480485916137695,
      "learning_rate": 4e-05,
      "loss": 8.3955,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5017428398132324,
      "learning_rate": 4.5e-05,
      "loss": 8.0734,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8936816453933716,
      "learning_rate": 5e-05,
      "loss": 7.9277,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9515222311019897,
      "learning_rate": 5.500000000000001e-05,
      "loss": 7.8227,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.311653971672058,
      "learning_rate": 6e-05,
      "loss": 7.7276,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1881475448608398,
      "learning_rate": 6.500000000000001e-05,
      "loss": 7.662,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4207669496536255,
      "learning_rate": 7e-05,
      "loss": 7.4657,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0706592798233032,
      "learning_rate": 7.500000000000001e-05,
      "loss": 7.3549,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1833094358444214,
      "learning_rate": 8e-05,
      "loss": 6.9902,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4383339881896973,
      "learning_rate": 8.5e-05,
      "loss": 6.8129,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1298564672470093,
      "learning_rate": 9e-05,
      "loss": 6.456,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2605056762695312,
      "learning_rate": 9.5e-05,
      "loss": 6.1794,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4736117124557495,
      "learning_rate": 0.0001,
      "loss": 5.8824,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1679015159606934,
      "learning_rate": 9.999999582259295e-05,
      "loss": 5.798,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.441890835762024,
      "learning_rate": 9.999998329037248e-05,
      "loss": 5.5512,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5507267713546753,
      "learning_rate": 9.999996240334068e-05,
      "loss": 5.3694,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4268945455551147,
      "learning_rate": 9.999993316150106e-05,
      "loss": 5.2394,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5284940004348755,
      "learning_rate": 9.999989556485852e-05,
      "loss": 5.107,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4177409410476685,
      "learning_rate": 9.999984961341928e-05,
      "loss": 4.9265,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4462488889694214,
      "learning_rate": 9.999979530719108e-05,
      "loss": 4.9018,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3914940357208252,
      "learning_rate": 9.999973264618297e-05,
      "loss": 4.7486,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.336748480796814,
      "learning_rate": 9.999966163040541e-05,
      "loss": 4.7297,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2854251861572266,
      "learning_rate": 9.99995822598703e-05,
      "loss": 4.5272,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3655463457107544,
      "learning_rate": 9.999949453459088e-05,
      "loss": 4.5451,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3519500494003296,
      "learning_rate": 9.999939845458179e-05,
      "loss": 4.3968,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1749165058135986,
      "learning_rate": 9.999929401985911e-05,
      "loss": 4.5802,
      "step": 165
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3074795007705688,
      "learning_rate": 9.99991812304403e-05,
      "loss": 4.3645,
      "step": 170
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2671828269958496,
      "learning_rate": 9.999906008634416e-05,
      "loss": 4.396,
      "step": 175
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2546554803848267,
      "learning_rate": 9.9998930587591e-05,
      "loss": 4.2651,
      "step": 180
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5431021451950073,
      "learning_rate": 9.99987927342024e-05,
      "loss": 4.2413,
      "step": 185
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4027025699615479,
      "learning_rate": 9.999864652620143e-05,
      "loss": 4.1918,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.257068395614624,
      "learning_rate": 9.999849196361251e-05,
      "loss": 4.1943,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3898354768753052,
      "learning_rate": 9.999832904646147e-05,
      "loss": 4.1811,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.08851957321167,
      "learning_rate": 9.99981577747755e-05,
      "loss": 4.0483,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0463107824325562,
      "learning_rate": 9.999797814858327e-05,
      "loss": 4.0914,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0840775966644287,
      "learning_rate": 9.999779016791477e-05,
      "loss": 4.0261,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0790539979934692,
      "learning_rate": 9.999759383280141e-05,
      "loss": 3.9192,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4761401414871216,
      "learning_rate": 9.9997389143276e-05,
      "loss": 4.0235,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1915615797042847,
      "learning_rate": 9.999717609937275e-05,
      "loss": 4.0038,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4661827087402344,
      "learning_rate": 9.999695470112725e-05,
      "loss": 3.9785,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3795654773712158,
      "learning_rate": 9.99967249485765e-05,
      "loss": 3.9284,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0616819858551025,
      "learning_rate": 9.999648684175888e-05,
      "loss": 3.8677,
      "step": 245
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0591880083084106,
      "learning_rate": 9.999624038071418e-05,
      "loss": 3.864,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9640903472900391,
      "learning_rate": 9.99959855654836e-05,
      "loss": 3.818,
      "step": 255
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0906050205230713,
      "learning_rate": 9.999572239610968e-05,
      "loss": 3.7889,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1482876539230347,
      "learning_rate": 9.999545087263645e-05,
      "loss": 3.8341,
      "step": 265
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.219773769378662,
      "learning_rate": 9.999517099510923e-05,
      "loss": 3.8079,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.071859359741211,
      "learning_rate": 9.999488276357481e-05,
      "loss": 3.7856,
      "step": 275
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2423732280731201,
      "learning_rate": 9.999458617808137e-05,
      "loss": 3.7476,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0655150413513184,
      "learning_rate": 9.999428123867843e-05,
      "loss": 3.8281,
      "step": 285
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0984116792678833,
      "learning_rate": 9.999396794541695e-05,
      "loss": 3.7885,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1136418581008911,
      "learning_rate": 9.999364629834931e-05,
      "loss": 3.719,
      "step": 295
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0645387172698975,
      "learning_rate": 9.999331629752924e-05,
      "loss": 3.7611,
      "step": 300
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.079219102859497,
      "learning_rate": 9.999297794301188e-05,
      "loss": 3.7523,
      "step": 305
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1035035848617554,
      "learning_rate": 9.999263123485376e-05,
      "loss": 3.7359,
      "step": 310
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.066282868385315,
      "learning_rate": 9.999227617311282e-05,
      "loss": 3.7014,
      "step": 315
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3502960205078125,
      "learning_rate": 9.99919127578484e-05,
      "loss": 3.6737,
      "step": 320
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0747475624084473,
      "learning_rate": 9.999154098912122e-05,
      "loss": 3.5929,
      "step": 325
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9328626990318298,
      "learning_rate": 9.999116086699338e-05,
      "loss": 3.7227,
      "step": 330
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9693546891212463,
      "learning_rate": 9.999077239152843e-05,
      "loss": 3.7017,
      "step": 335
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1984975337982178,
      "learning_rate": 9.999037556279127e-05,
      "loss": 3.693,
      "step": 340
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1442183256149292,
      "learning_rate": 9.998997038084818e-05,
      "loss": 3.5945,
      "step": 345
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9692398309707642,
      "learning_rate": 9.998955684576692e-05,
      "loss": 3.5316,
      "step": 350
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4262653589248657,
      "learning_rate": 9.998913495761654e-05,
      "loss": 3.5998,
      "step": 355
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2622092962265015,
      "learning_rate": 9.998870471646757e-05,
      "loss": 3.6043,
      "step": 360
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.882042646408081,
      "learning_rate": 9.99882661223919e-05,
      "loss": 3.6382,
      "step": 365
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.074329137802124,
      "learning_rate": 9.998781917546278e-05,
      "loss": 3.556,
      "step": 370
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1048412322998047,
      "learning_rate": 9.998736387575493e-05,
      "loss": 3.5562,
      "step": 375
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9298127889633179,
      "learning_rate": 9.998690022334442e-05,
      "loss": 3.4964,
      "step": 380
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1893562078475952,
      "learning_rate": 9.998642821830873e-05,
      "loss": 3.5048,
      "step": 385
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0300569534301758,
      "learning_rate": 9.99859478607267e-05,
      "loss": 3.6348,
      "step": 390
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0273958444595337,
      "learning_rate": 9.998545915067864e-05,
      "loss": 3.6249,
      "step": 395
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0632930994033813,
      "learning_rate": 9.998496208824618e-05,
      "loss": 3.5918,
      "step": 400
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0358432531356812,
      "learning_rate": 9.99844566735124e-05,
      "loss": 3.4392,
      "step": 405
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8709058165550232,
      "learning_rate": 9.998394290656172e-05,
      "loss": 3.544,
      "step": 410
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9104147553443909,
      "learning_rate": 9.998342078748002e-05,
      "loss": 3.5493,
      "step": 415
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9561464190483093,
      "learning_rate": 9.998289031635453e-05,
      "loss": 3.4562,
      "step": 420
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.02458655834198,
      "learning_rate": 9.998235149327389e-05,
      "loss": 3.5296,
      "step": 425
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9175445437431335,
      "learning_rate": 9.998180431832816e-05,
      "loss": 3.4571,
      "step": 430
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9985538125038147,
      "learning_rate": 9.998124879160872e-05,
      "loss": 3.5387,
      "step": 435
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9253854751586914,
      "learning_rate": 9.998068491320844e-05,
      "loss": 3.486,
      "step": 440
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9290066957473755,
      "learning_rate": 9.998011268322152e-05,
      "loss": 3.4881,
      "step": 445
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9410489201545715,
      "learning_rate": 9.997953210174361e-05,
      "loss": 3.5567,
      "step": 450
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0354490280151367,
      "learning_rate": 9.997894316887167e-05,
      "loss": 3.4156,
      "step": 455
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0677212476730347,
      "learning_rate": 9.997834588470414e-05,
      "loss": 3.4629,
      "step": 460
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9424716830253601,
      "learning_rate": 9.997774024934082e-05,
      "loss": 3.4913,
      "step": 465
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1338027715682983,
      "learning_rate": 9.997712626288292e-05,
      "loss": 3.4724,
      "step": 470
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9146742224693298,
      "learning_rate": 9.997650392543303e-05,
      "loss": 3.3756,
      "step": 475
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.958629846572876,
      "learning_rate": 9.997587323709512e-05,
      "loss": 3.4509,
      "step": 480
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0038388967514038,
      "learning_rate": 9.997523419797459e-05,
      "loss": 3.4265,
      "step": 485
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9675782918930054,
      "learning_rate": 9.997458680817824e-05,
      "loss": 3.4965,
      "step": 490
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8620448708534241,
      "learning_rate": 9.997393106781421e-05,
      "loss": 3.4828,
      "step": 495
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0572539567947388,
      "learning_rate": 9.99732669769921e-05,
      "loss": 3.4064,
      "step": 500
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9045765399932861,
      "learning_rate": 9.997259453582286e-05,
      "loss": 3.4255,
      "step": 505
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8029064536094666,
      "learning_rate": 9.997191374441887e-05,
      "loss": 3.4101,
      "step": 510
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9975433349609375,
      "learning_rate": 9.997122460289387e-05,
      "loss": 3.4226,
      "step": 515
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.949057936668396,
      "learning_rate": 9.997052711136301e-05,
      "loss": 3.2872,
      "step": 520
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.921711802482605,
      "learning_rate": 9.996982126994287e-05,
      "loss": 3.2754,
      "step": 525
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.923388659954071,
      "learning_rate": 9.996910707875136e-05,
      "loss": 3.401,
      "step": 530
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9914992451667786,
      "learning_rate": 9.996838453790784e-05,
      "loss": 3.4227,
      "step": 535
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.917880117893219,
      "learning_rate": 9.996765364753302e-05,
      "loss": 3.353,
      "step": 540
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.951382040977478,
      "learning_rate": 9.996691440774906e-05,
      "loss": 3.3718,
      "step": 545
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.899810791015625,
      "learning_rate": 9.996616681867945e-05,
      "loss": 3.3347,
      "step": 550
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8892326354980469,
      "learning_rate": 9.996541088044913e-05,
      "loss": 3.3206,
      "step": 555
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8602358102798462,
      "learning_rate": 9.996464659318442e-05,
      "loss": 3.2742,
      "step": 560
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8345521688461304,
      "learning_rate": 9.996387395701302e-05,
      "loss": 3.3675,
      "step": 565
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8011465668678284,
      "learning_rate": 9.996309297206403e-05,
      "loss": 3.312,
      "step": 570
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8164519667625427,
      "learning_rate": 9.996230363846797e-05,
      "loss": 3.3351,
      "step": 575
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8584182858467102,
      "learning_rate": 9.996150595635671e-05,
      "loss": 3.2416,
      "step": 580
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9361007213592529,
      "learning_rate": 9.996069992586355e-05,
      "loss": 3.2793,
      "step": 585
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8868554830551147,
      "learning_rate": 9.995988554712318e-05,
      "loss": 3.363,
      "step": 590
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9293203949928284,
      "learning_rate": 9.995906282027166e-05,
      "loss": 3.2911,
      "step": 595
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8787782192230225,
      "learning_rate": 9.995823174544649e-05,
      "loss": 3.3131,
      "step": 600
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8572219014167786,
      "learning_rate": 9.995739232278654e-05,
      "loss": 3.3307,
      "step": 605
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8700346946716309,
      "learning_rate": 9.995654455243204e-05,
      "loss": 3.3204,
      "step": 610
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8289816379547119,
      "learning_rate": 9.995568843452468e-05,
      "loss": 3.3191,
      "step": 615
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8450701236724854,
      "learning_rate": 9.99548239692075e-05,
      "loss": 3.3045,
      "step": 620
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7924940586090088,
      "learning_rate": 9.995395115662498e-05,
      "loss": 3.3449,
      "step": 625
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8301188945770264,
      "learning_rate": 9.995306999692293e-05,
      "loss": 3.3513,
      "step": 630
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8892745971679688,
      "learning_rate": 9.995218049024858e-05,
      "loss": 3.2963,
      "step": 635
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8111772537231445,
      "learning_rate": 9.995128263675058e-05,
      "loss": 3.1885,
      "step": 640
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8073171377182007,
      "learning_rate": 9.995037643657899e-05,
      "loss": 3.2304,
      "step": 645
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8065031170845032,
      "learning_rate": 9.994946188988516e-05,
      "loss": 3.223,
      "step": 650
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8282549977302551,
      "learning_rate": 9.994853899682197e-05,
      "loss": 3.2635,
      "step": 655
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7960890531539917,
      "learning_rate": 9.994760775754362e-05,
      "loss": 3.2334,
      "step": 660
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.799068808555603,
      "learning_rate": 9.994666817220571e-05,
      "loss": 3.1983,
      "step": 665
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8036028742790222,
      "learning_rate": 9.994572024096521e-05,
      "loss": 3.2184,
      "step": 670
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7873517274856567,
      "learning_rate": 9.994476396398055e-05,
      "loss": 3.1984,
      "step": 675
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0229746103286743,
      "learning_rate": 9.994379934141153e-05,
      "loss": 3.2116,
      "step": 680
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.801454484462738,
      "learning_rate": 9.994282637341932e-05,
      "loss": 3.245,
      "step": 685
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8292471170425415,
      "learning_rate": 9.994184506016651e-05,
      "loss": 3.2506,
      "step": 690
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7844598293304443,
      "learning_rate": 9.994085540181703e-05,
      "loss": 3.2664,
      "step": 695
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7493610382080078,
      "learning_rate": 9.99398573985363e-05,
      "loss": 3.2781,
      "step": 700
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7480127811431885,
      "learning_rate": 9.993885105049107e-05,
      "loss": 3.2788,
      "step": 705
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8006333112716675,
      "learning_rate": 9.993783635784947e-05,
      "loss": 3.222,
      "step": 710
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7927700877189636,
      "learning_rate": 9.993681332078108e-05,
      "loss": 3.2568,
      "step": 715
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8411673903465271,
      "learning_rate": 9.993578193945685e-05,
      "loss": 3.1995,
      "step": 720
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8330583572387695,
      "learning_rate": 9.99347422140491e-05,
      "loss": 3.2442,
      "step": 725
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7955582737922668,
      "learning_rate": 9.993369414473157e-05,
      "loss": 3.2495,
      "step": 730
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7797298431396484,
      "learning_rate": 9.993263773167939e-05,
      "loss": 3.2071,
      "step": 735
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.899322509765625,
      "learning_rate": 9.993157297506907e-05,
      "loss": 3.2912,
      "step": 740
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8390113711357117,
      "learning_rate": 9.993049987507856e-05,
      "loss": 3.1649,
      "step": 745
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7818528413772583,
      "learning_rate": 9.992941843188715e-05,
      "loss": 3.1865,
      "step": 750
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9829023480415344,
      "learning_rate": 9.992832864567554e-05,
      "loss": 3.1859,
      "step": 755
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8056334853172302,
      "learning_rate": 9.992723051662584e-05,
      "loss": 3.2312,
      "step": 760
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7419295907020569,
      "learning_rate": 9.992612404492153e-05,
      "loss": 3.2513,
      "step": 765
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8985966444015503,
      "learning_rate": 9.992500923074752e-05,
      "loss": 3.2025,
      "step": 770
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8233122825622559,
      "learning_rate": 9.992388607429006e-05,
      "loss": 3.1617,
      "step": 775
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8093167543411255,
      "learning_rate": 9.992275457573685e-05,
      "loss": 3.1838,
      "step": 780
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.888569712638855,
      "learning_rate": 9.992161473527695e-05,
      "loss": 3.2239,
      "step": 785
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7357171773910522,
      "learning_rate": 9.992046655310083e-05,
      "loss": 3.2421,
      "step": 790
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.75666743516922,
      "learning_rate": 9.991931002940034e-05,
      "loss": 3.1336,
      "step": 795
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8012424111366272,
      "learning_rate": 9.991814516436873e-05,
      "loss": 3.1684,
      "step": 800
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.744608998298645,
      "learning_rate": 9.991697195820065e-05,
      "loss": 3.1159,
      "step": 805
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8142192959785461,
      "learning_rate": 9.991579041109212e-05,
      "loss": 3.1412,
      "step": 810
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8927745223045349,
      "learning_rate": 9.991460052324061e-05,
      "loss": 3.2349,
      "step": 815
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8298518657684326,
      "learning_rate": 9.991340229484491e-05,
      "loss": 3.2343,
      "step": 820
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9085060358047485,
      "learning_rate": 9.991219572610526e-05,
      "loss": 3.2153,
      "step": 825
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.78221195936203,
      "learning_rate": 9.991098081722325e-05,
      "loss": 3.1436,
      "step": 830
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7076947093009949,
      "learning_rate": 9.990975756840193e-05,
      "loss": 3.1702,
      "step": 835
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7203157544136047,
      "learning_rate": 9.990852597984566e-05,
      "loss": 3.1828,
      "step": 840
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6831102967262268,
      "learning_rate": 9.990728605176025e-05,
      "loss": 3.1169,
      "step": 845
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8549178838729858,
      "learning_rate": 9.990603778435288e-05,
      "loss": 3.2031,
      "step": 850
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6966831684112549,
      "learning_rate": 9.990478117783214e-05,
      "loss": 3.1733,
      "step": 855
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8442224264144897,
      "learning_rate": 9.990351623240799e-05,
      "loss": 3.1511,
      "step": 860
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.71586674451828,
      "learning_rate": 9.990224294829181e-05,
      "loss": 3.1081,
      "step": 865
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8470659852027893,
      "learning_rate": 9.990096132569636e-05,
      "loss": 3.1643,
      "step": 870
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8034508228302002,
      "learning_rate": 9.98996713648358e-05,
      "loss": 3.2096,
      "step": 875
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7270082235336304,
      "learning_rate": 9.989837306592567e-05,
      "loss": 3.1837,
      "step": 880
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.812780499458313,
      "learning_rate": 9.989706642918291e-05,
      "loss": 3.1827,
      "step": 885
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7902137637138367,
      "learning_rate": 9.989575145482583e-05,
      "loss": 3.206,
      "step": 890
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6843730807304382,
      "learning_rate": 9.989442814307422e-05,
      "loss": 3.1825,
      "step": 895
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7117285132408142,
      "learning_rate": 9.989309649414914e-05,
      "loss": 3.0446,
      "step": 900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.730373740196228,
      "learning_rate": 9.989175650827314e-05,
      "loss": 3.0845,
      "step": 905
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.684935450553894,
      "learning_rate": 9.98904081856701e-05,
      "loss": 3.1005,
      "step": 910
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7128190994262695,
      "learning_rate": 9.988905152656534e-05,
      "loss": 3.1551,
      "step": 915
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7475801110267639,
      "learning_rate": 9.988768653118554e-05,
      "loss": 3.0456,
      "step": 920
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7273353934288025,
      "learning_rate": 9.988631319975881e-05,
      "loss": 3.231,
      "step": 925
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7224611043930054,
      "learning_rate": 9.988493153251459e-05,
      "loss": 3.0323,
      "step": 930
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7919772267341614,
      "learning_rate": 9.988354152968379e-05,
      "loss": 3.1056,
      "step": 935
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7950297594070435,
      "learning_rate": 9.988214319149865e-05,
      "loss": 3.1195,
      "step": 940
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7634941339492798,
      "learning_rate": 9.988073651819281e-05,
      "loss": 3.1204,
      "step": 945
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.733025074005127,
      "learning_rate": 9.987932151000137e-05,
      "loss": 3.0611,
      "step": 950
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7753676772117615,
      "learning_rate": 9.987789816716073e-05,
      "loss": 3.1784,
      "step": 955
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7051792740821838,
      "learning_rate": 9.987646648990876e-05,
      "loss": 3.0777,
      "step": 960
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7380036115646362,
      "learning_rate": 9.987502647848466e-05,
      "loss": 2.9928,
      "step": 965
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.677627444267273,
      "learning_rate": 9.987357813312905e-05,
      "loss": 3.1447,
      "step": 970
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7460645437240601,
      "learning_rate": 9.987212145408395e-05,
      "loss": 3.1194,
      "step": 975
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7605148553848267,
      "learning_rate": 9.987065644159278e-05,
      "loss": 3.0497,
      "step": 980
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7291200757026672,
      "learning_rate": 9.986918309590031e-05,
      "loss": 3.1405,
      "step": 985
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7350710034370422,
      "learning_rate": 9.986770141725277e-05,
      "loss": 3.0642,
      "step": 990
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6934482455253601,
      "learning_rate": 9.986621140589772e-05,
      "loss": 3.104,
      "step": 995
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6814989447593689,
      "learning_rate": 9.986471306208412e-05,
      "loss": 3.1211,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.719136655330658,
      "learning_rate": 9.986320638606235e-05,
      "loss": 3.1451,
      "step": 1005
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6745666265487671,
      "learning_rate": 9.986169137808419e-05,
      "loss": 3.0977,
      "step": 1010
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6611875295639038,
      "learning_rate": 9.986016803840275e-05,
      "loss": 3.1218,
      "step": 1015
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6404609680175781,
      "learning_rate": 9.985863636727262e-05,
      "loss": 3.0448,
      "step": 1020
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.716444194316864,
      "learning_rate": 9.985709636494971e-05,
      "loss": 3.092,
      "step": 1025
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7317157983779907,
      "learning_rate": 9.985554803169134e-05,
      "loss": 3.1329,
      "step": 1030
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7424524426460266,
      "learning_rate": 9.985399136775627e-05,
      "loss": 3.0947,
      "step": 1035
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7191465497016907,
      "learning_rate": 9.985242637340458e-05,
      "loss": 3.1214,
      "step": 1040
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6245012283325195,
      "learning_rate": 9.985085304889778e-05,
      "loss": 3.0382,
      "step": 1045
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6838718056678772,
      "learning_rate": 9.984927139449877e-05,
      "loss": 3.0203,
      "step": 1050
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7570410370826721,
      "learning_rate": 9.984768141047184e-05,
      "loss": 3.1188,
      "step": 1055
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6490225195884705,
      "learning_rate": 9.984608309708266e-05,
      "loss": 3.0522,
      "step": 1060
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7238661050796509,
      "learning_rate": 9.984447645459831e-05,
      "loss": 3.0832,
      "step": 1065
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7293949723243713,
      "learning_rate": 9.984286148328728e-05,
      "loss": 3.1248,
      "step": 1070
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7428945899009705,
      "learning_rate": 9.984123818341937e-05,
      "loss": 3.0261,
      "step": 1075
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7957409024238586,
      "learning_rate": 9.983960655526587e-05,
      "loss": 2.9772,
      "step": 1080
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6786031126976013,
      "learning_rate": 9.983796659909941e-05,
      "loss": 3.1094,
      "step": 1085
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7214710116386414,
      "learning_rate": 9.983631831519401e-05,
      "loss": 3.0555,
      "step": 1090
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7656241655349731,
      "learning_rate": 9.983466170382511e-05,
      "loss": 3.0879,
      "step": 1095
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6677206158638,
      "learning_rate": 9.983299676526949e-05,
      "loss": 3.0881,
      "step": 1100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7328168153762817,
      "learning_rate": 9.98313234998054e-05,
      "loss": 3.0394,
      "step": 1105
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8511549830436707,
      "learning_rate": 9.982964190771242e-05,
      "loss": 2.9994,
      "step": 1110
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7031623125076294,
      "learning_rate": 9.98279519892715e-05,
      "loss": 2.9641,
      "step": 1115
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.696881115436554,
      "learning_rate": 9.982625374476508e-05,
      "loss": 3.005,
      "step": 1120
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6312974691390991,
      "learning_rate": 9.98245471744769e-05,
      "loss": 3.0371,
      "step": 1125
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6982842087745667,
      "learning_rate": 9.982283227869211e-05,
      "loss": 3.0518,
      "step": 1130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6250959038734436,
      "learning_rate": 9.982110905769729e-05,
      "loss": 2.9928,
      "step": 1135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7204066514968872,
      "learning_rate": 9.981937751178037e-05,
      "loss": 2.9862,
      "step": 1140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6655579209327698,
      "learning_rate": 9.981763764123067e-05,
      "loss": 3.0563,
      "step": 1145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7126772999763489,
      "learning_rate": 9.981588944633893e-05,
      "loss": 3.0473,
      "step": 1150
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7115939259529114,
      "learning_rate": 9.981413292739727e-05,
      "loss": 2.9784,
      "step": 1155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7699306011199951,
      "learning_rate": 9.981236808469922e-05,
      "loss": 3.0088,
      "step": 1160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6607645750045776,
      "learning_rate": 9.981059491853964e-05,
      "loss": 3.0637,
      "step": 1165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6772843599319458,
      "learning_rate": 9.980881342921482e-05,
      "loss": 3.0459,
      "step": 1170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6567750573158264,
      "learning_rate": 9.980702361702246e-05,
      "loss": 3.0467,
      "step": 1175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6658007502555847,
      "learning_rate": 9.980522548226162e-05,
      "loss": 3.0533,
      "step": 1180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6799459457397461,
      "learning_rate": 9.980341902523279e-05,
      "loss": 3.0815,
      "step": 1185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6638361215591431,
      "learning_rate": 9.980160424623776e-05,
      "loss": 3.0846,
      "step": 1190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6707638502120972,
      "learning_rate": 9.979978114557984e-05,
      "loss": 2.9682,
      "step": 1195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6861352324485779,
      "learning_rate": 9.979794972356362e-05,
      "loss": 3.1166,
      "step": 1200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6533607244491577,
      "learning_rate": 9.979610998049516e-05,
      "loss": 3.0,
      "step": 1205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6316712498664856,
      "learning_rate": 9.979426191668182e-05,
      "loss": 2.9613,
      "step": 1210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8050622940063477,
      "learning_rate": 9.979240553243246e-05,
      "loss": 2.9386,
      "step": 1215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7335979342460632,
      "learning_rate": 9.979054082805726e-05,
      "loss": 3.041,
      "step": 1220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6639100313186646,
      "learning_rate": 9.978866780386777e-05,
      "loss": 2.9964,
      "step": 1225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7860399484634399,
      "learning_rate": 9.978678646017701e-05,
      "loss": 3.0143,
      "step": 1230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6329216361045837,
      "learning_rate": 9.978489679729934e-05,
      "loss": 2.9594,
      "step": 1235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8110703825950623,
      "learning_rate": 9.978299881555048e-05,
      "loss": 3.0195,
      "step": 1240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6823025345802307,
      "learning_rate": 9.97810925152476e-05,
      "loss": 3.0581,
      "step": 1245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6437945365905762,
      "learning_rate": 9.977917789670925e-05,
      "loss": 3.0165,
      "step": 1250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6594268083572388,
      "learning_rate": 9.977725496025535e-05,
      "loss": 2.9902,
      "step": 1255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6579127311706543,
      "learning_rate": 9.977532370620718e-05,
      "loss": 2.988,
      "step": 1260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6607731580734253,
      "learning_rate": 9.977338413488748e-05,
      "loss": 2.9516,
      "step": 1265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7003730535507202,
      "learning_rate": 9.977143624662035e-05,
      "loss": 2.951,
      "step": 1270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7102972269058228,
      "learning_rate": 9.976948004173125e-05,
      "loss": 2.9996,
      "step": 1275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6097055673599243,
      "learning_rate": 9.976751552054706e-05,
      "loss": 2.9934,
      "step": 1280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6477869153022766,
      "learning_rate": 9.976554268339607e-05,
      "loss": 2.9666,
      "step": 1285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6605212092399597,
      "learning_rate": 9.97635615306079e-05,
      "loss": 3.0396,
      "step": 1290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7195318937301636,
      "learning_rate": 9.976157206251361e-05,
      "loss": 2.9655,
      "step": 1295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7135999798774719,
      "learning_rate": 9.975957427944563e-05,
      "loss": 2.9693,
      "step": 1300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7039851546287537,
      "learning_rate": 9.975756818173778e-05,
      "loss": 3.0886,
      "step": 1305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7402403354644775,
      "learning_rate": 9.975555376972529e-05,
      "loss": 3.0293,
      "step": 1310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6558573842048645,
      "learning_rate": 9.975353104374474e-05,
      "loss": 3.0379,
      "step": 1315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.644749641418457,
      "learning_rate": 9.975150000413411e-05,
      "loss": 3.0074,
      "step": 1320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6222659349441528,
      "learning_rate": 9.97494606512328e-05,
      "loss": 2.9296,
      "step": 1325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6269394159317017,
      "learning_rate": 9.974741298538158e-05,
      "loss": 2.9583,
      "step": 1330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6192606687545776,
      "learning_rate": 9.974535700692257e-05,
      "loss": 2.9885,
      "step": 1335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6499235033988953,
      "learning_rate": 9.974329271619938e-05,
      "loss": 2.9639,
      "step": 1340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6857834458351135,
      "learning_rate": 9.97412201135569e-05,
      "loss": 2.9862,
      "step": 1345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6521609425544739,
      "learning_rate": 9.973913919934147e-05,
      "loss": 3.0181,
      "step": 1350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6418082118034363,
      "learning_rate": 9.973704997390079e-05,
      "loss": 2.9775,
      "step": 1355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6669169664382935,
      "learning_rate": 9.973495243758398e-05,
      "loss": 2.9623,
      "step": 1360
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6935232281684875,
      "learning_rate": 9.973284659074151e-05,
      "loss": 3.0217,
      "step": 1365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6414822936058044,
      "learning_rate": 9.973073243372528e-05,
      "loss": 2.901,
      "step": 1370
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6140183806419373,
      "learning_rate": 9.972860996688853e-05,
      "loss": 2.961,
      "step": 1375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6861962080001831,
      "learning_rate": 9.972647919058595e-05,
      "loss": 3.0122,
      "step": 1380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7161067128181458,
      "learning_rate": 9.972434010517358e-05,
      "loss": 2.9574,
      "step": 1385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.695543646812439,
      "learning_rate": 9.972219271100882e-05,
      "loss": 3.0453,
      "step": 1390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6502649784088135,
      "learning_rate": 9.972003700845052e-05,
      "loss": 3.0409,
      "step": 1395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6028346419334412,
      "learning_rate": 9.971787299785888e-05,
      "loss": 2.9599,
      "step": 1400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7154777646064758,
      "learning_rate": 9.97157006795955e-05,
      "loss": 2.9305,
      "step": 1405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6665739417076111,
      "learning_rate": 9.971352005402338e-05,
      "loss": 3.0172,
      "step": 1410
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6865460276603699,
      "learning_rate": 9.971133112150688e-05,
      "loss": 2.9361,
      "step": 1415
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6314422488212585,
      "learning_rate": 9.970913388241177e-05,
      "loss": 3.0167,
      "step": 1420
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6168122887611389,
      "learning_rate": 9.970692833710517e-05,
      "loss": 3.0098,
      "step": 1425
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6021905541419983,
      "learning_rate": 9.970471448595566e-05,
      "loss": 2.9467,
      "step": 1430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.683683454990387,
      "learning_rate": 9.970249232933315e-05,
      "loss": 2.9967,
      "step": 1435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6876024603843689,
      "learning_rate": 9.970026186760896e-05,
      "loss": 2.9991,
      "step": 1440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.679817259311676,
      "learning_rate": 9.969802310115577e-05,
      "loss": 2.8971,
      "step": 1445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6041108965873718,
      "learning_rate": 9.96957760303477e-05,
      "loss": 2.9687,
      "step": 1450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5891430377960205,
      "learning_rate": 9.969352065556022e-05,
      "loss": 2.95,
      "step": 1455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.608013391494751,
      "learning_rate": 9.969125697717017e-05,
      "loss": 3.0311,
      "step": 1460
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6581689715385437,
      "learning_rate": 9.968898499555582e-05,
      "loss": 2.9714,
      "step": 1465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6210693717002869,
      "learning_rate": 9.968670471109682e-05,
      "loss": 2.9347,
      "step": 1470
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6091535687446594,
      "learning_rate": 9.968441612417418e-05,
      "loss": 2.9703,
      "step": 1475
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6723106503486633,
      "learning_rate": 9.968211923517032e-05,
      "loss": 2.9132,
      "step": 1480
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6349649429321289,
      "learning_rate": 9.967981404446905e-05,
      "loss": 2.9389,
      "step": 1485
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6214143633842468,
      "learning_rate": 9.967750055245555e-05,
      "loss": 2.9583,
      "step": 1490
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.688920795917511,
      "learning_rate": 9.967517875951638e-05,
      "loss": 2.8977,
      "step": 1495
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6266434788703918,
      "learning_rate": 9.967284866603955e-05,
      "loss": 2.9175,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6625579595565796,
      "learning_rate": 9.967051027241436e-05,
      "loss": 2.9178,
      "step": 1505
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6194829344749451,
      "learning_rate": 9.966816357903155e-05,
      "loss": 2.921,
      "step": 1510
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6422644257545471,
      "learning_rate": 9.96658085862833e-05,
      "loss": 2.9717,
      "step": 1515
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.704857349395752,
      "learning_rate": 9.966344529456305e-05,
      "loss": 2.9047,
      "step": 1520
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6433600187301636,
      "learning_rate": 9.966107370426573e-05,
      "loss": 2.9399,
      "step": 1525
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6574586629867554,
      "learning_rate": 9.965869381578765e-05,
      "loss": 2.9416,
      "step": 1530
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6716108918190002,
      "learning_rate": 9.965630562952643e-05,
      "loss": 2.9253,
      "step": 1535
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6619248986244202,
      "learning_rate": 9.965390914588115e-05,
      "loss": 2.952,
      "step": 1540
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6187171936035156,
      "learning_rate": 9.965150436525225e-05,
      "loss": 2.8605,
      "step": 1545
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.600641131401062,
      "learning_rate": 9.964909128804159e-05,
      "loss": 2.9392,
      "step": 1550
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6540449261665344,
      "learning_rate": 9.964666991465234e-05,
      "loss": 2.9359,
      "step": 1555
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5999405384063721,
      "learning_rate": 9.964424024548914e-05,
      "loss": 2.9122,
      "step": 1560
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5879953503608704,
      "learning_rate": 9.964180228095793e-05,
      "loss": 2.9091,
      "step": 1565
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5614463686943054,
      "learning_rate": 9.963935602146612e-05,
      "loss": 2.9382,
      "step": 1570
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5792456865310669,
      "learning_rate": 9.963690146742248e-05,
      "loss": 2.9064,
      "step": 1575
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5562976002693176,
      "learning_rate": 9.963443861923715e-05,
      "loss": 2.8788,
      "step": 1580
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6366747617721558,
      "learning_rate": 9.963196747732164e-05,
      "loss": 2.8879,
      "step": 1585
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5753471851348877,
      "learning_rate": 9.96294880420889e-05,
      "loss": 2.8831,
      "step": 1590
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6217545866966248,
      "learning_rate": 9.96270003139532e-05,
      "loss": 2.9442,
      "step": 1595
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6165201663970947,
      "learning_rate": 9.962450429333026e-05,
      "loss": 2.9426,
      "step": 1600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5970577597618103,
      "learning_rate": 9.962199998063715e-05,
      "loss": 2.8635,
      "step": 1605
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5477282404899597,
      "learning_rate": 9.961948737629231e-05,
      "loss": 2.9228,
      "step": 1610
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5969791412353516,
      "learning_rate": 9.961696648071561e-05,
      "loss": 2.8945,
      "step": 1615
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6815431714057922,
      "learning_rate": 9.961443729432828e-05,
      "loss": 2.8612,
      "step": 1620
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6395023465156555,
      "learning_rate": 9.961189981755294e-05,
      "loss": 2.909,
      "step": 1625
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6199893355369568,
      "learning_rate": 9.960935405081357e-05,
      "loss": 2.8683,
      "step": 1630
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6070787906646729,
      "learning_rate": 9.960679999453559e-05,
      "loss": 2.9038,
      "step": 1635
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5745013356208801,
      "learning_rate": 9.960423764914575e-05,
      "loss": 2.9384,
      "step": 1640
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.621881365776062,
      "learning_rate": 9.960166701507221e-05,
      "loss": 2.9696,
      "step": 1645
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5809531211853027,
      "learning_rate": 9.959908809274453e-05,
      "loss": 2.9169,
      "step": 1650
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6392835974693298,
      "learning_rate": 9.959650088259362e-05,
      "loss": 2.8514,
      "step": 1655
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5887923836708069,
      "learning_rate": 9.959390538505181e-05,
      "loss": 2.9087,
      "step": 1660
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.602601170539856,
      "learning_rate": 9.959130160055279e-05,
      "loss": 2.8842,
      "step": 1665
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6565619111061096,
      "learning_rate": 9.958868952953163e-05,
      "loss": 2.9192,
      "step": 1670
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5655417442321777,
      "learning_rate": 9.95860691724248e-05,
      "loss": 2.8756,
      "step": 1675
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6591267585754395,
      "learning_rate": 9.958344052967018e-05,
      "loss": 2.8446,
      "step": 1680
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5729162693023682,
      "learning_rate": 9.958080360170698e-05,
      "loss": 2.8645,
      "step": 1685
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6372408866882324,
      "learning_rate": 9.957815838897585e-05,
      "loss": 2.8237,
      "step": 1690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5950170755386353,
      "learning_rate": 9.957550489191874e-05,
      "loss": 2.8355,
      "step": 1695
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6346689462661743,
      "learning_rate": 9.957284311097908e-05,
      "loss": 2.9415,
      "step": 1700
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6089036464691162,
      "learning_rate": 9.957017304660164e-05,
      "loss": 2.8608,
      "step": 1705
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5835061073303223,
      "learning_rate": 9.956749469923258e-05,
      "loss": 2.8758,
      "step": 1710
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5515388250350952,
      "learning_rate": 9.956480806931942e-05,
      "loss": 2.8003,
      "step": 1715
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5820745229721069,
      "learning_rate": 9.956211315731112e-05,
      "loss": 2.9871,
      "step": 1720
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5594883561134338,
      "learning_rate": 9.955940996365796e-05,
      "loss": 2.8379,
      "step": 1725
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6081607937812805,
      "learning_rate": 9.955669848881166e-05,
      "loss": 2.8462,
      "step": 1730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6159921884536743,
      "learning_rate": 9.955397873322527e-05,
      "loss": 2.842,
      "step": 1735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6121153831481934,
      "learning_rate": 9.955125069735327e-05,
      "loss": 2.8814,
      "step": 1740
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6904382109642029,
      "learning_rate": 9.95485143816515e-05,
      "loss": 2.9387,
      "step": 1745
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6267746090888977,
      "learning_rate": 9.95457697865772e-05,
      "loss": 2.8047,
      "step": 1750
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6184878945350647,
      "learning_rate": 9.954301691258894e-05,
      "loss": 2.8169,
      "step": 1755
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6034075021743774,
      "learning_rate": 9.954025576014674e-05,
      "loss": 2.9036,
      "step": 1760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6342995762825012,
      "learning_rate": 9.953748632971201e-05,
      "loss": 2.8179,
      "step": 1765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5981643199920654,
      "learning_rate": 9.953470862174748e-05,
      "loss": 2.9202,
      "step": 1770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5917516946792603,
      "learning_rate": 9.953192263671728e-05,
      "loss": 2.8991,
      "step": 1775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6334536075592041,
      "learning_rate": 9.952912837508697e-05,
      "loss": 2.8505,
      "step": 1780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5946343541145325,
      "learning_rate": 9.952632583732343e-05,
      "loss": 2.8333,
      "step": 1785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5668966770172119,
      "learning_rate": 9.952351502389498e-05,
      "loss": 2.9412,
      "step": 1790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5350172519683838,
      "learning_rate": 9.95206959352713e-05,
      "loss": 2.8654,
      "step": 1795
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6010748147964478,
      "learning_rate": 9.951786857192341e-05,
      "loss": 2.7936,
      "step": 1800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5743009448051453,
      "learning_rate": 9.951503293432378e-05,
      "loss": 2.8542,
      "step": 1805
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5684875249862671,
      "learning_rate": 9.951218902294625e-05,
      "loss": 2.8954,
      "step": 1810
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6118427515029907,
      "learning_rate": 9.9509336838266e-05,
      "loss": 2.7939,
      "step": 1815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5584838390350342,
      "learning_rate": 9.950647638075963e-05,
      "loss": 2.8456,
      "step": 1820
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5847982168197632,
      "learning_rate": 9.950360765090511e-05,
      "loss": 2.8305,
      "step": 1825
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6427006721496582,
      "learning_rate": 9.950073064918179e-05,
      "loss": 2.8014,
      "step": 1830
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5930681824684143,
      "learning_rate": 9.949784537607042e-05,
      "loss": 2.8446,
      "step": 1835
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6411857604980469,
      "learning_rate": 9.949495183205311e-05,
      "loss": 2.8655,
      "step": 1840
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6691299080848694,
      "learning_rate": 9.949205001761334e-05,
      "loss": 3.0106,
      "step": 1845
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5755900144577026,
      "learning_rate": 9.948913993323603e-05,
      "loss": 2.8584,
      "step": 1850
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.507483720779419,
      "learning_rate": 9.948622157940744e-05,
      "loss": 2.8202,
      "step": 1855
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6483911871910095,
      "learning_rate": 9.948329495661517e-05,
      "loss": 2.8245,
      "step": 1860
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5669857263565063,
      "learning_rate": 9.94803600653483e-05,
      "loss": 2.9178,
      "step": 1865
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5601768493652344,
      "learning_rate": 9.947741690609722e-05,
      "loss": 2.9037,
      "step": 1870
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5987757444381714,
      "learning_rate": 9.94744654793537e-05,
      "loss": 2.8606,
      "step": 1875
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6261595487594604,
      "learning_rate": 9.947150578561095e-05,
      "loss": 2.867,
      "step": 1880
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5849186182022095,
      "learning_rate": 9.946853782536351e-05,
      "loss": 2.8479,
      "step": 1885
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5586771368980408,
      "learning_rate": 9.946556159910732e-05,
      "loss": 2.8219,
      "step": 1890
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5895556807518005,
      "learning_rate": 9.946257710733966e-05,
      "loss": 2.8818,
      "step": 1895
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6172555685043335,
      "learning_rate": 9.945958435055927e-05,
      "loss": 2.9164,
      "step": 1900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6139718294143677,
      "learning_rate": 9.945658332926622e-05,
      "loss": 2.8453,
      "step": 1905
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5434176325798035,
      "learning_rate": 9.945357404396197e-05,
      "loss": 2.7844,
      "step": 1910
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5751949548721313,
      "learning_rate": 9.945055649514934e-05,
      "loss": 2.9183,
      "step": 1915
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5705636739730835,
      "learning_rate": 9.944753068333256e-05,
      "loss": 2.8565,
      "step": 1920
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5485510230064392,
      "learning_rate": 9.944449660901725e-05,
      "loss": 2.7568,
      "step": 1925
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.581463098526001,
      "learning_rate": 9.944145427271036e-05,
      "loss": 2.8766,
      "step": 1930
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5883970260620117,
      "learning_rate": 9.94384036749203e-05,
      "loss": 2.8213,
      "step": 1935
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6636074185371399,
      "learning_rate": 9.943534481615677e-05,
      "loss": 2.8651,
      "step": 1940
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5570939183235168,
      "learning_rate": 9.943227769693091e-05,
      "loss": 2.8054,
      "step": 1945
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5394867658615112,
      "learning_rate": 9.942920231775524e-05,
      "loss": 2.8336,
      "step": 1950
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6146237850189209,
      "learning_rate": 9.942611867914363e-05,
      "loss": 2.9048,
      "step": 1955
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5577776432037354,
      "learning_rate": 9.942302678161133e-05,
      "loss": 2.8261,
      "step": 1960
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5631608963012695,
      "learning_rate": 9.9419926625675e-05,
      "loss": 2.8453,
      "step": 1965
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6147571206092834,
      "learning_rate": 9.941681821185265e-05,
      "loss": 2.8382,
      "step": 1970
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5628846287727356,
      "learning_rate": 9.941370154066371e-05,
      "loss": 2.8356,
      "step": 1975
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5391845107078552,
      "learning_rate": 9.941057661262895e-05,
      "loss": 2.7942,
      "step": 1980
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5695251226425171,
      "learning_rate": 9.940744342827055e-05,
      "loss": 2.8991,
      "step": 1985
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.580085277557373,
      "learning_rate": 9.940430198811202e-05,
      "loss": 2.8737,
      "step": 1990
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6099917888641357,
      "learning_rate": 9.940115229267831e-05,
      "loss": 2.7844,
      "step": 1995
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5811432003974915,
      "learning_rate": 9.939799434249571e-05,
      "loss": 2.9515,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5743013024330139,
      "learning_rate": 9.93948281380919e-05,
      "loss": 2.8677,
      "step": 2005
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5809677839279175,
      "learning_rate": 9.939165367999597e-05,
      "loss": 2.7877,
      "step": 2010
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5865479111671448,
      "learning_rate": 9.938847096873831e-05,
      "loss": 2.8124,
      "step": 2015
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6373860836029053,
      "learning_rate": 9.938528000485078e-05,
      "loss": 2.8837,
      "step": 2020
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6045960187911987,
      "learning_rate": 9.938208078886655e-05,
      "loss": 2.8373,
      "step": 2025
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.551676869392395,
      "learning_rate": 9.937887332132023e-05,
      "loss": 2.8256,
      "step": 2030
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5304166078567505,
      "learning_rate": 9.937565760274776e-05,
      "loss": 2.7803,
      "step": 2035
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.542864203453064,
      "learning_rate": 9.937243363368645e-05,
      "loss": 2.8068,
      "step": 2040
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6012936234474182,
      "learning_rate": 9.936920141467505e-05,
      "loss": 2.7733,
      "step": 2045
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5768609642982483,
      "learning_rate": 9.936596094625364e-05,
      "loss": 2.8828,
      "step": 2050
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5563164949417114,
      "learning_rate": 9.936271222896367e-05,
      "loss": 2.8259,
      "step": 2055
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5481945872306824,
      "learning_rate": 9.935945526334803e-05,
      "loss": 2.881,
      "step": 2060
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5275861620903015,
      "learning_rate": 9.935619004995089e-05,
      "loss": 2.8387,
      "step": 2065
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5580055117607117,
      "learning_rate": 9.935291658931791e-05,
      "loss": 2.7523,
      "step": 2070
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5665960907936096,
      "learning_rate": 9.934963488199604e-05,
      "loss": 2.8775,
      "step": 2075
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6000376343727112,
      "learning_rate": 9.934634492853366e-05,
      "loss": 2.8525,
      "step": 2080
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5646090507507324,
      "learning_rate": 9.93430467294805e-05,
      "loss": 2.9046,
      "step": 2085
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5378292798995972,
      "learning_rate": 9.933974028538768e-05,
      "loss": 2.7721,
      "step": 2090
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5544887185096741,
      "learning_rate": 9.933642559680769e-05,
      "loss": 2.8671,
      "step": 2095
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5517820715904236,
      "learning_rate": 9.93331026642944e-05,
      "loss": 2.782,
      "step": 2100
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5245830416679382,
      "learning_rate": 9.932977148840307e-05,
      "loss": 2.8442,
      "step": 2105
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6178946495056152,
      "learning_rate": 9.93264320696903e-05,
      "loss": 2.7311,
      "step": 2110
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5883631706237793,
      "learning_rate": 9.932308440871413e-05,
      "loss": 2.8194,
      "step": 2115
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5216754674911499,
      "learning_rate": 9.931972850603394e-05,
      "loss": 2.8557,
      "step": 2120
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5626911520957947,
      "learning_rate": 9.931636436221048e-05,
      "loss": 2.7914,
      "step": 2125
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5545584559440613,
      "learning_rate": 9.931299197780586e-05,
      "loss": 2.891,
      "step": 2130
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5447973608970642,
      "learning_rate": 9.930961135338361e-05,
      "loss": 2.8056,
      "step": 2135
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5666818022727966,
      "learning_rate": 9.930622248950864e-05,
      "loss": 2.8309,
      "step": 2140
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5740319490432739,
      "learning_rate": 9.93028253867472e-05,
      "loss": 2.7855,
      "step": 2145
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5522761344909668,
      "learning_rate": 9.929942004566695e-05,
      "loss": 2.8302,
      "step": 2150
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5883065462112427,
      "learning_rate": 9.929600646683686e-05,
      "loss": 2.8356,
      "step": 2155
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.59709233045578,
      "learning_rate": 9.929258465082738e-05,
      "loss": 2.841,
      "step": 2160
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5535476803779602,
      "learning_rate": 9.928915459821027e-05,
      "loss": 2.8169,
      "step": 2165
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5374631285667419,
      "learning_rate": 9.928571630955865e-05,
      "loss": 2.8234,
      "step": 2170
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5283225178718567,
      "learning_rate": 9.928226978544708e-05,
      "loss": 2.7837,
      "step": 2175
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.580665647983551,
      "learning_rate": 9.927881502645146e-05,
      "loss": 2.8057,
      "step": 2180
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5601229667663574,
      "learning_rate": 9.927535203314904e-05,
      "loss": 2.7859,
      "step": 2185
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5217177867889404,
      "learning_rate": 9.927188080611849e-05,
      "loss": 2.8288,
      "step": 2190
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6021257042884827,
      "learning_rate": 9.926840134593984e-05,
      "loss": 2.8011,
      "step": 2195
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5314012765884399,
      "learning_rate": 9.92649136531945e-05,
      "loss": 2.7799,
      "step": 2200
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5148672461509705,
      "learning_rate": 9.926141772846525e-05,
      "loss": 2.7718,
      "step": 2205
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.542212963104248,
      "learning_rate": 9.925791357233623e-05,
      "loss": 2.7865,
      "step": 2210
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5221681594848633,
      "learning_rate": 9.925440118539298e-05,
      "loss": 2.8002,
      "step": 2215
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.538278341293335,
      "learning_rate": 9.925088056822241e-05,
      "loss": 2.8126,
      "step": 2220
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5310001969337463,
      "learning_rate": 9.924735172141281e-05,
      "loss": 2.7541,
      "step": 2225
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5602108240127563,
      "learning_rate": 9.924381464555381e-05,
      "loss": 2.8139,
      "step": 2230
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5399558544158936,
      "learning_rate": 9.924026934123647e-05,
      "loss": 2.8704,
      "step": 2235
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5503289103507996,
      "learning_rate": 9.92367158090532e-05,
      "loss": 2.7279,
      "step": 2240
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5047321319580078,
      "learning_rate": 9.923315404959775e-05,
      "loss": 2.8325,
      "step": 2245
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5400940775871277,
      "learning_rate": 9.92295840634653e-05,
      "loss": 2.8801,
      "step": 2250
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5465108156204224,
      "learning_rate": 9.922600585125237e-05,
      "loss": 2.8311,
      "step": 2255
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5470492243766785,
      "learning_rate": 9.922241941355688e-05,
      "loss": 2.8296,
      "step": 2260
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5418805480003357,
      "learning_rate": 9.921882475097811e-05,
      "loss": 2.8719,
      "step": 2265
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5812088251113892,
      "learning_rate": 9.921522186411671e-05,
      "loss": 2.8671,
      "step": 2270
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5513510704040527,
      "learning_rate": 9.92116107535747e-05,
      "loss": 2.7821,
      "step": 2275
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.544269323348999,
      "learning_rate": 9.920799141995549e-05,
      "loss": 2.7762,
      "step": 2280
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5325347185134888,
      "learning_rate": 9.920436386386387e-05,
      "loss": 2.7726,
      "step": 2285
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5488248467445374,
      "learning_rate": 9.920072808590596e-05,
      "loss": 2.8319,
      "step": 2290
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6011430621147156,
      "learning_rate": 9.91970840866893e-05,
      "loss": 2.7449,
      "step": 2295
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.528702974319458,
      "learning_rate": 9.919343186682282e-05,
      "loss": 2.7987,
      "step": 2300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6044089198112488,
      "learning_rate": 9.918977142691675e-05,
      "loss": 2.7626,
      "step": 2305
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6009995937347412,
      "learning_rate": 9.918610276758273e-05,
      "loss": 2.7499,
      "step": 2310
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5263553857803345,
      "learning_rate": 9.918242588943383e-05,
      "loss": 2.8506,
      "step": 2315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5225784182548523,
      "learning_rate": 9.91787407930844e-05,
      "loss": 2.7817,
      "step": 2320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5515724420547485,
      "learning_rate": 9.917504747915023e-05,
      "loss": 2.8503,
      "step": 2325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5208733677864075,
      "learning_rate": 9.917134594824843e-05,
      "loss": 2.8278,
      "step": 2330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5219298601150513,
      "learning_rate": 9.916763620099754e-05,
      "loss": 2.7408,
      "step": 2335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.610284686088562,
      "learning_rate": 9.916391823801743e-05,
      "loss": 2.7931,
      "step": 2340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5442891120910645,
      "learning_rate": 9.916019205992937e-05,
      "loss": 2.7897,
      "step": 2345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.577951967716217,
      "learning_rate": 9.915645766735597e-05,
      "loss": 2.7515,
      "step": 2350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5623117685317993,
      "learning_rate": 9.915271506092126e-05,
      "loss": 2.832,
      "step": 2355
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.53264981508255,
      "learning_rate": 9.91489642412506e-05,
      "loss": 2.8138,
      "step": 2360
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5395910739898682,
      "learning_rate": 9.914520520897072e-05,
      "loss": 2.7396,
      "step": 2365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.558599054813385,
      "learning_rate": 9.914143796470978e-05,
      "loss": 2.7516,
      "step": 2370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5247620940208435,
      "learning_rate": 9.913766250909726e-05,
      "loss": 2.7131,
      "step": 2375
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5144100785255432,
      "learning_rate": 9.9133878842764e-05,
      "loss": 2.767,
      "step": 2380
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5098922252655029,
      "learning_rate": 9.913008696634226e-05,
      "loss": 2.798,
      "step": 2385
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5244462490081787,
      "learning_rate": 9.912628688046564e-05,
      "loss": 2.7477,
      "step": 2390
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5114411115646362,
      "learning_rate": 9.912247858576914e-05,
      "loss": 2.8107,
      "step": 2395
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4963493049144745,
      "learning_rate": 9.911866208288906e-05,
      "loss": 2.7257,
      "step": 2400
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.522879958152771,
      "learning_rate": 9.911483737246319e-05,
      "loss": 2.7763,
      "step": 2405
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5592846274375916,
      "learning_rate": 9.911100445513058e-05,
      "loss": 2.776,
      "step": 2410
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5502878427505493,
      "learning_rate": 9.910716333153169e-05,
      "loss": 2.7474,
      "step": 2415
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5128346085548401,
      "learning_rate": 9.910331400230841e-05,
      "loss": 2.7539,
      "step": 2420
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5400784015655518,
      "learning_rate": 9.90994564681039e-05,
      "loss": 2.7445,
      "step": 2425
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5056523680686951,
      "learning_rate": 9.909559072956275e-05,
      "loss": 2.7503,
      "step": 2430
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.532008171081543,
      "learning_rate": 9.909171678733091e-05,
      "loss": 2.7365,
      "step": 2435
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5684343576431274,
      "learning_rate": 9.908783464205573e-05,
      "loss": 2.8201,
      "step": 2440
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.532482385635376,
      "learning_rate": 9.908394429438585e-05,
      "loss": 2.7705,
      "step": 2445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6100127696990967,
      "learning_rate": 9.908004574497139e-05,
      "loss": 2.7948,
      "step": 2450
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5573193430900574,
      "learning_rate": 9.907613899446372e-05,
      "loss": 2.7273,
      "step": 2455
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5323579907417297,
      "learning_rate": 9.90722240435157e-05,
      "loss": 2.6937,
      "step": 2460
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5475561618804932,
      "learning_rate": 9.906830089278148e-05,
      "loss": 2.7911,
      "step": 2465
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5400791764259338,
      "learning_rate": 9.906436954291661e-05,
      "loss": 2.7794,
      "step": 2470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5867957472801208,
      "learning_rate": 9.906042999457798e-05,
      "loss": 2.7594,
      "step": 2475
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5197498202323914,
      "learning_rate": 9.905648224842391e-05,
      "loss": 2.7976,
      "step": 2480
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5577018857002258,
      "learning_rate": 9.905252630511405e-05,
      "loss": 2.7133,
      "step": 2485
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5106146335601807,
      "learning_rate": 9.904856216530938e-05,
      "loss": 2.8102,
      "step": 2490
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5206735730171204,
      "learning_rate": 9.904458982967235e-05,
      "loss": 2.7839,
      "step": 2495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5027667284011841,
      "learning_rate": 9.904060929886668e-05,
      "loss": 2.7579,
      "step": 2500
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5722038745880127,
      "learning_rate": 9.903662057355751e-05,
      "loss": 2.8059,
      "step": 2505
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5326957106590271,
      "learning_rate": 9.903262365441137e-05,
      "loss": 2.7334,
      "step": 2510
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5312429070472717,
      "learning_rate": 9.90286185420961e-05,
      "loss": 2.8039,
      "step": 2515
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5297892093658447,
      "learning_rate": 9.902460523728094e-05,
      "loss": 2.7111,
      "step": 2520
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5346359014511108,
      "learning_rate": 9.902058374063651e-05,
      "loss": 2.7245,
      "step": 2525
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5481923818588257,
      "learning_rate": 9.90165540528348e-05,
      "loss": 2.7937,
      "step": 2530
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4908032715320587,
      "learning_rate": 9.901251617454913e-05,
      "loss": 2.7633,
      "step": 2535
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5241805911064148,
      "learning_rate": 9.900847010645422e-05,
      "loss": 2.6684,
      "step": 2540
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5085708498954773,
      "learning_rate": 9.900441584922616e-05,
      "loss": 2.7463,
      "step": 2545
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5117504596710205,
      "learning_rate": 9.90003534035424e-05,
      "loss": 2.7487,
      "step": 2550
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5160303115844727,
      "learning_rate": 9.899628277008175e-05,
      "loss": 2.7107,
      "step": 2555
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5155068039894104,
      "learning_rate": 9.899220394952441e-05,
      "loss": 2.838,
      "step": 2560
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6070230603218079,
      "learning_rate": 9.898811694255195e-05,
      "loss": 2.7346,
      "step": 2565
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.570728063583374,
      "learning_rate": 9.898402174984726e-05,
      "loss": 2.8189,
      "step": 2570
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48364219069480896,
      "learning_rate": 9.897991837209464e-05,
      "loss": 2.7161,
      "step": 2575
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49734756350517273,
      "learning_rate": 9.897580680997977e-05,
      "loss": 2.6834,
      "step": 2580
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5250561237335205,
      "learning_rate": 9.897168706418965e-05,
      "loss": 2.732,
      "step": 2585
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4989260733127594,
      "learning_rate": 9.89675591354127e-05,
      "loss": 2.7803,
      "step": 2590
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5097677111625671,
      "learning_rate": 9.896342302433866e-05,
      "loss": 2.7981,
      "step": 2595
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5408244132995605,
      "learning_rate": 9.895927873165867e-05,
      "loss": 2.7753,
      "step": 2600
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5786339044570923,
      "learning_rate": 9.895512625806522e-05,
      "loss": 2.7988,
      "step": 2605
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5229914784431458,
      "learning_rate": 9.895096560425218e-05,
      "loss": 2.6991,
      "step": 2610
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5528015494346619,
      "learning_rate": 9.894679677091477e-05,
      "loss": 2.6785,
      "step": 2615
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4930858314037323,
      "learning_rate": 9.89426197587496e-05,
      "loss": 2.7214,
      "step": 2620
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5052673816680908,
      "learning_rate": 9.893843456845462e-05,
      "loss": 2.7727,
      "step": 2625
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5403813123703003,
      "learning_rate": 9.893424120072916e-05,
      "loss": 2.7853,
      "step": 2630
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48861414194107056,
      "learning_rate": 9.893003965627394e-05,
      "loss": 2.7771,
      "step": 2635
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49067622423171997,
      "learning_rate": 9.8925829935791e-05,
      "loss": 2.8587,
      "step": 2640
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5005021095275879,
      "learning_rate": 9.892161203998375e-05,
      "loss": 2.6524,
      "step": 2645
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5215873718261719,
      "learning_rate": 9.891738596955703e-05,
      "loss": 2.8369,
      "step": 2650
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5165697336196899,
      "learning_rate": 9.891315172521696e-05,
      "loss": 2.7199,
      "step": 2655
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5254875421524048,
      "learning_rate": 9.890890930767111e-05,
      "loss": 2.8558,
      "step": 2660
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5124167799949646,
      "learning_rate": 9.890465871762832e-05,
      "loss": 2.8433,
      "step": 2665
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5394469499588013,
      "learning_rate": 9.89003999557989e-05,
      "loss": 2.738,
      "step": 2670
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4869963228702545,
      "learning_rate": 9.889613302289443e-05,
      "loss": 2.7105,
      "step": 2675
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5013901591300964,
      "learning_rate": 9.889185791962792e-05,
      "loss": 2.7858,
      "step": 2680
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5393935441970825,
      "learning_rate": 9.888757464671372e-05,
      "loss": 2.7074,
      "step": 2685
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5051852464675903,
      "learning_rate": 9.888328320486755e-05,
      "loss": 2.7306,
      "step": 2690
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5241373777389526,
      "learning_rate": 9.88789835948065e-05,
      "loss": 2.7678,
      "step": 2695
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5107607841491699,
      "learning_rate": 9.887467581724901e-05,
      "loss": 2.69,
      "step": 2700
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5003671646118164,
      "learning_rate": 9.887035987291488e-05,
      "loss": 2.797,
      "step": 2705
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5004083514213562,
      "learning_rate": 9.88660357625253e-05,
      "loss": 2.7355,
      "step": 2710
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5207839608192444,
      "learning_rate": 9.886170348680284e-05,
      "loss": 2.8232,
      "step": 2715
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5523451566696167,
      "learning_rate": 9.885736304647139e-05,
      "loss": 2.7221,
      "step": 2720
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5308597087860107,
      "learning_rate": 9.885301444225619e-05,
      "loss": 2.8157,
      "step": 2725
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5489399433135986,
      "learning_rate": 9.884865767488391e-05,
      "loss": 2.7532,
      "step": 2730
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5469006299972534,
      "learning_rate": 9.884429274508256e-05,
      "loss": 2.7114,
      "step": 2735
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5559083819389343,
      "learning_rate": 9.883991965358146e-05,
      "loss": 2.6875,
      "step": 2740
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5050241947174072,
      "learning_rate": 9.883553840111136e-05,
      "loss": 2.7414,
      "step": 2745
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5499570965766907,
      "learning_rate": 9.883114898840437e-05,
      "loss": 2.78,
      "step": 2750
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5262013673782349,
      "learning_rate": 9.882675141619391e-05,
      "loss": 2.7543,
      "step": 2755
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5299731492996216,
      "learning_rate": 9.882234568521482e-05,
      "loss": 2.7802,
      "step": 2760
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.568892776966095,
      "learning_rate": 9.881793179620328e-05,
      "loss": 2.7546,
      "step": 2765
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.528265655040741,
      "learning_rate": 9.881350974989682e-05,
      "loss": 2.702,
      "step": 2770
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48452872037887573,
      "learning_rate": 9.880907954703437e-05,
      "loss": 2.7568,
      "step": 2775
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5208167433738708,
      "learning_rate": 9.880464118835619e-05,
      "loss": 2.7151,
      "step": 2780
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7175845503807068,
      "learning_rate": 9.880019467460391e-05,
      "loss": 2.7325,
      "step": 2785
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5707563757896423,
      "learning_rate": 9.879574000652052e-05,
      "loss": 2.714,
      "step": 2790
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5005860328674316,
      "learning_rate": 9.87912771848504e-05,
      "loss": 2.7669,
      "step": 2795
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48325178027153015,
      "learning_rate": 9.878680621033925e-05,
      "loss": 2.7062,
      "step": 2800
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5340758562088013,
      "learning_rate": 9.878232708373415e-05,
      "loss": 2.7334,
      "step": 2805
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4927772581577301,
      "learning_rate": 9.877783980578357e-05,
      "loss": 2.6696,
      "step": 2810
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48546284437179565,
      "learning_rate": 9.877334437723728e-05,
      "loss": 2.7639,
      "step": 2815
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.492984801530838,
      "learning_rate": 9.87688407988465e-05,
      "loss": 2.7505,
      "step": 2820
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5286808013916016,
      "learning_rate": 9.876432907136372e-05,
      "loss": 2.6789,
      "step": 2825
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5092411637306213,
      "learning_rate": 9.875980919554287e-05,
      "loss": 2.6866,
      "step": 2830
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4731024205684662,
      "learning_rate": 9.875528117213915e-05,
      "loss": 2.6786,
      "step": 2835
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5068648457527161,
      "learning_rate": 9.875074500190922e-05,
      "loss": 2.7863,
      "step": 2840
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48445895314216614,
      "learning_rate": 9.874620068561104e-05,
      "loss": 2.6759,
      "step": 2845
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7175626754760742,
      "learning_rate": 9.874164822400396e-05,
      "loss": 2.7097,
      "step": 2850
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5763188004493713,
      "learning_rate": 9.873708761784866e-05,
      "loss": 2.7631,
      "step": 2855
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5808132886886597,
      "learning_rate": 9.873251886790722e-05,
      "loss": 2.7284,
      "step": 2860
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5005022883415222,
      "learning_rate": 9.872794197494307e-05,
      "loss": 2.7977,
      "step": 2865
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.536597490310669,
      "learning_rate": 9.872335693972094e-05,
      "loss": 2.7375,
      "step": 2870
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49845170974731445,
      "learning_rate": 9.871876376300702e-05,
      "loss": 2.7451,
      "step": 2875
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5192010998725891,
      "learning_rate": 9.871416244556881e-05,
      "loss": 2.7016,
      "step": 2880
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4982602894306183,
      "learning_rate": 9.870955298817517e-05,
      "loss": 2.7454,
      "step": 2885
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48857778310775757,
      "learning_rate": 9.87049353915963e-05,
      "loss": 2.7054,
      "step": 2890
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47679853439331055,
      "learning_rate": 9.87003096566038e-05,
      "loss": 2.7943,
      "step": 2895
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49647822976112366,
      "learning_rate": 9.869567578397062e-05,
      "loss": 2.6368,
      "step": 2900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5406832098960876,
      "learning_rate": 9.869103377447105e-05,
      "loss": 2.7239,
      "step": 2905
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49990880489349365,
      "learning_rate": 9.868638362888076e-05,
      "loss": 2.7157,
      "step": 2910
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5068215131759644,
      "learning_rate": 9.868172534797678e-05,
      "loss": 2.6698,
      "step": 2915
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46596211194992065,
      "learning_rate": 9.867705893253747e-05,
      "loss": 2.6729,
      "step": 2920
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5103382468223572,
      "learning_rate": 9.867238438334259e-05,
      "loss": 2.7789,
      "step": 2925
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47929635643959045,
      "learning_rate": 9.866770170117324e-05,
      "loss": 2.685,
      "step": 2930
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46291589736938477,
      "learning_rate": 9.866301088681185e-05,
      "loss": 2.6677,
      "step": 2935
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49163922667503357,
      "learning_rate": 9.865831194104227e-05,
      "loss": 2.7568,
      "step": 2940
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5066707730293274,
      "learning_rate": 9.865360486464967e-05,
      "loss": 2.7488,
      "step": 2945
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49747657775878906,
      "learning_rate": 9.864888965842056e-05,
      "loss": 2.7161,
      "step": 2950
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48043593764305115,
      "learning_rate": 9.864416632314287e-05,
      "loss": 2.7404,
      "step": 2955
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48907309770584106,
      "learning_rate": 9.863943485960583e-05,
      "loss": 2.6306,
      "step": 2960
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49248868227005005,
      "learning_rate": 9.863469526860005e-05,
      "loss": 2.6903,
      "step": 2965
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.536936342716217,
      "learning_rate": 9.862994755091752e-05,
      "loss": 2.6358,
      "step": 2970
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4952998757362366,
      "learning_rate": 9.862519170735154e-05,
      "loss": 2.7117,
      "step": 2975
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5472564697265625,
      "learning_rate": 9.86204277386968e-05,
      "loss": 2.7029,
      "step": 2980
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.540274441242218,
      "learning_rate": 9.861565564574935e-05,
      "loss": 2.6815,
      "step": 2985
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4996606111526489,
      "learning_rate": 9.861087542930658e-05,
      "loss": 2.6918,
      "step": 2990
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48429161310195923,
      "learning_rate": 9.860608709016726e-05,
      "loss": 2.6666,
      "step": 2995
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4707185626029968,
      "learning_rate": 9.860129062913147e-05,
      "loss": 2.7541,
      "step": 3000
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5414662957191467,
      "learning_rate": 9.859648604700072e-05,
      "loss": 2.6884,
      "step": 3005
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4731457233428955,
      "learning_rate": 9.859167334457781e-05,
      "loss": 2.6951,
      "step": 3010
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4958624839782715,
      "learning_rate": 9.858685252266696e-05,
      "loss": 2.6751,
      "step": 3015
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.508975625038147,
      "learning_rate": 9.858202358207367e-05,
      "loss": 2.7459,
      "step": 3020
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49762409925460815,
      "learning_rate": 9.857718652360486e-05,
      "loss": 2.706,
      "step": 3025
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48055601119995117,
      "learning_rate": 9.857234134806878e-05,
      "loss": 2.704,
      "step": 3030
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5035915374755859,
      "learning_rate": 9.856748805627503e-05,
      "loss": 2.6741,
      "step": 3035
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5016337037086487,
      "learning_rate": 9.856262664903462e-05,
      "loss": 2.6971,
      "step": 3040
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49414950609207153,
      "learning_rate": 9.855775712715981e-05,
      "loss": 2.6735,
      "step": 3045
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4953848719596863,
      "learning_rate": 9.855287949146432e-05,
      "loss": 2.7455,
      "step": 3050
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5040009617805481,
      "learning_rate": 9.854799374276318e-05,
      "loss": 2.7315,
      "step": 3055
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5024610757827759,
      "learning_rate": 9.854309988187277e-05,
      "loss": 2.7031,
      "step": 3060
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4802127480506897,
      "learning_rate": 9.853819790961086e-05,
      "loss": 2.7071,
      "step": 3065
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4796232283115387,
      "learning_rate": 9.853328782679652e-05,
      "loss": 2.6849,
      "step": 3070
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5054613947868347,
      "learning_rate": 9.852836963425021e-05,
      "loss": 2.7395,
      "step": 3075
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5207016468048096,
      "learning_rate": 9.852344333279375e-05,
      "loss": 2.7249,
      "step": 3080
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4914843738079071,
      "learning_rate": 9.851850892325032e-05,
      "loss": 2.7196,
      "step": 3085
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5107265114784241,
      "learning_rate": 9.851356640644443e-05,
      "loss": 2.7092,
      "step": 3090
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4639938771724701,
      "learning_rate": 9.850861578320196e-05,
      "loss": 2.7166,
      "step": 3095
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4959027171134949,
      "learning_rate": 9.850365705435013e-05,
      "loss": 2.6659,
      "step": 3100
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4598113000392914,
      "learning_rate": 9.849869022071753e-05,
      "loss": 2.6886,
      "step": 3105
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4987031817436218,
      "learning_rate": 9.849371528313411e-05,
      "loss": 2.6823,
      "step": 3110
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46973487734794617,
      "learning_rate": 9.848873224243114e-05,
      "loss": 2.6471,
      "step": 3115
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5113233923912048,
      "learning_rate": 9.848374109944128e-05,
      "loss": 2.7224,
      "step": 3120
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48511776328086853,
      "learning_rate": 9.847874185499855e-05,
      "loss": 2.7518,
      "step": 3125
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5560673475265503,
      "learning_rate": 9.847373450993828e-05,
      "loss": 2.6854,
      "step": 3130
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5296474695205688,
      "learning_rate": 9.846871906509719e-05,
      "loss": 2.6474,
      "step": 3135
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48699313402175903,
      "learning_rate": 9.846369552131334e-05,
      "loss": 2.6918,
      "step": 3140
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5564193725585938,
      "learning_rate": 9.845866387942614e-05,
      "loss": 2.6598,
      "step": 3145
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5348125100135803,
      "learning_rate": 9.845362414027637e-05,
      "loss": 2.7476,
      "step": 3150
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5079858303070068,
      "learning_rate": 9.844857630470614e-05,
      "loss": 2.721,
      "step": 3155
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48495516180992126,
      "learning_rate": 9.844352037355893e-05,
      "loss": 2.6732,
      "step": 3160
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5063591599464417,
      "learning_rate": 9.843845634767957e-05,
      "loss": 2.6333,
      "step": 3165
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5080045461654663,
      "learning_rate": 9.843338422791424e-05,
      "loss": 2.7329,
      "step": 3170
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.477338969707489,
      "learning_rate": 9.842830401511048e-05,
      "loss": 2.6821,
      "step": 3175
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47274377942085266,
      "learning_rate": 9.842321571011714e-05,
      "loss": 2.6767,
      "step": 3180
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46288689970970154,
      "learning_rate": 9.841811931378448e-05,
      "loss": 2.6227,
      "step": 3185
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48680469393730164,
      "learning_rate": 9.841301482696413e-05,
      "loss": 2.7341,
      "step": 3190
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4931415319442749,
      "learning_rate": 9.840790225050896e-05,
      "loss": 2.6955,
      "step": 3195
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49861282110214233,
      "learning_rate": 9.840278158527328e-05,
      "loss": 2.6907,
      "step": 3200
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48570066690444946,
      "learning_rate": 9.839765283211278e-05,
      "loss": 2.7504,
      "step": 3205
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4833829402923584,
      "learning_rate": 9.83925159918844e-05,
      "loss": 2.6789,
      "step": 3210
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47707071900367737,
      "learning_rate": 9.838737106544651e-05,
      "loss": 2.6517,
      "step": 3215
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4779832065105438,
      "learning_rate": 9.838221805365881e-05,
      "loss": 2.65,
      "step": 3220
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4906891882419586,
      "learning_rate": 9.837705695738236e-05,
      "loss": 2.6992,
      "step": 3225
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5201934576034546,
      "learning_rate": 9.837188777747954e-05,
      "loss": 2.5893,
      "step": 3230
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4803478717803955,
      "learning_rate": 9.836671051481411e-05,
      "loss": 2.6174,
      "step": 3235
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47541898488998413,
      "learning_rate": 9.836152517025116e-05,
      "loss": 2.6618,
      "step": 3240
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2099992036819458,
      "learning_rate": 9.835633174465715e-05,
      "loss": 2.6688,
      "step": 3245
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5176692605018616,
      "learning_rate": 9.83511302388999e-05,
      "loss": 2.6493,
      "step": 3250
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47160351276397705,
      "learning_rate": 9.834592065384853e-05,
      "loss": 2.6631,
      "step": 3255
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5062034726142883,
      "learning_rate": 9.834070299037357e-05,
      "loss": 2.7424,
      "step": 3260
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49870821833610535,
      "learning_rate": 9.833547724934687e-05,
      "loss": 2.6244,
      "step": 3265
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4933323264122009,
      "learning_rate": 9.833024343164159e-05,
      "loss": 2.7091,
      "step": 3270
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46616244316101074,
      "learning_rate": 9.832500153813234e-05,
      "loss": 2.6565,
      "step": 3275
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4776048958301544,
      "learning_rate": 9.831975156969498e-05,
      "loss": 2.6555,
      "step": 3280
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48551979660987854,
      "learning_rate": 9.831449352720677e-05,
      "loss": 2.6707,
      "step": 3285
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4840518832206726,
      "learning_rate": 9.830922741154631e-05,
      "loss": 2.6892,
      "step": 3290
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47098273038864136,
      "learning_rate": 9.830395322359357e-05,
      "loss": 2.7406,
      "step": 3295
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4619990587234497,
      "learning_rate": 9.829867096422983e-05,
      "loss": 2.7184,
      "step": 3300
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49326464533805847,
      "learning_rate": 9.829338063433773e-05,
      "loss": 2.6412,
      "step": 3305
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4801635146141052,
      "learning_rate": 9.828808223480126e-05,
      "loss": 2.6516,
      "step": 3310
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4465640187263489,
      "learning_rate": 9.828277576650578e-05,
      "loss": 2.6665,
      "step": 3315
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4658386707305908,
      "learning_rate": 9.827746123033796e-05,
      "loss": 2.6458,
      "step": 3320
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44876256585121155,
      "learning_rate": 9.827213862718587e-05,
      "loss": 2.6645,
      "step": 3325
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46239855885505676,
      "learning_rate": 9.826680795793887e-05,
      "loss": 2.5729,
      "step": 3330
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4640813171863556,
      "learning_rate": 9.826146922348773e-05,
      "loss": 2.5399,
      "step": 3335
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4730069935321808,
      "learning_rate": 9.825612242472448e-05,
      "loss": 2.6473,
      "step": 3340
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5075293183326721,
      "learning_rate": 9.82507675625426e-05,
      "loss": 2.7503,
      "step": 3345
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5289545655250549,
      "learning_rate": 9.824540463783682e-05,
      "loss": 2.6461,
      "step": 3350
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5168687105178833,
      "learning_rate": 9.824003365150329e-05,
      "loss": 2.7162,
      "step": 3355
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4965927004814148,
      "learning_rate": 9.82346546044395e-05,
      "loss": 2.7004,
      "step": 3360
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5472583770751953,
      "learning_rate": 9.822926749754424e-05,
      "loss": 2.7218,
      "step": 3365
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5117067694664001,
      "learning_rate": 9.822387233171771e-05,
      "loss": 2.7078,
      "step": 3370
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5062797665596008,
      "learning_rate": 9.821846910786136e-05,
      "loss": 2.7055,
      "step": 3375
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.500096321105957,
      "learning_rate": 9.821305782687812e-05,
      "loss": 2.6626,
      "step": 3380
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48254573345184326,
      "learning_rate": 9.820763848967215e-05,
      "loss": 2.6403,
      "step": 3385
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6973061561584473,
      "learning_rate": 9.820221109714902e-05,
      "loss": 2.6724,
      "step": 3390
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49116745591163635,
      "learning_rate": 9.81967756502156e-05,
      "loss": 2.6881,
      "step": 3395
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4825573265552521,
      "learning_rate": 9.819133214978017e-05,
      "loss": 2.6513,
      "step": 3400
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4736406207084656,
      "learning_rate": 9.818588059675229e-05,
      "loss": 2.7073,
      "step": 3405
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5035890936851501,
      "learning_rate": 9.818042099204292e-05,
      "loss": 2.6619,
      "step": 3410
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.476978600025177,
      "learning_rate": 9.817495333656433e-05,
      "loss": 2.7064,
      "step": 3415
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4948991537094116,
      "learning_rate": 9.816947763123013e-05,
      "loss": 2.6484,
      "step": 3420
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5480136275291443,
      "learning_rate": 9.816399387695529e-05,
      "loss": 2.7415,
      "step": 3425
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4766487777233124,
      "learning_rate": 9.815850207465616e-05,
      "loss": 2.624,
      "step": 3430
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4953169524669647,
      "learning_rate": 9.815300222525035e-05,
      "loss": 2.6982,
      "step": 3435
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47066280245780945,
      "learning_rate": 9.814749432965691e-05,
      "loss": 2.7124,
      "step": 3440
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.497530072927475,
      "learning_rate": 9.814197838879615e-05,
      "loss": 2.7082,
      "step": 3445
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4712545573711395,
      "learning_rate": 9.813645440358981e-05,
      "loss": 2.6907,
      "step": 3450
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46602267026901245,
      "learning_rate": 9.813092237496087e-05,
      "loss": 2.6037,
      "step": 3455
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4617881178855896,
      "learning_rate": 9.812538230383376e-05,
      "loss": 2.6789,
      "step": 3460
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4463293254375458,
      "learning_rate": 9.811983419113418e-05,
      "loss": 2.6395,
      "step": 3465
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4906013309955597,
      "learning_rate": 9.81142780377892e-05,
      "loss": 2.6314,
      "step": 3470
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.466228723526001,
      "learning_rate": 9.810871384472724e-05,
      "loss": 2.6641,
      "step": 3475
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49875035881996155,
      "learning_rate": 9.810314161287806e-05,
      "loss": 2.7015,
      "step": 3480
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46299684047698975,
      "learning_rate": 9.809756134317276e-05,
      "loss": 2.667,
      "step": 3485
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4750029146671295,
      "learning_rate": 9.809197303654376e-05,
      "loss": 2.6918,
      "step": 3490
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4703943133354187,
      "learning_rate": 9.808637669392487e-05,
      "loss": 2.6235,
      "step": 3495
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4732406437397003,
      "learning_rate": 9.808077231625119e-05,
      "loss": 2.5908,
      "step": 3500
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4645402133464813,
      "learning_rate": 9.807515990445922e-05,
      "loss": 2.6308,
      "step": 3505
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4524895250797272,
      "learning_rate": 9.806953945948676e-05,
      "loss": 2.6906,
      "step": 3510
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4707656800746918,
      "learning_rate": 9.806391098227297e-05,
      "loss": 2.6799,
      "step": 3515
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4807879328727722,
      "learning_rate": 9.805827447375835e-05,
      "loss": 2.6708,
      "step": 3520
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43682295083999634,
      "learning_rate": 9.805262993488473e-05,
      "loss": 2.5642,
      "step": 3525
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4825822412967682,
      "learning_rate": 9.804697736659532e-05,
      "loss": 2.6912,
      "step": 3530
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4722949266433716,
      "learning_rate": 9.804131676983458e-05,
      "loss": 2.6486,
      "step": 3535
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.445079505443573,
      "learning_rate": 9.803564814554844e-05,
      "loss": 2.5946,
      "step": 3540
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47372156381607056,
      "learning_rate": 9.802997149468408e-05,
      "loss": 2.6103,
      "step": 3545
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45374569296836853,
      "learning_rate": 9.802428681819003e-05,
      "loss": 2.6537,
      "step": 3550
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5472946763038635,
      "learning_rate": 9.801859411701621e-05,
      "loss": 2.6421,
      "step": 3555
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45874202251434326,
      "learning_rate": 9.801289339211384e-05,
      "loss": 2.7055,
      "step": 3560
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43282902240753174,
      "learning_rate": 9.800718464443547e-05,
      "loss": 2.6529,
      "step": 3565
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4467175602912903,
      "learning_rate": 9.800146787493503e-05,
      "loss": 2.5799,
      "step": 3570
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47334006428718567,
      "learning_rate": 9.799574308456775e-05,
      "loss": 2.5828,
      "step": 3575
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4792570173740387,
      "learning_rate": 9.799001027429026e-05,
      "loss": 2.6568,
      "step": 3580
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.480743944644928,
      "learning_rate": 9.798426944506048e-05,
      "loss": 2.6448,
      "step": 3585
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4752001464366913,
      "learning_rate": 9.797852059783766e-05,
      "loss": 2.6269,
      "step": 3590
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4876648187637329,
      "learning_rate": 9.79727637335824e-05,
      "loss": 2.6916,
      "step": 3595
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.508886992931366,
      "learning_rate": 9.79669988532567e-05,
      "loss": 2.661,
      "step": 3600
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48778828978538513,
      "learning_rate": 9.79612259578238e-05,
      "loss": 2.7352,
      "step": 3605
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4681411385536194,
      "learning_rate": 9.795544504824836e-05,
      "loss": 2.6286,
      "step": 3610
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5289109945297241,
      "learning_rate": 9.794965612549635e-05,
      "loss": 2.5134,
      "step": 3615
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49147579073905945,
      "learning_rate": 9.794385919053506e-05,
      "loss": 2.6198,
      "step": 3620
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4840907156467438,
      "learning_rate": 9.793805424433313e-05,
      "loss": 2.6291,
      "step": 3625
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5015655159950256,
      "learning_rate": 9.793224128786058e-05,
      "loss": 2.6761,
      "step": 3630
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43456682562828064,
      "learning_rate": 9.79264203220887e-05,
      "loss": 2.623,
      "step": 3635
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4789671003818512,
      "learning_rate": 9.792059134799016e-05,
      "loss": 2.6379,
      "step": 3640
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4429008364677429,
      "learning_rate": 9.791475436653896e-05,
      "loss": 2.5808,
      "step": 3645
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4656991958618164,
      "learning_rate": 9.790890937871046e-05,
      "loss": 2.6635,
      "step": 3650
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5119509100914001,
      "learning_rate": 9.79030563854813e-05,
      "loss": 2.6261,
      "step": 3655
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45926767587661743,
      "learning_rate": 9.789719538782952e-05,
      "loss": 2.7199,
      "step": 3660
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49072903394699097,
      "learning_rate": 9.789132638673445e-05,
      "loss": 2.6295,
      "step": 3665
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43878254294395447,
      "learning_rate": 9.78854493831768e-05,
      "loss": 2.6857,
      "step": 3670
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45711296796798706,
      "learning_rate": 9.787956437813859e-05,
      "loss": 2.607,
      "step": 3675
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48649832606315613,
      "learning_rate": 9.787367137260317e-05,
      "loss": 2.7041,
      "step": 3680
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4556087553501129,
      "learning_rate": 9.786777036755525e-05,
      "loss": 2.5783,
      "step": 3685
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4662274718284607,
      "learning_rate": 9.786186136398085e-05,
      "loss": 2.6103,
      "step": 3690
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49316126108169556,
      "learning_rate": 9.785594436286738e-05,
      "loss": 2.6962,
      "step": 3695
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48223158717155457,
      "learning_rate": 9.78500193652035e-05,
      "loss": 2.6108,
      "step": 3700
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4835610091686249,
      "learning_rate": 9.784408637197929e-05,
      "loss": 2.6817,
      "step": 3705
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45839619636535645,
      "learning_rate": 9.783814538418612e-05,
      "loss": 2.5736,
      "step": 3710
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4699246883392334,
      "learning_rate": 9.783219640281671e-05,
      "loss": 2.6594,
      "step": 3715
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46189993619918823,
      "learning_rate": 9.78262394288651e-05,
      "loss": 2.6855,
      "step": 3720
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48868775367736816,
      "learning_rate": 9.782027446332671e-05,
      "loss": 2.6944,
      "step": 3725
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4716065526008606,
      "learning_rate": 9.781430150719822e-05,
      "loss": 2.6659,
      "step": 3730
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4976038932800293,
      "learning_rate": 9.780832056147772e-05,
      "loss": 2.6989,
      "step": 3735
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4828726053237915,
      "learning_rate": 9.78023316271646e-05,
      "loss": 2.5659,
      "step": 3740
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5046469569206238,
      "learning_rate": 9.779633470525957e-05,
      "loss": 2.6673,
      "step": 3745
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47908851504325867,
      "learning_rate": 9.77903297967647e-05,
      "loss": 2.6144,
      "step": 3750
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5016385912895203,
      "learning_rate": 9.778431690268341e-05,
      "loss": 2.5885,
      "step": 3755
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44735029339790344,
      "learning_rate": 9.777829602402043e-05,
      "loss": 2.6906,
      "step": 3760
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49224552512168884,
      "learning_rate": 9.777226716178178e-05,
      "loss": 2.6349,
      "step": 3765
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4854823350906372,
      "learning_rate": 9.776623031697492e-05,
      "loss": 2.574,
      "step": 3770
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4576481580734253,
      "learning_rate": 9.776018549060855e-05,
      "loss": 2.585,
      "step": 3775
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45245352387428284,
      "learning_rate": 9.775413268369274e-05,
      "loss": 2.6453,
      "step": 3780
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43974950909614563,
      "learning_rate": 9.77480718972389e-05,
      "loss": 2.5957,
      "step": 3785
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4594436585903168,
      "learning_rate": 9.774200313225976e-05,
      "loss": 2.6759,
      "step": 3790
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4643900394439697,
      "learning_rate": 9.773592638976939e-05,
      "loss": 2.7133,
      "step": 3795
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45019522309303284,
      "learning_rate": 9.77298416707832e-05,
      "loss": 2.5763,
      "step": 3800
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47717463970184326,
      "learning_rate": 9.772374897631791e-05,
      "loss": 2.5734,
      "step": 3805
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45437413454055786,
      "learning_rate": 9.771764830739159e-05,
      "loss": 2.5927,
      "step": 3810
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47948700189590454,
      "learning_rate": 9.771153966502365e-05,
      "loss": 2.637,
      "step": 3815
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4616056978702545,
      "learning_rate": 9.77054230502348e-05,
      "loss": 2.6252,
      "step": 3820
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.442796915769577,
      "learning_rate": 9.769929846404711e-05,
      "loss": 2.6894,
      "step": 3825
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4560488760471344,
      "learning_rate": 9.769316590748401e-05,
      "loss": 2.606,
      "step": 3830
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4483989179134369,
      "learning_rate": 9.768702538157018e-05,
      "loss": 2.6305,
      "step": 3835
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.531042754650116,
      "learning_rate": 9.76808768873317e-05,
      "loss": 2.55,
      "step": 3840
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45492008328437805,
      "learning_rate": 9.767472042579595e-05,
      "loss": 2.6491,
      "step": 3845
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4479203522205353,
      "learning_rate": 9.766855599799168e-05,
      "loss": 2.5882,
      "step": 3850
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48066672682762146,
      "learning_rate": 9.76623836049489e-05,
      "loss": 2.5639,
      "step": 3855
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4672229588031769,
      "learning_rate": 9.765620324769903e-05,
      "loss": 2.6143,
      "step": 3860
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4680553078651428,
      "learning_rate": 9.765001492727476e-05,
      "loss": 2.6633,
      "step": 3865
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5400980710983276,
      "learning_rate": 9.764381864471017e-05,
      "loss": 2.6448,
      "step": 3870
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4546037018299103,
      "learning_rate": 9.76376144010406e-05,
      "loss": 2.6191,
      "step": 3875
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.503840446472168,
      "learning_rate": 9.763140219730278e-05,
      "loss": 2.7058,
      "step": 3880
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44492724537849426,
      "learning_rate": 9.762518203453472e-05,
      "loss": 2.6871,
      "step": 3885
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4676051735877991,
      "learning_rate": 9.761895391377581e-05,
      "loss": 2.6043,
      "step": 3890
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4854621887207031,
      "learning_rate": 9.761271783606674e-05,
      "loss": 2.6078,
      "step": 3895
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5062069296836853,
      "learning_rate": 9.760647380244952e-05,
      "loss": 2.7287,
      "step": 3900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4572903513908386,
      "learning_rate": 9.760022181396753e-05,
      "loss": 2.6592,
      "step": 3905
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4892476797103882,
      "learning_rate": 9.759396187166543e-05,
      "loss": 2.692,
      "step": 3910
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5426668524742126,
      "learning_rate": 9.758769397658926e-05,
      "loss": 2.6239,
      "step": 3915
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47848665714263916,
      "learning_rate": 9.758141812978634e-05,
      "loss": 2.6364,
      "step": 3920
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4641406834125519,
      "learning_rate": 9.757513433230534e-05,
      "loss": 2.5936,
      "step": 3925
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45468270778656006,
      "learning_rate": 9.756884258519627e-05,
      "loss": 2.576,
      "step": 3930
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47367316484451294,
      "learning_rate": 9.756254288951046e-05,
      "loss": 2.6091,
      "step": 3935
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5169945359230042,
      "learning_rate": 9.755623524630055e-05,
      "loss": 2.5781,
      "step": 3940
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49775031208992004,
      "learning_rate": 9.754991965662055e-05,
      "loss": 2.5992,
      "step": 3945
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4470565915107727,
      "learning_rate": 9.754359612152574e-05,
      "loss": 2.6533,
      "step": 3950
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4485219717025757,
      "learning_rate": 9.753726464207278e-05,
      "loss": 2.6848,
      "step": 3955
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47798603773117065,
      "learning_rate": 9.753092521931963e-05,
      "loss": 2.5366,
      "step": 3960
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4426920413970947,
      "learning_rate": 9.752457785432558e-05,
      "loss": 2.6568,
      "step": 3965
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4463879466056824,
      "learning_rate": 9.751822254815126e-05,
      "loss": 2.7119,
      "step": 3970
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45935317873954773,
      "learning_rate": 9.751185930185861e-05,
      "loss": 2.6394,
      "step": 3975
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4466797411441803,
      "learning_rate": 9.750548811651091e-05,
      "loss": 2.6127,
      "step": 3980
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4424043595790863,
      "learning_rate": 9.749910899317276e-05,
      "loss": 2.5992,
      "step": 3985
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4667627811431885,
      "learning_rate": 9.74927219329101e-05,
      "loss": 2.6434,
      "step": 3990
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4289703965187073,
      "learning_rate": 9.748632693679016e-05,
      "loss": 2.5862,
      "step": 3995
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42398667335510254,
      "learning_rate": 9.747992400588153e-05,
      "loss": 2.6169,
      "step": 4000
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4347371757030487,
      "learning_rate": 9.747351314125413e-05,
      "loss": 2.5686,
      "step": 4005
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4412640929222107,
      "learning_rate": 9.746709434397916e-05,
      "loss": 2.6096,
      "step": 4010
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4930550754070282,
      "learning_rate": 9.746066761512921e-05,
      "loss": 2.6181,
      "step": 4015
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5224257111549377,
      "learning_rate": 9.745423295577815e-05,
      "loss": 2.6227,
      "step": 4020
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4766988456249237,
      "learning_rate": 9.744779036700118e-05,
      "loss": 2.5941,
      "step": 4025
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4755936563014984,
      "learning_rate": 9.744133984987485e-05,
      "loss": 2.6804,
      "step": 4030
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45348021388053894,
      "learning_rate": 9.743488140547701e-05,
      "loss": 2.6266,
      "step": 4035
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4818703532218933,
      "learning_rate": 9.742841503488682e-05,
      "loss": 2.6197,
      "step": 4040
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45373526215553284,
      "learning_rate": 9.742194073918482e-05,
      "loss": 2.4804,
      "step": 4045
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4407598078250885,
      "learning_rate": 9.741545851945282e-05,
      "loss": 2.6297,
      "step": 4050
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4519897997379303,
      "learning_rate": 9.740896837677398e-05,
      "loss": 2.6786,
      "step": 4055
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48574450612068176,
      "learning_rate": 9.740247031223279e-05,
      "loss": 2.6203,
      "step": 4060
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4529514014720917,
      "learning_rate": 9.739596432691502e-05,
      "loss": 2.5712,
      "step": 4065
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44618985056877136,
      "learning_rate": 9.738945042190785e-05,
      "loss": 2.6526,
      "step": 4070
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4756530821323395,
      "learning_rate": 9.738292859829968e-05,
      "loss": 2.611,
      "step": 4075
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.450326144695282,
      "learning_rate": 9.73763988571803e-05,
      "loss": 2.6034,
      "step": 4080
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43218350410461426,
      "learning_rate": 9.73698611996408e-05,
      "loss": 2.6039,
      "step": 4085
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5299659967422485,
      "learning_rate": 9.736331562677361e-05,
      "loss": 2.6434,
      "step": 4090
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47889408469200134,
      "learning_rate": 9.735676213967247e-05,
      "loss": 2.5496,
      "step": 4095
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45277661085128784,
      "learning_rate": 9.735020073943244e-05,
      "loss": 2.6182,
      "step": 4100
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4685920774936676,
      "learning_rate": 9.734363142714988e-05,
      "loss": 2.622,
      "step": 4105
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6399181485176086,
      "learning_rate": 9.733705420392254e-05,
      "loss": 2.6599,
      "step": 4110
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4386637806892395,
      "learning_rate": 9.733046907084944e-05,
      "loss": 2.6315,
      "step": 4115
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46826547384262085,
      "learning_rate": 9.732387602903091e-05,
      "loss": 2.5766,
      "step": 4120
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44339719414711,
      "learning_rate": 9.731727507956862e-05,
      "loss": 2.6128,
      "step": 4125
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4858640134334564,
      "learning_rate": 9.73106662235656e-05,
      "loss": 2.5658,
      "step": 4130
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4632338285446167,
      "learning_rate": 9.730404946212614e-05,
      "loss": 2.688,
      "step": 4135
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4441925585269928,
      "learning_rate": 9.729742479635589e-05,
      "loss": 2.6571,
      "step": 4140
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4671192765235901,
      "learning_rate": 9.729079222736177e-05,
      "loss": 2.5957,
      "step": 4145
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45600539445877075,
      "learning_rate": 9.728415175625211e-05,
      "loss": 2.5289,
      "step": 4150
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4570378065109253,
      "learning_rate": 9.727750338413647e-05,
      "loss": 2.5405,
      "step": 4155
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4476661682128906,
      "learning_rate": 9.727084711212579e-05,
      "loss": 2.7237,
      "step": 4160
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4454682171344757,
      "learning_rate": 9.726418294133229e-05,
      "loss": 2.6134,
      "step": 4165
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4574211537837982,
      "learning_rate": 9.725751087286954e-05,
      "loss": 2.6517,
      "step": 4170
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4431537389755249,
      "learning_rate": 9.725083090785242e-05,
      "loss": 2.6027,
      "step": 4175
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4735550880432129,
      "learning_rate": 9.724414304739712e-05,
      "loss": 2.5341,
      "step": 4180
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44680875539779663,
      "learning_rate": 9.723744729262116e-05,
      "loss": 2.5869,
      "step": 4185
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47717389464378357,
      "learning_rate": 9.723074364464337e-05,
      "loss": 2.5566,
      "step": 4190
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4797333776950836,
      "learning_rate": 9.722403210458393e-05,
      "loss": 2.6359,
      "step": 4195
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47550147771835327,
      "learning_rate": 9.721731267356426e-05,
      "loss": 2.613,
      "step": 4200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4269581437110901,
      "learning_rate": 9.721058535270722e-05,
      "loss": 2.6001,
      "step": 4205
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48927977681159973,
      "learning_rate": 9.720385014313686e-05,
      "loss": 2.6872,
      "step": 4210
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43932586908340454,
      "learning_rate": 9.719710704597865e-05,
      "loss": 2.6198,
      "step": 4215
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4678601622581482,
      "learning_rate": 9.719035606235931e-05,
      "loss": 2.7118,
      "step": 4220
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43632757663726807,
      "learning_rate": 9.718359719340693e-05,
      "loss": 2.6709,
      "step": 4225
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4342629313468933,
      "learning_rate": 9.717683044025086e-05,
      "loss": 2.6531,
      "step": 4230
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4579364061355591,
      "learning_rate": 9.717005580402184e-05,
      "loss": 2.5283,
      "step": 4235
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4306492209434509,
      "learning_rate": 9.716327328585183e-05,
      "loss": 2.5784,
      "step": 4240
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44478484988212585,
      "learning_rate": 9.715648288687421e-05,
      "loss": 2.6207,
      "step": 4245
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5008513331413269,
      "learning_rate": 9.714968460822363e-05,
      "loss": 2.6389,
      "step": 4250
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4930800497531891,
      "learning_rate": 9.714287845103602e-05,
      "loss": 2.5984,
      "step": 4255
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4785988926887512,
      "learning_rate": 9.71360644164487e-05,
      "loss": 2.6706,
      "step": 4260
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44186335802078247,
      "learning_rate": 9.712924250560025e-05,
      "loss": 2.7151,
      "step": 4265
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4594005048274994,
      "learning_rate": 9.712241271963059e-05,
      "loss": 2.7409,
      "step": 4270
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4290255606174469,
      "learning_rate": 9.711557505968097e-05,
      "loss": 2.5715,
      "step": 4275
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4482232332229614,
      "learning_rate": 9.710872952689391e-05,
      "loss": 2.5991,
      "step": 4280
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46905651688575745,
      "learning_rate": 9.71018761224133e-05,
      "loss": 2.5885,
      "step": 4285
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.411348819732666,
      "learning_rate": 9.709501484738429e-05,
      "loss": 2.699,
      "step": 4290
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4318615794181824,
      "learning_rate": 9.70881457029534e-05,
      "loss": 2.6194,
      "step": 4295
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44581833481788635,
      "learning_rate": 9.708126869026841e-05,
      "loss": 2.6426,
      "step": 4300
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4325377643108368,
      "learning_rate": 9.707438381047846e-05,
      "loss": 2.5623,
      "step": 4305
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44676464796066284,
      "learning_rate": 9.7067491064734e-05,
      "loss": 2.6526,
      "step": 4310
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44363999366760254,
      "learning_rate": 9.706059045418676e-05,
      "loss": 2.585,
      "step": 4315
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6325715780258179,
      "learning_rate": 9.705368197998982e-05,
      "loss": 2.6139,
      "step": 4320
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4435577094554901,
      "learning_rate": 9.704676564329756e-05,
      "loss": 2.5785,
      "step": 4325
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4155755937099457,
      "learning_rate": 9.703984144526565e-05,
      "loss": 2.5921,
      "step": 4330
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44520145654678345,
      "learning_rate": 9.703290938705114e-05,
      "loss": 2.5367,
      "step": 4335
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43349963426589966,
      "learning_rate": 9.702596946981232e-05,
      "loss": 2.6454,
      "step": 4340
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47369253635406494,
      "learning_rate": 9.701902169470883e-05,
      "loss": 2.5691,
      "step": 4345
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4687567949295044,
      "learning_rate": 9.701206606290164e-05,
      "loss": 2.6442,
      "step": 4350
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47281041741371155,
      "learning_rate": 9.700510257555295e-05,
      "loss": 2.5152,
      "step": 4355
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44984105229377747,
      "learning_rate": 9.69981312338264e-05,
      "loss": 2.6065,
      "step": 4360
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45966631174087524,
      "learning_rate": 9.699115203888686e-05,
      "loss": 2.6109,
      "step": 4365
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4269980192184448,
      "learning_rate": 9.69841649919005e-05,
      "loss": 2.566,
      "step": 4370
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4424937069416046,
      "learning_rate": 9.697717009403485e-05,
      "loss": 2.643,
      "step": 4375
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44990742206573486,
      "learning_rate": 9.697016734645873e-05,
      "loss": 2.5539,
      "step": 4380
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4754108786582947,
      "learning_rate": 9.696315675034227e-05,
      "loss": 2.5048,
      "step": 4385
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4157191514968872,
      "learning_rate": 9.69561383068569e-05,
      "loss": 2.5281,
      "step": 4390
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4168970286846161,
      "learning_rate": 9.694911201717541e-05,
      "loss": 2.5817,
      "step": 4395
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4412231147289276,
      "learning_rate": 9.694207788247186e-05,
      "loss": 2.5363,
      "step": 4400
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4404055178165436,
      "learning_rate": 9.693503590392159e-05,
      "loss": 2.5046,
      "step": 4405
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4428286552429199,
      "learning_rate": 9.692798608270132e-05,
      "loss": 2.6115,
      "step": 4410
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4464128613471985,
      "learning_rate": 9.692092841998904e-05,
      "loss": 2.6895,
      "step": 4415
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4283076822757721,
      "learning_rate": 9.691386291696407e-05,
      "loss": 2.6267,
      "step": 4420
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45709434151649475,
      "learning_rate": 9.690678957480702e-05,
      "loss": 2.6341,
      "step": 4425
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43516820669174194,
      "learning_rate": 9.689970839469981e-05,
      "loss": 2.5694,
      "step": 4430
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44499579071998596,
      "learning_rate": 9.689261937782571e-05,
      "loss": 2.6526,
      "step": 4435
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44828152656555176,
      "learning_rate": 9.688552252536924e-05,
      "loss": 2.5632,
      "step": 4440
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4838024079799652,
      "learning_rate": 9.687841783851626e-05,
      "loss": 2.5649,
      "step": 4445
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45438045263290405,
      "learning_rate": 9.687130531845394e-05,
      "loss": 2.6173,
      "step": 4450
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4475763440132141,
      "learning_rate": 9.686418496637076e-05,
      "loss": 2.599,
      "step": 4455
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45761996507644653,
      "learning_rate": 9.685705678345652e-05,
      "loss": 2.6634,
      "step": 4460
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45161736011505127,
      "learning_rate": 9.684992077090227e-05,
      "loss": 2.5866,
      "step": 4465
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46867048740386963,
      "learning_rate": 9.684277692990045e-05,
      "loss": 2.5888,
      "step": 4470
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43986010551452637,
      "learning_rate": 9.683562526164475e-05,
      "loss": 2.5694,
      "step": 4475
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4543614089488983,
      "learning_rate": 9.68284657673302e-05,
      "loss": 2.5304,
      "step": 4480
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46665850281715393,
      "learning_rate": 9.682129844815311e-05,
      "loss": 2.5911,
      "step": 4485
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44252681732177734,
      "learning_rate": 9.681412330531112e-05,
      "loss": 2.5747,
      "step": 4490
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46593552827835083,
      "learning_rate": 9.680694034000318e-05,
      "loss": 2.5384,
      "step": 4495
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45436468720436096,
      "learning_rate": 9.67997495534295e-05,
      "loss": 2.605,
      "step": 4500
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4338791072368622,
      "learning_rate": 9.679255094679169e-05,
      "loss": 2.611,
      "step": 4505
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45518308877944946,
      "learning_rate": 9.678534452129256e-05,
      "loss": 2.5403,
      "step": 4510
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45483773946762085,
      "learning_rate": 9.677813027813632e-05,
      "loss": 2.5958,
      "step": 4515
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4658750891685486,
      "learning_rate": 9.677090821852842e-05,
      "loss": 2.5427,
      "step": 4520
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44976553320884705,
      "learning_rate": 9.676367834367562e-05,
      "loss": 2.5522,
      "step": 4525
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45395180583000183,
      "learning_rate": 9.675644065478602e-05,
      "loss": 2.646,
      "step": 4530
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43791964650154114,
      "learning_rate": 9.674919515306904e-05,
      "loss": 2.6335,
      "step": 4535
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4656250774860382,
      "learning_rate": 9.674194183973533e-05,
      "loss": 2.5441,
      "step": 4540
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42730534076690674,
      "learning_rate": 9.673468071599694e-05,
      "loss": 2.6062,
      "step": 4545
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4155958294868469,
      "learning_rate": 9.672741178306713e-05,
      "loss": 2.5697,
      "step": 4550
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4698600471019745,
      "learning_rate": 9.672013504216054e-05,
      "loss": 2.6021,
      "step": 4555
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4398573338985443,
      "learning_rate": 9.671285049449307e-05,
      "loss": 2.497,
      "step": 4560
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42937836050987244,
      "learning_rate": 9.670555814128197e-05,
      "loss": 2.5585,
      "step": 4565
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44492435455322266,
      "learning_rate": 9.669825798374574e-05,
      "loss": 2.4526,
      "step": 4570
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4336588978767395,
      "learning_rate": 9.66909500231042e-05,
      "loss": 2.5417,
      "step": 4575
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44524091482162476,
      "learning_rate": 9.668363426057852e-05,
      "loss": 2.5814,
      "step": 4580
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4591561555862427,
      "learning_rate": 9.66763106973911e-05,
      "loss": 2.6098,
      "step": 4585
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.439449280500412,
      "learning_rate": 9.666897933476568e-05,
      "loss": 2.5368,
      "step": 4590
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4499751925468445,
      "learning_rate": 9.666164017392734e-05,
      "loss": 2.5751,
      "step": 4595
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44472751021385193,
      "learning_rate": 9.665429321610239e-05,
      "loss": 2.6201,
      "step": 4600
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4532940089702606,
      "learning_rate": 9.66469384625185e-05,
      "loss": 2.627,
      "step": 4605
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45582395792007446,
      "learning_rate": 9.66395759144046e-05,
      "loss": 2.622,
      "step": 4610
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4440876543521881,
      "learning_rate": 9.6632205572991e-05,
      "loss": 2.5933,
      "step": 4615
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4551061987876892,
      "learning_rate": 9.662482743950918e-05,
      "loss": 2.5208,
      "step": 4620
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4432244598865509,
      "learning_rate": 9.661744151519206e-05,
      "loss": 2.4986,
      "step": 4625
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4441525340080261,
      "learning_rate": 9.661004780127373e-05,
      "loss": 2.6,
      "step": 4630
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4624142050743103,
      "learning_rate": 9.660264629898974e-05,
      "loss": 2.577,
      "step": 4635
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4363066554069519,
      "learning_rate": 9.65952370095768e-05,
      "loss": 2.6385,
      "step": 4640
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40413373708724976,
      "learning_rate": 9.6587819934273e-05,
      "loss": 2.506,
      "step": 4645
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4493168592453003,
      "learning_rate": 9.658039507431768e-05,
      "loss": 2.6651,
      "step": 4650
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44339892268180847,
      "learning_rate": 9.657296243095151e-05,
      "loss": 2.5818,
      "step": 4655
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4092075526714325,
      "learning_rate": 9.656552200541648e-05,
      "loss": 2.5264,
      "step": 4660
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4383605718612671,
      "learning_rate": 9.655807379895583e-05,
      "loss": 2.6208,
      "step": 4665
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43384894728660583,
      "learning_rate": 9.655061781281413e-05,
      "loss": 2.5516,
      "step": 4670
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47866761684417725,
      "learning_rate": 9.654315404823729e-05,
      "loss": 2.5913,
      "step": 4675
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4314783811569214,
      "learning_rate": 9.653568250647242e-05,
      "loss": 2.5883,
      "step": 4680
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4521118402481079,
      "learning_rate": 9.652820318876802e-05,
      "loss": 2.6091,
      "step": 4685
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44227656722068787,
      "learning_rate": 9.652071609637385e-05,
      "loss": 2.4979,
      "step": 4690
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4537944495677948,
      "learning_rate": 9.651322123054097e-05,
      "loss": 2.5216,
      "step": 4695
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43361300230026245,
      "learning_rate": 9.650571859252175e-05,
      "loss": 2.6356,
      "step": 4700
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43910056352615356,
      "learning_rate": 9.649820818356985e-05,
      "loss": 2.6282,
      "step": 4705
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.48898497223854065,
      "learning_rate": 9.649069000494025e-05,
      "loss": 2.5668,
      "step": 4710
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4533195495605469,
      "learning_rate": 9.648316405788916e-05,
      "loss": 2.6415,
      "step": 4715
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45246779918670654,
      "learning_rate": 9.647563034367419e-05,
      "loss": 2.5469,
      "step": 4720
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42156991362571716,
      "learning_rate": 9.646808886355417e-05,
      "loss": 2.6168,
      "step": 4725
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4632546603679657,
      "learning_rate": 9.646053961878925e-05,
      "loss": 2.6067,
      "step": 4730
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.446268767118454,
      "learning_rate": 9.64529826106409e-05,
      "loss": 2.5601,
      "step": 4735
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42847803235054016,
      "learning_rate": 9.644541784037186e-05,
      "loss": 2.5542,
      "step": 4740
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49971336126327515,
      "learning_rate": 9.643784530924616e-05,
      "loss": 2.589,
      "step": 4745
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4438745677471161,
      "learning_rate": 9.643026501852916e-05,
      "loss": 2.5395,
      "step": 4750
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46424031257629395,
      "learning_rate": 9.64226769694875e-05,
      "loss": 2.5505,
      "step": 4755
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5024244785308838,
      "learning_rate": 9.641508116338908e-05,
      "loss": 2.5667,
      "step": 4760
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5033005475997925,
      "learning_rate": 9.640747760150319e-05,
      "loss": 2.562,
      "step": 4765
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4211534261703491,
      "learning_rate": 9.639986628510032e-05,
      "loss": 2.5875,
      "step": 4770
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43362003564834595,
      "learning_rate": 9.639224721545229e-05,
      "loss": 2.6015,
      "step": 4775
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4554220139980316,
      "learning_rate": 9.638462039383223e-05,
      "loss": 2.5234,
      "step": 4780
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4177861213684082,
      "learning_rate": 9.637698582151454e-05,
      "loss": 2.5345,
      "step": 4785
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5100805759429932,
      "learning_rate": 9.636934349977493e-05,
      "loss": 2.4968,
      "step": 4790
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43652793765068054,
      "learning_rate": 9.636169342989043e-05,
      "loss": 2.5233,
      "step": 4795
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45252400636672974,
      "learning_rate": 9.635403561313932e-05,
      "loss": 2.5636,
      "step": 4800
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4129639267921448,
      "learning_rate": 9.634637005080119e-05,
      "loss": 2.5861,
      "step": 4805
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4277005195617676,
      "learning_rate": 9.633869674415693e-05,
      "loss": 2.5632,
      "step": 4810
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4276449382305145,
      "learning_rate": 9.633101569448873e-05,
      "loss": 2.4764,
      "step": 4815
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4279254674911499,
      "learning_rate": 9.632332690308004e-05,
      "loss": 2.5572,
      "step": 4820
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43134957551956177,
      "learning_rate": 9.631563037121565e-05,
      "loss": 2.5435,
      "step": 4825
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43422985076904297,
      "learning_rate": 9.63079261001816e-05,
      "loss": 2.64,
      "step": 4830
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.407306969165802,
      "learning_rate": 9.630021409126529e-05,
      "loss": 2.5684,
      "step": 4835
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46722519397735596,
      "learning_rate": 9.629249434575531e-05,
      "loss": 2.481,
      "step": 4840
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4341907501220703,
      "learning_rate": 9.628476686494164e-05,
      "loss": 2.6014,
      "step": 4845
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4359488785266876,
      "learning_rate": 9.627703165011551e-05,
      "loss": 2.5593,
      "step": 4850
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4350602924823761,
      "learning_rate": 9.626928870256944e-05,
      "loss": 2.5169,
      "step": 4855
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44680458307266235,
      "learning_rate": 9.626153802359724e-05,
      "loss": 2.6267,
      "step": 4860
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44038641452789307,
      "learning_rate": 9.625377961449402e-05,
      "loss": 2.6353,
      "step": 4865
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4001147150993347,
      "learning_rate": 9.62460134765562e-05,
      "loss": 2.578,
      "step": 4870
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4296320378780365,
      "learning_rate": 9.623823961108145e-05,
      "loss": 2.5781,
      "step": 4875
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46491768956184387,
      "learning_rate": 9.623045801936876e-05,
      "loss": 2.5954,
      "step": 4880
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42220571637153625,
      "learning_rate": 9.622266870271842e-05,
      "loss": 2.5583,
      "step": 4885
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4110945463180542,
      "learning_rate": 9.6214871662432e-05,
      "loss": 2.5336,
      "step": 4890
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4264915883541107,
      "learning_rate": 9.620706689981232e-05,
      "loss": 2.5948,
      "step": 4895
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4374304413795471,
      "learning_rate": 9.619925441616354e-05,
      "loss": 2.6063,
      "step": 4900
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.47758716344833374,
      "learning_rate": 9.619143421279112e-05,
      "loss": 2.5828,
      "step": 4905
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44148120284080505,
      "learning_rate": 9.618360629100176e-05,
      "loss": 2.5563,
      "step": 4910
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4855572283267975,
      "learning_rate": 9.61757706521035e-05,
      "loss": 2.5738,
      "step": 4915
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41679733991622925,
      "learning_rate": 9.616792729740564e-05,
      "loss": 2.5862,
      "step": 4920
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44374290108680725,
      "learning_rate": 9.616007622821877e-05,
      "loss": 2.5352,
      "step": 4925
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4310937821865082,
      "learning_rate": 9.615221744585476e-05,
      "loss": 2.587,
      "step": 4930
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42021870613098145,
      "learning_rate": 9.614435095162679e-05,
      "loss": 2.651,
      "step": 4935
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4361599385738373,
      "learning_rate": 9.613647674684936e-05,
      "loss": 2.6061,
      "step": 4940
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4402942359447479,
      "learning_rate": 9.612859483283818e-05,
      "loss": 2.4921,
      "step": 4945
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44033828377723694,
      "learning_rate": 9.612070521091029e-05,
      "loss": 2.5402,
      "step": 4950
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43730807304382324,
      "learning_rate": 9.611280788238403e-05,
      "loss": 2.5777,
      "step": 4955
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4447525441646576,
      "learning_rate": 9.610490284857901e-05,
      "loss": 2.5933,
      "step": 4960
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.440430223941803,
      "learning_rate": 9.609699011081613e-05,
      "loss": 2.5995,
      "step": 4965
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4453335106372833,
      "learning_rate": 9.608906967041757e-05,
      "loss": 2.5986,
      "step": 4970
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46487849950790405,
      "learning_rate": 9.608114152870684e-05,
      "loss": 2.5707,
      "step": 4975
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4300251305103302,
      "learning_rate": 9.607320568700866e-05,
      "loss": 2.5138,
      "step": 4980
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.422080934047699,
      "learning_rate": 9.606526214664911e-05,
      "loss": 2.4524,
      "step": 4985
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.438266396522522,
      "learning_rate": 9.605731090895551e-05,
      "loss": 2.5602,
      "step": 4990
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48401594161987305,
      "learning_rate": 9.604935197525649e-05,
      "loss": 2.5351,
      "step": 4995
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4296211302280426,
      "learning_rate": 9.604138534688196e-05,
      "loss": 2.5061,
      "step": 5000
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4456224739551544,
      "learning_rate": 9.60334110251631e-05,
      "loss": 2.637,
      "step": 5005
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42975178360939026,
      "learning_rate": 9.602542901143241e-05,
      "loss": 2.5038,
      "step": 5010
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42101457715034485,
      "learning_rate": 9.601743930702363e-05,
      "loss": 2.5973,
      "step": 5015
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43370506167411804,
      "learning_rate": 9.600944191327184e-05,
      "loss": 2.6334,
      "step": 5020
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4759250581264496,
      "learning_rate": 9.600143683151335e-05,
      "loss": 2.6223,
      "step": 5025
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44351065158843994,
      "learning_rate": 9.59934240630858e-05,
      "loss": 2.6192,
      "step": 5030
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4410642087459564,
      "learning_rate": 9.598540360932806e-05,
      "loss": 2.501,
      "step": 5035
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45048609375953674,
      "learning_rate": 9.597737547158035e-05,
      "loss": 2.6066,
      "step": 5040
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4392760992050171,
      "learning_rate": 9.596933965118413e-05,
      "loss": 2.5769,
      "step": 5045
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41479748487472534,
      "learning_rate": 9.596129614948217e-05,
      "loss": 2.5528,
      "step": 5050
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4195507764816284,
      "learning_rate": 9.595324496781849e-05,
      "loss": 2.6871,
      "step": 5055
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42513927817344666,
      "learning_rate": 9.594518610753841e-05,
      "loss": 2.5837,
      "step": 5060
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4424985647201538,
      "learning_rate": 9.593711956998855e-05,
      "loss": 2.5886,
      "step": 5065
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42417261004447937,
      "learning_rate": 9.59290453565168e-05,
      "loss": 2.5443,
      "step": 5070
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4336182773113251,
      "learning_rate": 9.592096346847232e-05,
      "loss": 2.5878,
      "step": 5075
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.435522198677063,
      "learning_rate": 9.591287390720555e-05,
      "loss": 2.524,
      "step": 5080
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42820778489112854,
      "learning_rate": 9.590477667406828e-05,
      "loss": 2.515,
      "step": 5085
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.411154568195343,
      "learning_rate": 9.589667177041347e-05,
      "loss": 2.5546,
      "step": 5090
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.420138955116272,
      "learning_rate": 9.588855919759544e-05,
      "loss": 2.5607,
      "step": 5095
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40477409958839417,
      "learning_rate": 9.588043895696978e-05,
      "loss": 2.6088,
      "step": 5100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4475378096103668,
      "learning_rate": 9.587231104989334e-05,
      "loss": 2.6231,
      "step": 5105
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42219778895378113,
      "learning_rate": 9.586417547772426e-05,
      "loss": 2.5497,
      "step": 5110
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4136791229248047,
      "learning_rate": 9.585603224182196e-05,
      "loss": 2.5629,
      "step": 5115
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43930384516716003,
      "learning_rate": 9.584788134354718e-05,
      "loss": 2.5686,
      "step": 5120
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49346476793289185,
      "learning_rate": 9.583972278426186e-05,
      "loss": 2.5471,
      "step": 5125
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4333929419517517,
      "learning_rate": 9.583155656532929e-05,
      "loss": 2.56,
      "step": 5130
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4307059347629547,
      "learning_rate": 9.582338268811402e-05,
      "loss": 2.6179,
      "step": 5135
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.423790842294693,
      "learning_rate": 9.581520115398185e-05,
      "loss": 2.6053,
      "step": 5140
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4523265063762665,
      "learning_rate": 9.580701196429991e-05,
      "loss": 2.6476,
      "step": 5145
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43349704146385193,
      "learning_rate": 9.579881512043655e-05,
      "loss": 2.5992,
      "step": 5150
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4233090877532959,
      "learning_rate": 9.579061062376146e-05,
      "loss": 2.5276,
      "step": 5155
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4375646114349365,
      "learning_rate": 9.578239847564559e-05,
      "loss": 2.5077,
      "step": 5160
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41347214579582214,
      "learning_rate": 9.577417867746113e-05,
      "loss": 2.5751,
      "step": 5165
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40949127078056335,
      "learning_rate": 9.576595123058158e-05,
      "loss": 2.5428,
      "step": 5170
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4130486249923706,
      "learning_rate": 9.575771613638175e-05,
      "loss": 2.5056,
      "step": 5175
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43017107248306274,
      "learning_rate": 9.574947339623765e-05,
      "loss": 2.5282,
      "step": 5180
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4884456694126129,
      "learning_rate": 9.574122301152665e-05,
      "loss": 2.5874,
      "step": 5185
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4529236853122711,
      "learning_rate": 9.573296498362733e-05,
      "loss": 2.5139,
      "step": 5190
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4587782621383667,
      "learning_rate": 9.572469931391959e-05,
      "loss": 2.4721,
      "step": 5195
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4889732599258423,
      "learning_rate": 9.571642600378459e-05,
      "loss": 2.516,
      "step": 5200
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43909093737602234,
      "learning_rate": 9.570814505460475e-05,
      "loss": 2.4538,
      "step": 5205
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4286167323589325,
      "learning_rate": 9.569985646776382e-05,
      "loss": 2.574,
      "step": 5210
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4089890420436859,
      "learning_rate": 9.569156024464677e-05,
      "loss": 2.6044,
      "step": 5215
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4282733201980591,
      "learning_rate": 9.568325638663989e-05,
      "loss": 2.6229,
      "step": 5220
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4149312674999237,
      "learning_rate": 9.567494489513068e-05,
      "loss": 2.551,
      "step": 5225
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4285503923892975,
      "learning_rate": 9.5666625771508e-05,
      "loss": 2.5864,
      "step": 5230
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4194752275943756,
      "learning_rate": 9.565829901716194e-05,
      "loss": 2.527,
      "step": 5235
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4287759065628052,
      "learning_rate": 9.564996463348384e-05,
      "loss": 2.5907,
      "step": 5240
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4518637955188751,
      "learning_rate": 9.564162262186638e-05,
      "loss": 2.4847,
      "step": 5245
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42148494720458984,
      "learning_rate": 9.563327298370347e-05,
      "loss": 2.5951,
      "step": 5250
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4316718876361847,
      "learning_rate": 9.56249157203903e-05,
      "loss": 2.5824,
      "step": 5255
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45053818821907043,
      "learning_rate": 9.561655083332333e-05,
      "loss": 2.5813,
      "step": 5260
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43530189990997314,
      "learning_rate": 9.56081783239003e-05,
      "loss": 2.4909,
      "step": 5265
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4058390259742737,
      "learning_rate": 9.559979819352024e-05,
      "loss": 2.5989,
      "step": 5270
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43891143798828125,
      "learning_rate": 9.559141044358342e-05,
      "loss": 2.5636,
      "step": 5275
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4275728762149811,
      "learning_rate": 9.558301507549143e-05,
      "loss": 2.6547,
      "step": 5280
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4459207355976105,
      "learning_rate": 9.557461209064708e-05,
      "loss": 2.596,
      "step": 5285
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4032520055770874,
      "learning_rate": 9.556620149045447e-05,
      "loss": 2.5369,
      "step": 5290
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44391438364982605,
      "learning_rate": 9.555778327631903e-05,
      "loss": 2.4768,
      "step": 5295
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42260050773620605,
      "learning_rate": 9.554935744964734e-05,
      "loss": 2.5229,
      "step": 5300
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4283473789691925,
      "learning_rate": 9.55409240118474e-05,
      "loss": 2.5996,
      "step": 5305
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43201181292533875,
      "learning_rate": 9.553248296432835e-05,
      "loss": 2.6291,
      "step": 5310
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45650261640548706,
      "learning_rate": 9.552403430850068e-05,
      "loss": 2.5919,
      "step": 5315
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42137759923934937,
      "learning_rate": 9.551557804577612e-05,
      "loss": 2.6472,
      "step": 5320
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41298148036003113,
      "learning_rate": 9.550711417756767e-05,
      "loss": 2.5127,
      "step": 5325
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42275604605674744,
      "learning_rate": 9.549864270528965e-05,
      "loss": 2.4682,
      "step": 5330
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43461498618125916,
      "learning_rate": 9.549016363035758e-05,
      "loss": 2.6472,
      "step": 5335
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41209277510643005,
      "learning_rate": 9.548167695418828e-05,
      "loss": 2.5747,
      "step": 5340
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42800062894821167,
      "learning_rate": 9.547318267819986e-05,
      "loss": 2.6269,
      "step": 5345
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4129374921321869,
      "learning_rate": 9.546468080381167e-05,
      "loss": 2.5402,
      "step": 5350
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47431594133377075,
      "learning_rate": 9.545617133244436e-05,
      "loss": 2.5142,
      "step": 5355
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42093464732170105,
      "learning_rate": 9.544765426551981e-05,
      "loss": 2.46,
      "step": 5360
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46347492933273315,
      "learning_rate": 9.543912960446117e-05,
      "loss": 2.5507,
      "step": 5365
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4226077198982239,
      "learning_rate": 9.543059735069293e-05,
      "loss": 2.5043,
      "step": 5370
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43685558438301086,
      "learning_rate": 9.542205750564077e-05,
      "loss": 2.6058,
      "step": 5375
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4290757477283478,
      "learning_rate": 9.541351007073167e-05,
      "loss": 2.5762,
      "step": 5380
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.467733770608902,
      "learning_rate": 9.540495504739387e-05,
      "loss": 2.4916,
      "step": 5385
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5029346346855164,
      "learning_rate": 9.53963924370569e-05,
      "loss": 2.5409,
      "step": 5390
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4434680640697479,
      "learning_rate": 9.538782224115151e-05,
      "loss": 2.601,
      "step": 5395
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43221840262413025,
      "learning_rate": 9.537924446110976e-05,
      "loss": 2.5426,
      "step": 5400
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4162624180316925,
      "learning_rate": 9.537065909836497e-05,
      "loss": 2.5096,
      "step": 5405
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.441475510597229,
      "learning_rate": 9.536206615435174e-05,
      "loss": 2.551,
      "step": 5410
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4654933214187622,
      "learning_rate": 9.535346563050591e-05,
      "loss": 2.5766,
      "step": 5415
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4277030825614929,
      "learning_rate": 9.534485752826456e-05,
      "loss": 2.4602,
      "step": 5420
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44299617409706116,
      "learning_rate": 9.53362418490661e-05,
      "loss": 2.4586,
      "step": 5425
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42347922921180725,
      "learning_rate": 9.53276185943502e-05,
      "loss": 2.4589,
      "step": 5430
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.466861367225647,
      "learning_rate": 9.531898776555773e-05,
      "loss": 2.5418,
      "step": 5435
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4284459054470062,
      "learning_rate": 9.531034936413091e-05,
      "loss": 2.5211,
      "step": 5440
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4222709834575653,
      "learning_rate": 9.530170339151314e-05,
      "loss": 2.507,
      "step": 5445
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44306549429893494,
      "learning_rate": 9.529304984914917e-05,
      "loss": 2.515,
      "step": 5450
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4236861765384674,
      "learning_rate": 9.528438873848496e-05,
      "loss": 2.496,
      "step": 5455
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4135424792766571,
      "learning_rate": 9.527572006096773e-05,
      "loss": 2.5361,
      "step": 5460
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4338001608848572,
      "learning_rate": 9.526704381804601e-05,
      "loss": 2.5258,
      "step": 5465
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41204121708869934,
      "learning_rate": 9.525836001116957e-05,
      "loss": 2.4493,
      "step": 5470
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42316964268684387,
      "learning_rate": 9.524966864178943e-05,
      "loss": 2.4919,
      "step": 5475
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43602845072746277,
      "learning_rate": 9.524096971135789e-05,
      "loss": 2.538,
      "step": 5480
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43299898505210876,
      "learning_rate": 9.52322632213285e-05,
      "loss": 2.558,
      "step": 5485
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4431018531322479,
      "learning_rate": 9.52235491731561e-05,
      "loss": 2.4424,
      "step": 5490
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.440341055393219,
      "learning_rate": 9.521482756829675e-05,
      "loss": 2.5063,
      "step": 5495
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4384385049343109,
      "learning_rate": 9.520609840820782e-05,
      "loss": 2.5645,
      "step": 5500
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42498359084129333,
      "learning_rate": 9.51973616943479e-05,
      "loss": 2.6056,
      "step": 5505
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4267031252384186,
      "learning_rate": 9.518861742817689e-05,
      "loss": 2.5699,
      "step": 5510
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4264640808105469,
      "learning_rate": 9.51798656111559e-05,
      "loss": 2.5362,
      "step": 5515
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45963427424430847,
      "learning_rate": 9.517110624474733e-05,
      "loss": 2.5916,
      "step": 5520
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43770313262939453,
      "learning_rate": 9.516233933041485e-05,
      "loss": 2.5805,
      "step": 5525
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.346693754196167,
      "learning_rate": 9.515356486962337e-05,
      "loss": 2.5612,
      "step": 5530
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41638097167015076,
      "learning_rate": 9.514478286383907e-05,
      "loss": 2.4294,
      "step": 5535
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4298093914985657,
      "learning_rate": 9.513599331452939e-05,
      "loss": 2.5034,
      "step": 5540
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46153974533081055,
      "learning_rate": 9.512719622316304e-05,
      "loss": 2.5732,
      "step": 5545
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4192149341106415,
      "learning_rate": 9.511839159120997e-05,
      "loss": 2.4883,
      "step": 5550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4279155731201172,
      "learning_rate": 9.51095794201414e-05,
      "loss": 2.557,
      "step": 5555
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42389431595802307,
      "learning_rate": 9.510075971142983e-05,
      "loss": 2.5985,
      "step": 5560
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4731607437133789,
      "learning_rate": 9.509193246654896e-05,
      "loss": 2.4947,
      "step": 5565
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4982207715511322,
      "learning_rate": 9.508309768697384e-05,
      "loss": 2.533,
      "step": 5570
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3958958685398102,
      "learning_rate": 9.507425537418068e-05,
      "loss": 2.551,
      "step": 5575
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44019371271133423,
      "learning_rate": 9.506540552964704e-05,
      "loss": 2.4823,
      "step": 5580
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44092121720314026,
      "learning_rate": 9.505654815485168e-05,
      "loss": 2.5443,
      "step": 5585
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44609081745147705,
      "learning_rate": 9.504768325127464e-05,
      "loss": 2.5527,
      "step": 5590
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39678052067756653,
      "learning_rate": 9.503881082039719e-05,
      "loss": 2.5776,
      "step": 5595
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4180943965911865,
      "learning_rate": 9.50299308637019e-05,
      "loss": 2.5484,
      "step": 5600
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42471492290496826,
      "learning_rate": 9.502104338267257e-05,
      "loss": 2.6276,
      "step": 5605
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45049214363098145,
      "learning_rate": 9.501214837879428e-05,
      "loss": 2.5514,
      "step": 5610
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4619409739971161,
      "learning_rate": 9.500324585355333e-05,
      "loss": 2.4741,
      "step": 5615
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43255501985549927,
      "learning_rate": 9.499433580843732e-05,
      "loss": 2.5411,
      "step": 5620
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45360127091407776,
      "learning_rate": 9.498541824493506e-05,
      "loss": 2.6313,
      "step": 5625
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45646053552627563,
      "learning_rate": 9.497649316453667e-05,
      "loss": 2.4675,
      "step": 5630
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4518727660179138,
      "learning_rate": 9.496756056873348e-05,
      "loss": 2.4872,
      "step": 5635
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4333740770816803,
      "learning_rate": 9.49586204590181e-05,
      "loss": 2.5402,
      "step": 5640
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41719987988471985,
      "learning_rate": 9.494967283688438e-05,
      "loss": 2.6018,
      "step": 5645
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41066136956214905,
      "learning_rate": 9.494071770382745e-05,
      "loss": 2.4999,
      "step": 5650
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3989674150943756,
      "learning_rate": 9.493175506134367e-05,
      "loss": 2.5493,
      "step": 5655
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40897610783576965,
      "learning_rate": 9.492278491093067e-05,
      "loss": 2.5707,
      "step": 5660
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44172096252441406,
      "learning_rate": 9.491380725408731e-05,
      "loss": 2.4938,
      "step": 5665
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41193971037864685,
      "learning_rate": 9.490482209231374e-05,
      "loss": 2.5362,
      "step": 5670
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41603243350982666,
      "learning_rate": 9.489582942711135e-05,
      "loss": 2.594,
      "step": 5675
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4025401473045349,
      "learning_rate": 9.488682925998277e-05,
      "loss": 2.5129,
      "step": 5680
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4339984059333801,
      "learning_rate": 9.48778215924319e-05,
      "loss": 2.578,
      "step": 5685
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42660099267959595,
      "learning_rate": 9.48688064259639e-05,
      "loss": 2.4979,
      "step": 5690
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43978646397590637,
      "learning_rate": 9.485978376208513e-05,
      "loss": 2.5004,
      "step": 5695
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4124844968318939,
      "learning_rate": 9.485075360230328e-05,
      "loss": 2.5435,
      "step": 5700
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4172883927822113,
      "learning_rate": 9.484171594812728e-05,
      "loss": 2.4969,
      "step": 5705
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42192554473876953,
      "learning_rate": 9.483267080106723e-05,
      "loss": 2.5246,
      "step": 5710
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4383605718612671,
      "learning_rate": 9.482361816263456e-05,
      "loss": 2.5761,
      "step": 5715
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4235226809978485,
      "learning_rate": 9.481455803434195e-05,
      "loss": 2.4969,
      "step": 5720
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45171162486076355,
      "learning_rate": 9.48054904177033e-05,
      "loss": 2.5089,
      "step": 5725
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4334436058998108,
      "learning_rate": 9.479641531423377e-05,
      "loss": 2.5009,
      "step": 5730
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41093477606773376,
      "learning_rate": 9.47873327254498e-05,
      "loss": 2.5176,
      "step": 5735
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45536521077156067,
      "learning_rate": 9.477824265286904e-05,
      "loss": 2.5228,
      "step": 5740
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4394491910934448,
      "learning_rate": 9.476914509801039e-05,
      "loss": 2.4562,
      "step": 5745
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44092652201652527,
      "learning_rate": 9.476004006239406e-05,
      "loss": 2.5846,
      "step": 5750
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45242786407470703,
      "learning_rate": 9.475092754754142e-05,
      "loss": 2.4997,
      "step": 5755
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41245540976524353,
      "learning_rate": 9.474180755497517e-05,
      "loss": 2.5748,
      "step": 5760
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4283467233181,
      "learning_rate": 9.473268008621922e-05,
      "loss": 2.5342,
      "step": 5765
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4353565275669098,
      "learning_rate": 9.472354514279874e-05,
      "loss": 2.5031,
      "step": 5770
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41626742482185364,
      "learning_rate": 9.471440272624012e-05,
      "loss": 2.483,
      "step": 5775
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41810423135757446,
      "learning_rate": 9.470525283807105e-05,
      "loss": 2.5137,
      "step": 5780
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4157010316848755,
      "learning_rate": 9.469609547982044e-05,
      "loss": 2.3825,
      "step": 5785
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4133671224117279,
      "learning_rate": 9.468693065301843e-05,
      "loss": 2.4825,
      "step": 5790
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4114730656147003,
      "learning_rate": 9.467775835919646e-05,
      "loss": 2.5706,
      "step": 5795
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46839776635169983,
      "learning_rate": 9.466857859988715e-05,
      "loss": 2.506,
      "step": 5800
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40094703435897827,
      "learning_rate": 9.465939137662444e-05,
      "loss": 2.5342,
      "step": 5805
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42107337713241577,
      "learning_rate": 9.465019669094346e-05,
      "loss": 2.5366,
      "step": 5810
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.440297931432724,
      "learning_rate": 9.464099454438062e-05,
      "loss": 2.4869,
      "step": 5815
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4051035940647125,
      "learning_rate": 9.463178493847355e-05,
      "loss": 2.4872,
      "step": 5820
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43100571632385254,
      "learning_rate": 9.462256787476113e-05,
      "loss": 2.5312,
      "step": 5825
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40120089054107666,
      "learning_rate": 9.461334335478354e-05,
      "loss": 2.5642,
      "step": 5830
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3974968194961548,
      "learning_rate": 9.460411138008214e-05,
      "loss": 2.4707,
      "step": 5835
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3964753746986389,
      "learning_rate": 9.459487195219954e-05,
      "loss": 2.503,
      "step": 5840
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4223116934299469,
      "learning_rate": 9.458562507267963e-05,
      "loss": 2.5136,
      "step": 5845
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4102421998977661,
      "learning_rate": 9.457637074306754e-05,
      "loss": 2.4646,
      "step": 5850
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46379172801971436,
      "learning_rate": 9.456710896490961e-05,
      "loss": 2.4874,
      "step": 5855
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42001834511756897,
      "learning_rate": 9.455783973975348e-05,
      "loss": 2.4373,
      "step": 5860
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44193732738494873,
      "learning_rate": 9.454856306914797e-05,
      "loss": 2.506,
      "step": 5865
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42565906047821045,
      "learning_rate": 9.453927895464321e-05,
      "loss": 2.5155,
      "step": 5870
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.437845915555954,
      "learning_rate": 9.452998739779051e-05,
      "loss": 2.5829,
      "step": 5875
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44047537446022034,
      "learning_rate": 9.452068840014246e-05,
      "loss": 2.5216,
      "step": 5880
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43011894822120667,
      "learning_rate": 9.451138196325291e-05,
      "loss": 2.484,
      "step": 5885
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43121519684791565,
      "learning_rate": 9.45020680886769e-05,
      "loss": 2.5268,
      "step": 5890
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43038368225097656,
      "learning_rate": 9.449274677797078e-05,
      "loss": 2.5133,
      "step": 5895
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4766116738319397,
      "learning_rate": 9.448341803269208e-05,
      "loss": 2.4613,
      "step": 5900
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42387720942497253,
      "learning_rate": 9.44740818543996e-05,
      "loss": 2.4983,
      "step": 5905
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4513736665248871,
      "learning_rate": 9.44647382446534e-05,
      "loss": 2.542,
      "step": 5910
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40445801615715027,
      "learning_rate": 9.445538720501474e-05,
      "loss": 2.504,
      "step": 5915
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4689542055130005,
      "learning_rate": 9.444602873704616e-05,
      "loss": 2.444,
      "step": 5920
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4157983064651489,
      "learning_rate": 9.443666284231141e-05,
      "loss": 2.4803,
      "step": 5925
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4009268879890442,
      "learning_rate": 9.44272895223755e-05,
      "loss": 2.4565,
      "step": 5930
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4230417311191559,
      "learning_rate": 9.441790877880469e-05,
      "loss": 2.515,
      "step": 5935
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42531412839889526,
      "learning_rate": 9.440852061316645e-05,
      "loss": 2.5409,
      "step": 5940
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4243563413619995,
      "learning_rate": 9.439912502702952e-05,
      "loss": 2.6145,
      "step": 5945
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45384442806243896,
      "learning_rate": 9.438972202196387e-05,
      "loss": 2.5042,
      "step": 5950
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4335915744304657,
      "learning_rate": 9.43803115995407e-05,
      "loss": 2.5155,
      "step": 5955
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4366929531097412,
      "learning_rate": 9.437089376133246e-05,
      "loss": 2.4769,
      "step": 5960
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4301469922065735,
      "learning_rate": 9.436146850891283e-05,
      "loss": 2.531,
      "step": 5965
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44979313015937805,
      "learning_rate": 9.435203584385673e-05,
      "loss": 2.4787,
      "step": 5970
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4250316321849823,
      "learning_rate": 9.434259576774034e-05,
      "loss": 2.5235,
      "step": 5975
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45758256316185,
      "learning_rate": 9.433314828214103e-05,
      "loss": 2.5553,
      "step": 5980
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4245993494987488,
      "learning_rate": 9.43236933886375e-05,
      "loss": 2.6072,
      "step": 5985
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41855931282043457,
      "learning_rate": 9.431423108880957e-05,
      "loss": 2.4333,
      "step": 5990
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4170442223548889,
      "learning_rate": 9.430476138423838e-05,
      "loss": 2.4354,
      "step": 5995
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3966996967792511,
      "learning_rate": 9.429528427650627e-05,
      "loss": 2.4938,
      "step": 6000
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40678006410598755,
      "learning_rate": 9.428579976719684e-05,
      "loss": 2.4928,
      "step": 6005
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41005849838256836,
      "learning_rate": 9.427630785789491e-05,
      "loss": 2.5519,
      "step": 6010
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42331594228744507,
      "learning_rate": 9.426680855018654e-05,
      "loss": 2.4933,
      "step": 6015
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4261105954647064,
      "learning_rate": 9.425730184565907e-05,
      "loss": 2.4857,
      "step": 6020
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41394850611686707,
      "learning_rate": 9.424778774590096e-05,
      "loss": 2.4394,
      "step": 6025
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40519389510154724,
      "learning_rate": 9.423826625250204e-05,
      "loss": 2.6161,
      "step": 6030
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42782995104789734,
      "learning_rate": 9.422873736705329e-05,
      "loss": 2.5314,
      "step": 6035
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.448547899723053,
      "learning_rate": 9.421920109114696e-05,
      "loss": 2.4241,
      "step": 6040
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40832725167274475,
      "learning_rate": 9.420965742637653e-05,
      "loss": 2.4734,
      "step": 6045
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4063279330730438,
      "learning_rate": 9.420010637433669e-05,
      "loss": 2.5917,
      "step": 6050
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3904539942741394,
      "learning_rate": 9.419054793662342e-05,
      "loss": 2.491,
      "step": 6055
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4014335572719574,
      "learning_rate": 9.418098211483387e-05,
      "loss": 2.4911,
      "step": 6060
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41656240820884705,
      "learning_rate": 9.417140891056646e-05,
      "loss": 2.5156,
      "step": 6065
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4271679222583771,
      "learning_rate": 9.416182832542085e-05,
      "loss": 2.5758,
      "step": 6070
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4332314133644104,
      "learning_rate": 9.41522403609979e-05,
      "loss": 2.4711,
      "step": 6075
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4536721408367157,
      "learning_rate": 9.414264501889975e-05,
      "loss": 2.4694,
      "step": 6080
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4274296760559082,
      "learning_rate": 9.41330423007297e-05,
      "loss": 2.4903,
      "step": 6085
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42725637555122375,
      "learning_rate": 9.412343220809239e-05,
      "loss": 2.4362,
      "step": 6090
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43876078724861145,
      "learning_rate": 9.41138147425936e-05,
      "loss": 2.4969,
      "step": 6095
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4261208772659302,
      "learning_rate": 9.410418990584034e-05,
      "loss": 2.5327,
      "step": 6100
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4307292699813843,
      "learning_rate": 9.409455769944093e-05,
      "loss": 2.434,
      "step": 6105
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4401223063468933,
      "learning_rate": 9.408491812500486e-05,
      "loss": 2.4204,
      "step": 6110
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41935160756111145,
      "learning_rate": 9.407527118414287e-05,
      "loss": 2.5014,
      "step": 6115
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4154062867164612,
      "learning_rate": 9.406561687846693e-05,
      "loss": 2.5266,
      "step": 6120
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40132319927215576,
      "learning_rate": 9.405595520959025e-05,
      "loss": 2.5213,
      "step": 6125
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41811245679855347,
      "learning_rate": 9.404628617912723e-05,
      "loss": 2.4938,
      "step": 6130
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44753867387771606,
      "learning_rate": 9.403660978869353e-05,
      "loss": 2.5974,
      "step": 6135
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4059959352016449,
      "learning_rate": 9.402692603990608e-05,
      "loss": 2.5081,
      "step": 6140
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40733474493026733,
      "learning_rate": 9.401723493438295e-05,
      "loss": 2.5523,
      "step": 6145
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42706725001335144,
      "learning_rate": 9.400753647374352e-05,
      "loss": 2.5225,
      "step": 6150
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46776634454727173,
      "learning_rate": 9.399783065960835e-05,
      "loss": 2.5229,
      "step": 6155
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4308278262615204,
      "learning_rate": 9.398811749359924e-05,
      "loss": 2.5423,
      "step": 6160
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44025126099586487,
      "learning_rate": 9.397839697733924e-05,
      "loss": 2.5529,
      "step": 6165
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4009905755519867,
      "learning_rate": 9.396866911245261e-05,
      "loss": 2.5243,
      "step": 6170
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.418181449174881,
      "learning_rate": 9.395893390056484e-05,
      "loss": 2.504,
      "step": 6175
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42255473136901855,
      "learning_rate": 9.394919134330264e-05,
      "loss": 2.5456,
      "step": 6180
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40619102120399475,
      "learning_rate": 9.393944144229396e-05,
      "loss": 2.5342,
      "step": 6185
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4330846071243286,
      "learning_rate": 9.392968419916797e-05,
      "loss": 2.5732,
      "step": 6190
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4079923629760742,
      "learning_rate": 9.391991961555507e-05,
      "loss": 2.5135,
      "step": 6195
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40318113565444946,
      "learning_rate": 9.39101476930869e-05,
      "loss": 2.4885,
      "step": 6200
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40537670254707336,
      "learning_rate": 9.390036843339629e-05,
      "loss": 2.5122,
      "step": 6205
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4027025103569031,
      "learning_rate": 9.389058183811732e-05,
      "loss": 2.591,
      "step": 6210
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4152447283267975,
      "learning_rate": 9.38807879088853e-05,
      "loss": 2.5412,
      "step": 6215
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41206973791122437,
      "learning_rate": 9.387098664733676e-05,
      "loss": 2.5009,
      "step": 6220
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3989275395870209,
      "learning_rate": 9.386117805510945e-05,
      "loss": 2.6119,
      "step": 6225
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.39186277985572815,
      "learning_rate": 9.385136213384237e-05,
      "loss": 2.5298,
      "step": 6230
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40288278460502625,
      "learning_rate": 9.38415388851757e-05,
      "loss": 2.5343,
      "step": 6235
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4469844698905945,
      "learning_rate": 9.383170831075086e-05,
      "loss": 2.518,
      "step": 6240
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3867177963256836,
      "learning_rate": 9.382187041221054e-05,
      "loss": 2.4931,
      "step": 6245
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43714529275894165,
      "learning_rate": 9.381202519119858e-05,
      "loss": 2.5436,
      "step": 6250
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4237934350967407,
      "learning_rate": 9.38021726493601e-05,
      "loss": 2.474,
      "step": 6255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4037536084651947,
      "learning_rate": 9.37923127883414e-05,
      "loss": 2.5152,
      "step": 6260
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40903440117836,
      "learning_rate": 9.378244560979007e-05,
      "loss": 2.476,
      "step": 6265
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4252000153064728,
      "learning_rate": 9.377257111535482e-05,
      "loss": 2.5809,
      "step": 6270
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4150964915752411,
      "learning_rate": 9.376268930668569e-05,
      "loss": 2.5586,
      "step": 6275
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3887867033481598,
      "learning_rate": 9.375280018543387e-05,
      "loss": 2.357,
      "step": 6280
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42077672481536865,
      "learning_rate": 9.374290375325181e-05,
      "loss": 2.4795,
      "step": 6285
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41463419795036316,
      "learning_rate": 9.373300001179316e-05,
      "loss": 2.4307,
      "step": 6290
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41650158166885376,
      "learning_rate": 9.37230889627128e-05,
      "loss": 2.5546,
      "step": 6295
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40060704946517944,
      "learning_rate": 9.371317060766682e-05,
      "loss": 2.4832,
      "step": 6300
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4150589406490326,
      "learning_rate": 9.370324494831254e-05,
      "loss": 2.5249,
      "step": 6305
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42641910910606384,
      "learning_rate": 9.369331198630852e-05,
      "loss": 2.4731,
      "step": 6310
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4160577952861786,
      "learning_rate": 9.36833717233145e-05,
      "loss": 2.4472,
      "step": 6315
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3839844763278961,
      "learning_rate": 9.367342416099147e-05,
      "loss": 2.5612,
      "step": 6320
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.39857152104377747,
      "learning_rate": 9.366346930100163e-05,
      "loss": 2.4434,
      "step": 6325
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3996109366416931,
      "learning_rate": 9.36535071450084e-05,
      "loss": 2.498,
      "step": 6330
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41312164068222046,
      "learning_rate": 9.364353769467643e-05,
      "loss": 2.4848,
      "step": 6335
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4294893741607666,
      "learning_rate": 9.363356095167156e-05,
      "loss": 2.5063,
      "step": 6340
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4264080822467804,
      "learning_rate": 9.362357691766088e-05,
      "loss": 2.4386,
      "step": 6345
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40423136949539185,
      "learning_rate": 9.361358559431266e-05,
      "loss": 2.5322,
      "step": 6350
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40751513838768005,
      "learning_rate": 9.360358698329646e-05,
      "loss": 2.4761,
      "step": 6355
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4352635145187378,
      "learning_rate": 9.359358108628298e-05,
      "loss": 2.4545,
      "step": 6360
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41734710335731506,
      "learning_rate": 9.358356790494416e-05,
      "loss": 2.3676,
      "step": 6365
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4370875060558319,
      "learning_rate": 9.357354744095317e-05,
      "loss": 2.4345,
      "step": 6370
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39757782220840454,
      "learning_rate": 9.35635196959844e-05,
      "loss": 2.4105,
      "step": 6375
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4083496928215027,
      "learning_rate": 9.355348467171347e-05,
      "loss": 2.3961,
      "step": 6380
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4267670214176178,
      "learning_rate": 9.354344236981716e-05,
      "loss": 2.5423,
      "step": 6385
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3821193277835846,
      "learning_rate": 9.353339279197351e-05,
      "loss": 2.5138,
      "step": 6390
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3872281014919281,
      "learning_rate": 9.352333593986176e-05,
      "loss": 2.5522,
      "step": 6395
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40003877878189087,
      "learning_rate": 9.35132718151624e-05,
      "loss": 2.5294,
      "step": 6400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3964272737503052,
      "learning_rate": 9.35032004195571e-05,
      "loss": 2.4613,
      "step": 6405
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4151190221309662,
      "learning_rate": 9.349312175472873e-05,
      "loss": 2.5898,
      "step": 6410
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41053223609924316,
      "learning_rate": 9.348303582236142e-05,
      "loss": 2.4357,
      "step": 6415
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39696887135505676,
      "learning_rate": 9.347294262414047e-05,
      "loss": 2.4269,
      "step": 6420
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.392342209815979,
      "learning_rate": 9.346284216175245e-05,
      "loss": 2.4678,
      "step": 6425
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4583955407142639,
      "learning_rate": 9.345273443688509e-05,
      "loss": 2.4904,
      "step": 6430
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3801257908344269,
      "learning_rate": 9.344261945122734e-05,
      "loss": 2.5148,
      "step": 6435
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4021928310394287,
      "learning_rate": 9.343249720646939e-05,
      "loss": 2.4796,
      "step": 6440
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4125891625881195,
      "learning_rate": 9.342236770430264e-05,
      "loss": 2.4888,
      "step": 6445
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41641175746917725,
      "learning_rate": 9.341223094641967e-05,
      "loss": 2.4861,
      "step": 6450
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3939574658870697,
      "learning_rate": 9.340208693451431e-05,
      "loss": 2.404,
      "step": 6455
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3980167508125305,
      "learning_rate": 9.339193567028157e-05,
      "loss": 2.5161,
      "step": 6460
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.481575071811676,
      "learning_rate": 9.338177715541772e-05,
      "loss": 2.4358,
      "step": 6465
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40373948216438293,
      "learning_rate": 9.337161139162019e-05,
      "loss": 2.5084,
      "step": 6470
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42019030451774597,
      "learning_rate": 9.336143838058761e-05,
      "loss": 2.5189,
      "step": 6475
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41629448533058167,
      "learning_rate": 9.335125812401993e-05,
      "loss": 2.4289,
      "step": 6480
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42983371019363403,
      "learning_rate": 9.334107062361816e-05,
      "loss": 2.5139,
      "step": 6485
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41714584827423096,
      "learning_rate": 9.333087588108462e-05,
      "loss": 2.474,
      "step": 6490
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4060211777687073,
      "learning_rate": 9.332067389812282e-05,
      "loss": 2.4609,
      "step": 6495
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4235735833644867,
      "learning_rate": 9.331046467643746e-05,
      "loss": 2.4865,
      "step": 6500
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.386515736579895,
      "learning_rate": 9.330024821773448e-05,
      "loss": 2.567,
      "step": 6505
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42633599042892456,
      "learning_rate": 9.329002452372102e-05,
      "loss": 2.5047,
      "step": 6510
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4093624949455261,
      "learning_rate": 9.327979359610537e-05,
      "loss": 2.5174,
      "step": 6515
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41147568821907043,
      "learning_rate": 9.326955543659714e-05,
      "loss": 2.5322,
      "step": 6520
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3943870961666107,
      "learning_rate": 9.325931004690706e-05,
      "loss": 2.4689,
      "step": 6525
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4247778058052063,
      "learning_rate": 9.324905742874709e-05,
      "loss": 2.4517,
      "step": 6530
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44162097573280334,
      "learning_rate": 9.323879758383042e-05,
      "loss": 2.4522,
      "step": 6535
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43372026085853577,
      "learning_rate": 9.322853051387141e-05,
      "loss": 2.5564,
      "step": 6540
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4204712510108948,
      "learning_rate": 9.321825622058568e-05,
      "loss": 2.4726,
      "step": 6545
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38143154978752136,
      "learning_rate": 9.320797470569001e-05,
      "loss": 2.4427,
      "step": 6550
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4082615375518799,
      "learning_rate": 9.319768597090242e-05,
      "loss": 2.4437,
      "step": 6555
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3912806212902069,
      "learning_rate": 9.318739001794208e-05,
      "loss": 2.5644,
      "step": 6560
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4614407420158386,
      "learning_rate": 9.317708684852942e-05,
      "loss": 2.5289,
      "step": 6565
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41859331727027893,
      "learning_rate": 9.31667764643861e-05,
      "loss": 2.4063,
      "step": 6570
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.406552255153656,
      "learning_rate": 9.31564588672349e-05,
      "loss": 2.4755,
      "step": 6575
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41230008006095886,
      "learning_rate": 9.314613405879986e-05,
      "loss": 2.4945,
      "step": 6580
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4150064289569855,
      "learning_rate": 9.313580204080623e-05,
      "loss": 2.4476,
      "step": 6585
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40132513642311096,
      "learning_rate": 9.312546281498047e-05,
      "loss": 2.4392,
      "step": 6590
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42388832569122314,
      "learning_rate": 9.311511638305018e-05,
      "loss": 2.4992,
      "step": 6595
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40124279260635376,
      "learning_rate": 9.310476274674425e-05,
      "loss": 2.5194,
      "step": 6600
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3785821497440338,
      "learning_rate": 9.30944019077927e-05,
      "loss": 2.3636,
      "step": 6605
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39865389466285706,
      "learning_rate": 9.30840338679268e-05,
      "loss": 2.4969,
      "step": 6610
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3903028666973114,
      "learning_rate": 9.307365862887905e-05,
      "loss": 2.5146,
      "step": 6615
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42383021116256714,
      "learning_rate": 9.306327619238306e-05,
      "loss": 2.4566,
      "step": 6620
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4117472767829895,
      "learning_rate": 9.305288656017372e-05,
      "loss": 2.4875,
      "step": 6625
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39715898036956787,
      "learning_rate": 9.304248973398711e-05,
      "loss": 2.4869,
      "step": 6630
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39965611696243286,
      "learning_rate": 9.303208571556048e-05,
      "loss": 2.4757,
      "step": 6635
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4152176082134247,
      "learning_rate": 9.30216745066323e-05,
      "loss": 2.4227,
      "step": 6640
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4305969774723053,
      "learning_rate": 9.301125610894226e-05,
      "loss": 2.4524,
      "step": 6645
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4086262881755829,
      "learning_rate": 9.300083052423124e-05,
      "loss": 2.4589,
      "step": 6650
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40399169921875,
      "learning_rate": 9.29903977542413e-05,
      "loss": 2.5175,
      "step": 6655
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4040873348712921,
      "learning_rate": 9.297995780071571e-05,
      "loss": 2.5039,
      "step": 6660
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4366387724876404,
      "learning_rate": 9.296951066539899e-05,
      "loss": 2.4543,
      "step": 6665
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42486152052879333,
      "learning_rate": 9.295905635003677e-05,
      "loss": 2.5034,
      "step": 6670
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3942788243293762,
      "learning_rate": 9.294859485637595e-05,
      "loss": 2.4418,
      "step": 6675
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44938531517982483,
      "learning_rate": 9.29381261861646e-05,
      "loss": 2.4913,
      "step": 6680
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41577833890914917,
      "learning_rate": 9.292765034115198e-05,
      "loss": 2.4944,
      "step": 6685
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41087108850479126,
      "learning_rate": 9.291716732308862e-05,
      "loss": 2.5074,
      "step": 6690
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4055137634277344,
      "learning_rate": 9.290667713372613e-05,
      "loss": 2.5196,
      "step": 6695
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.402158260345459,
      "learning_rate": 9.289617977481741e-05,
      "loss": 2.3931,
      "step": 6700
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4236752390861511,
      "learning_rate": 9.288567524811654e-05,
      "loss": 2.5039,
      "step": 6705
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40303048491477966,
      "learning_rate": 9.287516355537876e-05,
      "loss": 2.5081,
      "step": 6710
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42499154806137085,
      "learning_rate": 9.286464469836056e-05,
      "loss": 2.4807,
      "step": 6715
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40572255849838257,
      "learning_rate": 9.28541186788196e-05,
      "loss": 2.4949,
      "step": 6720
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3995773196220398,
      "learning_rate": 9.284358549851471e-05,
      "loss": 2.4298,
      "step": 6725
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4103162884712219,
      "learning_rate": 9.283304515920598e-05,
      "loss": 2.6176,
      "step": 6730
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4019069969654083,
      "learning_rate": 9.282249766265464e-05,
      "loss": 2.5231,
      "step": 6735
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4199429154396057,
      "learning_rate": 9.281194301062315e-05,
      "loss": 2.4357,
      "step": 6740
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41695234179496765,
      "learning_rate": 9.280138120487514e-05,
      "loss": 2.548,
      "step": 6745
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4047410786151886,
      "learning_rate": 9.279081224717547e-05,
      "loss": 2.573,
      "step": 6750
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4423235058784485,
      "learning_rate": 9.278023613929014e-05,
      "loss": 2.5232,
      "step": 6755
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.393957257270813,
      "learning_rate": 9.276965288298642e-05,
      "loss": 2.515,
      "step": 6760
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41010457277297974,
      "learning_rate": 9.27590624800327e-05,
      "loss": 2.4435,
      "step": 6765
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4141731560230255,
      "learning_rate": 9.27484649321986e-05,
      "loss": 2.553,
      "step": 6770
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3933008313179016,
      "learning_rate": 9.273786024125496e-05,
      "loss": 2.4988,
      "step": 6775
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39083534479141235,
      "learning_rate": 9.272724840897375e-05,
      "loss": 2.4299,
      "step": 6780
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.36795780062675476,
      "learning_rate": 9.271662943712818e-05,
      "loss": 2.4045,
      "step": 6785
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4150278866291046,
      "learning_rate": 9.270600332749264e-05,
      "loss": 2.6007,
      "step": 6790
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4101994037628174,
      "learning_rate": 9.269537008184273e-05,
      "loss": 2.4757,
      "step": 6795
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4270690381526947,
      "learning_rate": 9.268472970195519e-05,
      "loss": 2.5531,
      "step": 6800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39698436856269836,
      "learning_rate": 9.267408218960803e-05,
      "loss": 2.5205,
      "step": 6805
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39866212010383606,
      "learning_rate": 9.266342754658036e-05,
      "loss": 2.4665,
      "step": 6810
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45891427993774414,
      "learning_rate": 9.265276577465259e-05,
      "loss": 2.4147,
      "step": 6815
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40048113465309143,
      "learning_rate": 9.264209687560621e-05,
      "loss": 2.4001,
      "step": 6820
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4080037772655487,
      "learning_rate": 9.263142085122399e-05,
      "loss": 2.4784,
      "step": 6825
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4163697063922882,
      "learning_rate": 9.262073770328983e-05,
      "loss": 2.4996,
      "step": 6830
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45572197437286377,
      "learning_rate": 9.261004743358886e-05,
      "loss": 2.472,
      "step": 6835
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3865598738193512,
      "learning_rate": 9.259935004390737e-05,
      "loss": 2.4831,
      "step": 6840
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4053834080696106,
      "learning_rate": 9.258864553603286e-05,
      "loss": 2.3806,
      "step": 6845
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4177885949611664,
      "learning_rate": 9.257793391175402e-05,
      "loss": 2.536,
      "step": 6850
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44557592272758484,
      "learning_rate": 9.256721517286073e-05,
      "loss": 2.4406,
      "step": 6855
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43028703331947327,
      "learning_rate": 9.255648932114403e-05,
      "loss": 2.4218,
      "step": 6860
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41610559821128845,
      "learning_rate": 9.254575635839618e-05,
      "loss": 2.5575,
      "step": 6865
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40284594893455505,
      "learning_rate": 9.253501628641061e-05,
      "loss": 2.4608,
      "step": 6870
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3899051249027252,
      "learning_rate": 9.252426910698198e-05,
      "loss": 2.5068,
      "step": 6875
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4059048891067505,
      "learning_rate": 9.251351482190604e-05,
      "loss": 2.431,
      "step": 6880
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40844786167144775,
      "learning_rate": 9.250275343297985e-05,
      "loss": 2.5015,
      "step": 6885
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41207408905029297,
      "learning_rate": 9.249198494200158e-05,
      "loss": 2.5209,
      "step": 6890
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40799108147621155,
      "learning_rate": 9.24812093507706e-05,
      "loss": 2.4553,
      "step": 6895
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40902918577194214,
      "learning_rate": 9.247042666108747e-05,
      "loss": 2.4576,
      "step": 6900
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38770580291748047,
      "learning_rate": 9.245963687475395e-05,
      "loss": 2.5256,
      "step": 6905
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39547327160835266,
      "learning_rate": 9.244883999357293e-05,
      "loss": 2.3979,
      "step": 6910
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40360960364341736,
      "learning_rate": 9.24380360193486e-05,
      "loss": 2.4377,
      "step": 6915
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40597623586654663,
      "learning_rate": 9.242722495388621e-05,
      "loss": 2.558,
      "step": 6920
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43082284927368164,
      "learning_rate": 9.241640679899226e-05,
      "loss": 2.4889,
      "step": 6925
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39266934990882874,
      "learning_rate": 9.240558155647443e-05,
      "loss": 2.465,
      "step": 6930
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3945293128490448,
      "learning_rate": 9.239474922814157e-05,
      "loss": 2.3726,
      "step": 6935
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3854297697544098,
      "learning_rate": 9.238390981580375e-05,
      "loss": 2.4983,
      "step": 6940
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38555726408958435,
      "learning_rate": 9.237306332127215e-05,
      "loss": 2.4391,
      "step": 6945
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39616504311561584,
      "learning_rate": 9.236220974635923e-05,
      "loss": 2.4542,
      "step": 6950
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4014972448348999,
      "learning_rate": 9.235134909287853e-05,
      "loss": 2.4029,
      "step": 6955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3937961161136627,
      "learning_rate": 9.234048136264485e-05,
      "loss": 2.4783,
      "step": 6960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4152255058288574,
      "learning_rate": 9.232960655747416e-05,
      "loss": 2.5398,
      "step": 6965
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3789069652557373,
      "learning_rate": 9.231872467918358e-05,
      "loss": 2.4414,
      "step": 6970
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4551713466644287,
      "learning_rate": 9.230783572959143e-05,
      "loss": 2.4992,
      "step": 6975
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3882410228252411,
      "learning_rate": 9.229693971051723e-05,
      "loss": 2.4515,
      "step": 6980
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39079245924949646,
      "learning_rate": 9.228603662378166e-05,
      "loss": 2.4793,
      "step": 6985
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4034731388092041,
      "learning_rate": 9.227512647120656e-05,
      "loss": 2.3899,
      "step": 6990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42846351861953735,
      "learning_rate": 9.2264209254615e-05,
      "loss": 2.4092,
      "step": 6995
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3986421525478363,
      "learning_rate": 9.225328497583122e-05,
      "loss": 2.4603,
      "step": 7000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4112190008163452,
      "learning_rate": 9.224235363668058e-05,
      "loss": 2.349,
      "step": 7005
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3994402289390564,
      "learning_rate": 9.223141523898972e-05,
      "loss": 2.4841,
      "step": 7010
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38165953755378723,
      "learning_rate": 9.222046978458639e-05,
      "loss": 2.4766,
      "step": 7015
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45038318634033203,
      "learning_rate": 9.22095172752995e-05,
      "loss": 2.3763,
      "step": 7020
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40207475423812866,
      "learning_rate": 9.21985577129592e-05,
      "loss": 2.502,
      "step": 7025
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43114516139030457,
      "learning_rate": 9.218759109939681e-05,
      "loss": 2.4619,
      "step": 7030
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4226589798927307,
      "learning_rate": 9.217661743644477e-05,
      "loss": 2.4741,
      "step": 7035
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4208976626396179,
      "learning_rate": 9.216563672593677e-05,
      "loss": 2.4687,
      "step": 7040
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3867039382457733,
      "learning_rate": 9.215464896970763e-05,
      "loss": 2.4787,
      "step": 7045
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41848090291023254,
      "learning_rate": 9.214365416959336e-05,
      "loss": 2.4536,
      "step": 7050
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.419969379901886,
      "learning_rate": 9.213265232743119e-05,
      "loss": 2.5348,
      "step": 7055
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39378392696380615,
      "learning_rate": 9.212164344505942e-05,
      "loss": 2.5152,
      "step": 7060
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41367271542549133,
      "learning_rate": 9.211062752431765e-05,
      "loss": 2.5219,
      "step": 7065
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41978511214256287,
      "learning_rate": 9.209960456704656e-05,
      "loss": 2.5023,
      "step": 7070
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39591288566589355,
      "learning_rate": 9.208857457508809e-05,
      "loss": 2.4739,
      "step": 7075
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4167196452617645,
      "learning_rate": 9.207753755028526e-05,
      "loss": 2.4903,
      "step": 7080
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4126549959182739,
      "learning_rate": 9.206649349448235e-05,
      "loss": 2.4406,
      "step": 7085
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41538065671920776,
      "learning_rate": 9.205544240952477e-05,
      "loss": 2.4314,
      "step": 7090
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3931046724319458,
      "learning_rate": 9.204438429725911e-05,
      "loss": 2.4103,
      "step": 7095
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4163888096809387,
      "learning_rate": 9.203331915953314e-05,
      "loss": 2.4946,
      "step": 7100
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4093388617038727,
      "learning_rate": 9.202224699819584e-05,
      "loss": 2.4538,
      "step": 7105
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40619924664497375,
      "learning_rate": 9.201116781509726e-05,
      "loss": 2.4036,
      "step": 7110
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42623546719551086,
      "learning_rate": 9.200008161208875e-05,
      "loss": 2.4822,
      "step": 7115
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4365828335285187,
      "learning_rate": 9.198898839102273e-05,
      "loss": 2.5071,
      "step": 7120
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40054094791412354,
      "learning_rate": 9.197788815375287e-05,
      "loss": 2.478,
      "step": 7125
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41042304039001465,
      "learning_rate": 9.196678090213395e-05,
      "loss": 2.3998,
      "step": 7130
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4371322691440582,
      "learning_rate": 9.195566663802198e-05,
      "loss": 2.4764,
      "step": 7135
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39167118072509766,
      "learning_rate": 9.194454536327408e-05,
      "loss": 2.5036,
      "step": 7140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4181944727897644,
      "learning_rate": 9.19334170797486e-05,
      "loss": 2.5241,
      "step": 7145
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41757503151893616,
      "learning_rate": 9.192228178930503e-05,
      "loss": 2.4005,
      "step": 7150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39922234416007996,
      "learning_rate": 9.191113949380402e-05,
      "loss": 2.5389,
      "step": 7155
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3947104513645172,
      "learning_rate": 9.189999019510742e-05,
      "loss": 2.4559,
      "step": 7160
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40675970911979675,
      "learning_rate": 9.188883389507823e-05,
      "loss": 2.484,
      "step": 7165
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4377189874649048,
      "learning_rate": 9.187767059558064e-05,
      "loss": 2.4937,
      "step": 7170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40966561436653137,
      "learning_rate": 9.186650029847998e-05,
      "loss": 2.4648,
      "step": 7175
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.412021666765213,
      "learning_rate": 9.185532300564276e-05,
      "loss": 2.5376,
      "step": 7180
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41294631361961365,
      "learning_rate": 9.184413871893668e-05,
      "loss": 2.4583,
      "step": 7185
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38553792238235474,
      "learning_rate": 9.183294744023058e-05,
      "loss": 2.4212,
      "step": 7190
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3888728618621826,
      "learning_rate": 9.18217491713945e-05,
      "loss": 2.4954,
      "step": 7195
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4032144844532013,
      "learning_rate": 9.181054391429962e-05,
      "loss": 2.4922,
      "step": 7200
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4157060384750366,
      "learning_rate": 9.179933167081828e-05,
      "loss": 2.4723,
      "step": 7205
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39756959676742554,
      "learning_rate": 9.178811244282404e-05,
      "loss": 2.5112,
      "step": 7210
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40855342149734497,
      "learning_rate": 9.177688623219154e-05,
      "loss": 2.4684,
      "step": 7215
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38615483045578003,
      "learning_rate": 9.176565304079669e-05,
      "loss": 2.5304,
      "step": 7220
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3969455659389496,
      "learning_rate": 9.175441287051648e-05,
      "loss": 2.4469,
      "step": 7225
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4011453092098236,
      "learning_rate": 9.174316572322912e-05,
      "loss": 2.3898,
      "step": 7230
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38569316267967224,
      "learning_rate": 9.173191160081395e-05,
      "loss": 2.3702,
      "step": 7235
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3952690660953522,
      "learning_rate": 9.17206505051515e-05,
      "loss": 2.4789,
      "step": 7240
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.414072185754776,
      "learning_rate": 9.170938243812349e-05,
      "loss": 2.4745,
      "step": 7245
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4285476505756378,
      "learning_rate": 9.16981074016127e-05,
      "loss": 2.4668,
      "step": 7250
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41575586795806885,
      "learning_rate": 9.16868253975032e-05,
      "loss": 2.4619,
      "step": 7255
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4029485881328583,
      "learning_rate": 9.167553642768017e-05,
      "loss": 2.4618,
      "step": 7260
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38122501969337463,
      "learning_rate": 9.166424049402993e-05,
      "loss": 2.49,
      "step": 7265
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39007601141929626,
      "learning_rate": 9.165293759844e-05,
      "loss": 2.4632,
      "step": 7270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39515021443367004,
      "learning_rate": 9.164162774279905e-05,
      "loss": 2.4185,
      "step": 7275
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42379826307296753,
      "learning_rate": 9.163031092899693e-05,
      "loss": 2.5713,
      "step": 7280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42731860280036926,
      "learning_rate": 9.161898715892463e-05,
      "loss": 2.489,
      "step": 7285
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39561155438423157,
      "learning_rate": 9.160765643447429e-05,
      "loss": 2.4595,
      "step": 7290
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5382345914840698,
      "learning_rate": 9.159631875753925e-05,
      "loss": 2.4598,
      "step": 7295
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38561880588531494,
      "learning_rate": 9.1584974130014e-05,
      "loss": 2.4297,
      "step": 7300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4297572076320648,
      "learning_rate": 9.157362255379419e-05,
      "loss": 2.4396,
      "step": 7305
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3955624997615814,
      "learning_rate": 9.156226403077657e-05,
      "loss": 2.5,
      "step": 7310
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3898204267024994,
      "learning_rate": 9.155089856285917e-05,
      "loss": 2.4656,
      "step": 7315
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4285324513912201,
      "learning_rate": 9.153952615194111e-05,
      "loss": 2.4398,
      "step": 7320
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.374349981546402,
      "learning_rate": 9.152814679992265e-05,
      "loss": 2.4164,
      "step": 7325
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46969830989837646,
      "learning_rate": 9.151676050870526e-05,
      "loss": 2.3697,
      "step": 7330
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40863069891929626,
      "learning_rate": 9.150536728019154e-05,
      "loss": 2.3973,
      "step": 7335
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39526456594467163,
      "learning_rate": 9.149396711628523e-05,
      "loss": 2.4505,
      "step": 7340
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39476099610328674,
      "learning_rate": 9.14825600188913e-05,
      "loss": 2.4839,
      "step": 7345
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41229116916656494,
      "learning_rate": 9.147114598991583e-05,
      "loss": 2.4849,
      "step": 7350
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3948456943035126,
      "learning_rate": 9.145972503126603e-05,
      "loss": 2.4436,
      "step": 7355
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5061823129653931,
      "learning_rate": 9.14482971448503e-05,
      "loss": 2.5278,
      "step": 7360
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4219697415828705,
      "learning_rate": 9.143686233257824e-05,
      "loss": 2.4438,
      "step": 7365
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4192996919155121,
      "learning_rate": 9.142542059636053e-05,
      "loss": 2.4828,
      "step": 7370
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42134177684783936,
      "learning_rate": 9.141397193810903e-05,
      "loss": 2.4646,
      "step": 7375
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40898656845092773,
      "learning_rate": 9.14025163597368e-05,
      "loss": 2.4233,
      "step": 7380
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3854668140411377,
      "learning_rate": 9.1391053863158e-05,
      "loss": 2.3923,
      "step": 7385
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40889132022857666,
      "learning_rate": 9.137958445028799e-05,
      "loss": 2.3807,
      "step": 7390
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3943890929222107,
      "learning_rate": 9.136810812304326e-05,
      "loss": 2.4971,
      "step": 7395
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39766353368759155,
      "learning_rate": 9.135662488334147e-05,
      "loss": 2.4688,
      "step": 7400
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3935258686542511,
      "learning_rate": 9.13451347331014e-05,
      "loss": 2.4859,
      "step": 7405
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39748895168304443,
      "learning_rate": 9.133363767424303e-05,
      "loss": 2.527,
      "step": 7410
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40117040276527405,
      "learning_rate": 9.132213370868748e-05,
      "loss": 2.4341,
      "step": 7415
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39426368474960327,
      "learning_rate": 9.131062283835701e-05,
      "loss": 2.4324,
      "step": 7420
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3992118835449219,
      "learning_rate": 9.129910506517503e-05,
      "loss": 2.4448,
      "step": 7425
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4169372022151947,
      "learning_rate": 9.128758039106616e-05,
      "loss": 2.3969,
      "step": 7430
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38945749402046204,
      "learning_rate": 9.127604881795609e-05,
      "loss": 2.4319,
      "step": 7435
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4352794885635376,
      "learning_rate": 9.126451034777173e-05,
      "loss": 2.4249,
      "step": 7440
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3903143107891083,
      "learning_rate": 9.12529649824411e-05,
      "loss": 2.4109,
      "step": 7445
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42277947068214417,
      "learning_rate": 9.124141272389337e-05,
      "loss": 2.4489,
      "step": 7450
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4254138767719269,
      "learning_rate": 9.122985357405892e-05,
      "loss": 2.4478,
      "step": 7455
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38699761033058167,
      "learning_rate": 9.121828753486921e-05,
      "loss": 2.4817,
      "step": 7460
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39487603306770325,
      "learning_rate": 9.120671460825691e-05,
      "loss": 2.4567,
      "step": 7465
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4114915132522583,
      "learning_rate": 9.119513479615577e-05,
      "loss": 2.502,
      "step": 7470
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.415498822927475,
      "learning_rate": 9.118354810050079e-05,
      "loss": 2.4612,
      "step": 7475
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37394261360168457,
      "learning_rate": 9.117195452322801e-05,
      "loss": 2.5197,
      "step": 7480
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4107445776462555,
      "learning_rate": 9.11603540662747e-05,
      "loss": 2.5574,
      "step": 7485
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4057503044605255,
      "learning_rate": 9.114874673157927e-05,
      "loss": 2.4822,
      "step": 7490
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42777949571609497,
      "learning_rate": 9.113713252108123e-05,
      "loss": 2.4489,
      "step": 7495
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9310176372528076,
      "learning_rate": 9.112551143672127e-05,
      "loss": 2.4277,
      "step": 7500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4200431704521179,
      "learning_rate": 9.111388348044125e-05,
      "loss": 2.4246,
      "step": 7505
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39214301109313965,
      "learning_rate": 9.110224865418417e-05,
      "loss": 2.4397,
      "step": 7510
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4254666566848755,
      "learning_rate": 9.109060695989413e-05,
      "loss": 2.4339,
      "step": 7515
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39131057262420654,
      "learning_rate": 9.107895839951643e-05,
      "loss": 2.4624,
      "step": 7520
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3794894814491272,
      "learning_rate": 9.106730297499752e-05,
      "loss": 2.4231,
      "step": 7525
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3956964910030365,
      "learning_rate": 9.105564068828497e-05,
      "loss": 2.5144,
      "step": 7530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4020974338054657,
      "learning_rate": 9.104397154132747e-05,
      "loss": 2.3952,
      "step": 7535
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3773643672466278,
      "learning_rate": 9.103229553607493e-05,
      "loss": 2.3999,
      "step": 7540
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38404762744903564,
      "learning_rate": 9.102061267447835e-05,
      "loss": 2.4322,
      "step": 7545
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.413204550743103,
      "learning_rate": 9.100892295848989e-05,
      "loss": 2.4049,
      "step": 7550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4551413953304291,
      "learning_rate": 9.099722639006287e-05,
      "loss": 2.4401,
      "step": 7555
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39578938484191895,
      "learning_rate": 9.098552297115175e-05,
      "loss": 2.4159,
      "step": 7560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41858065128326416,
      "learning_rate": 9.097381270371212e-05,
      "loss": 2.4789,
      "step": 7565
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37782928347587585,
      "learning_rate": 9.09620955897007e-05,
      "loss": 2.4496,
      "step": 7570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4226630628108978,
      "learning_rate": 9.09503716310754e-05,
      "loss": 2.4785,
      "step": 7575
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40337392687797546,
      "learning_rate": 9.093864082979525e-05,
      "loss": 2.4594,
      "step": 7580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40132245421409607,
      "learning_rate": 9.092690318782042e-05,
      "loss": 2.3864,
      "step": 7585
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4206814765930176,
      "learning_rate": 9.091515870711221e-05,
      "loss": 2.425,
      "step": 7590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3952077627182007,
      "learning_rate": 9.09034073896331e-05,
      "loss": 2.4025,
      "step": 7595
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3989850580692291,
      "learning_rate": 9.08916492373467e-05,
      "loss": 2.4433,
      "step": 7600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4138714373111725,
      "learning_rate": 9.087988425221772e-05,
      "loss": 2.3895,
      "step": 7605
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37805548310279846,
      "learning_rate": 9.086811243621206e-05,
      "loss": 2.4748,
      "step": 7610
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3849479854106903,
      "learning_rate": 9.085633379129677e-05,
      "loss": 2.5205,
      "step": 7615
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3831448256969452,
      "learning_rate": 9.084454831943998e-05,
      "loss": 2.496,
      "step": 7620
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3968905210494995,
      "learning_rate": 9.083275602261101e-05,
      "loss": 2.4109,
      "step": 7625
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39275988936424255,
      "learning_rate": 9.082095690278034e-05,
      "loss": 2.4542,
      "step": 7630
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41345614194869995,
      "learning_rate": 9.080915096191951e-05,
      "loss": 2.4422,
      "step": 7635
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4079107940196991,
      "learning_rate": 9.07973382020013e-05,
      "loss": 2.4482,
      "step": 7640
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40640273690223694,
      "learning_rate": 9.078551862499953e-05,
      "loss": 2.4715,
      "step": 7645
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3898564577102661,
      "learning_rate": 9.077369223288923e-05,
      "loss": 2.4719,
      "step": 7650
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.407083123922348,
      "learning_rate": 9.076185902764656e-05,
      "loss": 2.5239,
      "step": 7655
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4200746417045593,
      "learning_rate": 9.075001901124878e-05,
      "loss": 2.3933,
      "step": 7660
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3988346457481384,
      "learning_rate": 9.073817218567431e-05,
      "loss": 2.5276,
      "step": 7665
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4164753556251526,
      "learning_rate": 9.072631855290275e-05,
      "loss": 2.4219,
      "step": 7670
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40487903356552124,
      "learning_rate": 9.071445811491475e-05,
      "loss": 2.4764,
      "step": 7675
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39667224884033203,
      "learning_rate": 9.070259087369218e-05,
      "loss": 2.4432,
      "step": 7680
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4147872030735016,
      "learning_rate": 9.069071683121799e-05,
      "loss": 2.4859,
      "step": 7685
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4013163447380066,
      "learning_rate": 9.067883598947628e-05,
      "loss": 2.4567,
      "step": 7690
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.403656542301178,
      "learning_rate": 9.066694835045233e-05,
      "loss": 2.3462,
      "step": 7695
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3975553512573242,
      "learning_rate": 9.06550539161325e-05,
      "loss": 2.4735,
      "step": 7700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3841886818408966,
      "learning_rate": 9.06431526885043e-05,
      "loss": 2.3849,
      "step": 7705
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41878432035446167,
      "learning_rate": 9.063124466955639e-05,
      "loss": 2.4511,
      "step": 7710
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41313835978507996,
      "learning_rate": 9.061932986127857e-05,
      "loss": 2.3794,
      "step": 7715
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39815065264701843,
      "learning_rate": 9.060740826566172e-05,
      "loss": 2.4376,
      "step": 7720
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39675381779670715,
      "learning_rate": 9.059547988469792e-05,
      "loss": 2.4463,
      "step": 7725
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3877049684524536,
      "learning_rate": 9.058354472038037e-05,
      "loss": 2.4181,
      "step": 7730
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3900602459907532,
      "learning_rate": 9.057160277470337e-05,
      "loss": 2.4078,
      "step": 7735
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4026559293270111,
      "learning_rate": 9.055965404966238e-05,
      "loss": 2.5143,
      "step": 7740
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41245371103286743,
      "learning_rate": 9.054769854725399e-05,
      "loss": 2.4735,
      "step": 7745
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4037948250770569,
      "learning_rate": 9.053573626947593e-05,
      "loss": 2.4688,
      "step": 7750
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38033199310302734,
      "learning_rate": 9.052376721832702e-05,
      "loss": 2.4482,
      "step": 7755
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3902681767940521,
      "learning_rate": 9.051179139580727e-05,
      "loss": 2.3472,
      "step": 7760
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38712871074676514,
      "learning_rate": 9.049980880391782e-05,
      "loss": 2.4534,
      "step": 7765
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3924829065799713,
      "learning_rate": 9.048781944466087e-05,
      "loss": 2.4302,
      "step": 7770
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3825702667236328,
      "learning_rate": 9.04758233200398e-05,
      "loss": 2.3441,
      "step": 7775
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3791918456554413,
      "learning_rate": 9.046382043205914e-05,
      "loss": 2.3983,
      "step": 7780
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38685643672943115,
      "learning_rate": 9.045181078272453e-05,
      "loss": 2.4415,
      "step": 7785
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4493215084075928,
      "learning_rate": 9.043979437404273e-05,
      "loss": 2.3988,
      "step": 7790
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37873566150665283,
      "learning_rate": 9.042777120802164e-05,
      "loss": 2.4229,
      "step": 7795
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38450950384140015,
      "learning_rate": 9.041574128667027e-05,
      "loss": 2.4242,
      "step": 7800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38752907514572144,
      "learning_rate": 9.040370461199879e-05,
      "loss": 2.4835,
      "step": 7805
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40568819642066956,
      "learning_rate": 9.039166118601849e-05,
      "loss": 2.4407,
      "step": 7810
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3879481554031372,
      "learning_rate": 9.037961101074175e-05,
      "loss": 2.4496,
      "step": 7815
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3978101313114166,
      "learning_rate": 9.036755408818217e-05,
      "loss": 2.4157,
      "step": 7820
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3911791741847992,
      "learning_rate": 9.035549042035434e-05,
      "loss": 2.4557,
      "step": 7825
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3949492573738098,
      "learning_rate": 9.03434200092741e-05,
      "loss": 2.5737,
      "step": 7830
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39901968836784363,
      "learning_rate": 9.033134285695837e-05,
      "loss": 2.3882,
      "step": 7835
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38547447323799133,
      "learning_rate": 9.031925896542521e-05,
      "loss": 2.3994,
      "step": 7840
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.392621785402298,
      "learning_rate": 9.030716833669375e-05,
      "loss": 2.4044,
      "step": 7845
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4203197658061981,
      "learning_rate": 9.029507097278433e-05,
      "loss": 2.43,
      "step": 7850
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4066040515899658,
      "learning_rate": 9.028296687571834e-05,
      "loss": 2.4304,
      "step": 7855
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4183081090450287,
      "learning_rate": 9.027085604751835e-05,
      "loss": 2.5119,
      "step": 7860
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4105401635169983,
      "learning_rate": 9.025873849020805e-05,
      "loss": 2.4161,
      "step": 7865
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40350955724716187,
      "learning_rate": 9.024661420581222e-05,
      "loss": 2.5019,
      "step": 7870
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.47972285747528076,
      "learning_rate": 9.023448319635676e-05,
      "loss": 2.4281,
      "step": 7875
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41599181294441223,
      "learning_rate": 9.022234546386877e-05,
      "loss": 2.4509,
      "step": 7880
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3925907611846924,
      "learning_rate": 9.021020101037637e-05,
      "loss": 2.4335,
      "step": 7885
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3878338932991028,
      "learning_rate": 9.019804983790888e-05,
      "loss": 2.4655,
      "step": 7890
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3818398714065552,
      "learning_rate": 9.018589194849672e-05,
      "loss": 2.4375,
      "step": 7895
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.39287206530570984,
      "learning_rate": 9.01737273441714e-05,
      "loss": 2.4115,
      "step": 7900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42343229055404663,
      "learning_rate": 9.016155602696559e-05,
      "loss": 2.4037,
      "step": 7905
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4054190516471863,
      "learning_rate": 9.01493779989131e-05,
      "loss": 2.4045,
      "step": 7910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4303849935531616,
      "learning_rate": 9.013719326204881e-05,
      "loss": 2.3577,
      "step": 7915
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41618824005126953,
      "learning_rate": 9.012500181840874e-05,
      "loss": 2.481,
      "step": 7920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4085509777069092,
      "learning_rate": 9.011280367003003e-05,
      "loss": 2.434,
      "step": 7925
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42664098739624023,
      "learning_rate": 9.010059881895099e-05,
      "loss": 2.4717,
      "step": 7930
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3786696195602417,
      "learning_rate": 9.008838726721094e-05,
      "loss": 2.4786,
      "step": 7935
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42319145798683167,
      "learning_rate": 9.007616901685042e-05,
      "loss": 2.3803,
      "step": 7940
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4108833074569702,
      "learning_rate": 9.006394406991108e-05,
      "loss": 2.3555,
      "step": 7945
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41623106598854065,
      "learning_rate": 9.00517124284356e-05,
      "loss": 2.4385,
      "step": 7950
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40810003876686096,
      "learning_rate": 9.003947409446789e-05,
      "loss": 2.4484,
      "step": 7955
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3966154456138611,
      "learning_rate": 9.002722907005292e-05,
      "loss": 2.4573,
      "step": 7960
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4091435372829437,
      "learning_rate": 9.001497735723679e-05,
      "loss": 2.5706,
      "step": 7965
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3871368169784546,
      "learning_rate": 9.000271895806669e-05,
      "loss": 2.4345,
      "step": 7970
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42019370198249817,
      "learning_rate": 8.9990453874591e-05,
      "loss": 2.4949,
      "step": 7975
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3826068341732025,
      "learning_rate": 8.997818210885912e-05,
      "loss": 2.51,
      "step": 7980
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4699552059173584,
      "learning_rate": 8.996590366292164e-05,
      "loss": 2.4486,
      "step": 7985
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4041611850261688,
      "learning_rate": 8.995361853883026e-05,
      "loss": 2.4052,
      "step": 7990
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4570592939853668,
      "learning_rate": 8.994132673863774e-05,
      "loss": 2.4423,
      "step": 7995
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4097904860973358,
      "learning_rate": 8.992902826439802e-05,
      "loss": 2.4708,
      "step": 8000
    }
  ],
  "logging_steps": 5,
  "max_steps": 38527,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 2.126010100010189e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}