{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.999936517341346,
  "eval_steps": 500,
  "global_step": 70884,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1493.0737256705327,
      "learning_rate": 1.4106361969248132e-09,
      "loss": 16.0844,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 1668.6822325141973,
      "learning_rate": 7.053180984624066e-09,
      "loss": 16.2728,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 1593.5609005055294,
      "learning_rate": 1.4106361969248132e-08,
      "loss": 15.8795,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1740.7994774605,
      "learning_rate": 2.11595429538722e-08,
      "loss": 16.7429,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1745.1855857808887,
      "learning_rate": 2.8212723938496263e-08,
      "loss": 16.1763,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1600.9559741761996,
      "learning_rate": 3.526590492312033e-08,
      "loss": 16.3725,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 1617.3961352296326,
      "learning_rate": 4.23190859077444e-08,
      "loss": 15.4492,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1621.5530943892318,
      "learning_rate": 4.937226689236846e-08,
      "loss": 15.4405,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 1325.9925021820668,
      "learning_rate": 5.6425447876992527e-08,
      "loss": 14.6386,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 1327.0691324050445,
      "learning_rate": 6.34786288616166e-08,
      "loss": 13.4356,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 1454.622792295747,
      "learning_rate": 7.053180984624066e-08,
      "loss": 13.2331,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 1135.4587201610175,
      "learning_rate": 7.758499083086472e-08,
      "loss": 12.2934,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 976.6086360192892,
      "learning_rate": 8.46381718154888e-08,
      "loss": 10.5764,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 787.7464747036888,
      "learning_rate": 9.169135280011286e-08,
      "loss": 9.5626,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 944.5961012828227,
      "learning_rate": 9.874453378473692e-08,
      "loss": 9.28,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 876.3350246610729,
      "learning_rate": 1.0579771476936099e-07,
      "loss": 8.7555,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 471.3586319835879,
      "learning_rate": 1.1285089575398505e-07,
      "loss": 7.3749,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 539.2888488233264,
      "learning_rate": 1.1990407673860913e-07,
      "loss": 6.7965,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 319.6686952446837,
      "learning_rate": 1.269572577232332e-07,
      "loss": 6.2039,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 323.0983845688617,
      "learning_rate": 1.3401043870785725e-07,
      "loss": 5.902,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 277.4164551652692,
      "learning_rate": 1.4106361969248132e-07,
      "loss": 5.6253,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 211.89978719226133,
      "learning_rate": 1.481168006771054e-07,
      "loss": 5.4443,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 278.59462836711947,
      "learning_rate": 1.5516998166172944e-07,
      "loss": 5.2659,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 250.10995490045883,
      "learning_rate": 1.622231626463535e-07,
      "loss": 5.0685,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 196.42141486889983,
      "learning_rate": 1.692763436309776e-07,
      "loss": 4.8159,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 191.3540979266941,
      "learning_rate": 1.7632952461560164e-07,
      "loss": 4.6016,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 198.3644471959955,
      "learning_rate": 1.8338270560022571e-07,
      "loss": 4.4773,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 174.16448574781828,
      "learning_rate": 1.9043588658484976e-07,
      "loss": 4.3955,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 185.01038252455538,
      "learning_rate": 1.9748906756947384e-07,
      "loss": 4.3058,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 140.31424944177627,
      "learning_rate": 2.045422485540979e-07,
      "loss": 3.974,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 142.34834095434633,
      "learning_rate": 2.1159542953872198e-07,
      "loss": 3.9014,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 141.51462944025857,
      "learning_rate": 2.1864861052334603e-07,
      "loss": 3.7357,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 112.06028834828933,
      "learning_rate": 2.257017915079701e-07,
      "loss": 3.6341,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 111.50920364748472,
      "learning_rate": 2.3275497249259415e-07,
      "loss": 3.471,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 90.71732178503606,
      "learning_rate": 2.3980815347721825e-07,
      "loss": 3.2521,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 100.63799360618768,
      "learning_rate": 2.468613344618423e-07,
      "loss": 3.1855,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 108.26725108354506,
      "learning_rate": 2.539145154464664e-07,
      "loss": 3.1287,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 69.6341645301703,
      "learning_rate": 2.609676964310904e-07,
      "loss": 3.0425,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 82.59536181881954,
      "learning_rate": 2.680208774157145e-07,
      "loss": 2.79,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 60.62454649003572,
      "learning_rate": 2.7507405840033857e-07,
      "loss": 2.9082,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 64.45850724694489,
      "learning_rate": 2.8212723938496265e-07,
      "loss": 2.8368,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 55.92476953087811,
      "learning_rate": 2.8918042036958667e-07,
      "loss": 2.9402,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 65.7903141986139,
      "learning_rate": 2.962336013542108e-07,
      "loss": 2.7063,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 63.13230710643131,
      "learning_rate": 3.032867823388348e-07,
      "loss": 2.711,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 58.0309123612463,
      "learning_rate": 3.103399633234589e-07,
      "loss": 2.5864,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 51.934921917753016,
      "learning_rate": 3.1739314430808296e-07,
      "loss": 2.6546,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 56.9400406309904,
      "learning_rate": 3.24446325292707e-07,
      "loss": 2.5507,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 48.52809979625148,
      "learning_rate": 3.314995062773311e-07,
      "loss": 2.5385,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 47.13451743977923,
      "learning_rate": 3.385526872619552e-07,
      "loss": 2.5494,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 48.93552123234581,
      "learning_rate": 3.4560586824657926e-07,
      "loss": 2.4727,
      "step": 245
    },
    {
      "epoch": 0.01,
      "grad_norm": 38.741755532923,
      "learning_rate": 3.526590492312033e-07,
      "loss": 2.4113,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 39.61271782523791,
      "learning_rate": 3.5971223021582736e-07,
      "loss": 2.4041,
      "step": 255
    },
    {
      "epoch": 0.01,
      "grad_norm": 38.984524568518786,
      "learning_rate": 3.6676541120045143e-07,
      "loss": 2.2889,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 34.863973070307736,
      "learning_rate": 3.738185921850755e-07,
      "loss": 2.3173,
      "step": 265
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.63886144618183,
      "learning_rate": 3.808717731696995e-07,
      "loss": 2.3947,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 33.63961303197873,
      "learning_rate": 3.879249541543236e-07,
      "loss": 2.2098,
      "step": 275
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.37629625616045,
      "learning_rate": 3.9497813513894767e-07,
      "loss": 2.3853,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 32.870580045008275,
      "learning_rate": 4.020313161235718e-07,
      "loss": 2.2034,
      "step": 285
    },
    {
      "epoch": 0.01,
      "grad_norm": 36.17392923347136,
      "learning_rate": 4.090844971081958e-07,
      "loss": 2.2471,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 35.86672938213747,
      "learning_rate": 4.161376780928199e-07,
      "loss": 2.2797,
      "step": 295
    },
    {
      "epoch": 0.01,
      "grad_norm": 31.491322137657292,
      "learning_rate": 4.2319085907744397e-07,
      "loss": 2.3222,
      "step": 300
    },
    {
      "epoch": 0.01,
      "grad_norm": 30.417845046914852,
      "learning_rate": 4.3024404006206804e-07,
      "loss": 2.2601,
      "step": 305
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.230195546192633,
      "learning_rate": 4.3729722104669207e-07,
      "loss": 2.1369,
      "step": 310
    },
    {
      "epoch": 0.01,
      "grad_norm": 26.06903397162156,
      "learning_rate": 4.4435040203131614e-07,
      "loss": 2.1749,
      "step": 315
    },
    {
      "epoch": 0.01,
      "grad_norm": 35.515790618558846,
      "learning_rate": 4.514035830159402e-07,
      "loss": 2.131,
      "step": 320
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.39959370215983,
      "learning_rate": 4.584567640005643e-07,
      "loss": 2.2326,
      "step": 325
    },
    {
      "epoch": 0.01,
      "grad_norm": 25.83729551385949,
      "learning_rate": 4.655099449851883e-07,
      "loss": 2.1105,
      "step": 330
    },
    {
      "epoch": 0.01,
      "grad_norm": 27.433628109552256,
      "learning_rate": 4.725631259698124e-07,
      "loss": 2.121,
      "step": 335
    },
    {
      "epoch": 0.01,
      "grad_norm": 23.869493383953355,
      "learning_rate": 4.796163069544365e-07,
      "loss": 2.0329,
      "step": 340
    },
    {
      "epoch": 0.01,
      "grad_norm": 21.20926380105703,
      "learning_rate": 4.866694879390605e-07,
      "loss": 2.1412,
      "step": 345
    },
    {
      "epoch": 0.01,
      "grad_norm": 20.23856754938487,
      "learning_rate": 4.937226689236846e-07,
      "loss": 2.0711,
      "step": 350
    },
    {
      "epoch": 0.02,
      "grad_norm": 24.5510550881969,
      "learning_rate": 5.007758499083087e-07,
      "loss": 1.9521,
      "step": 355
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.308996549844295,
      "learning_rate": 5.078290308929328e-07,
      "loss": 2.1002,
      "step": 360
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.650458319960855,
      "learning_rate": 5.148822118775568e-07,
      "loss": 2.0129,
      "step": 365
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.35608125514246,
      "learning_rate": 5.219353928621808e-07,
      "loss": 1.9728,
      "step": 370
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.391329162858245,
      "learning_rate": 5.28988573846805e-07,
      "loss": 2.0186,
      "step": 375
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.913264467280225,
      "learning_rate": 5.36041754831429e-07,
      "loss": 2.072,
      "step": 380
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.017640641423437,
      "learning_rate": 5.430949358160531e-07,
      "loss": 2.0241,
      "step": 385
    },
    {
      "epoch": 0.02,
      "grad_norm": 29.61053230838206,
      "learning_rate": 5.501481168006771e-07,
      "loss": 2.06,
      "step": 390
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.66707782214414,
      "learning_rate": 5.572012977853012e-07,
      "loss": 1.9234,
      "step": 395
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.56925408153988,
      "learning_rate": 5.642544787699253e-07,
      "loss": 1.9593,
      "step": 400
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.541478346089974,
      "learning_rate": 5.713076597545494e-07,
      "loss": 2.0793,
      "step": 405
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.50830085835403,
      "learning_rate": 5.783608407391733e-07,
      "loss": 1.9824,
      "step": 410
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.980919852203144,
      "learning_rate": 5.854140217237975e-07,
      "loss": 1.962,
      "step": 415
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.68117938946925,
      "learning_rate": 5.924672027084216e-07,
      "loss": 2.0047,
      "step": 420
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.26570158613031,
      "learning_rate": 5.995203836930456e-07,
      "loss": 1.9533,
      "step": 425
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.78662520957527,
      "learning_rate": 6.065735646776696e-07,
      "loss": 1.8809,
      "step": 430
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.416830761222496,
      "learning_rate": 6.136267456622938e-07,
      "loss": 1.8558,
      "step": 435
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.03954637757054,
      "learning_rate": 6.206799266469178e-07,
      "loss": 1.9291,
      "step": 440
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.7112234084775,
      "learning_rate": 6.277331076315418e-07,
      "loss": 1.793,
      "step": 445
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.82696832616069,
      "learning_rate": 6.347862886161659e-07,
      "loss": 1.9045,
      "step": 450
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.126201478944438,
      "learning_rate": 6.4183946960079e-07,
      "loss": 1.8342,
      "step": 455
    },
    {
      "epoch": 0.02,
      "grad_norm": 23.293631330318448,
      "learning_rate": 6.48892650585414e-07,
      "loss": 1.903,
      "step": 460
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.593676527550635,
      "learning_rate": 6.559458315700382e-07,
      "loss": 1.791,
      "step": 465
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.786452970894214,
      "learning_rate": 6.629990125546622e-07,
      "loss": 1.8003,
      "step": 470
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.804670739432435,
      "learning_rate": 6.700521935392864e-07,
      "loss": 1.8244,
      "step": 475
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.48348527780603,
      "learning_rate": 6.771053745239104e-07,
      "loss": 2.0048,
      "step": 480
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.40123583023564,
      "learning_rate": 6.841585555085344e-07,
      "loss": 1.8236,
      "step": 485
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.173168434695988,
      "learning_rate": 6.912117364931585e-07,
      "loss": 1.8029,
      "step": 490
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.208246288455292,
      "learning_rate": 6.982649174777825e-07,
      "loss": 1.8292,
      "step": 495
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.55138615149806,
      "learning_rate": 7.053180984624066e-07,
      "loss": 1.7629,
      "step": 500
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.549071943792157,
      "learning_rate": 7.123712794470307e-07,
      "loss": 1.7353,
      "step": 505
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.19229022979663,
      "learning_rate": 7.194244604316547e-07,
      "loss": 1.8791,
      "step": 510
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.990919931737333,
      "learning_rate": 7.264776414162788e-07,
      "loss": 1.7342,
      "step": 515
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.196458223019654,
      "learning_rate": 7.335308224009029e-07,
      "loss": 1.8321,
      "step": 520
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.491910452768337,
      "learning_rate": 7.405840033855269e-07,
      "loss": 1.7515,
      "step": 525
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.409490880670386,
      "learning_rate": 7.47637184370151e-07,
      "loss": 1.7221,
      "step": 530
    },
    {
      "epoch": 0.02,
      "grad_norm": 17.21027525557651,
      "learning_rate": 7.54690365354775e-07,
      "loss": 1.7706,
      "step": 535
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.462134586321856,
      "learning_rate": 7.61743546339399e-07,
      "loss": 1.7895,
      "step": 540
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.51094643801263,
      "learning_rate": 7.687967273240232e-07,
      "loss": 1.7663,
      "step": 545
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.31997025951962,
      "learning_rate": 7.758499083086472e-07,
      "loss": 1.7756,
      "step": 550
    },
    {
      "epoch": 0.02,
      "grad_norm": 15.645818406479359,
      "learning_rate": 7.829030892932714e-07,
      "loss": 1.7598,
      "step": 555
    },
    {
      "epoch": 0.02,
      "grad_norm": 20.36937660151297,
      "learning_rate": 7.899562702778953e-07,
      "loss": 1.7537,
      "step": 560
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.716849406801753,
      "learning_rate": 7.970094512625194e-07,
      "loss": 1.7234,
      "step": 565
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.677226207354845,
      "learning_rate": 8.040626322471436e-07,
      "loss": 1.789,
      "step": 570
    },
    {
      "epoch": 0.02,
      "grad_norm": 19.521214063483317,
      "learning_rate": 8.111158132317676e-07,
      "loss": 1.6659,
      "step": 575
    },
    {
      "epoch": 0.02,
      "grad_norm": 18.897013710329233,
      "learning_rate": 8.181689942163916e-07,
      "loss": 1.6887,
      "step": 580
    },
    {
      "epoch": 0.02,
      "grad_norm": 25.51822899243616,
      "learning_rate": 8.252221752010158e-07,
      "loss": 1.7394,
      "step": 585
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.055805022967824,
      "learning_rate": 8.322753561856398e-07,
      "loss": 1.7094,
      "step": 590
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.56903423819068,
      "learning_rate": 8.393285371702639e-07,
      "loss": 1.7717,
      "step": 595
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.14916579452123,
      "learning_rate": 8.463817181548879e-07,
      "loss": 1.7228,
      "step": 600
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.453570175592414,
      "learning_rate": 8.53434899139512e-07,
      "loss": 1.6956,
      "step": 605
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.41331527875509,
      "learning_rate": 8.604880801241361e-07,
      "loss": 1.6637,
      "step": 610
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.160251384145718,
      "learning_rate": 8.675412611087601e-07,
      "loss": 1.7307,
      "step": 615
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.755458063023116,
      "learning_rate": 8.745944420933841e-07,
      "loss": 1.6759,
      "step": 620
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.873437079778068,
      "learning_rate": 8.816476230780083e-07,
      "loss": 1.8,
      "step": 625
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.670246364127115,
      "learning_rate": 8.887008040626323e-07,
      "loss": 1.6605,
      "step": 630
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.697858705196312,
      "learning_rate": 8.957539850472564e-07,
      "loss": 1.637,
      "step": 635
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.385126253131975,
      "learning_rate": 9.028071660318804e-07,
      "loss": 1.7152,
      "step": 640
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.44914280911491,
      "learning_rate": 9.098603470165044e-07,
      "loss": 1.7758,
      "step": 645
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.681771535276187,
      "learning_rate": 9.169135280011286e-07,
      "loss": 1.6362,
      "step": 650
    },
    {
      "epoch": 0.03,
      "grad_norm": 68.24376713762558,
      "learning_rate": 9.239667089857526e-07,
      "loss": 1.68,
      "step": 655
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.15036407245028,
      "learning_rate": 9.310198899703766e-07,
      "loss": 1.6402,
      "step": 660
    },
    {
      "epoch": 0.03,
      "grad_norm": 25.189743493363206,
      "learning_rate": 9.380730709550008e-07,
      "loss": 1.6397,
      "step": 665
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.042629468292205,
      "learning_rate": 9.451262519396248e-07,
      "loss": 1.6398,
      "step": 670
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.995753481259305,
      "learning_rate": 9.52179432924249e-07,
      "loss": 1.647,
      "step": 675
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.09703797004552,
      "learning_rate": 9.59232613908873e-07,
      "loss": 1.6328,
      "step": 680
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.576412545883827,
      "learning_rate": 9.66285794893497e-07,
      "loss": 1.66,
      "step": 685
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.316074398747404,
      "learning_rate": 9.73338975878121e-07,
      "loss": 1.5745,
      "step": 690
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.177981255253448,
      "learning_rate": 9.80392156862745e-07,
      "loss": 1.6405,
      "step": 695
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.350109267760402,
      "learning_rate": 9.87445337847369e-07,
      "loss": 1.6253,
      "step": 700
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.709539765625115,
      "learning_rate": 9.944985188319933e-07,
      "loss": 1.6578,
      "step": 705
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.576124272552956,
      "learning_rate": 1.0015516998166174e-06,
      "loss": 1.5545,
      "step": 710
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.170981917972304,
      "learning_rate": 1.0086048808012416e-06,
      "loss": 1.6033,
      "step": 715
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.280043822001478,
      "learning_rate": 1.0156580617858656e-06,
      "loss": 1.7083,
      "step": 720
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.21071445705185,
      "learning_rate": 1.0227112427704894e-06,
      "loss": 1.5985,
      "step": 725
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.11314449596276,
      "learning_rate": 1.0297644237551137e-06,
      "loss": 1.6236,
      "step": 730
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.49088382137558,
      "learning_rate": 1.0368176047397377e-06,
      "loss": 1.6458,
      "step": 735
    },
    {
      "epoch": 0.03,
      "grad_norm": 23.477084210504625,
      "learning_rate": 1.0438707857243617e-06,
      "loss": 1.619,
      "step": 740
    },
    {
      "epoch": 0.03,
      "grad_norm": 18.294983555366425,
      "learning_rate": 1.050923966708986e-06,
      "loss": 1.604,
      "step": 745
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.52057771612792,
      "learning_rate": 1.05797714769361e-06,
      "loss": 1.5461,
      "step": 750
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.346423808447515,
      "learning_rate": 1.065030328678234e-06,
      "loss": 1.5353,
      "step": 755
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.200436885611502,
      "learning_rate": 1.072083509662858e-06,
      "loss": 1.6506,
      "step": 760
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.807350294042166,
      "learning_rate": 1.079136690647482e-06,
      "loss": 1.6094,
      "step": 765
    },
    {
      "epoch": 0.03,
      "grad_norm": 31.180522720043733,
      "learning_rate": 1.0861898716321062e-06,
      "loss": 1.6395,
      "step": 770
    },
    {
      "epoch": 0.03,
      "grad_norm": 21.33280205078513,
      "learning_rate": 1.0932430526167303e-06,
      "loss": 1.5827,
      "step": 775
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.890135262047313,
      "learning_rate": 1.1002962336013543e-06,
      "loss": 1.5186,
      "step": 780
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.326513851295944,
      "learning_rate": 1.1073494145859783e-06,
      "loss": 1.6546,
      "step": 785
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.32273931917068,
      "learning_rate": 1.1144025955706023e-06,
      "loss": 1.5597,
      "step": 790
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.047288821920386,
      "learning_rate": 1.1214557765552266e-06,
      "loss": 1.5502,
      "step": 795
    },
    {
      "epoch": 0.03,
      "grad_norm": 15.680953956325846,
      "learning_rate": 1.1285089575398506e-06,
      "loss": 1.6232,
      "step": 800
    },
    {
      "epoch": 0.03,
      "grad_norm": 28.445383445826906,
      "learning_rate": 1.1355621385244746e-06,
      "loss": 1.5552,
      "step": 805
    },
    {
      "epoch": 0.03,
      "grad_norm": 14.706228347694973,
      "learning_rate": 1.1426153195090988e-06,
      "loss": 1.5749,
      "step": 810
    },
    {
      "epoch": 0.03,
      "grad_norm": 16.150970886796813,
      "learning_rate": 1.1496685004937226e-06,
      "loss": 1.6283,
      "step": 815
    },
    {
      "epoch": 0.03,
      "grad_norm": 17.330211320075424,
      "learning_rate": 1.1567216814783467e-06,
      "loss": 1.5405,
      "step": 820
    },
    {
      "epoch": 0.03,
      "grad_norm": 13.595097796083746,
      "learning_rate": 1.163774862462971e-06,
      "loss": 1.531,
      "step": 825
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.969918818831422,
      "learning_rate": 1.170828043447595e-06,
      "loss": 1.5029,
      "step": 830
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.331658046415496,
      "learning_rate": 1.1778812244322192e-06,
      "loss": 1.4709,
      "step": 835
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.983987382276847,
      "learning_rate": 1.1849344054168432e-06,
      "loss": 1.5851,
      "step": 840
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.32997600480952,
      "learning_rate": 1.1919875864014672e-06,
      "loss": 1.5011,
      "step": 845
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.608800626612172,
      "learning_rate": 1.1990407673860912e-06,
      "loss": 1.5744,
      "step": 850
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.43423848123491,
      "learning_rate": 1.2060939483707152e-06,
      "loss": 1.5913,
      "step": 855
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.563368927957066,
      "learning_rate": 1.2131471293553393e-06,
      "loss": 1.4959,
      "step": 860
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.962098801441446,
      "learning_rate": 1.2202003103399635e-06,
      "loss": 1.5536,
      "step": 865
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.663938218494293,
      "learning_rate": 1.2272534913245875e-06,
      "loss": 1.5721,
      "step": 870
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.808277124221602,
      "learning_rate": 1.2343066723092115e-06,
      "loss": 1.6077,
      "step": 875
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.660652928445412,
      "learning_rate": 1.2413598532938356e-06,
      "loss": 1.5393,
      "step": 880
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.406825454097827,
      "learning_rate": 1.2484130342784596e-06,
      "loss": 1.5348,
      "step": 885
    },
    {
      "epoch": 0.04,
      "grad_norm": 12.965892407297877,
      "learning_rate": 1.2554662152630836e-06,
      "loss": 1.5528,
      "step": 890
    },
    {
      "epoch": 0.04,
      "grad_norm": 21.573662392243765,
      "learning_rate": 1.2625193962477078e-06,
      "loss": 1.5669,
      "step": 895
    },
    {
      "epoch": 0.04,
      "grad_norm": 23.675814715786238,
      "learning_rate": 1.2695725772323319e-06,
      "loss": 1.5172,
      "step": 900
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.171504982437398,
      "learning_rate": 1.2766257582169559e-06,
      "loss": 1.5242,
      "step": 905
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.92744254846148,
      "learning_rate": 1.28367893920158e-06,
      "loss": 1.5454,
      "step": 910
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.538907137463239,
      "learning_rate": 1.2907321201862041e-06,
      "loss": 1.5368,
      "step": 915
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.720558770000086,
      "learning_rate": 1.297785301170828e-06,
      "loss": 1.5307,
      "step": 920
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.74935766220052,
      "learning_rate": 1.3048384821554522e-06,
      "loss": 1.5091,
      "step": 925
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.55288877610659,
      "learning_rate": 1.3118916631400764e-06,
      "loss": 1.5422,
      "step": 930
    },
    {
      "epoch": 0.04,
      "grad_norm": 25.569895521724,
      "learning_rate": 1.3189448441247004e-06,
      "loss": 1.5712,
      "step": 935
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.420460603450163,
      "learning_rate": 1.3259980251093244e-06,
      "loss": 1.4793,
      "step": 940
    },
    {
      "epoch": 0.04,
      "grad_norm": 20.78721483805772,
      "learning_rate": 1.3330512060939485e-06,
      "loss": 1.5645,
      "step": 945
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.791878511339808,
      "learning_rate": 1.3401043870785727e-06,
      "loss": 1.5616,
      "step": 950
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.202414812578114,
      "learning_rate": 1.3471575680631965e-06,
      "loss": 1.4889,
      "step": 955
    },
    {
      "epoch": 0.04,
      "grad_norm": 18.27629934592644,
      "learning_rate": 1.3542107490478207e-06,
      "loss": 1.5141,
      "step": 960
    },
    {
      "epoch": 0.04,
      "grad_norm": 32.4150671492386,
      "learning_rate": 1.3612639300324448e-06,
      "loss": 1.5331,
      "step": 965
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.093751622915649,
      "learning_rate": 1.3683171110170688e-06,
      "loss": 1.4729,
      "step": 970
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.958939198385906,
      "learning_rate": 1.3753702920016928e-06,
      "loss": 1.5687,
      "step": 975
    },
    {
      "epoch": 0.04,
      "grad_norm": 30.95918343385252,
      "learning_rate": 1.382423472986317e-06,
      "loss": 1.5015,
      "step": 980
    },
    {
      "epoch": 0.04,
      "grad_norm": 25.006202414226543,
      "learning_rate": 1.3894766539709409e-06,
      "loss": 1.513,
      "step": 985
    },
    {
      "epoch": 0.04,
      "grad_norm": 29.915355187277456,
      "learning_rate": 1.396529834955565e-06,
      "loss": 1.493,
      "step": 990
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.19724693772254,
      "learning_rate": 1.403583015940189e-06,
      "loss": 1.6291,
      "step": 995
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.419354806927592,
      "learning_rate": 1.4106361969248131e-06,
      "loss": 1.5671,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "grad_norm": 13.511881848270505,
      "learning_rate": 1.4176893779094371e-06,
      "loss": 1.5093,
      "step": 1005
    },
    {
      "epoch": 0.04,
      "grad_norm": 19.15809744907682,
      "learning_rate": 1.4247425588940614e-06,
      "loss": 1.4915,
      "step": 1010
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.73348327692445,
      "learning_rate": 1.4317957398786856e-06,
      "loss": 1.4952,
      "step": 1015
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.140021739161664,
      "learning_rate": 1.4388489208633094e-06,
      "loss": 1.5194,
      "step": 1020
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.432519692489358,
      "learning_rate": 1.4459021018479337e-06,
      "loss": 1.5706,
      "step": 1025
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.332989363553716,
      "learning_rate": 1.4529552828325577e-06,
      "loss": 1.5046,
      "step": 1030
    },
    {
      "epoch": 0.04,
      "grad_norm": 16.583952967506214,
      "learning_rate": 1.4600084638171815e-06,
      "loss": 1.4802,
      "step": 1035
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.459657012064126,
      "learning_rate": 1.4670616448018057e-06,
      "loss": 1.4653,
      "step": 1040
    },
    {
      "epoch": 0.04,
      "grad_norm": 17.122154121342444,
      "learning_rate": 1.47411482578643e-06,
      "loss": 1.5526,
      "step": 1045
    },
    {
      "epoch": 0.04,
      "grad_norm": 14.329292660217874,
      "learning_rate": 1.4811680067710538e-06,
      "loss": 1.4963,
      "step": 1050
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.230617845206329,
      "learning_rate": 1.488221187755678e-06,
      "loss": 1.4018,
      "step": 1055
    },
    {
      "epoch": 0.04,
      "grad_norm": 15.873361842831489,
      "learning_rate": 1.495274368740302e-06,
      "loss": 1.4458,
      "step": 1060
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.782794216072803,
      "learning_rate": 1.502327549724926e-06,
      "loss": 1.591,
      "step": 1065
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.206349777440574,
      "learning_rate": 1.50938073070955e-06,
      "loss": 1.4889,
      "step": 1070
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.225104279659387,
      "learning_rate": 1.5164339116941743e-06,
      "loss": 1.512,
      "step": 1075
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.346180509521528,
      "learning_rate": 1.523487092678798e-06,
      "loss": 1.4694,
      "step": 1080
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.063404822419741,
      "learning_rate": 1.5305402736634223e-06,
      "loss": 1.4155,
      "step": 1085
    },
    {
      "epoch": 0.05,
      "grad_norm": 20.455007784432723,
      "learning_rate": 1.5375934546480464e-06,
      "loss": 1.4692,
      "step": 1090
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.57413377412271,
      "learning_rate": 1.5446466356326706e-06,
      "loss": 1.5055,
      "step": 1095
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.85600466250613,
      "learning_rate": 1.5516998166172944e-06,
      "loss": 1.4905,
      "step": 1100
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.199361241676119,
      "learning_rate": 1.5587529976019186e-06,
      "loss": 1.4647,
      "step": 1105
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.771270353435733,
      "learning_rate": 1.5658061785865429e-06,
      "loss": 1.4676,
      "step": 1110
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.749034314939355,
      "learning_rate": 1.5728593595711667e-06,
      "loss": 1.4262,
      "step": 1115
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.00053305157364,
      "learning_rate": 1.5799125405557907e-06,
      "loss": 1.4795,
      "step": 1120
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.003896234045099,
      "learning_rate": 1.586965721540415e-06,
      "loss": 1.4993,
      "step": 1125
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.095279458152357,
      "learning_rate": 1.5940189025250387e-06,
      "loss": 1.4814,
      "step": 1130
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.543162407942619,
      "learning_rate": 1.601072083509663e-06,
      "loss": 1.4315,
      "step": 1135
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.42311101247395,
      "learning_rate": 1.6081252644942872e-06,
      "loss": 1.4234,
      "step": 1140
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.350351773931624,
      "learning_rate": 1.615178445478911e-06,
      "loss": 1.4529,
      "step": 1145
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.06791995678334,
      "learning_rate": 1.6222316264635352e-06,
      "loss": 1.4306,
      "step": 1150
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.044063353747298,
      "learning_rate": 1.6292848074481593e-06,
      "loss": 1.4614,
      "step": 1155
    },
    {
      "epoch": 0.05,
      "grad_norm": 12.965247686511777,
      "learning_rate": 1.6363379884327833e-06,
      "loss": 1.4448,
      "step": 1160
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.39413251454674,
      "learning_rate": 1.6433911694174073e-06,
      "loss": 1.4909,
      "step": 1165
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.501289202231618,
      "learning_rate": 1.6504443504020315e-06,
      "loss": 1.4759,
      "step": 1170
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.004850610466043,
      "learning_rate": 1.6574975313866556e-06,
      "loss": 1.4527,
      "step": 1175
    },
    {
      "epoch": 0.05,
      "grad_norm": 21.049825858253758,
      "learning_rate": 1.6645507123712796e-06,
      "loss": 1.4794,
      "step": 1180
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.371088915938834,
      "learning_rate": 1.6716038933559036e-06,
      "loss": 1.5064,
      "step": 1185
    },
    {
      "epoch": 0.05,
      "grad_norm": 19.919662542711166,
      "learning_rate": 1.6786570743405278e-06,
      "loss": 1.5015,
      "step": 1190
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.06116754997933,
      "learning_rate": 1.6857102553251516e-06,
      "loss": 1.4229,
      "step": 1195
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.006772543757613,
      "learning_rate": 1.6927634363097759e-06,
      "loss": 1.4358,
      "step": 1200
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.28120385613161,
      "learning_rate": 1.6998166172944001e-06,
      "loss": 1.4693,
      "step": 1205
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.285364585101334,
      "learning_rate": 1.706869798279024e-06,
      "loss": 1.4717,
      "step": 1210
    },
    {
      "epoch": 0.05,
      "grad_norm": 15.389444883981302,
      "learning_rate": 1.713922979263648e-06,
      "loss": 1.4671,
      "step": 1215
    },
    {
      "epoch": 0.05,
      "grad_norm": 18.43444710155935,
      "learning_rate": 1.7209761602482722e-06,
      "loss": 1.4491,
      "step": 1220
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.553744604035002,
      "learning_rate": 1.728029341232896e-06,
      "loss": 1.4807,
      "step": 1225
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.255376552689096,
      "learning_rate": 1.7350825222175202e-06,
      "loss": 1.4971,
      "step": 1230
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.162027337280843,
      "learning_rate": 1.7421357032021445e-06,
      "loss": 1.4039,
      "step": 1235
    },
    {
      "epoch": 0.05,
      "grad_norm": 13.735772683725203,
      "learning_rate": 1.7491888841867683e-06,
      "loss": 1.4397,
      "step": 1240
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.31259459324704,
      "learning_rate": 1.7562420651713925e-06,
      "loss": 1.4632,
      "step": 1245
    },
    {
      "epoch": 0.05,
      "grad_norm": 11.860225896706122,
      "learning_rate": 1.7632952461560165e-06,
      "loss": 1.4312,
      "step": 1250
    },
    {
      "epoch": 0.05,
      "grad_norm": 16.60544134913032,
      "learning_rate": 1.7703484271406407e-06,
      "loss": 1.4046,
      "step": 1255
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.664553111453099,
      "learning_rate": 1.7774016081252646e-06,
      "loss": 1.4493,
      "step": 1260
    },
    {
      "epoch": 0.05,
      "grad_norm": 24.980869760313446,
      "learning_rate": 1.7844547891098888e-06,
      "loss": 1.4209,
      "step": 1265
    },
    {
      "epoch": 0.05,
      "grad_norm": 17.49625044017745,
      "learning_rate": 1.7915079700945128e-06,
      "loss": 1.4141,
      "step": 1270
    },
    {
      "epoch": 0.05,
      "grad_norm": 40.34521596546542,
      "learning_rate": 1.7985611510791368e-06,
      "loss": 1.4496,
      "step": 1275
    },
    {
      "epoch": 0.05,
      "grad_norm": 22.73153569950643,
      "learning_rate": 1.8056143320637609e-06,
      "loss": 1.4257,
      "step": 1280
    },
    {
      "epoch": 0.05,
      "grad_norm": 25.069577277635705,
      "learning_rate": 1.812667513048385e-06,
      "loss": 1.4042,
      "step": 1285
    },
    {
      "epoch": 0.05,
      "grad_norm": 14.071015188265285,
      "learning_rate": 1.819720694033009e-06,
      "loss": 1.3545,
      "step": 1290
    },
    {
      "epoch": 0.05,
      "grad_norm": 27.667539679244467,
      "learning_rate": 1.8267738750176331e-06,
      "loss": 1.4392,
      "step": 1295
    },
    {
      "epoch": 0.06,
      "grad_norm": 28.486727021793588,
      "learning_rate": 1.8338270560022571e-06,
      "loss": 1.4715,
      "step": 1300
    },
    {
      "epoch": 0.06,
      "grad_norm": 18.27418774106074,
      "learning_rate": 1.8408802369868812e-06,
      "loss": 1.4184,
      "step": 1305
    },
    {
      "epoch": 0.06,
      "grad_norm": 18.594453745449353,
      "learning_rate": 1.8479334179715052e-06,
      "loss": 1.4119,
      "step": 1310
    },
    {
      "epoch": 0.06,
      "grad_norm": 21.381590391450835,
      "learning_rate": 1.8549865989561294e-06,
      "loss": 1.4217,
      "step": 1315
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.413120286400547,
      "learning_rate": 1.8620397799407532e-06,
      "loss": 1.4436,
      "step": 1320
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.893693175352412,
      "learning_rate": 1.8690929609253775e-06,
      "loss": 1.4022,
      "step": 1325
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.493176005244209,
      "learning_rate": 1.8761461419100017e-06,
      "loss": 1.4129,
      "step": 1330
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.326519977003366,
      "learning_rate": 1.8831993228946257e-06,
      "loss": 1.418,
      "step": 1335
    },
    {
      "epoch": 0.06,
      "grad_norm": 13.237437459256197,
      "learning_rate": 1.8902525038792495e-06,
      "loss": 1.3908,
      "step": 1340
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.147333913625673,
      "learning_rate": 1.8973056848638738e-06,
      "loss": 1.4188,
      "step": 1345
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.48135737731555,
      "learning_rate": 1.904358865848498e-06,
      "loss": 1.4249,
      "step": 1350
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.271710806477344,
      "learning_rate": 1.911412046833122e-06,
      "loss": 1.4079,
      "step": 1355
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.87900935508388,
      "learning_rate": 1.918465227817746e-06,
      "loss": 1.4289,
      "step": 1360
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.74854488831577,
      "learning_rate": 1.92551840880237e-06,
      "loss": 1.4398,
      "step": 1365
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.158899131153982,
      "learning_rate": 1.932571589786994e-06,
      "loss": 1.436,
      "step": 1370
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.39229688205433,
      "learning_rate": 1.939624770771618e-06,
      "loss": 1.4569,
      "step": 1375
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.25788023716984,
      "learning_rate": 1.946677951756242e-06,
      "loss": 1.3673,
      "step": 1380
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.551985889523067,
      "learning_rate": 1.953731132740866e-06,
      "loss": 1.4129,
      "step": 1385
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.019618245577336,
      "learning_rate": 1.96078431372549e-06,
      "loss": 1.3786,
      "step": 1390
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.958370211529191,
      "learning_rate": 1.9678374947101146e-06,
      "loss": 1.428,
      "step": 1395
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.942414051256707,
      "learning_rate": 1.974890675694738e-06,
      "loss": 1.3626,
      "step": 1400
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.110012624743174,
      "learning_rate": 1.9819438566793627e-06,
      "loss": 1.4725,
      "step": 1405
    },
    {
      "epoch": 0.06,
      "grad_norm": 19.496193291138677,
      "learning_rate": 1.9889970376639867e-06,
      "loss": 1.3902,
      "step": 1410
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.957177884901764,
      "learning_rate": 1.9960502186486107e-06,
      "loss": 1.3647,
      "step": 1415
    },
    {
      "epoch": 0.06,
      "grad_norm": 23.97029717591676,
      "learning_rate": 2.0031033996332347e-06,
      "loss": 1.4448,
      "step": 1420
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.661277134514194,
      "learning_rate": 2.0101565806178587e-06,
      "loss": 1.38,
      "step": 1425
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.466348958740856,
      "learning_rate": 2.017209761602483e-06,
      "loss": 1.4105,
      "step": 1430
    },
    {
      "epoch": 0.06,
      "grad_norm": 14.18943404010148,
      "learning_rate": 2.0242629425871068e-06,
      "loss": 1.4121,
      "step": 1435
    },
    {
      "epoch": 0.06,
      "grad_norm": 26.197151032062003,
      "learning_rate": 2.0313161235717312e-06,
      "loss": 1.419,
      "step": 1440
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.08878236179872,
      "learning_rate": 2.0383693045563552e-06,
      "loss": 1.4033,
      "step": 1445
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.89064350358611,
      "learning_rate": 2.045422485540979e-06,
      "loss": 1.4637,
      "step": 1450
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.70613178958977,
      "learning_rate": 2.0524756665256033e-06,
      "loss": 1.353,
      "step": 1455
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.791535916409925,
      "learning_rate": 2.0595288475102273e-06,
      "loss": 1.3793,
      "step": 1460
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.147748419636127,
      "learning_rate": 2.0665820284948513e-06,
      "loss": 1.4328,
      "step": 1465
    },
    {
      "epoch": 0.06,
      "grad_norm": 49.261388594284156,
      "learning_rate": 2.0736352094794754e-06,
      "loss": 1.4484,
      "step": 1470
    },
    {
      "epoch": 0.06,
      "grad_norm": 63.17069815777682,
      "learning_rate": 2.0806883904640994e-06,
      "loss": 1.3834,
      "step": 1475
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.663593522273047,
      "learning_rate": 2.0877415714487234e-06,
      "loss": 1.4638,
      "step": 1480
    },
    {
      "epoch": 0.06,
      "grad_norm": 75.92609631954323,
      "learning_rate": 2.0947947524333474e-06,
      "loss": 1.3994,
      "step": 1485
    },
    {
      "epoch": 0.06,
      "grad_norm": 22.106485004542424,
      "learning_rate": 2.101847933417972e-06,
      "loss": 1.3998,
      "step": 1490
    },
    {
      "epoch": 0.06,
      "grad_norm": 40.84734254107766,
      "learning_rate": 2.108901114402596e-06,
      "loss": 1.448,
      "step": 1495
    },
    {
      "epoch": 0.06,
      "grad_norm": 39.78835295369607,
      "learning_rate": 2.11595429538722e-06,
      "loss": 1.3962,
      "step": 1500
    },
    {
      "epoch": 0.06,
      "grad_norm": 29.183531521251187,
      "learning_rate": 2.123007476371844e-06,
      "loss": 1.4144,
      "step": 1505
    },
    {
      "epoch": 0.06,
      "grad_norm": 36.42115195566105,
      "learning_rate": 2.130060657356468e-06,
      "loss": 1.432,
      "step": 1510
    },
    {
      "epoch": 0.06,
      "grad_norm": 16.0577774070433,
      "learning_rate": 2.137113838341092e-06,
      "loss": 1.3785,
      "step": 1515
    },
    {
      "epoch": 0.06,
      "grad_norm": 29.85453094374671,
      "learning_rate": 2.144167019325716e-06,
      "loss": 1.4085,
      "step": 1520
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.714264755030822,
      "learning_rate": 2.1512202003103404e-06,
      "loss": 1.4052,
      "step": 1525
    },
    {
      "epoch": 0.06,
      "grad_norm": 27.39846878215763,
      "learning_rate": 2.158273381294964e-06,
      "loss": 1.4022,
      "step": 1530
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.085106952588045,
      "learning_rate": 2.1653265622795885e-06,
      "loss": 1.433,
      "step": 1535
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.702911786029883,
      "learning_rate": 2.1723797432642125e-06,
      "loss": 1.405,
      "step": 1540
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.21316478545504,
      "learning_rate": 2.179432924248836e-06,
      "loss": 1.4159,
      "step": 1545
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.891125178606297,
      "learning_rate": 2.1864861052334605e-06,
      "loss": 1.399,
      "step": 1550
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.408338062001597,
      "learning_rate": 2.1935392862180846e-06,
      "loss": 1.4339,
      "step": 1555
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.287379305314193,
      "learning_rate": 2.2005924672027086e-06,
      "loss": 1.3584,
      "step": 1560
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.05706675877235,
      "learning_rate": 2.2076456481873326e-06,
      "loss": 1.4022,
      "step": 1565
    },
    {
      "epoch": 0.07,
      "grad_norm": 26.829991457932888,
      "learning_rate": 2.2146988291719566e-06,
      "loss": 1.4012,
      "step": 1570
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.6522840404752,
      "learning_rate": 2.221752010156581e-06,
      "loss": 1.3991,
      "step": 1575
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.827408295721925,
      "learning_rate": 2.2288051911412047e-06,
      "loss": 1.3692,
      "step": 1580
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.130862768303327,
      "learning_rate": 2.235858372125829e-06,
      "loss": 1.4243,
      "step": 1585
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.080414871071657,
      "learning_rate": 2.242911553110453e-06,
      "loss": 1.3906,
      "step": 1590
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.720817904904198,
      "learning_rate": 2.249964734095077e-06,
      "loss": 1.3741,
      "step": 1595
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.169883267513889,
      "learning_rate": 2.257017915079701e-06,
      "loss": 1.3741,
      "step": 1600
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.592925311905242,
      "learning_rate": 2.264071096064325e-06,
      "loss": 1.3667,
      "step": 1605
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.32664364869199,
      "learning_rate": 2.2711242770489492e-06,
      "loss": 1.3673,
      "step": 1610
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.983222923205547,
      "learning_rate": 2.2781774580335732e-06,
      "loss": 1.3391,
      "step": 1615
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.084498525731727,
      "learning_rate": 2.2852306390181977e-06,
      "loss": 1.358,
      "step": 1620
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.721355413481922,
      "learning_rate": 2.2922838200028213e-06,
      "loss": 1.4439,
      "step": 1625
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.22726316745244,
      "learning_rate": 2.2993370009874453e-06,
      "loss": 1.3166,
      "step": 1630
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.448353076993822,
      "learning_rate": 2.3063901819720697e-06,
      "loss": 1.4671,
      "step": 1635
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.21732388378683,
      "learning_rate": 2.3134433629566933e-06,
      "loss": 1.3832,
      "step": 1640
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.162029900989616,
      "learning_rate": 2.3204965439413178e-06,
      "loss": 1.3298,
      "step": 1645
    },
    {
      "epoch": 0.07,
      "grad_norm": 20.131760362607487,
      "learning_rate": 2.327549724925942e-06,
      "loss": 1.3596,
      "step": 1650
    },
    {
      "epoch": 0.07,
      "grad_norm": 17.75664903033911,
      "learning_rate": 2.334602905910566e-06,
      "loss": 1.3084,
      "step": 1655
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.599873670259976,
      "learning_rate": 2.34165608689519e-06,
      "loss": 1.3819,
      "step": 1660
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.33043534627968,
      "learning_rate": 2.348709267879814e-06,
      "loss": 1.3836,
      "step": 1665
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.114395708736549,
      "learning_rate": 2.3557624488644383e-06,
      "loss": 1.3784,
      "step": 1670
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.223743901122075,
      "learning_rate": 2.362815629849062e-06,
      "loss": 1.3816,
      "step": 1675
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.026634719268413,
      "learning_rate": 2.3698688108336864e-06,
      "loss": 1.3273,
      "step": 1680
    },
    {
      "epoch": 0.07,
      "grad_norm": 18.017701591744665,
      "learning_rate": 2.3769219918183104e-06,
      "loss": 1.3369,
      "step": 1685
    },
    {
      "epoch": 0.07,
      "grad_norm": 13.683754087387241,
      "learning_rate": 2.3839751728029344e-06,
      "loss": 1.409,
      "step": 1690
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.348704753527892,
      "learning_rate": 2.3910283537875584e-06,
      "loss": 1.4118,
      "step": 1695
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.542889735095187,
      "learning_rate": 2.3980815347721824e-06,
      "loss": 1.3366,
      "step": 1700
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.025686891831606,
      "learning_rate": 2.4051347157568065e-06,
      "loss": 1.3229,
      "step": 1705
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.533083554714429,
      "learning_rate": 2.4121878967414305e-06,
      "loss": 1.3833,
      "step": 1710
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.737349718281207,
      "learning_rate": 2.4192410777260545e-06,
      "loss": 1.3506,
      "step": 1715
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.545444844779153,
      "learning_rate": 2.4262942587106785e-06,
      "loss": 1.4374,
      "step": 1720
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.782696376867012,
      "learning_rate": 2.4333474396953025e-06,
      "loss": 1.3368,
      "step": 1725
    },
    {
      "epoch": 0.07,
      "grad_norm": 23.511542513936423,
      "learning_rate": 2.440400620679927e-06,
      "loss": 1.3859,
      "step": 1730
    },
    {
      "epoch": 0.07,
      "grad_norm": 22.21156240631286,
      "learning_rate": 2.447453801664551e-06,
      "loss": 1.4004,
      "step": 1735
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.81637150409835,
      "learning_rate": 2.454506982649175e-06,
      "loss": 1.3957,
      "step": 1740
    },
    {
      "epoch": 0.07,
      "grad_norm": 12.369915233483516,
      "learning_rate": 2.461560163633799e-06,
      "loss": 1.3727,
      "step": 1745
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.648441458739722,
      "learning_rate": 2.468613344618423e-06,
      "loss": 1.363,
      "step": 1750
    },
    {
      "epoch": 0.07,
      "grad_norm": 10.931682600920412,
      "learning_rate": 2.475666525603047e-06,
      "loss": 1.3587,
      "step": 1755
    },
    {
      "epoch": 0.07,
      "grad_norm": 34.330012053919155,
      "learning_rate": 2.482719706587671e-06,
      "loss": 1.3958,
      "step": 1760
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.735310404626148,
      "learning_rate": 2.4897728875722956e-06,
      "loss": 1.3612,
      "step": 1765
    },
    {
      "epoch": 0.07,
      "grad_norm": 16.384813410643346,
      "learning_rate": 2.496826068556919e-06,
      "loss": 1.3112,
      "step": 1770
    },
    {
      "epoch": 0.08,
      "grad_norm": 26.838655170379702,
      "learning_rate": 2.5038792495415436e-06,
      "loss": 1.3532,
      "step": 1775
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.706171587310045,
      "learning_rate": 2.510932430526167e-06,
      "loss": 1.391,
      "step": 1780
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.126954660108268,
      "learning_rate": 2.5179856115107916e-06,
      "loss": 1.3312,
      "step": 1785
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.9804836218818,
      "learning_rate": 2.5250387924954157e-06,
      "loss": 1.3977,
      "step": 1790
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.558145099953244,
      "learning_rate": 2.5320919734800397e-06,
      "loss": 1.3647,
      "step": 1795
    },
    {
      "epoch": 0.08,
      "grad_norm": 31.61058690933841,
      "learning_rate": 2.5391451544646637e-06,
      "loss": 1.3761,
      "step": 1800
    },
    {
      "epoch": 0.08,
      "grad_norm": 30.24686401792589,
      "learning_rate": 2.546198335449288e-06,
      "loss": 1.3049,
      "step": 1805
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.651728714815397,
      "learning_rate": 2.5532515164339118e-06,
      "loss": 1.3646,
      "step": 1810
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.507949081652326,
      "learning_rate": 2.5603046974185358e-06,
      "loss": 1.3453,
      "step": 1815
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.895282000896792,
      "learning_rate": 2.56735787840316e-06,
      "loss": 1.399,
      "step": 1820
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.897609064572269,
      "learning_rate": 2.5744110593877842e-06,
      "loss": 1.3668,
      "step": 1825
    },
    {
      "epoch": 0.08,
      "grad_norm": 13.37757192102203,
      "learning_rate": 2.5814642403724083e-06,
      "loss": 1.2878,
      "step": 1830
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.8907097254796,
      "learning_rate": 2.5885174213570323e-06,
      "loss": 1.3834,
      "step": 1835
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.168138053426295,
      "learning_rate": 2.595570602341656e-06,
      "loss": 1.3573,
      "step": 1840
    },
    {
      "epoch": 0.08,
      "grad_norm": 16.758859909730333,
      "learning_rate": 2.6026237833262803e-06,
      "loss": 1.3796,
      "step": 1845
    },
    {
      "epoch": 0.08,
      "grad_norm": 46.55914050794329,
      "learning_rate": 2.6096769643109043e-06,
      "loss": 1.4054,
      "step": 1850
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.077422825170535,
      "learning_rate": 2.6167301452955284e-06,
      "loss": 1.3622,
      "step": 1855
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.461754902624616,
      "learning_rate": 2.623783326280153e-06,
      "loss": 1.4014,
      "step": 1860
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.819181630091125,
      "learning_rate": 2.630836507264777e-06,
      "loss": 1.3178,
      "step": 1865
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.989631124341651,
      "learning_rate": 2.637889688249401e-06,
      "loss": 1.394,
      "step": 1870
    },
    {
      "epoch": 0.08,
      "grad_norm": 63.08141055617478,
      "learning_rate": 2.6449428692340245e-06,
      "loss": 1.3432,
      "step": 1875
    },
    {
      "epoch": 0.08,
      "grad_norm": 95.68367095251443,
      "learning_rate": 2.651996050218649e-06,
      "loss": 1.3665,
      "step": 1880
    },
    {
      "epoch": 0.08,
      "grad_norm": 58.67303782481335,
      "learning_rate": 2.659049231203273e-06,
      "loss": 1.3322,
      "step": 1885
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.602893765265211,
      "learning_rate": 2.666102412187897e-06,
      "loss": 1.3146,
      "step": 1890
    },
    {
      "epoch": 0.08,
      "grad_norm": 18.2506181142999,
      "learning_rate": 2.673155593172521e-06,
      "loss": 1.3438,
      "step": 1895
    },
    {
      "epoch": 0.08,
      "grad_norm": 20.76463669688865,
      "learning_rate": 2.6802087741571454e-06,
      "loss": 1.3194,
      "step": 1900
    },
    {
      "epoch": 0.08,
      "grad_norm": 29.568610153820664,
      "learning_rate": 2.687261955141769e-06,
      "loss": 1.3281,
      "step": 1905
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.562512088861933,
      "learning_rate": 2.694315136126393e-06,
      "loss": 1.4566,
      "step": 1910
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.227205242969235,
      "learning_rate": 2.701368317111017e-06,
      "loss": 1.2923,
      "step": 1915
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.867047764220658,
      "learning_rate": 2.7084214980956415e-06,
      "loss": 1.3746,
      "step": 1920
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.276227086261787,
      "learning_rate": 2.7154746790802655e-06,
      "loss": 1.3505,
      "step": 1925
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.823383199296096,
      "learning_rate": 2.7225278600648895e-06,
      "loss": 1.3455,
      "step": 1930
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.504931851462494,
      "learning_rate": 2.729581041049514e-06,
      "loss": 1.3695,
      "step": 1935
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.210047329054863,
      "learning_rate": 2.7366342220341376e-06,
      "loss": 1.3894,
      "step": 1940
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.882538088996677,
      "learning_rate": 2.7436874030187616e-06,
      "loss": 1.4123,
      "step": 1945
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.991949556805977,
      "learning_rate": 2.7507405840033856e-06,
      "loss": 1.3439,
      "step": 1950
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.073250901092063,
      "learning_rate": 2.75779376498801e-06,
      "loss": 1.3315,
      "step": 1955
    },
    {
      "epoch": 0.08,
      "grad_norm": 22.564408712583464,
      "learning_rate": 2.764846945972634e-06,
      "loss": 1.3305,
      "step": 1960
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.597199108169024,
      "learning_rate": 2.771900126957258e-06,
      "loss": 1.3554,
      "step": 1965
    },
    {
      "epoch": 0.08,
      "grad_norm": 17.32623367921614,
      "learning_rate": 2.7789533079418817e-06,
      "loss": 1.3102,
      "step": 1970
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.800012276730401,
      "learning_rate": 2.786006488926506e-06,
      "loss": 1.3767,
      "step": 1975
    },
    {
      "epoch": 0.08,
      "grad_norm": 14.087888765565582,
      "learning_rate": 2.79305966991113e-06,
      "loss": 1.3371,
      "step": 1980
    },
    {
      "epoch": 0.08,
      "grad_norm": 24.082486116092745,
      "learning_rate": 2.800112850895754e-06,
      "loss": 1.3409,
      "step": 1985
    },
    {
      "epoch": 0.08,
      "grad_norm": 21.58854975477442,
      "learning_rate": 2.807166031880378e-06,
      "loss": 1.3658,
      "step": 1990
    },
    {
      "epoch": 0.08,
      "grad_norm": 10.939736646976643,
      "learning_rate": 2.8142192128650027e-06,
      "loss": 1.3868,
      "step": 1995
    },
    {
      "epoch": 0.08,
      "grad_norm": 15.184029861709678,
      "learning_rate": 2.8212723938496263e-06,
      "loss": 1.3823,
      "step": 2000
    },
    {
      "epoch": 0.08,
      "grad_norm": 12.331610968276019,
      "learning_rate": 2.8283255748342503e-06,
      "loss": 1.3486,
      "step": 2005
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.841051214282952,
      "learning_rate": 2.8353787558188743e-06,
      "loss": 1.3259,
      "step": 2010
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.349318652539013,
      "learning_rate": 2.8424319368034987e-06,
      "loss": 1.312,
      "step": 2015
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.757183957633567,
      "learning_rate": 2.8494851177881228e-06,
      "loss": 1.3936,
      "step": 2020
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.332244529140562,
      "learning_rate": 2.8565382987727468e-06,
      "loss": 1.3405,
      "step": 2025
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.58824888325681,
      "learning_rate": 2.8635914797573712e-06,
      "loss": 1.3458,
      "step": 2030
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.159526982596624,
      "learning_rate": 2.870644660741995e-06,
      "loss": 1.3629,
      "step": 2035
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.955794904060317,
      "learning_rate": 2.877697841726619e-06,
      "loss": 1.3312,
      "step": 2040
    },
    {
      "epoch": 0.09,
      "grad_norm": 26.59938857728824,
      "learning_rate": 2.884751022711243e-06,
      "loss": 1.3191,
      "step": 2045
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.677332217292815,
      "learning_rate": 2.8918042036958673e-06,
      "loss": 1.3026,
      "step": 2050
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.252022834666661,
      "learning_rate": 2.8988573846804913e-06,
      "loss": 1.3127,
      "step": 2055
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.27122412233228,
      "learning_rate": 2.9059105656651154e-06,
      "loss": 1.3196,
      "step": 2060
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.728472202693842,
      "learning_rate": 2.912963746649739e-06,
      "loss": 1.3256,
      "step": 2065
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.852605127202235,
      "learning_rate": 2.920016927634363e-06,
      "loss": 1.3346,
      "step": 2070
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.759742743483143,
      "learning_rate": 2.9270701086189874e-06,
      "loss": 1.3265,
      "step": 2075
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.968779214939827,
      "learning_rate": 2.9341232896036114e-06,
      "loss": 1.3381,
      "step": 2080
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.78630714297429,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 1.3242,
      "step": 2085
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.027645849501496,
      "learning_rate": 2.94822965157286e-06,
      "loss": 1.3256,
      "step": 2090
    },
    {
      "epoch": 0.09,
      "grad_norm": 22.951803114724818,
      "learning_rate": 2.955282832557484e-06,
      "loss": 1.3203,
      "step": 2095
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.702937636560858,
      "learning_rate": 2.9623360135421075e-06,
      "loss": 1.3017,
      "step": 2100
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.60012649866337,
      "learning_rate": 2.9693891945267315e-06,
      "loss": 1.3347,
      "step": 2105
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.825619993648441,
      "learning_rate": 2.976442375511356e-06,
      "loss": 1.3325,
      "step": 2110
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.108079068771541,
      "learning_rate": 2.98349555649598e-06,
      "loss": 1.2901,
      "step": 2115
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.716657113286335,
      "learning_rate": 2.990548737480604e-06,
      "loss": 1.3074,
      "step": 2120
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.610418128877384,
      "learning_rate": 2.9976019184652285e-06,
      "loss": 1.3635,
      "step": 2125
    },
    {
      "epoch": 0.09,
      "grad_norm": 24.76809179113239,
      "learning_rate": 3.004655099449852e-06,
      "loss": 1.3328,
      "step": 2130
    },
    {
      "epoch": 0.09,
      "grad_norm": 20.945111195759797,
      "learning_rate": 3.011708280434476e-06,
      "loss": 1.3319,
      "step": 2135
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.009822169288428,
      "learning_rate": 3.0187614614191e-06,
      "loss": 1.3325,
      "step": 2140
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.687229067901399,
      "learning_rate": 3.0258146424037246e-06,
      "loss": 1.3766,
      "step": 2145
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.42831294095289,
      "learning_rate": 3.0328678233883486e-06,
      "loss": 1.3923,
      "step": 2150
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.283394824788045,
      "learning_rate": 3.0399210043729726e-06,
      "loss": 1.3352,
      "step": 2155
    },
    {
      "epoch": 0.09,
      "grad_norm": 23.440699492833733,
      "learning_rate": 3.046974185357596e-06,
      "loss": 1.3345,
      "step": 2160
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.996365943900992,
      "learning_rate": 3.0540273663422202e-06,
      "loss": 1.3508,
      "step": 2165
    },
    {
      "epoch": 0.09,
      "grad_norm": 15.47001623803198,
      "learning_rate": 3.0610805473268447e-06,
      "loss": 1.3227,
      "step": 2170
    },
    {
      "epoch": 0.09,
      "grad_norm": 14.84822049653563,
      "learning_rate": 3.0681337283114687e-06,
      "loss": 1.3236,
      "step": 2175
    },
    {
      "epoch": 0.09,
      "grad_norm": 37.11621833214017,
      "learning_rate": 3.0751869092960927e-06,
      "loss": 1.3245,
      "step": 2180
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.987006755001374,
      "learning_rate": 3.082240090280717e-06,
      "loss": 1.3151,
      "step": 2185
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.023382053687076,
      "learning_rate": 3.089293271265341e-06,
      "loss": 1.3297,
      "step": 2190
    },
    {
      "epoch": 0.09,
      "grad_norm": 19.444832333451078,
      "learning_rate": 3.0963464522499648e-06,
      "loss": 1.2886,
      "step": 2195
    },
    {
      "epoch": 0.09,
      "grad_norm": 36.22815938531628,
      "learning_rate": 3.103399633234589e-06,
      "loss": 1.3297,
      "step": 2200
    },
    {
      "epoch": 0.09,
      "grad_norm": 12.813592227335402,
      "learning_rate": 3.1104528142192132e-06,
      "loss": 1.2639,
      "step": 2205
    },
    {
      "epoch": 0.09,
      "grad_norm": 18.004689886867,
      "learning_rate": 3.1175059952038373e-06,
      "loss": 1.3373,
      "step": 2210
    },
    {
      "epoch": 0.09,
      "grad_norm": 21.642658462844512,
      "learning_rate": 3.1245591761884613e-06,
      "loss": 1.2647,
      "step": 2215
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.879065452649801,
      "learning_rate": 3.1316123571730857e-06,
      "loss": 1.2782,
      "step": 2220
    },
    {
      "epoch": 0.09,
      "grad_norm": 10.136567194354573,
      "learning_rate": 3.1386655381577093e-06,
      "loss": 1.2772,
      "step": 2225
    },
    {
      "epoch": 0.09,
      "grad_norm": 11.41394232869268,
      "learning_rate": 3.1457187191423333e-06,
      "loss": 1.3092,
      "step": 2230
    },
    {
      "epoch": 0.09,
      "grad_norm": 13.892433684124184,
      "learning_rate": 3.1527719001269574e-06,
      "loss": 1.3778,
      "step": 2235
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.115420760888988,
      "learning_rate": 3.1598250811115814e-06,
      "loss": 1.3333,
      "step": 2240
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.224038237090316,
      "learning_rate": 3.166878262096206e-06,
      "loss": 1.3445,
      "step": 2245
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.409227556066273,
      "learning_rate": 3.17393144308083e-06,
      "loss": 1.3295,
      "step": 2250
    },
    {
      "epoch": 0.1,
      "grad_norm": 41.98539016712513,
      "learning_rate": 3.180984624065454e-06,
      "loss": 1.3305,
      "step": 2255
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.958212387725098,
      "learning_rate": 3.1880378050500775e-06,
      "loss": 1.3386,
      "step": 2260
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.036202828711685,
      "learning_rate": 3.195090986034702e-06,
      "loss": 1.3773,
      "step": 2265
    },
    {
      "epoch": 0.1,
      "grad_norm": 25.47410159501302,
      "learning_rate": 3.202144167019326e-06,
      "loss": 1.2733,
      "step": 2270
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.787436738839402,
      "learning_rate": 3.20919734800395e-06,
      "loss": 1.2663,
      "step": 2275
    },
    {
      "epoch": 0.1,
      "grad_norm": 21.62668559821898,
      "learning_rate": 3.2162505289885744e-06,
      "loss": 1.307,
      "step": 2280
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.521918095272499,
      "learning_rate": 3.2233037099731984e-06,
      "loss": 1.3043,
      "step": 2285
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.8136363287776,
      "learning_rate": 3.230356890957822e-06,
      "loss": 1.2959,
      "step": 2290
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.647769776866582,
      "learning_rate": 3.237410071942446e-06,
      "loss": 1.3262,
      "step": 2295
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.097590129920224,
      "learning_rate": 3.2444632529270705e-06,
      "loss": 1.3142,
      "step": 2300
    },
    {
      "epoch": 0.1,
      "grad_norm": 27.07091015193857,
      "learning_rate": 3.2515164339116945e-06,
      "loss": 1.318,
      "step": 2305
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.830005880702661,
      "learning_rate": 3.2585696148963185e-06,
      "loss": 1.3984,
      "step": 2310
    },
    {
      "epoch": 0.1,
      "grad_norm": 17.729141083621435,
      "learning_rate": 3.265622795880943e-06,
      "loss": 1.3169,
      "step": 2315
    },
    {
      "epoch": 0.1,
      "grad_norm": 25.65087307063273,
      "learning_rate": 3.2726759768655666e-06,
      "loss": 1.3101,
      "step": 2320
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.50130264357366,
      "learning_rate": 3.2797291578501906e-06,
      "loss": 1.3188,
      "step": 2325
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.058211371622484,
      "learning_rate": 3.2867823388348146e-06,
      "loss": 1.315,
      "step": 2330
    },
    {
      "epoch": 0.1,
      "grad_norm": 26.44583527393522,
      "learning_rate": 3.2938355198194386e-06,
      "loss": 1.2715,
      "step": 2335
    },
    {
      "epoch": 0.1,
      "grad_norm": 26.53474816416927,
      "learning_rate": 3.300888700804063e-06,
      "loss": 1.294,
      "step": 2340
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.482297997858216,
      "learning_rate": 3.307941881788687e-06,
      "loss": 1.2977,
      "step": 2345
    },
    {
      "epoch": 0.1,
      "grad_norm": 40.74591393337569,
      "learning_rate": 3.314995062773311e-06,
      "loss": 1.3364,
      "step": 2350
    },
    {
      "epoch": 0.1,
      "grad_norm": 23.502221090287733,
      "learning_rate": 3.3220482437579347e-06,
      "loss": 1.3135,
      "step": 2355
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.981683970958356,
      "learning_rate": 3.329101424742559e-06,
      "loss": 1.312,
      "step": 2360
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.800464338505064,
      "learning_rate": 3.336154605727183e-06,
      "loss": 1.2884,
      "step": 2365
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.931513536142512,
      "learning_rate": 3.343207786711807e-06,
      "loss": 1.2874,
      "step": 2370
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.909096968031482,
      "learning_rate": 3.3502609676964317e-06,
      "loss": 1.3575,
      "step": 2375
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.609320976043588,
      "learning_rate": 3.3573141486810557e-06,
      "loss": 1.3592,
      "step": 2380
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.072613831161481,
      "learning_rate": 3.3643673296656793e-06,
      "loss": 1.2885,
      "step": 2385
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.6865970695013,
      "learning_rate": 3.3714205106503033e-06,
      "loss": 1.3017,
      "step": 2390
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.351657609639986,
      "learning_rate": 3.3784736916349277e-06,
      "loss": 1.3181,
      "step": 2395
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.534039631385772,
      "learning_rate": 3.3855268726195518e-06,
      "loss": 1.296,
      "step": 2400
    },
    {
      "epoch": 0.1,
      "grad_norm": 12.80918629498463,
      "learning_rate": 3.3925800536041758e-06,
      "loss": 1.3042,
      "step": 2405
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.621829378155411,
      "learning_rate": 3.3996332345888002e-06,
      "loss": 1.2943,
      "step": 2410
    },
    {
      "epoch": 0.1,
      "grad_norm": 16.243774254738863,
      "learning_rate": 3.4066864155734242e-06,
      "loss": 1.3059,
      "step": 2415
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.9337711654213,
      "learning_rate": 3.413739596558048e-06,
      "loss": 1.3214,
      "step": 2420
    },
    {
      "epoch": 0.1,
      "grad_norm": 14.242545930087541,
      "learning_rate": 3.420792777542672e-06,
      "loss": 1.302,
      "step": 2425
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.95223396111189,
      "learning_rate": 3.427845958527296e-06,
      "loss": 1.2684,
      "step": 2430
    },
    {
      "epoch": 0.1,
      "grad_norm": 11.658892738145807,
      "learning_rate": 3.4348991395119203e-06,
      "loss": 1.333,
      "step": 2435
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.000832226226128,
      "learning_rate": 3.4419523204965444e-06,
      "loss": 1.2924,
      "step": 2440
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.810483633571383,
      "learning_rate": 3.4490055014811684e-06,
      "loss": 1.2569,
      "step": 2445
    },
    {
      "epoch": 0.1,
      "grad_norm": 13.1723050506315,
      "learning_rate": 3.456058682465792e-06,
      "loss": 1.2777,
      "step": 2450
    },
    {
      "epoch": 0.1,
      "grad_norm": 19.440211606303848,
      "learning_rate": 3.4631118634504164e-06,
      "loss": 1.3192,
      "step": 2455
    },
    {
      "epoch": 0.1,
      "grad_norm": 10.661384646268813,
      "learning_rate": 3.4701650444350404e-06,
      "loss": 1.2841,
      "step": 2460
    },
    {
      "epoch": 0.1,
      "grad_norm": 22.661837522337574,
      "learning_rate": 3.4772182254196645e-06,
      "loss": 1.2775,
      "step": 2465
    },
    {
      "epoch": 0.1,
      "grad_norm": 40.77966132645562,
      "learning_rate": 3.484271406404289e-06,
      "loss": 1.3205,
      "step": 2470
    },
    {
      "epoch": 0.1,
      "grad_norm": 32.84843369664207,
      "learning_rate": 3.491324587388913e-06,
      "loss": 1.3226,
      "step": 2475
    },
    {
      "epoch": 0.1,
      "grad_norm": 15.842673711941524,
      "learning_rate": 3.4983777683735365e-06,
      "loss": 1.321,
      "step": 2480
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.240344484235214,
      "learning_rate": 3.5054309493581605e-06,
      "loss": 1.2861,
      "step": 2485
    },
    {
      "epoch": 0.11,
      "grad_norm": 21.947691749994437,
      "learning_rate": 3.512484130342785e-06,
      "loss": 1.2759,
      "step": 2490
    },
    {
      "epoch": 0.11,
      "grad_norm": 33.778062904136135,
      "learning_rate": 3.519537311327409e-06,
      "loss": 1.2348,
      "step": 2495
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.287923457877422,
      "learning_rate": 3.526590492312033e-06,
      "loss": 1.2993,
      "step": 2500
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.17228249699254,
      "learning_rate": 3.533643673296657e-06,
      "loss": 1.2981,
      "step": 2505
    },
    {
      "epoch": 0.11,
      "grad_norm": 12.469848850375767,
      "learning_rate": 3.5406968542812815e-06,
      "loss": 1.3183,
      "step": 2510
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.146219051893173,
      "learning_rate": 3.547750035265905e-06,
      "loss": 1.2639,
      "step": 2515
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.930408084241922,
      "learning_rate": 3.554803216250529e-06,
      "loss": 1.2769,
      "step": 2520
    },
    {
      "epoch": 0.11,
      "grad_norm": 35.88907171960186,
      "learning_rate": 3.561856397235153e-06,
      "loss": 1.2934,
      "step": 2525
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.89872823061288,
      "learning_rate": 3.5689095782197776e-06,
      "loss": 1.3344,
      "step": 2530
    },
    {
      "epoch": 0.11,
      "grad_norm": 23.836740683850383,
      "learning_rate": 3.5759627592044016e-06,
      "loss": 1.3004,
      "step": 2535
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.70401344073436,
      "learning_rate": 3.5830159401890256e-06,
      "loss": 1.3062,
      "step": 2540
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.62922016464355,
      "learning_rate": 3.5900691211736492e-06,
      "loss": 1.2808,
      "step": 2545
    },
    {
      "epoch": 0.11,
      "grad_norm": 52.97337032778942,
      "learning_rate": 3.5971223021582737e-06,
      "loss": 1.2735,
      "step": 2550
    },
    {
      "epoch": 0.11,
      "grad_norm": 97.69711334654265,
      "learning_rate": 3.6041754831428977e-06,
      "loss": 1.3837,
      "step": 2555
    },
    {
      "epoch": 0.11,
      "grad_norm": 177.5056324394591,
      "learning_rate": 3.6112286641275217e-06,
      "loss": 1.4126,
      "step": 2560
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.872006882419626,
      "learning_rate": 3.618281845112146e-06,
      "loss": 1.3315,
      "step": 2565
    },
    {
      "epoch": 0.11,
      "grad_norm": 39.71345828288374,
      "learning_rate": 3.62533502609677e-06,
      "loss": 1.3122,
      "step": 2570
    },
    {
      "epoch": 0.11,
      "grad_norm": 17.058967582104138,
      "learning_rate": 3.632388207081394e-06,
      "loss": 1.3473,
      "step": 2575
    },
    {
      "epoch": 0.11,
      "grad_norm": 36.7107874365188,
      "learning_rate": 3.639441388066018e-06,
      "loss": 1.3599,
      "step": 2580
    },
    {
      "epoch": 0.11,
      "grad_norm": 29.22607155225058,
      "learning_rate": 3.6464945690506422e-06,
      "loss": 1.3297,
      "step": 2585
    },
    {
      "epoch": 0.11,
      "grad_norm": 20.53354404909482,
      "learning_rate": 3.6535477500352663e-06,
      "loss": 1.2778,
      "step": 2590
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.86006827561748,
      "learning_rate": 3.6606009310198903e-06,
      "loss": 1.3521,
      "step": 2595
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.948199895854632,
      "learning_rate": 3.6676541120045143e-06,
      "loss": 1.2731,
      "step": 2600
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.71692402370183,
      "learning_rate": 3.6747072929891387e-06,
      "loss": 1.298,
      "step": 2605
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.062536899430931,
      "learning_rate": 3.6817604739737623e-06,
      "loss": 1.3149,
      "step": 2610
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.102253390657898,
      "learning_rate": 3.6888136549583864e-06,
      "loss": 1.274,
      "step": 2615
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.642812977658556,
      "learning_rate": 3.6958668359430104e-06,
      "loss": 1.265,
      "step": 2620
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.625803262954769,
      "learning_rate": 3.702920016927635e-06,
      "loss": 1.3035,
      "step": 2625
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.030825998116676,
      "learning_rate": 3.709973197912259e-06,
      "loss": 1.2392,
      "step": 2630
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.259740106001445,
      "learning_rate": 3.717026378896883e-06,
      "loss": 1.3278,
      "step": 2635
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.769848126676907,
      "learning_rate": 3.7240795598815065e-06,
      "loss": 1.3199,
      "step": 2640
    },
    {
      "epoch": 0.11,
      "grad_norm": 14.339422438110708,
      "learning_rate": 3.731132740866131e-06,
      "loss": 1.3215,
      "step": 2645
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.364269568645252,
      "learning_rate": 3.738185921850755e-06,
      "loss": 1.2795,
      "step": 2650
    },
    {
      "epoch": 0.11,
      "grad_norm": 11.673151505722714,
      "learning_rate": 3.745239102835379e-06,
      "loss": 1.2879,
      "step": 2655
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.561347279035587,
      "learning_rate": 3.7522922838200034e-06,
      "loss": 1.2966,
      "step": 2660
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.317004878154123,
      "learning_rate": 3.7593454648046274e-06,
      "loss": 1.2809,
      "step": 2665
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.269771080603558,
      "learning_rate": 3.7663986457892514e-06,
      "loss": 1.2365,
      "step": 2670
    },
    {
      "epoch": 0.11,
      "grad_norm": 18.41904430908563,
      "learning_rate": 3.773451826773875e-06,
      "loss": 1.2937,
      "step": 2675
    },
    {
      "epoch": 0.11,
      "grad_norm": 26.903715456061647,
      "learning_rate": 3.780505007758499e-06,
      "loss": 1.3293,
      "step": 2680
    },
    {
      "epoch": 0.11,
      "grad_norm": 34.5838086355772,
      "learning_rate": 3.7875581887431235e-06,
      "loss": 1.273,
      "step": 2685
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.564786127157202,
      "learning_rate": 3.7946113697277475e-06,
      "loss": 1.3264,
      "step": 2690
    },
    {
      "epoch": 0.11,
      "grad_norm": 15.016589345921918,
      "learning_rate": 3.8016645507123715e-06,
      "loss": 1.3289,
      "step": 2695
    },
    {
      "epoch": 0.11,
      "grad_norm": 16.225391620897856,
      "learning_rate": 3.808717731696996e-06,
      "loss": 1.3281,
      "step": 2700
    },
    {
      "epoch": 0.11,
      "grad_norm": 9.170812110441807,
      "learning_rate": 3.81577091268162e-06,
      "loss": 1.2689,
      "step": 2705
    },
    {
      "epoch": 0.11,
      "grad_norm": 13.85476329953772,
      "learning_rate": 3.822824093666244e-06,
      "loss": 1.2498,
      "step": 2710
    },
    {
      "epoch": 0.11,
      "grad_norm": 10.980331687023341,
      "learning_rate": 3.829877274650868e-06,
      "loss": 1.3877,
      "step": 2715
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.9311116074833,
      "learning_rate": 3.836930455635492e-06,
      "loss": 1.2732,
      "step": 2720
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.65648536273734,
      "learning_rate": 3.843983636620116e-06,
      "loss": 1.327,
      "step": 2725
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.658055479106222,
      "learning_rate": 3.85103681760474e-06,
      "loss": 1.2391,
      "step": 2730
    },
    {
      "epoch": 0.12,
      "grad_norm": 15.267489207192607,
      "learning_rate": 3.8580899985893646e-06,
      "loss": 1.2683,
      "step": 2735
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.675434205406328,
      "learning_rate": 3.865143179573988e-06,
      "loss": 1.3038,
      "step": 2740
    },
    {
      "epoch": 0.12,
      "grad_norm": 43.07447486043644,
      "learning_rate": 3.872196360558612e-06,
      "loss": 1.2872,
      "step": 2745
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.09540935857782,
      "learning_rate": 3.879249541543236e-06,
      "loss": 1.284,
      "step": 2750
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.329659232647545,
      "learning_rate": 3.886302722527861e-06,
      "loss": 1.3234,
      "step": 2755
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.4422946502081,
      "learning_rate": 3.893355903512484e-06,
      "loss": 1.2493,
      "step": 2760
    },
    {
      "epoch": 0.12,
      "grad_norm": 18.457155038757037,
      "learning_rate": 3.900409084497109e-06,
      "loss": 1.3149,
      "step": 2765
    },
    {
      "epoch": 0.12,
      "grad_norm": 19.066991949591987,
      "learning_rate": 3.907462265481732e-06,
      "loss": 1.3173,
      "step": 2770
    },
    {
      "epoch": 0.12,
      "grad_norm": 59.33881414335634,
      "learning_rate": 3.914515446466357e-06,
      "loss": 1.2631,
      "step": 2775
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.26059385079084,
      "learning_rate": 3.92156862745098e-06,
      "loss": 1.2353,
      "step": 2780
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.063165722995702,
      "learning_rate": 3.928621808435605e-06,
      "loss": 1.2735,
      "step": 2785
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.797982570435053,
      "learning_rate": 3.935674989420229e-06,
      "loss": 1.3306,
      "step": 2790
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.183148780642624,
      "learning_rate": 3.942728170404853e-06,
      "loss": 1.3298,
      "step": 2795
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.016617200365182,
      "learning_rate": 3.949781351389476e-06,
      "loss": 1.2983,
      "step": 2800
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.836752394258982,
      "learning_rate": 3.956834532374101e-06,
      "loss": 1.3208,
      "step": 2805
    },
    {
      "epoch": 0.12,
      "grad_norm": 13.785214669426912,
      "learning_rate": 3.963887713358725e-06,
      "loss": 1.288,
      "step": 2810
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.25549862823478,
      "learning_rate": 3.970940894343349e-06,
      "loss": 1.2841,
      "step": 2815
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.864212812801505,
      "learning_rate": 3.977994075327973e-06,
      "loss": 1.2843,
      "step": 2820
    },
    {
      "epoch": 0.12,
      "grad_norm": 17.604115953866582,
      "learning_rate": 3.985047256312598e-06,
      "loss": 1.2703,
      "step": 2825
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.906829408716234,
      "learning_rate": 3.992100437297221e-06,
      "loss": 1.344,
      "step": 2830
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.549679676413833,
      "learning_rate": 3.999153618281845e-06,
      "loss": 1.2821,
      "step": 2835
    },
    {
      "epoch": 0.12,
      "grad_norm": 9.740129414659062,
      "learning_rate": 4.0062067992664694e-06,
      "loss": 1.2387,
      "step": 2840
    },
    {
      "epoch": 0.12,
      "grad_norm": 29.495403707854127,
      "learning_rate": 4.013259980251094e-06,
      "loss": 1.2691,
      "step": 2845
    },
    {
      "epoch": 0.12,
      "grad_norm": 20.795329749765493,
      "learning_rate": 4.0203131612357175e-06,
      "loss": 1.2712,
      "step": 2850
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.695114485986377,
      "learning_rate": 4.027366342220342e-06,
      "loss": 1.3347,
      "step": 2855
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.058774070795085,
      "learning_rate": 4.034419523204966e-06,
      "loss": 1.3013,
      "step": 2860
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.711121636674607,
      "learning_rate": 4.04147270418959e-06,
      "loss": 1.2703,
      "step": 2865
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.891345574399109,
      "learning_rate": 4.0485258851742136e-06,
      "loss": 1.2876,
      "step": 2870
    },
    {
      "epoch": 0.12,
      "grad_norm": 11.003490723314501,
      "learning_rate": 4.055579066158838e-06,
      "loss": 1.2898,
      "step": 2875
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.492300588020184,
      "learning_rate": 4.0626322471434624e-06,
      "loss": 1.2768,
      "step": 2880
    },
    {
      "epoch": 0.12,
      "grad_norm": 21.064936218751935,
      "learning_rate": 4.069685428128086e-06,
      "loss": 1.2826,
      "step": 2885
    },
    {
      "epoch": 0.12,
      "grad_norm": 30.262012568820364,
      "learning_rate": 4.0767386091127105e-06,
      "loss": 1.2122,
      "step": 2890
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.1170416281707,
      "learning_rate": 4.083791790097334e-06,
      "loss": 1.3159,
      "step": 2895
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.470922649541901,
      "learning_rate": 4.090844971081958e-06,
      "loss": 1.2255,
      "step": 2900
    },
    {
      "epoch": 0.12,
      "grad_norm": 92.94573686231264,
      "learning_rate": 4.097898152066582e-06,
      "loss": 1.3357,
      "step": 2905
    },
    {
      "epoch": 0.12,
      "grad_norm": 57.55677343074897,
      "learning_rate": 4.1049513330512066e-06,
      "loss": 1.3017,
      "step": 2910
    },
    {
      "epoch": 0.12,
      "grad_norm": 35.54071332346504,
      "learning_rate": 4.11200451403583e-06,
      "loss": 1.3637,
      "step": 2915
    },
    {
      "epoch": 0.12,
      "grad_norm": 22.551752557418194,
      "learning_rate": 4.119057695020455e-06,
      "loss": 1.321,
      "step": 2920
    },
    {
      "epoch": 0.12,
      "grad_norm": 27.835512236537443,
      "learning_rate": 4.126110876005079e-06,
      "loss": 1.2858,
      "step": 2925
    },
    {
      "epoch": 0.12,
      "grad_norm": 67.08799625233122,
      "learning_rate": 4.133164056989703e-06,
      "loss": 1.3391,
      "step": 2930
    },
    {
      "epoch": 0.12,
      "grad_norm": 80.82188934335447,
      "learning_rate": 4.140217237974326e-06,
      "loss": 1.2762,
      "step": 2935
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.595367431257326,
      "learning_rate": 4.147270418958951e-06,
      "loss": 1.3076,
      "step": 2940
    },
    {
      "epoch": 0.12,
      "grad_norm": 17.486191032838914,
      "learning_rate": 4.154323599943575e-06,
      "loss": 1.3045,
      "step": 2945
    },
    {
      "epoch": 0.12,
      "grad_norm": 47.41869802391103,
      "learning_rate": 4.161376780928199e-06,
      "loss": 1.2969,
      "step": 2950
    },
    {
      "epoch": 0.13,
      "grad_norm": 43.55408239434461,
      "learning_rate": 4.168429961912823e-06,
      "loss": 1.3257,
      "step": 2955
    },
    {
      "epoch": 0.13,
      "grad_norm": 55.65334416941587,
      "learning_rate": 4.175483142897447e-06,
      "loss": 1.3091,
      "step": 2960
    },
    {
      "epoch": 0.13,
      "grad_norm": 42.206429191583105,
      "learning_rate": 4.182536323882071e-06,
      "loss": 1.3263,
      "step": 2965
    },
    {
      "epoch": 0.13,
      "grad_norm": 29.374198431314877,
      "learning_rate": 4.189589504866695e-06,
      "loss": 1.2673,
      "step": 2970
    },
    {
      "epoch": 0.13,
      "grad_norm": 39.630515590905446,
      "learning_rate": 4.196642685851319e-06,
      "loss": 1.2646,
      "step": 2975
    },
    {
      "epoch": 0.13,
      "grad_norm": 31.923398379906896,
      "learning_rate": 4.203695866835944e-06,
      "loss": 1.2679,
      "step": 2980
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.24286810361188,
      "learning_rate": 4.210749047820567e-06,
      "loss": 1.2238,
      "step": 2985
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.078710885293793,
      "learning_rate": 4.217802228805192e-06,
      "loss": 1.282,
      "step": 2990
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.636571569090822,
      "learning_rate": 4.224855409789815e-06,
      "loss": 1.2593,
      "step": 2995
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.233324328700752,
      "learning_rate": 4.23190859077444e-06,
      "loss": 1.2586,
      "step": 3000
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.301653033520429,
      "learning_rate": 4.238961771759063e-06,
      "loss": 1.2428,
      "step": 3005
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.465812370291173,
      "learning_rate": 4.246014952743688e-06,
      "loss": 1.2813,
      "step": 3010
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.080936940122832,
      "learning_rate": 4.253068133728312e-06,
      "loss": 1.2739,
      "step": 3015
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.422997949175425,
      "learning_rate": 4.260121314712936e-06,
      "loss": 1.2907,
      "step": 3020
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.159826171454963,
      "learning_rate": 4.2671744956975595e-06,
      "loss": 1.3148,
      "step": 3025
    },
    {
      "epoch": 0.13,
      "grad_norm": 14.376910478041388,
      "learning_rate": 4.274227676682184e-06,
      "loss": 1.3304,
      "step": 3030
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.327628521308107,
      "learning_rate": 4.281280857666808e-06,
      "loss": 1.2522,
      "step": 3035
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.969648129784302,
      "learning_rate": 4.288334038651432e-06,
      "loss": 1.3042,
      "step": 3040
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.076411629526614,
      "learning_rate": 4.295387219636056e-06,
      "loss": 1.3039,
      "step": 3045
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.21469773823197,
      "learning_rate": 4.302440400620681e-06,
      "loss": 1.2547,
      "step": 3050
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.881882420196233,
      "learning_rate": 4.3094935816053045e-06,
      "loss": 1.2522,
      "step": 3055
    },
    {
      "epoch": 0.13,
      "grad_norm": 17.573570285439917,
      "learning_rate": 4.316546762589928e-06,
      "loss": 1.2581,
      "step": 3060
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.65540407408589,
      "learning_rate": 4.3235999435745525e-06,
      "loss": 1.3096,
      "step": 3065
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.143615295859695,
      "learning_rate": 4.330653124559177e-06,
      "loss": 1.2808,
      "step": 3070
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.781906250128856,
      "learning_rate": 4.3377063055438005e-06,
      "loss": 1.1971,
      "step": 3075
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.198103716610406,
      "learning_rate": 4.344759486528425e-06,
      "loss": 1.196,
      "step": 3080
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.132663311125032,
      "learning_rate": 4.351812667513049e-06,
      "loss": 1.2969,
      "step": 3085
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.392770585766172,
      "learning_rate": 4.358865848497672e-06,
      "loss": 1.2861,
      "step": 3090
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.90090090073315,
      "learning_rate": 4.365919029482297e-06,
      "loss": 1.2451,
      "step": 3095
    },
    {
      "epoch": 0.13,
      "grad_norm": 23.32261999386539,
      "learning_rate": 4.372972210466921e-06,
      "loss": 1.2464,
      "step": 3100
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.003312527261658,
      "learning_rate": 4.380025391451545e-06,
      "loss": 1.3055,
      "step": 3105
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.888227524955175,
      "learning_rate": 4.387078572436169e-06,
      "loss": 1.3067,
      "step": 3110
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.514075282076226,
      "learning_rate": 4.3941317534207936e-06,
      "loss": 1.2589,
      "step": 3115
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.639588585325548,
      "learning_rate": 4.401184934405417e-06,
      "loss": 1.2855,
      "step": 3120
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.186294494085404,
      "learning_rate": 4.408238115390041e-06,
      "loss": 1.2921,
      "step": 3125
    },
    {
      "epoch": 0.13,
      "grad_norm": 24.52326846217789,
      "learning_rate": 4.415291296374665e-06,
      "loss": 1.2579,
      "step": 3130
    },
    {
      "epoch": 0.13,
      "grad_norm": 19.329861694298764,
      "learning_rate": 4.42234447735929e-06,
      "loss": 1.2616,
      "step": 3135
    },
    {
      "epoch": 0.13,
      "grad_norm": 15.993999159722897,
      "learning_rate": 4.429397658343913e-06,
      "loss": 1.3041,
      "step": 3140
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.158261046071381,
      "learning_rate": 4.436450839328538e-06,
      "loss": 1.2986,
      "step": 3145
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.027813476577078,
      "learning_rate": 4.443504020313162e-06,
      "loss": 1.3188,
      "step": 3150
    },
    {
      "epoch": 0.13,
      "grad_norm": 13.649258720805259,
      "learning_rate": 4.450557201297786e-06,
      "loss": 1.2454,
      "step": 3155
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.438215584608244,
      "learning_rate": 4.457610382282409e-06,
      "loss": 1.3254,
      "step": 3160
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.7496660816618,
      "learning_rate": 4.464663563267034e-06,
      "loss": 1.2546,
      "step": 3165
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.496524193277665,
      "learning_rate": 4.471716744251658e-06,
      "loss": 1.2651,
      "step": 3170
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.72691437805028,
      "learning_rate": 4.478769925236282e-06,
      "loss": 1.2721,
      "step": 3175
    },
    {
      "epoch": 0.13,
      "grad_norm": 12.14529539156146,
      "learning_rate": 4.485823106220906e-06,
      "loss": 1.2601,
      "step": 3180
    },
    {
      "epoch": 0.13,
      "grad_norm": 16.457732292094185,
      "learning_rate": 4.49287628720553e-06,
      "loss": 1.3189,
      "step": 3185
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.650494838167653,
      "learning_rate": 4.499929468190154e-06,
      "loss": 1.2754,
      "step": 3190
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.323631987087946,
      "learning_rate": 4.506982649174778e-06,
      "loss": 1.2467,
      "step": 3195
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.011830811647577,
      "learning_rate": 4.514035830159402e-06,
      "loss": 1.2685,
      "step": 3200
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.424442499264432,
      "learning_rate": 4.521089011144027e-06,
      "loss": 1.2704,
      "step": 3205
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.371961437359987,
      "learning_rate": 4.52814219212865e-06,
      "loss": 1.3226,
      "step": 3210
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.835651916357772,
      "learning_rate": 4.535195373113275e-06,
      "loss": 1.3091,
      "step": 3215
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.41319274073559,
      "learning_rate": 4.5422485540978984e-06,
      "loss": 1.2568,
      "step": 3220
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.246287636513452,
      "learning_rate": 4.549301735082523e-06,
      "loss": 1.2485,
      "step": 3225
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.570100016693516,
      "learning_rate": 4.5563549160671465e-06,
      "loss": 1.2255,
      "step": 3230
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.59834089852237,
      "learning_rate": 4.563408097051771e-06,
      "loss": 1.2537,
      "step": 3235
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.954268303022719,
      "learning_rate": 4.570461278036395e-06,
      "loss": 1.2545,
      "step": 3240
    },
    {
      "epoch": 0.14,
      "grad_norm": 14.10277506496896,
      "learning_rate": 4.577514459021019e-06,
      "loss": 1.2341,
      "step": 3245
    },
    {
      "epoch": 0.14,
      "grad_norm": 43.27968295823003,
      "learning_rate": 4.5845676400056426e-06,
      "loss": 1.2698,
      "step": 3250
    },
    {
      "epoch": 0.14,
      "grad_norm": 29.975467048193586,
      "learning_rate": 4.591620820990267e-06,
      "loss": 1.2689,
      "step": 3255
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.975237616706838,
      "learning_rate": 4.598674001974891e-06,
      "loss": 1.2746,
      "step": 3260
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.665770893676218,
      "learning_rate": 4.605727182959515e-06,
      "loss": 1.2147,
      "step": 3265
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.42988049029319,
      "learning_rate": 4.6127803639441395e-06,
      "loss": 1.2523,
      "step": 3270
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.018191519156783,
      "learning_rate": 4.619833544928763e-06,
      "loss": 1.2749,
      "step": 3275
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.441603665958906,
      "learning_rate": 4.626886725913387e-06,
      "loss": 1.2419,
      "step": 3280
    },
    {
      "epoch": 0.14,
      "grad_norm": 51.78100777693077,
      "learning_rate": 4.633939906898011e-06,
      "loss": 1.2423,
      "step": 3285
    },
    {
      "epoch": 0.14,
      "grad_norm": 45.86870898692082,
      "learning_rate": 4.6409930878826356e-06,
      "loss": 1.2786,
      "step": 3290
    },
    {
      "epoch": 0.14,
      "grad_norm": 17.147759385681542,
      "learning_rate": 4.648046268867259e-06,
      "loss": 1.2626,
      "step": 3295
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.82689736554392,
      "learning_rate": 4.655099449851884e-06,
      "loss": 1.2384,
      "step": 3300
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.455156457105254,
      "learning_rate": 4.662152630836508e-06,
      "loss": 1.2676,
      "step": 3305
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.82375637626874,
      "learning_rate": 4.669205811821132e-06,
      "loss": 1.2364,
      "step": 3310
    },
    {
      "epoch": 0.14,
      "grad_norm": 21.085889861009786,
      "learning_rate": 4.676258992805755e-06,
      "loss": 1.2543,
      "step": 3315
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.272611268477121,
      "learning_rate": 4.68331217379038e-06,
      "loss": 1.2722,
      "step": 3320
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.911231828036165,
      "learning_rate": 4.690365354775004e-06,
      "loss": 1.2736,
      "step": 3325
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.693666764497696,
      "learning_rate": 4.697418535759628e-06,
      "loss": 1.1832,
      "step": 3330
    },
    {
      "epoch": 0.14,
      "grad_norm": 23.968076222558604,
      "learning_rate": 4.704471716744252e-06,
      "loss": 1.2391,
      "step": 3335
    },
    {
      "epoch": 0.14,
      "grad_norm": 11.985398361901478,
      "learning_rate": 4.711524897728877e-06,
      "loss": 1.3609,
      "step": 3340
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.990054681752294,
      "learning_rate": 4.7185780787135e-06,
      "loss": 1.2265,
      "step": 3345
    },
    {
      "epoch": 0.14,
      "grad_norm": 15.1148727401013,
      "learning_rate": 4.725631259698124e-06,
      "loss": 1.2943,
      "step": 3350
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.22872713748695,
      "learning_rate": 4.732684440682748e-06,
      "loss": 1.2198,
      "step": 3355
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.445795893427473,
      "learning_rate": 4.739737621667373e-06,
      "loss": 1.2157,
      "step": 3360
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.162977383487279,
      "learning_rate": 4.746790802651996e-06,
      "loss": 1.2655,
      "step": 3365
    },
    {
      "epoch": 0.14,
      "grad_norm": 24.696746213762975,
      "learning_rate": 4.753843983636621e-06,
      "loss": 1.2587,
      "step": 3370
    },
    {
      "epoch": 0.14,
      "grad_norm": 28.10235490138781,
      "learning_rate": 4.760897164621244e-06,
      "loss": 1.2917,
      "step": 3375
    },
    {
      "epoch": 0.14,
      "grad_norm": 20.756286209291872,
      "learning_rate": 4.767950345605869e-06,
      "loss": 1.2959,
      "step": 3380
    },
    {
      "epoch": 0.14,
      "grad_norm": 16.20353117004509,
      "learning_rate": 4.775003526590492e-06,
      "loss": 1.2195,
      "step": 3385
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.189508691118403,
      "learning_rate": 4.782056707575117e-06,
      "loss": 1.2833,
      "step": 3390
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.428530737443314,
      "learning_rate": 4.789109888559741e-06,
      "loss": 1.2794,
      "step": 3395
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.66301189906629,
      "learning_rate": 4.796163069544365e-06,
      "loss": 1.2871,
      "step": 3400
    },
    {
      "epoch": 0.14,
      "grad_norm": 18.450614951492735,
      "learning_rate": 4.803216250528989e-06,
      "loss": 1.2987,
      "step": 3405
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.246211335901181,
      "learning_rate": 4.810269431513613e-06,
      "loss": 1.223,
      "step": 3410
    },
    {
      "epoch": 0.14,
      "grad_norm": 13.113485988685436,
      "learning_rate": 4.817322612498237e-06,
      "loss": 1.2844,
      "step": 3415
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.292734527639393,
      "learning_rate": 4.824375793482861e-06,
      "loss": 1.2387,
      "step": 3420
    },
    {
      "epoch": 0.14,
      "grad_norm": 47.78148910178047,
      "learning_rate": 4.831428974467485e-06,
      "loss": 1.3424,
      "step": 3425
    },
    {
      "epoch": 0.15,
      "grad_norm": 13.890277669697383,
      "learning_rate": 4.838482155452109e-06,
      "loss": 1.2322,
      "step": 3430
    },
    {
      "epoch": 0.15,
      "grad_norm": 17.933300497127828,
      "learning_rate": 4.8455353364367335e-06,
      "loss": 1.2728,
      "step": 3435
    },
    {
      "epoch": 0.15,
      "grad_norm": 22.84721691773643,
      "learning_rate": 4.852588517421357e-06,
      "loss": 1.2291,
      "step": 3440
    },
    {
      "epoch": 0.15,
      "grad_norm": 30.554845364799714,
      "learning_rate": 4.8596416984059815e-06,
      "loss": 1.2025,
      "step": 3445
    },
    {
      "epoch": 0.15,
      "grad_norm": 28.508765751777105,
      "learning_rate": 4.866694879390605e-06,
      "loss": 1.238,
      "step": 3450
    },
    {
      "epoch": 0.15,
      "grad_norm": 49.73294767090246,
      "learning_rate": 4.8737480603752295e-06,
      "loss": 1.2829,
      "step": 3455
    },
    {
      "epoch": 0.15,
      "grad_norm": 39.587702658575665,
      "learning_rate": 4.880801241359854e-06,
      "loss": 1.2542,
      "step": 3460
    },
    {
      "epoch": 0.15,
      "grad_norm": 39.1003622377763,
      "learning_rate": 4.887854422344478e-06,
      "loss": 1.249,
      "step": 3465
    },
    {
      "epoch": 0.15,
      "grad_norm": 27.845414136751447,
      "learning_rate": 4.894907603329102e-06,
      "loss": 1.2086,
      "step": 3470
    },
    {
      "epoch": 0.15,
      "grad_norm": 30.498720085476286,
      "learning_rate": 4.901960784313726e-06,
      "loss": 1.2771,
      "step": 3475
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.786855760143926,
      "learning_rate": 4.90901396529835e-06,
      "loss": 1.2083,
      "step": 3480
    },
    {
      "epoch": 0.15,
      "grad_norm": 36.24396094525753,
      "learning_rate": 4.916067146282974e-06,
      "loss": 1.2578,
      "step": 3485
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.951722024434854,
      "learning_rate": 4.923120327267598e-06,
      "loss": 1.2938,
      "step": 3490
    },
    {
      "epoch": 0.15,
      "grad_norm": 28.62541316775776,
      "learning_rate": 4.9301735082522226e-06,
      "loss": 1.2567,
      "step": 3495
    },
    {
      "epoch": 0.15,
      "grad_norm": 11.409404146967361,
      "learning_rate": 4.937226689236846e-06,
      "loss": 1.2897,
      "step": 3500
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.438793537906815,
      "learning_rate": 4.94427987022147e-06,
      "loss": 1.2747,
      "step": 3505
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.473561091047692,
      "learning_rate": 4.951333051206094e-06,
      "loss": 1.2367,
      "step": 3510
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.08078614053848,
      "learning_rate": 4.958386232190719e-06,
      "loss": 1.2957,
      "step": 3515
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.04613897016436,
      "learning_rate": 4.965439413175342e-06,
      "loss": 1.2551,
      "step": 3520
    },
    {
      "epoch": 0.15,
      "grad_norm": 25.4806371035209,
      "learning_rate": 4.972492594159967e-06,
      "loss": 1.278,
      "step": 3525
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.982113715913048,
      "learning_rate": 4.979545775144591e-06,
      "loss": 1.251,
      "step": 3530
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.33016544355097,
      "learning_rate": 4.986598956129215e-06,
      "loss": 1.3304,
      "step": 3535
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.201567875703628,
      "learning_rate": 4.993652137113838e-06,
      "loss": 1.2802,
      "step": 3540
    },
    {
      "epoch": 0.15,
      "grad_norm": 28.618573898180685,
      "learning_rate": 5.000705318098464e-06,
      "loss": 1.2785,
      "step": 3545
    },
    {
      "epoch": 0.15,
      "grad_norm": 16.972373426929096,
      "learning_rate": 5.007758499083087e-06,
      "loss": 1.26,
      "step": 3550
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.659160389774978,
      "learning_rate": 5.014811680067711e-06,
      "loss": 1.2654,
      "step": 3555
    },
    {
      "epoch": 0.15,
      "grad_norm": 20.954522437050688,
      "learning_rate": 5.021864861052334e-06,
      "loss": 1.2917,
      "step": 3560
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.99441827169673,
      "learning_rate": 5.028918042036959e-06,
      "loss": 1.2494,
      "step": 3565
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.753998533601283,
      "learning_rate": 5.035971223021583e-06,
      "loss": 1.2595,
      "step": 3570
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.763756611272813,
      "learning_rate": 5.043024404006207e-06,
      "loss": 1.2143,
      "step": 3575
    },
    {
      "epoch": 0.15,
      "grad_norm": 27.84702459003064,
      "learning_rate": 5.050077584990831e-06,
      "loss": 1.2716,
      "step": 3580
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.125008774044057,
      "learning_rate": 5.057130765975456e-06,
      "loss": 1.3131,
      "step": 3585
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.479878126083905,
      "learning_rate": 5.064183946960079e-06,
      "loss": 1.2692,
      "step": 3590
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.550931355429048,
      "learning_rate": 5.071237127944704e-06,
      "loss": 1.169,
      "step": 3595
    },
    {
      "epoch": 0.15,
      "grad_norm": 28.478250059292655,
      "learning_rate": 5.0782903089293274e-06,
      "loss": 1.2417,
      "step": 3600
    },
    {
      "epoch": 0.15,
      "grad_norm": 33.8453585371968,
      "learning_rate": 5.085343489913952e-06,
      "loss": 1.2972,
      "step": 3605
    },
    {
      "epoch": 0.15,
      "grad_norm": 29.07758781376888,
      "learning_rate": 5.092396670898576e-06,
      "loss": 1.2244,
      "step": 3610
    },
    {
      "epoch": 0.15,
      "grad_norm": 26.208980724520302,
      "learning_rate": 5.099449851883199e-06,
      "loss": 1.2853,
      "step": 3615
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.281973497746348,
      "learning_rate": 5.1065030328678235e-06,
      "loss": 1.2211,
      "step": 3620
    },
    {
      "epoch": 0.15,
      "grad_norm": 10.128605749862677,
      "learning_rate": 5.113556213852447e-06,
      "loss": 1.2624,
      "step": 3625
    },
    {
      "epoch": 0.15,
      "grad_norm": 14.276090059754628,
      "learning_rate": 5.1206093948370716e-06,
      "loss": 1.3062,
      "step": 3630
    },
    {
      "epoch": 0.15,
      "grad_norm": 27.184153460761387,
      "learning_rate": 5.127662575821696e-06,
      "loss": 1.2807,
      "step": 3635
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.81039743907487,
      "learning_rate": 5.13471575680632e-06,
      "loss": 1.2522,
      "step": 3640
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.300584431242333,
      "learning_rate": 5.141768937790944e-06,
      "loss": 1.292,
      "step": 3645
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.248704558390312,
      "learning_rate": 5.1488221187755685e-06,
      "loss": 1.2602,
      "step": 3650
    },
    {
      "epoch": 0.15,
      "grad_norm": 19.57079862561798,
      "learning_rate": 5.155875299760192e-06,
      "loss": 1.2518,
      "step": 3655
    },
    {
      "epoch": 0.15,
      "grad_norm": 23.026987515970035,
      "learning_rate": 5.1629284807448165e-06,
      "loss": 1.2566,
      "step": 3660
    },
    {
      "epoch": 0.16,
      "grad_norm": 53.242706156792124,
      "learning_rate": 5.169981661729441e-06,
      "loss": 1.2915,
      "step": 3665
    },
    {
      "epoch": 0.16,
      "grad_norm": 22.11577869661808,
      "learning_rate": 5.1770348427140646e-06,
      "loss": 1.2245,
      "step": 3670
    },
    {
      "epoch": 0.16,
      "grad_norm": 33.852659895231234,
      "learning_rate": 5.184088023698689e-06,
      "loss": 1.3023,
      "step": 3675
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.268584865926686,
      "learning_rate": 5.191141204683312e-06,
      "loss": 1.2568,
      "step": 3680
    },
    {
      "epoch": 0.16,
      "grad_norm": 24.821961553526766,
      "learning_rate": 5.198194385667936e-06,
      "loss": 1.2488,
      "step": 3685
    },
    {
      "epoch": 0.16,
      "grad_norm": 24.550111315927886,
      "learning_rate": 5.205247566652561e-06,
      "loss": 1.2473,
      "step": 3690
    },
    {
      "epoch": 0.16,
      "grad_norm": 36.929398031390676,
      "learning_rate": 5.212300747637184e-06,
      "loss": 1.3003,
      "step": 3695
    },
    {
      "epoch": 0.16,
      "grad_norm": 18.268144552737915,
      "learning_rate": 5.219353928621809e-06,
      "loss": 1.2627,
      "step": 3700
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.96899890985399,
      "learning_rate": 5.226407109606433e-06,
      "loss": 1.245,
      "step": 3705
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.226320022504392,
      "learning_rate": 5.233460290591057e-06,
      "loss": 1.2526,
      "step": 3710
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.51427221258509,
      "learning_rate": 5.240513471575681e-06,
      "loss": 1.2423,
      "step": 3715
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.481946559437045,
      "learning_rate": 5.247566652560306e-06,
      "loss": 1.2731,
      "step": 3720
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.848281322736836,
      "learning_rate": 5.254619833544929e-06,
      "loss": 1.2452,
      "step": 3725
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.668540879766415,
      "learning_rate": 5.261673014529554e-06,
      "loss": 1.2329,
      "step": 3730
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.922959391456253,
      "learning_rate": 5.268726195514178e-06,
      "loss": 1.2924,
      "step": 3735
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.70242172595704,
      "learning_rate": 5.275779376498802e-06,
      "loss": 1.2388,
      "step": 3740
    },
    {
      "epoch": 0.16,
      "grad_norm": 41.14582247683044,
      "learning_rate": 5.282832557483425e-06,
      "loss": 1.2864,
      "step": 3745
    },
    {
      "epoch": 0.16,
      "grad_norm": 18.062823082268306,
      "learning_rate": 5.289885738468049e-06,
      "loss": 1.2721,
      "step": 3750
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.554832773331313,
      "learning_rate": 5.296938919452673e-06,
      "loss": 1.2511,
      "step": 3755
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.359285159621118,
      "learning_rate": 5.303992100437298e-06,
      "loss": 1.2218,
      "step": 3760
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.732104143396588,
      "learning_rate": 5.311045281421921e-06,
      "loss": 1.2697,
      "step": 3765
    },
    {
      "epoch": 0.16,
      "grad_norm": 37.23806119058716,
      "learning_rate": 5.318098462406546e-06,
      "loss": 1.2736,
      "step": 3770
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.693945486538784,
      "learning_rate": 5.3251516433911694e-06,
      "loss": 1.2675,
      "step": 3775
    },
    {
      "epoch": 0.16,
      "grad_norm": 34.828089864169726,
      "learning_rate": 5.332204824375794e-06,
      "loss": 1.2769,
      "step": 3780
    },
    {
      "epoch": 0.16,
      "grad_norm": 11.89472386477093,
      "learning_rate": 5.339258005360418e-06,
      "loss": 1.2025,
      "step": 3785
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.172766677676536,
      "learning_rate": 5.346311186345042e-06,
      "loss": 1.2306,
      "step": 3790
    },
    {
      "epoch": 0.16,
      "grad_norm": 13.04400025314236,
      "learning_rate": 5.353364367329666e-06,
      "loss": 1.2336,
      "step": 3795
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.921863451743317,
      "learning_rate": 5.360417548314291e-06,
      "loss": 1.193,
      "step": 3800
    },
    {
      "epoch": 0.16,
      "grad_norm": 14.260150370917506,
      "learning_rate": 5.367470729298914e-06,
      "loss": 1.2395,
      "step": 3805
    },
    {
      "epoch": 0.16,
      "grad_norm": 9.316570987730108,
      "learning_rate": 5.374523910283538e-06,
      "loss": 1.238,
      "step": 3810
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.98288944969821,
      "learning_rate": 5.381577091268162e-06,
      "loss": 1.2943,
      "step": 3815
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.302114164743593,
      "learning_rate": 5.388630272252786e-06,
      "loss": 1.2426,
      "step": 3820
    },
    {
      "epoch": 0.16,
      "grad_norm": 17.591483688383303,
      "learning_rate": 5.3956834532374105e-06,
      "loss": 1.2775,
      "step": 3825
    },
    {
      "epoch": 0.16,
      "grad_norm": 16.87300482551865,
      "learning_rate": 5.402736634222034e-06,
      "loss": 1.2676,
      "step": 3830
    },
    {
      "epoch": 0.16,
      "grad_norm": 24.844493245385017,
      "learning_rate": 5.4097898152066585e-06,
      "loss": 1.2578,
      "step": 3835
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.493827131088455,
      "learning_rate": 5.416842996191283e-06,
      "loss": 1.2441,
      "step": 3840
    },
    {
      "epoch": 0.16,
      "grad_norm": 18.862839267695904,
      "learning_rate": 5.423896177175907e-06,
      "loss": 1.2904,
      "step": 3845
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.51445164871292,
      "learning_rate": 5.430949358160531e-06,
      "loss": 1.2498,
      "step": 3850
    },
    {
      "epoch": 0.16,
      "grad_norm": 15.788641397206522,
      "learning_rate": 5.4380025391451555e-06,
      "loss": 1.2382,
      "step": 3855
    },
    {
      "epoch": 0.16,
      "grad_norm": 42.66853067803099,
      "learning_rate": 5.445055720129779e-06,
      "loss": 1.2885,
      "step": 3860
    },
    {
      "epoch": 0.16,
      "grad_norm": 39.38940734172004,
      "learning_rate": 5.4521089011144035e-06,
      "loss": 1.2558,
      "step": 3865
    },
    {
      "epoch": 0.16,
      "grad_norm": 35.30926800047287,
      "learning_rate": 5.459162082099028e-06,
      "loss": 1.2539,
      "step": 3870
    },
    {
      "epoch": 0.16,
      "grad_norm": 27.121868433439538,
      "learning_rate": 5.466215263083651e-06,
      "loss": 1.3113,
      "step": 3875
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.508865159697022,
      "learning_rate": 5.473268444068275e-06,
      "loss": 1.2632,
      "step": 3880
    },
    {
      "epoch": 0.16,
      "grad_norm": 29.946751869951267,
      "learning_rate": 5.480321625052899e-06,
      "loss": 1.2403,
      "step": 3885
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.98567733859749,
      "learning_rate": 5.487374806037523e-06,
      "loss": 1.2952,
      "step": 3890
    },
    {
      "epoch": 0.16,
      "grad_norm": 19.969233454032903,
      "learning_rate": 5.494427987022148e-06,
      "loss": 1.2286,
      "step": 3895
    },
    {
      "epoch": 0.17,
      "grad_norm": 13.200495931920694,
      "learning_rate": 5.501481168006771e-06,
      "loss": 1.2731,
      "step": 3900
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.287753658655994,
      "learning_rate": 5.508534348991396e-06,
      "loss": 1.2445,
      "step": 3905
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.579639814833723,
      "learning_rate": 5.51558752997602e-06,
      "loss": 1.2881,
      "step": 3910
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.895564369578244,
      "learning_rate": 5.522640710960644e-06,
      "loss": 1.2609,
      "step": 3915
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.593585199742005,
      "learning_rate": 5.529693891945268e-06,
      "loss": 1.258,
      "step": 3920
    },
    {
      "epoch": 0.17,
      "grad_norm": 24.154349961486478,
      "learning_rate": 5.536747072929893e-06,
      "loss": 1.2697,
      "step": 3925
    },
    {
      "epoch": 0.17,
      "grad_norm": 17.826512647308057,
      "learning_rate": 5.543800253914516e-06,
      "loss": 1.2619,
      "step": 3930
    },
    {
      "epoch": 0.17,
      "grad_norm": 29.347512545346877,
      "learning_rate": 5.55085343489914e-06,
      "loss": 1.2648,
      "step": 3935
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.323806259722165,
      "learning_rate": 5.557906615883763e-06,
      "loss": 1.289,
      "step": 3940
    },
    {
      "epoch": 0.17,
      "grad_norm": 38.02571334385203,
      "learning_rate": 5.564959796868388e-06,
      "loss": 1.2503,
      "step": 3945
    },
    {
      "epoch": 0.17,
      "grad_norm": 32.545735306795066,
      "learning_rate": 5.572012977853012e-06,
      "loss": 1.2366,
      "step": 3950
    },
    {
      "epoch": 0.17,
      "grad_norm": 43.70304004193697,
      "learning_rate": 5.579066158837636e-06,
      "loss": 1.3085,
      "step": 3955
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.908420084401357,
      "learning_rate": 5.58611933982226e-06,
      "loss": 1.2645,
      "step": 3960
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.127772130455492,
      "learning_rate": 5.593172520806884e-06,
      "loss": 1.2205,
      "step": 3965
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.196720356489546,
      "learning_rate": 5.600225701791508e-06,
      "loss": 1.2641,
      "step": 3970
    },
    {
      "epoch": 0.17,
      "grad_norm": 51.99348947062528,
      "learning_rate": 5.607278882776133e-06,
      "loss": 1.2814,
      "step": 3975
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.450848090893036,
      "learning_rate": 5.614332063760756e-06,
      "loss": 1.2949,
      "step": 3980
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.347372539755218,
      "learning_rate": 5.621385244745381e-06,
      "loss": 1.267,
      "step": 3985
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.067400551415359,
      "learning_rate": 5.628438425730005e-06,
      "loss": 1.2383,
      "step": 3990
    },
    {
      "epoch": 0.17,
      "grad_norm": 24.396665192217117,
      "learning_rate": 5.635491606714629e-06,
      "loss": 1.2776,
      "step": 3995
    },
    {
      "epoch": 0.17,
      "grad_norm": 37.38209595993001,
      "learning_rate": 5.6425447876992525e-06,
      "loss": 1.234,
      "step": 4000
    },
    {
      "epoch": 0.17,
      "grad_norm": 33.86989983375221,
      "learning_rate": 5.649597968683876e-06,
      "loss": 1.2029,
      "step": 4005
    },
    {
      "epoch": 0.17,
      "grad_norm": 40.47299111083213,
      "learning_rate": 5.6566511496685005e-06,
      "loss": 1.2872,
      "step": 4010
    },
    {
      "epoch": 0.17,
      "grad_norm": 21.74925283256198,
      "learning_rate": 5.663704330653125e-06,
      "loss": 1.2281,
      "step": 4015
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.054164494760023,
      "learning_rate": 5.670757511637749e-06,
      "loss": 1.2325,
      "step": 4020
    },
    {
      "epoch": 0.17,
      "grad_norm": 28.96586779119221,
      "learning_rate": 5.677810692622373e-06,
      "loss": 1.2334,
      "step": 4025
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.218959428611983,
      "learning_rate": 5.6848638736069975e-06,
      "loss": 1.2272,
      "step": 4030
    },
    {
      "epoch": 0.17,
      "grad_norm": 43.124976304431065,
      "learning_rate": 5.691917054591621e-06,
      "loss": 1.25,
      "step": 4035
    },
    {
      "epoch": 0.17,
      "grad_norm": 93.30468133380172,
      "learning_rate": 5.6989702355762455e-06,
      "loss": 1.2803,
      "step": 4040
    },
    {
      "epoch": 0.17,
      "grad_norm": 102.43333571715009,
      "learning_rate": 5.70602341656087e-06,
      "loss": 1.2331,
      "step": 4045
    },
    {
      "epoch": 0.17,
      "grad_norm": 112.03418005270605,
      "learning_rate": 5.7130765975454936e-06,
      "loss": 1.2503,
      "step": 4050
    },
    {
      "epoch": 0.17,
      "grad_norm": 103.91390832861327,
      "learning_rate": 5.720129778530118e-06,
      "loss": 1.2606,
      "step": 4055
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.093811751761326,
      "learning_rate": 5.7271829595147425e-06,
      "loss": 1.283,
      "step": 4060
    },
    {
      "epoch": 0.17,
      "grad_norm": 26.93172767941601,
      "learning_rate": 5.734236140499365e-06,
      "loss": 1.2921,
      "step": 4065
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.177941944522155,
      "learning_rate": 5.74128932148399e-06,
      "loss": 1.2806,
      "step": 4070
    },
    {
      "epoch": 0.17,
      "grad_norm": 29.185596883098917,
      "learning_rate": 5.748342502468613e-06,
      "loss": 1.2417,
      "step": 4075
    },
    {
      "epoch": 0.17,
      "grad_norm": 12.051744305337033,
      "learning_rate": 5.755395683453238e-06,
      "loss": 1.2265,
      "step": 4080
    },
    {
      "epoch": 0.17,
      "grad_norm": 14.057221593617186,
      "learning_rate": 5.762448864437862e-06,
      "loss": 1.2887,
      "step": 4085
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.714811304989844,
      "learning_rate": 5.769502045422486e-06,
      "loss": 1.2208,
      "step": 4090
    },
    {
      "epoch": 0.17,
      "grad_norm": 32.512722645252836,
      "learning_rate": 5.77655522640711e-06,
      "loss": 1.2873,
      "step": 4095
    },
    {
      "epoch": 0.17,
      "grad_norm": 26.843848577592496,
      "learning_rate": 5.783608407391735e-06,
      "loss": 1.278,
      "step": 4100
    },
    {
      "epoch": 0.17,
      "grad_norm": 15.46005842074122,
      "learning_rate": 5.790661588376358e-06,
      "loss": 1.2269,
      "step": 4105
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.109716322565651,
      "learning_rate": 5.797714769360983e-06,
      "loss": 1.2814,
      "step": 4110
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.4481152441406975,
      "learning_rate": 5.804767950345607e-06,
      "loss": 1.2517,
      "step": 4115
    },
    {
      "epoch": 0.17,
      "grad_norm": 18.327012496458657,
      "learning_rate": 5.811821131330231e-06,
      "loss": 1.2537,
      "step": 4120
    },
    {
      "epoch": 0.17,
      "grad_norm": 10.017276109128982,
      "learning_rate": 5.818874312314855e-06,
      "loss": 1.2466,
      "step": 4125
    },
    {
      "epoch": 0.17,
      "grad_norm": 23.34561879987263,
      "learning_rate": 5.825927493299478e-06,
      "loss": 1.2587,
      "step": 4130
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.53598448137646,
      "learning_rate": 5.832980674284102e-06,
      "loss": 1.2579,
      "step": 4135
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.188914641970314,
      "learning_rate": 5.840033855268726e-06,
      "loss": 1.2466,
      "step": 4140
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.657768210567315,
      "learning_rate": 5.84708703625335e-06,
      "loss": 1.261,
      "step": 4145
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.617412272378381,
      "learning_rate": 5.854140217237975e-06,
      "loss": 1.2028,
      "step": 4150
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.837707382428771,
      "learning_rate": 5.8611933982225984e-06,
      "loss": 1.2658,
      "step": 4155
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.968075865608485,
      "learning_rate": 5.868246579207223e-06,
      "loss": 1.23,
      "step": 4160
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.627901389233413,
      "learning_rate": 5.875299760191847e-06,
      "loss": 1.2327,
      "step": 4165
    },
    {
      "epoch": 0.18,
      "grad_norm": 26.98579082603569,
      "learning_rate": 5.882352941176471e-06,
      "loss": 1.2567,
      "step": 4170
    },
    {
      "epoch": 0.18,
      "grad_norm": 25.929086318887546,
      "learning_rate": 5.889406122161095e-06,
      "loss": 1.332,
      "step": 4175
    },
    {
      "epoch": 0.18,
      "grad_norm": 37.879129805501314,
      "learning_rate": 5.89645930314572e-06,
      "loss": 1.2862,
      "step": 4180
    },
    {
      "epoch": 0.18,
      "grad_norm": 42.23786240661063,
      "learning_rate": 5.903512484130343e-06,
      "loss": 1.2187,
      "step": 4185
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.567635063094574,
      "learning_rate": 5.910565665114968e-06,
      "loss": 1.3477,
      "step": 4190
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.925037559476838,
      "learning_rate": 5.917618846099591e-06,
      "loss": 1.2657,
      "step": 4195
    },
    {
      "epoch": 0.18,
      "grad_norm": 10.755363814871327,
      "learning_rate": 5.924672027084215e-06,
      "loss": 1.2147,
      "step": 4200
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.013765422404685,
      "learning_rate": 5.9317252080688395e-06,
      "loss": 1.2226,
      "step": 4205
    },
    {
      "epoch": 0.18,
      "grad_norm": 51.29638914020649,
      "learning_rate": 5.938778389053463e-06,
      "loss": 1.297,
      "step": 4210
    },
    {
      "epoch": 0.18,
      "grad_norm": 57.20210681239939,
      "learning_rate": 5.9458315700380875e-06,
      "loss": 1.2776,
      "step": 4215
    },
    {
      "epoch": 0.18,
      "grad_norm": 48.442410441464226,
      "learning_rate": 5.952884751022712e-06,
      "loss": 1.2401,
      "step": 4220
    },
    {
      "epoch": 0.18,
      "grad_norm": 53.06027896405559,
      "learning_rate": 5.9599379320073356e-06,
      "loss": 1.2872,
      "step": 4225
    },
    {
      "epoch": 0.18,
      "grad_norm": 64.11701466539499,
      "learning_rate": 5.96699111299196e-06,
      "loss": 1.2644,
      "step": 4230
    },
    {
      "epoch": 0.18,
      "grad_norm": 105.4405397871233,
      "learning_rate": 5.9740442939765845e-06,
      "loss": 1.3301,
      "step": 4235
    },
    {
      "epoch": 0.18,
      "grad_norm": 42.761016692177606,
      "learning_rate": 5.981097474961208e-06,
      "loss": 1.2817,
      "step": 4240
    },
    {
      "epoch": 0.18,
      "grad_norm": 132.0515501036898,
      "learning_rate": 5.9881506559458325e-06,
      "loss": 1.311,
      "step": 4245
    },
    {
      "epoch": 0.18,
      "grad_norm": 61.23783719416133,
      "learning_rate": 5.995203836930457e-06,
      "loss": 1.2831,
      "step": 4250
    },
    {
      "epoch": 0.18,
      "grad_norm": 28.3842841786828,
      "learning_rate": 6.00225701791508e-06,
      "loss": 1.3447,
      "step": 4255
    },
    {
      "epoch": 0.18,
      "grad_norm": 34.50311445783356,
      "learning_rate": 6.009310198899704e-06,
      "loss": 1.2465,
      "step": 4260
    },
    {
      "epoch": 0.18,
      "grad_norm": 17.35310049489662,
      "learning_rate": 6.016363379884328e-06,
      "loss": 1.2437,
      "step": 4265
    },
    {
      "epoch": 0.18,
      "grad_norm": 76.98519172566743,
      "learning_rate": 6.023416560868952e-06,
      "loss": 1.2907,
      "step": 4270
    },
    {
      "epoch": 0.18,
      "grad_norm": 19.65288762857618,
      "learning_rate": 6.030469741853577e-06,
      "loss": 1.2327,
      "step": 4275
    },
    {
      "epoch": 0.18,
      "grad_norm": 32.78279868866156,
      "learning_rate": 6.0375229228382e-06,
      "loss": 1.2968,
      "step": 4280
    },
    {
      "epoch": 0.18,
      "grad_norm": 25.930535333766777,
      "learning_rate": 6.044576103822825e-06,
      "loss": 1.2477,
      "step": 4285
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.053943243412213,
      "learning_rate": 6.051629284807449e-06,
      "loss": 1.2264,
      "step": 4290
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.71260152586023,
      "learning_rate": 6.058682465792073e-06,
      "loss": 1.2098,
      "step": 4295
    },
    {
      "epoch": 0.18,
      "grad_norm": 18.252220329487802,
      "learning_rate": 6.065735646776697e-06,
      "loss": 1.2694,
      "step": 4300
    },
    {
      "epoch": 0.18,
      "grad_norm": 11.463870529926188,
      "learning_rate": 6.072788827761321e-06,
      "loss": 1.2522,
      "step": 4305
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.845249553515464,
      "learning_rate": 6.079842008745945e-06,
      "loss": 1.2397,
      "step": 4310
    },
    {
      "epoch": 0.18,
      "grad_norm": 28.731247425252732,
      "learning_rate": 6.08689518973057e-06,
      "loss": 1.2727,
      "step": 4315
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.861653343680853,
      "learning_rate": 6.093948370715192e-06,
      "loss": 1.2236,
      "step": 4320
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.009147926480956,
      "learning_rate": 6.101001551699817e-06,
      "loss": 1.3193,
      "step": 4325
    },
    {
      "epoch": 0.18,
      "grad_norm": 15.038702926390313,
      "learning_rate": 6.1080547326844404e-06,
      "loss": 1.2023,
      "step": 4330
    },
    {
      "epoch": 0.18,
      "grad_norm": 22.729923867980226,
      "learning_rate": 6.115107913669065e-06,
      "loss": 1.2688,
      "step": 4335
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.138365001346727,
      "learning_rate": 6.122161094653689e-06,
      "loss": 1.2214,
      "step": 4340
    },
    {
      "epoch": 0.18,
      "grad_norm": 23.344159254135658,
      "learning_rate": 6.129214275638313e-06,
      "loss": 1.2622,
      "step": 4345
    },
    {
      "epoch": 0.18,
      "grad_norm": 21.288581889146414,
      "learning_rate": 6.136267456622937e-06,
      "loss": 1.2784,
      "step": 4350
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.081998317456286,
      "learning_rate": 6.143320637607562e-06,
      "loss": 1.2332,
      "step": 4355
    },
    {
      "epoch": 0.18,
      "grad_norm": 30.968987023163816,
      "learning_rate": 6.150373818592185e-06,
      "loss": 1.25,
      "step": 4360
    },
    {
      "epoch": 0.18,
      "grad_norm": 12.037895046922486,
      "learning_rate": 6.15742699957681e-06,
      "loss": 1.208,
      "step": 4365
    },
    {
      "epoch": 0.18,
      "grad_norm": 16.508743796682314,
      "learning_rate": 6.164480180561434e-06,
      "loss": 1.2668,
      "step": 4370
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.205816434345365,
      "learning_rate": 6.171533361546058e-06,
      "loss": 1.2226,
      "step": 4375
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.685875478301838,
      "learning_rate": 6.178586542530682e-06,
      "loss": 1.269,
      "step": 4380
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.116569000450617,
      "learning_rate": 6.185639723515305e-06,
      "loss": 1.262,
      "step": 4385
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.502824477099253,
      "learning_rate": 6.1926929044999295e-06,
      "loss": 1.2214,
      "step": 4390
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.737765036321906,
      "learning_rate": 6.199746085484554e-06,
      "loss": 1.1868,
      "step": 4395
    },
    {
      "epoch": 0.19,
      "grad_norm": 27.895069088164966,
      "learning_rate": 6.206799266469178e-06,
      "loss": 1.1919,
      "step": 4400
    },
    {
      "epoch": 0.19,
      "grad_norm": 26.63082356931384,
      "learning_rate": 6.213852447453802e-06,
      "loss": 1.2798,
      "step": 4405
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.70896586815328,
      "learning_rate": 6.2209056284384265e-06,
      "loss": 1.2699,
      "step": 4410
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.802451489350744,
      "learning_rate": 6.22795880942305e-06,
      "loss": 1.2491,
      "step": 4415
    },
    {
      "epoch": 0.19,
      "grad_norm": 11.398986693480657,
      "learning_rate": 6.2350119904076745e-06,
      "loss": 1.2717,
      "step": 4420
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.426557816658939,
      "learning_rate": 6.242065171392299e-06,
      "loss": 1.2773,
      "step": 4425
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.62247766700659,
      "learning_rate": 6.2491183523769226e-06,
      "loss": 1.2359,
      "step": 4430
    },
    {
      "epoch": 0.19,
      "grad_norm": 14.395373937470046,
      "learning_rate": 6.256171533361547e-06,
      "loss": 1.212,
      "step": 4435
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.204125230316254,
      "learning_rate": 6.2632247143461715e-06,
      "loss": 1.2624,
      "step": 4440
    },
    {
      "epoch": 0.19,
      "grad_norm": 19.63742095193836,
      "learning_rate": 6.270277895330795e-06,
      "loss": 1.253,
      "step": 4445
    },
    {
      "epoch": 0.19,
      "grad_norm": 22.14414972404637,
      "learning_rate": 6.277331076315419e-06,
      "loss": 1.2406,
      "step": 4450
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.421229613948904,
      "learning_rate": 6.284384257300042e-06,
      "loss": 1.2789,
      "step": 4455
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.687595585259304,
      "learning_rate": 6.291437438284667e-06,
      "loss": 1.2064,
      "step": 4460
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.25114093177178,
      "learning_rate": 6.298490619269291e-06,
      "loss": 1.2548,
      "step": 4465
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.01156957125535,
      "learning_rate": 6.305543800253915e-06,
      "loss": 1.2434,
      "step": 4470
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.118558932084882,
      "learning_rate": 6.312596981238539e-06,
      "loss": 1.2618,
      "step": 4475
    },
    {
      "epoch": 0.19,
      "grad_norm": 15.110716073860914,
      "learning_rate": 6.319650162223163e-06,
      "loss": 1.2546,
      "step": 4480
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.070031692809758,
      "learning_rate": 6.326703343207787e-06,
      "loss": 1.2033,
      "step": 4485
    },
    {
      "epoch": 0.19,
      "grad_norm": 31.72662568896546,
      "learning_rate": 6.333756524192412e-06,
      "loss": 1.2307,
      "step": 4490
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.302550119851436,
      "learning_rate": 6.340809705177035e-06,
      "loss": 1.2006,
      "step": 4495
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.83523842445794,
      "learning_rate": 6.34786288616166e-06,
      "loss": 1.2525,
      "step": 4500
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.850077169953218,
      "learning_rate": 6.354916067146284e-06,
      "loss": 1.2732,
      "step": 4505
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.170526122594419,
      "learning_rate": 6.361969248130908e-06,
      "loss": 1.2551,
      "step": 4510
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.306824503737438,
      "learning_rate": 6.369022429115531e-06,
      "loss": 1.2374,
      "step": 4515
    },
    {
      "epoch": 0.19,
      "grad_norm": 20.42985922621295,
      "learning_rate": 6.376075610100155e-06,
      "loss": 1.2819,
      "step": 4520
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.818664414829446,
      "learning_rate": 6.383128791084779e-06,
      "loss": 1.2594,
      "step": 4525
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.589621421728435,
      "learning_rate": 6.390181972069404e-06,
      "loss": 1.2265,
      "step": 4530
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.080515844986637,
      "learning_rate": 6.3972351530540274e-06,
      "loss": 1.2599,
      "step": 4535
    },
    {
      "epoch": 0.19,
      "grad_norm": 31.567375012082916,
      "learning_rate": 6.404288334038652e-06,
      "loss": 1.2733,
      "step": 4540
    },
    {
      "epoch": 0.19,
      "grad_norm": 23.8849475472864,
      "learning_rate": 6.411341515023276e-06,
      "loss": 1.2053,
      "step": 4545
    },
    {
      "epoch": 0.19,
      "grad_norm": 26.215994132172288,
      "learning_rate": 6.4183946960079e-06,
      "loss": 1.2129,
      "step": 4550
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.03711583044702,
      "learning_rate": 6.425447876992524e-06,
      "loss": 1.2505,
      "step": 4555
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.32762863686287,
      "learning_rate": 6.432501057977149e-06,
      "loss": 1.1944,
      "step": 4560
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.11227662469492,
      "learning_rate": 6.439554238961772e-06,
      "loss": 1.2329,
      "step": 4565
    },
    {
      "epoch": 0.19,
      "grad_norm": 12.141301098976342,
      "learning_rate": 6.446607419946397e-06,
      "loss": 1.2284,
      "step": 4570
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.574675606298673,
      "learning_rate": 6.45366060093102e-06,
      "loss": 1.1957,
      "step": 4575
    },
    {
      "epoch": 0.19,
      "grad_norm": 21.162667594146992,
      "learning_rate": 6.460713781915644e-06,
      "loss": 1.2674,
      "step": 4580
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.597960415367137,
      "learning_rate": 6.4677669629002685e-06,
      "loss": 1.1878,
      "step": 4585
    },
    {
      "epoch": 0.19,
      "grad_norm": 17.41587230228707,
      "learning_rate": 6.474820143884892e-06,
      "loss": 1.2607,
      "step": 4590
    },
    {
      "epoch": 0.19,
      "grad_norm": 18.275329721851886,
      "learning_rate": 6.4818733248695165e-06,
      "loss": 1.2465,
      "step": 4595
    },
    {
      "epoch": 0.19,
      "grad_norm": 16.4181329184499,
      "learning_rate": 6.488926505854141e-06,
      "loss": 1.2292,
      "step": 4600
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.186165091829702,
      "learning_rate": 6.4959796868387646e-06,
      "loss": 1.2496,
      "step": 4605
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.7939122458535,
      "learning_rate": 6.503032867823389e-06,
      "loss": 1.2421,
      "step": 4610
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.062086815233554,
      "learning_rate": 6.5100860488080135e-06,
      "loss": 1.2602,
      "step": 4615
    },
    {
      "epoch": 0.2,
      "grad_norm": 42.912857207803484,
      "learning_rate": 6.517139229792637e-06,
      "loss": 1.2236,
      "step": 4620
    },
    {
      "epoch": 0.2,
      "grad_norm": 17.971370881200183,
      "learning_rate": 6.5241924107772615e-06,
      "loss": 1.2416,
      "step": 4625
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.229527833171424,
      "learning_rate": 6.531245591761886e-06,
      "loss": 1.2604,
      "step": 4630
    },
    {
      "epoch": 0.2,
      "grad_norm": 32.07525245323547,
      "learning_rate": 6.5382987727465095e-06,
      "loss": 1.2442,
      "step": 4635
    },
    {
      "epoch": 0.2,
      "grad_norm": 26.722295172677985,
      "learning_rate": 6.545351953731133e-06,
      "loss": 1.2481,
      "step": 4640
    },
    {
      "epoch": 0.2,
      "grad_norm": 58.8620086264494,
      "learning_rate": 6.552405134715757e-06,
      "loss": 1.2551,
      "step": 4645
    },
    {
      "epoch": 0.2,
      "grad_norm": 27.105148956024465,
      "learning_rate": 6.559458315700381e-06,
      "loss": 1.1924,
      "step": 4650
    },
    {
      "epoch": 0.2,
      "grad_norm": 37.3996716839102,
      "learning_rate": 6.566511496685005e-06,
      "loss": 1.2396,
      "step": 4655
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.385584996283885,
      "learning_rate": 6.573564677669629e-06,
      "loss": 1.2529,
      "step": 4660
    },
    {
      "epoch": 0.2,
      "grad_norm": 47.27767529483241,
      "learning_rate": 6.580617858654254e-06,
      "loss": 1.2648,
      "step": 4665
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.111911533829485,
      "learning_rate": 6.587671039638877e-06,
      "loss": 1.2372,
      "step": 4670
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.580053924808482,
      "learning_rate": 6.594724220623502e-06,
      "loss": 1.2551,
      "step": 4675
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.07039795529478,
      "learning_rate": 6.601777401608126e-06,
      "loss": 1.2241,
      "step": 4680
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.487965893517133,
      "learning_rate": 6.60883058259275e-06,
      "loss": 1.2567,
      "step": 4685
    },
    {
      "epoch": 0.2,
      "grad_norm": 33.2559515027743,
      "learning_rate": 6.615883763577374e-06,
      "loss": 1.2927,
      "step": 4690
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.423399354802449,
      "learning_rate": 6.622936944561999e-06,
      "loss": 1.3167,
      "step": 4695
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.371956106241857,
      "learning_rate": 6.629990125546622e-06,
      "loss": 1.2472,
      "step": 4700
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.159259831697371,
      "learning_rate": 6.637043306531246e-06,
      "loss": 1.2318,
      "step": 4705
    },
    {
      "epoch": 0.2,
      "grad_norm": 10.773569273743618,
      "learning_rate": 6.6440964875158694e-06,
      "loss": 1.2548,
      "step": 4710
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.07173662577065,
      "learning_rate": 6.651149668500494e-06,
      "loss": 1.2435,
      "step": 4715
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.558139234649046,
      "learning_rate": 6.658202849485118e-06,
      "loss": 1.2273,
      "step": 4720
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.609711847821076,
      "learning_rate": 6.665256030469742e-06,
      "loss": 1.2342,
      "step": 4725
    },
    {
      "epoch": 0.2,
      "grad_norm": 17.724996011129768,
      "learning_rate": 6.672309211454366e-06,
      "loss": 1.2351,
      "step": 4730
    },
    {
      "epoch": 0.2,
      "grad_norm": 28.492096353961507,
      "learning_rate": 6.679362392438991e-06,
      "loss": 1.2502,
      "step": 4735
    },
    {
      "epoch": 0.2,
      "grad_norm": 26.32408910441213,
      "learning_rate": 6.686415573423614e-06,
      "loss": 1.2265,
      "step": 4740
    },
    {
      "epoch": 0.2,
      "grad_norm": 28.482098451695897,
      "learning_rate": 6.693468754408239e-06,
      "loss": 1.207,
      "step": 4745
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.893845252990738,
      "learning_rate": 6.700521935392863e-06,
      "loss": 1.2437,
      "step": 4750
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.485924479755573,
      "learning_rate": 6.707575116377487e-06,
      "loss": 1.2426,
      "step": 4755
    },
    {
      "epoch": 0.2,
      "grad_norm": 22.93788744863069,
      "learning_rate": 6.714628297362111e-06,
      "loss": 1.2499,
      "step": 4760
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.739103812664775,
      "learning_rate": 6.721681478346736e-06,
      "loss": 1.2205,
      "step": 4765
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.269578899710321,
      "learning_rate": 6.7287346593313585e-06,
      "loss": 1.2172,
      "step": 4770
    },
    {
      "epoch": 0.2,
      "grad_norm": 21.467756816152093,
      "learning_rate": 6.735787840315983e-06,
      "loss": 1.2394,
      "step": 4775
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.38554829800853,
      "learning_rate": 6.742841021300607e-06,
      "loss": 1.2815,
      "step": 4780
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.939605925465878,
      "learning_rate": 6.749894202285231e-06,
      "loss": 1.2115,
      "step": 4785
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.987850108768736,
      "learning_rate": 6.7569473832698555e-06,
      "loss": 1.2877,
      "step": 4790
    },
    {
      "epoch": 0.2,
      "grad_norm": 18.723757123880965,
      "learning_rate": 6.764000564254479e-06,
      "loss": 1.2158,
      "step": 4795
    },
    {
      "epoch": 0.2,
      "grad_norm": 23.979174431867218,
      "learning_rate": 6.7710537452391035e-06,
      "loss": 1.2259,
      "step": 4800
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.884194215475357,
      "learning_rate": 6.778106926223728e-06,
      "loss": 1.2614,
      "step": 4805
    },
    {
      "epoch": 0.2,
      "grad_norm": 23.695292180317153,
      "learning_rate": 6.7851601072083516e-06,
      "loss": 1.246,
      "step": 4810
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.64286558091429,
      "learning_rate": 6.792213288192976e-06,
      "loss": 1.2334,
      "step": 4815
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.659020871876756,
      "learning_rate": 6.7992664691776004e-06,
      "loss": 1.291,
      "step": 4820
    },
    {
      "epoch": 0.2,
      "grad_norm": 31.832973137697884,
      "learning_rate": 6.806319650162224e-06,
      "loss": 1.2274,
      "step": 4825
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.279745241859139,
      "learning_rate": 6.8133728311468485e-06,
      "loss": 1.2535,
      "step": 4830
    },
    {
      "epoch": 0.2,
      "grad_norm": 15.281529448929419,
      "learning_rate": 6.820426012131471e-06,
      "loss": 1.2132,
      "step": 4835
    },
    {
      "epoch": 0.2,
      "grad_norm": 14.303496407960964,
      "learning_rate": 6.827479193116096e-06,
      "loss": 1.2138,
      "step": 4840
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.414439438900157,
      "learning_rate": 6.834532374100719e-06,
      "loss": 1.2321,
      "step": 4845
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.958303837167044,
      "learning_rate": 6.841585555085344e-06,
      "loss": 1.211,
      "step": 4850
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.024877483435045,
      "learning_rate": 6.848638736069968e-06,
      "loss": 1.2443,
      "step": 4855
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.751619488887554,
      "learning_rate": 6.855691917054592e-06,
      "loss": 1.2187,
      "step": 4860
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.00348990973335,
      "learning_rate": 6.862745098039216e-06,
      "loss": 1.1981,
      "step": 4865
    },
    {
      "epoch": 0.21,
      "grad_norm": 21.61449142807839,
      "learning_rate": 6.869798279023841e-06,
      "loss": 1.2252,
      "step": 4870
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.811329482264858,
      "learning_rate": 6.876851460008464e-06,
      "loss": 1.2429,
      "step": 4875
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.402305332264874,
      "learning_rate": 6.883904640993089e-06,
      "loss": 1.2804,
      "step": 4880
    },
    {
      "epoch": 0.21,
      "grad_norm": 38.30895830986675,
      "learning_rate": 6.890957821977713e-06,
      "loss": 1.2292,
      "step": 4885
    },
    {
      "epoch": 0.21,
      "grad_norm": 28.730711036505173,
      "learning_rate": 6.898011002962337e-06,
      "loss": 1.2022,
      "step": 4890
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.76296723662178,
      "learning_rate": 6.90506418394696e-06,
      "loss": 1.2084,
      "step": 4895
    },
    {
      "epoch": 0.21,
      "grad_norm": 44.36626252904021,
      "learning_rate": 6.912117364931584e-06,
      "loss": 1.2366,
      "step": 4900
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.203170313920902,
      "learning_rate": 6.919170545916208e-06,
      "loss": 1.2631,
      "step": 4905
    },
    {
      "epoch": 0.21,
      "grad_norm": 14.336109267660797,
      "learning_rate": 6.926223726900833e-06,
      "loss": 1.23,
      "step": 4910
    },
    {
      "epoch": 0.21,
      "grad_norm": 19.385169329304997,
      "learning_rate": 6.9332769078854564e-06,
      "loss": 1.2368,
      "step": 4915
    },
    {
      "epoch": 0.21,
      "grad_norm": 24.41686949163609,
      "learning_rate": 6.940330088870081e-06,
      "loss": 1.2557,
      "step": 4920
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.654871942404327,
      "learning_rate": 6.947383269854705e-06,
      "loss": 1.2092,
      "step": 4925
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.219187386260392,
      "learning_rate": 6.954436450839329e-06,
      "loss": 1.2369,
      "step": 4930
    },
    {
      "epoch": 0.21,
      "grad_norm": 15.045082138446046,
      "learning_rate": 6.961489631823953e-06,
      "loss": 1.2341,
      "step": 4935
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.615703407655676,
      "learning_rate": 6.968542812808578e-06,
      "loss": 1.2507,
      "step": 4940
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.6145670914862915,
      "learning_rate": 6.975595993793201e-06,
      "loss": 1.258,
      "step": 4945
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.66544246130067,
      "learning_rate": 6.982649174777826e-06,
      "loss": 1.2604,
      "step": 4950
    },
    {
      "epoch": 0.21,
      "grad_norm": 34.324393318617716,
      "learning_rate": 6.98970235576245e-06,
      "loss": 1.2708,
      "step": 4955
    },
    {
      "epoch": 0.21,
      "grad_norm": 37.89415729417008,
      "learning_rate": 6.996755536747073e-06,
      "loss": 1.2206,
      "step": 4960
    },
    {
      "epoch": 0.21,
      "grad_norm": 37.586419068795685,
      "learning_rate": 7.0038087177316975e-06,
      "loss": 1.2375,
      "step": 4965
    },
    {
      "epoch": 0.21,
      "grad_norm": 41.293465323629775,
      "learning_rate": 7.010861898716321e-06,
      "loss": 1.2625,
      "step": 4970
    },
    {
      "epoch": 0.21,
      "grad_norm": 27.21765098464862,
      "learning_rate": 7.0179150797009455e-06,
      "loss": 1.1929,
      "step": 4975
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.305652818541972,
      "learning_rate": 7.02496826068557e-06,
      "loss": 1.2309,
      "step": 4980
    },
    {
      "epoch": 0.21,
      "grad_norm": 28.64509065057073,
      "learning_rate": 7.0320214416701936e-06,
      "loss": 1.2249,
      "step": 4985
    },
    {
      "epoch": 0.21,
      "grad_norm": 85.12393967497654,
      "learning_rate": 7.039074622654818e-06,
      "loss": 1.2481,
      "step": 4990
    },
    {
      "epoch": 0.21,
      "grad_norm": 135.8987539009934,
      "learning_rate": 7.0461278036394425e-06,
      "loss": 1.2196,
      "step": 4995
    },
    {
      "epoch": 0.21,
      "grad_norm": 208.07194196806995,
      "learning_rate": 7.053180984624066e-06,
      "loss": 1.274,
      "step": 5000
    },
    {
      "epoch": 0.21,
      "grad_norm": 71.16760199887554,
      "learning_rate": 7.0602341656086905e-06,
      "loss": 1.3879,
      "step": 5005
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.447640267720217,
      "learning_rate": 7.067287346593314e-06,
      "loss": 1.3144,
      "step": 5010
    },
    {
      "epoch": 0.21,
      "grad_norm": 31.382670180597444,
      "learning_rate": 7.0743405275779385e-06,
      "loss": 1.2637,
      "step": 5015
    },
    {
      "epoch": 0.21,
      "grad_norm": 45.52636450521939,
      "learning_rate": 7.081393708562563e-06,
      "loss": 1.3118,
      "step": 5020
    },
    {
      "epoch": 0.21,
      "grad_norm": 43.957279036532334,
      "learning_rate": 7.088446889547186e-06,
      "loss": 1.3008,
      "step": 5025
    },
    {
      "epoch": 0.21,
      "grad_norm": 43.739346603212894,
      "learning_rate": 7.09550007053181e-06,
      "loss": 1.2719,
      "step": 5030
    },
    {
      "epoch": 0.21,
      "grad_norm": 30.362015467775393,
      "learning_rate": 7.102553251516434e-06,
      "loss": 1.2414,
      "step": 5035
    },
    {
      "epoch": 0.21,
      "grad_norm": 26.810874193829417,
      "learning_rate": 7.109606432501058e-06,
      "loss": 1.2776,
      "step": 5040
    },
    {
      "epoch": 0.21,
      "grad_norm": 16.04965719279758,
      "learning_rate": 7.116659613485683e-06,
      "loss": 1.2267,
      "step": 5045
    },
    {
      "epoch": 0.21,
      "grad_norm": 22.832321649941303,
      "learning_rate": 7.123712794470306e-06,
      "loss": 1.2251,
      "step": 5050
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.52005669267623,
      "learning_rate": 7.130765975454931e-06,
      "loss": 1.2555,
      "step": 5055
    },
    {
      "epoch": 0.21,
      "grad_norm": 24.340708031011626,
      "learning_rate": 7.137819156439555e-06,
      "loss": 1.2791,
      "step": 5060
    },
    {
      "epoch": 0.21,
      "grad_norm": 18.919712129577587,
      "learning_rate": 7.144872337424179e-06,
      "loss": 1.2454,
      "step": 5065
    },
    {
      "epoch": 0.21,
      "grad_norm": 12.191742788980482,
      "learning_rate": 7.151925518408803e-06,
      "loss": 1.2504,
      "step": 5070
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.2707869372775065,
      "learning_rate": 7.158978699393428e-06,
      "loss": 1.2128,
      "step": 5075
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.483696456508929,
      "learning_rate": 7.166031880378051e-06,
      "loss": 1.2503,
      "step": 5080
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.904851570847637,
      "learning_rate": 7.173085061362676e-06,
      "loss": 1.2405,
      "step": 5085
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.27068299831643,
      "learning_rate": 7.1801382423472984e-06,
      "loss": 1.2255,
      "step": 5090
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.086551940068498,
      "learning_rate": 7.187191423331923e-06,
      "loss": 1.197,
      "step": 5095
    },
    {
      "epoch": 0.22,
      "grad_norm": 30.351386199272618,
      "learning_rate": 7.194244604316547e-06,
      "loss": 1.2506,
      "step": 5100
    },
    {
      "epoch": 0.22,
      "grad_norm": 25.136033625238245,
      "learning_rate": 7.201297785301171e-06,
      "loss": 1.2367,
      "step": 5105
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.41315672808275,
      "learning_rate": 7.208350966285795e-06,
      "loss": 1.1984,
      "step": 5110
    },
    {
      "epoch": 0.22,
      "grad_norm": 28.215296752301334,
      "learning_rate": 7.21540414727042e-06,
      "loss": 1.2647,
      "step": 5115
    },
    {
      "epoch": 0.22,
      "grad_norm": 35.28103881347845,
      "learning_rate": 7.222457328255043e-06,
      "loss": 1.2398,
      "step": 5120
    },
    {
      "epoch": 0.22,
      "grad_norm": 27.378602433879518,
      "learning_rate": 7.229510509239668e-06,
      "loss": 1.231,
      "step": 5125
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.7609425202486,
      "learning_rate": 7.236563690224292e-06,
      "loss": 1.2306,
      "step": 5130
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.048799249449326,
      "learning_rate": 7.243616871208916e-06,
      "loss": 1.226,
      "step": 5135
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.697886494754847,
      "learning_rate": 7.25067005219354e-06,
      "loss": 1.2333,
      "step": 5140
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.943073511017728,
      "learning_rate": 7.257723233178165e-06,
      "loss": 1.2357,
      "step": 5145
    },
    {
      "epoch": 0.22,
      "grad_norm": 17.29181708016029,
      "learning_rate": 7.264776414162788e-06,
      "loss": 1.2389,
      "step": 5150
    },
    {
      "epoch": 0.22,
      "grad_norm": 9.651865809048662,
      "learning_rate": 7.271829595147412e-06,
      "loss": 1.2218,
      "step": 5155
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.64169189621745,
      "learning_rate": 7.278882776132036e-06,
      "loss": 1.2158,
      "step": 5160
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.11774091818124,
      "learning_rate": 7.28593595711666e-06,
      "loss": 1.2415,
      "step": 5165
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.771656199240232,
      "learning_rate": 7.2929891381012845e-06,
      "loss": 1.2472,
      "step": 5170
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.540803359450853,
      "learning_rate": 7.300042319085908e-06,
      "loss": 1.31,
      "step": 5175
    },
    {
      "epoch": 0.22,
      "grad_norm": 16.61472642623873,
      "learning_rate": 7.3070955000705325e-06,
      "loss": 1.3246,
      "step": 5180
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.1878758336833,
      "learning_rate": 7.314148681055156e-06,
      "loss": 1.2258,
      "step": 5185
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.221182429738512,
      "learning_rate": 7.3212018620397806e-06,
      "loss": 1.2575,
      "step": 5190
    },
    {
      "epoch": 0.22,
      "grad_norm": 22.50872224561072,
      "learning_rate": 7.328255043024405e-06,
      "loss": 1.2197,
      "step": 5195
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.6994195201989,
      "learning_rate": 7.335308224009029e-06,
      "loss": 1.234,
      "step": 5200
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.94860403481735,
      "learning_rate": 7.342361404993653e-06,
      "loss": 1.2017,
      "step": 5205
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.60764621923028,
      "learning_rate": 7.3494145859782775e-06,
      "loss": 1.1956,
      "step": 5210
    },
    {
      "epoch": 0.22,
      "grad_norm": 12.868630296089705,
      "learning_rate": 7.3564677669629e-06,
      "loss": 1.2232,
      "step": 5215
    },
    {
      "epoch": 0.22,
      "grad_norm": 19.60381910504052,
      "learning_rate": 7.363520947947525e-06,
      "loss": 1.25,
      "step": 5220
    },
    {
      "epoch": 0.22,
      "grad_norm": 28.170648681732985,
      "learning_rate": 7.370574128932148e-06,
      "loss": 1.2503,
      "step": 5225
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.710517283293103,
      "learning_rate": 7.377627309916773e-06,
      "loss": 1.2602,
      "step": 5230
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.156412950775163,
      "learning_rate": 7.384680490901397e-06,
      "loss": 1.2194,
      "step": 5235
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.346549065589771,
      "learning_rate": 7.391733671886021e-06,
      "loss": 1.2053,
      "step": 5240
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.05275653516866,
      "learning_rate": 7.398786852870645e-06,
      "loss": 1.2435,
      "step": 5245
    },
    {
      "epoch": 0.22,
      "grad_norm": 14.780765152837011,
      "learning_rate": 7.40584003385527e-06,
      "loss": 1.2374,
      "step": 5250
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.578334485018415,
      "learning_rate": 7.412893214839893e-06,
      "loss": 1.1733,
      "step": 5255
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.089542001474847,
      "learning_rate": 7.419946395824518e-06,
      "loss": 1.2053,
      "step": 5260
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.055737575437313,
      "learning_rate": 7.426999576809142e-06,
      "loss": 1.2428,
      "step": 5265
    },
    {
      "epoch": 0.22,
      "grad_norm": 33.28259990502187,
      "learning_rate": 7.434052757793766e-06,
      "loss": 1.1782,
      "step": 5270
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.629775398546094,
      "learning_rate": 7.44110593877839e-06,
      "loss": 1.327,
      "step": 5275
    },
    {
      "epoch": 0.22,
      "grad_norm": 21.091615298178418,
      "learning_rate": 7.448159119763013e-06,
      "loss": 1.2721,
      "step": 5280
    },
    {
      "epoch": 0.22,
      "grad_norm": 59.36783809856298,
      "learning_rate": 7.455212300747637e-06,
      "loss": 1.2424,
      "step": 5285
    },
    {
      "epoch": 0.22,
      "grad_norm": 20.788459245771993,
      "learning_rate": 7.462265481732262e-06,
      "loss": 1.2114,
      "step": 5290
    },
    {
      "epoch": 0.22,
      "grad_norm": 25.50948987670773,
      "learning_rate": 7.469318662716885e-06,
      "loss": 1.2384,
      "step": 5295
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.722448159244014,
      "learning_rate": 7.47637184370151e-06,
      "loss": 1.2698,
      "step": 5300
    },
    {
      "epoch": 0.22,
      "grad_norm": 15.375770571889714,
      "learning_rate": 7.483425024686134e-06,
      "loss": 1.241,
      "step": 5305
    },
    {
      "epoch": 0.22,
      "grad_norm": 13.258407829634491,
      "learning_rate": 7.490478205670758e-06,
      "loss": 1.271,
      "step": 5310
    },
    {
      "epoch": 0.22,
      "grad_norm": 27.29466633067146,
      "learning_rate": 7.497531386655382e-06,
      "loss": 1.2247,
      "step": 5315
    },
    {
      "epoch": 0.23,
      "grad_norm": 27.39793150373791,
      "learning_rate": 7.504584567640007e-06,
      "loss": 1.2847,
      "step": 5320
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.625293576187255,
      "learning_rate": 7.51163774862463e-06,
      "loss": 1.2409,
      "step": 5325
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.499825077190442,
      "learning_rate": 7.518690929609255e-06,
      "loss": 1.2379,
      "step": 5330
    },
    {
      "epoch": 0.23,
      "grad_norm": 44.09292451305061,
      "learning_rate": 7.525744110593879e-06,
      "loss": 1.284,
      "step": 5335
    },
    {
      "epoch": 0.23,
      "grad_norm": 21.285965713590198,
      "learning_rate": 7.532797291578503e-06,
      "loss": 1.2467,
      "step": 5340
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.644627220773152,
      "learning_rate": 7.5398504725631265e-06,
      "loss": 1.1858,
      "step": 5345
    },
    {
      "epoch": 0.23,
      "grad_norm": 19.663194191382065,
      "learning_rate": 7.54690365354775e-06,
      "loss": 1.2171,
      "step": 5350
    },
    {
      "epoch": 0.23,
      "grad_norm": 28.902397205479353,
      "learning_rate": 7.5539568345323745e-06,
      "loss": 1.2271,
      "step": 5355
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.945930006474786,
      "learning_rate": 7.561010015516998e-06,
      "loss": 1.2322,
      "step": 5360
    },
    {
      "epoch": 0.23,
      "grad_norm": 19.414902357142513,
      "learning_rate": 7.5680631965016226e-06,
      "loss": 1.2074,
      "step": 5365
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.475661376914047,
      "learning_rate": 7.575116377486247e-06,
      "loss": 1.2363,
      "step": 5370
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.999917219708243,
      "learning_rate": 7.582169558470871e-06,
      "loss": 1.2628,
      "step": 5375
    },
    {
      "epoch": 0.23,
      "grad_norm": 38.849038339519595,
      "learning_rate": 7.589222739455495e-06,
      "loss": 1.2485,
      "step": 5380
    },
    {
      "epoch": 0.23,
      "grad_norm": 32.841423621602274,
      "learning_rate": 7.5962759204401195e-06,
      "loss": 1.2107,
      "step": 5385
    },
    {
      "epoch": 0.23,
      "grad_norm": 35.69942150166393,
      "learning_rate": 7.603329101424743e-06,
      "loss": 1.2589,
      "step": 5390
    },
    {
      "epoch": 0.23,
      "grad_norm": 29.696559181952644,
      "learning_rate": 7.6103822824093675e-06,
      "loss": 1.2622,
      "step": 5395
    },
    {
      "epoch": 0.23,
      "grad_norm": 24.542615685281255,
      "learning_rate": 7.617435463393992e-06,
      "loss": 1.2789,
      "step": 5400
    },
    {
      "epoch": 0.23,
      "grad_norm": 37.03208988309521,
      "learning_rate": 7.624488644378616e-06,
      "loss": 1.2489,
      "step": 5405
    },
    {
      "epoch": 0.23,
      "grad_norm": 30.78302835451295,
      "learning_rate": 7.63154182536324e-06,
      "loss": 1.246,
      "step": 5410
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.69008865206168,
      "learning_rate": 7.638595006347863e-06,
      "loss": 1.2748,
      "step": 5415
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.61441193591923,
      "learning_rate": 7.645648187332488e-06,
      "loss": 1.2625,
      "step": 5420
    },
    {
      "epoch": 0.23,
      "grad_norm": 10.080083707736156,
      "learning_rate": 7.652701368317112e-06,
      "loss": 1.2568,
      "step": 5425
    },
    {
      "epoch": 0.23,
      "grad_norm": 14.197557034230073,
      "learning_rate": 7.659754549301735e-06,
      "loss": 1.2328,
      "step": 5430
    },
    {
      "epoch": 0.23,
      "grad_norm": 17.757497801017163,
      "learning_rate": 7.66680773028636e-06,
      "loss": 1.2327,
      "step": 5435
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.237937606222026,
      "learning_rate": 7.673860911270984e-06,
      "loss": 1.2932,
      "step": 5440
    },
    {
      "epoch": 0.23,
      "grad_norm": 26.596692719642242,
      "learning_rate": 7.680914092255608e-06,
      "loss": 1.2155,
      "step": 5445
    },
    {
      "epoch": 0.23,
      "grad_norm": 29.475003336477855,
      "learning_rate": 7.687967273240231e-06,
      "loss": 1.2566,
      "step": 5450
    },
    {
      "epoch": 0.23,
      "grad_norm": 47.35898381503337,
      "learning_rate": 7.695020454224857e-06,
      "loss": 1.2041,
      "step": 5455
    },
    {
      "epoch": 0.23,
      "grad_norm": 41.63328925060982,
      "learning_rate": 7.70207363520948e-06,
      "loss": 1.199,
      "step": 5460
    },
    {
      "epoch": 0.23,
      "grad_norm": 48.10794058752825,
      "learning_rate": 7.709126816194104e-06,
      "loss": 1.226,
      "step": 5465
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.496571253075856,
      "learning_rate": 7.716179997178729e-06,
      "loss": 1.1894,
      "step": 5470
    },
    {
      "epoch": 0.23,
      "grad_norm": 24.74008078473553,
      "learning_rate": 7.723233178163351e-06,
      "loss": 1.2284,
      "step": 5475
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.09872035218063,
      "learning_rate": 7.730286359147976e-06,
      "loss": 1.292,
      "step": 5480
    },
    {
      "epoch": 0.23,
      "grad_norm": 7.937516307018183,
      "learning_rate": 7.7373395401326e-06,
      "loss": 1.2219,
      "step": 5485
    },
    {
      "epoch": 0.23,
      "grad_norm": 15.708157872424051,
      "learning_rate": 7.744392721117224e-06,
      "loss": 1.2542,
      "step": 5490
    },
    {
      "epoch": 0.23,
      "grad_norm": 32.82070351482054,
      "learning_rate": 7.751445902101849e-06,
      "loss": 1.2413,
      "step": 5495
    },
    {
      "epoch": 0.23,
      "grad_norm": 45.352652386867184,
      "learning_rate": 7.758499083086472e-06,
      "loss": 1.2461,
      "step": 5500
    },
    {
      "epoch": 0.23,
      "grad_norm": 23.682529662802693,
      "learning_rate": 7.765552264071096e-06,
      "loss": 1.2513,
      "step": 5505
    },
    {
      "epoch": 0.23,
      "grad_norm": 13.026179777053374,
      "learning_rate": 7.772605445055721e-06,
      "loss": 1.2086,
      "step": 5510
    },
    {
      "epoch": 0.23,
      "grad_norm": 20.451632989280416,
      "learning_rate": 7.779658626040345e-06,
      "loss": 1.2593,
      "step": 5515
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.645583127940919,
      "learning_rate": 7.786711807024968e-06,
      "loss": 1.2267,
      "step": 5520
    },
    {
      "epoch": 0.23,
      "grad_norm": 18.17246112408486,
      "learning_rate": 7.793764988009594e-06,
      "loss": 1.2759,
      "step": 5525
    },
    {
      "epoch": 0.23,
      "grad_norm": 57.88720436051201,
      "learning_rate": 7.800818168994217e-06,
      "loss": 1.2603,
      "step": 5530
    },
    {
      "epoch": 0.23,
      "grad_norm": 32.99982230022284,
      "learning_rate": 7.807871349978841e-06,
      "loss": 1.2291,
      "step": 5535
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.907954453742963,
      "learning_rate": 7.814924530963465e-06,
      "loss": 1.2425,
      "step": 5540
    },
    {
      "epoch": 0.23,
      "grad_norm": 9.755041842092039,
      "learning_rate": 7.821977711948088e-06,
      "loss": 1.2465,
      "step": 5545
    },
    {
      "epoch": 0.23,
      "grad_norm": 11.108776012805647,
      "learning_rate": 7.829030892932713e-06,
      "loss": 1.1827,
      "step": 5550
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.051574743245526,
      "learning_rate": 7.836084073917337e-06,
      "loss": 1.2208,
      "step": 5555
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.29578232029804,
      "learning_rate": 7.84313725490196e-06,
      "loss": 1.2522,
      "step": 5560
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.437022803264277,
      "learning_rate": 7.850190435886586e-06,
      "loss": 1.2787,
      "step": 5565
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.813191051935494,
      "learning_rate": 7.85724361687121e-06,
      "loss": 1.23,
      "step": 5570
    },
    {
      "epoch": 0.24,
      "grad_norm": 20.982176454917052,
      "learning_rate": 7.864296797855833e-06,
      "loss": 1.21,
      "step": 5575
    },
    {
      "epoch": 0.24,
      "grad_norm": 38.91391833811104,
      "learning_rate": 7.871349978840458e-06,
      "loss": 1.2275,
      "step": 5580
    },
    {
      "epoch": 0.24,
      "grad_norm": 25.652714703934958,
      "learning_rate": 7.878403159825082e-06,
      "loss": 1.2645,
      "step": 5585
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.586967924272614,
      "learning_rate": 7.885456340809706e-06,
      "loss": 1.2101,
      "step": 5590
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.759683978221286,
      "learning_rate": 7.892509521794331e-06,
      "loss": 1.2695,
      "step": 5595
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.594071720655114,
      "learning_rate": 7.899562702778953e-06,
      "loss": 1.2081,
      "step": 5600
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.557672653758722,
      "learning_rate": 7.906615883763578e-06,
      "loss": 1.2413,
      "step": 5605
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.303184324542405,
      "learning_rate": 7.913669064748202e-06,
      "loss": 1.2112,
      "step": 5610
    },
    {
      "epoch": 0.24,
      "grad_norm": 16.093487008474828,
      "learning_rate": 7.920722245732825e-06,
      "loss": 1.2247,
      "step": 5615
    },
    {
      "epoch": 0.24,
      "grad_norm": 24.14998033564898,
      "learning_rate": 7.92777542671745e-06,
      "loss": 1.2365,
      "step": 5620
    },
    {
      "epoch": 0.24,
      "grad_norm": 57.246181883266686,
      "learning_rate": 7.934828607702074e-06,
      "loss": 1.256,
      "step": 5625
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.939623103628394,
      "learning_rate": 7.941881788686698e-06,
      "loss": 1.2494,
      "step": 5630
    },
    {
      "epoch": 0.24,
      "grad_norm": 26.782811993135432,
      "learning_rate": 7.948934969671323e-06,
      "loss": 1.2393,
      "step": 5635
    },
    {
      "epoch": 0.24,
      "grad_norm": 80.56850881605354,
      "learning_rate": 7.955988150655947e-06,
      "loss": 1.2408,
      "step": 5640
    },
    {
      "epoch": 0.24,
      "grad_norm": 81.83284909740311,
      "learning_rate": 7.96304133164057e-06,
      "loss": 1.2026,
      "step": 5645
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.815992929128662,
      "learning_rate": 7.970094512625196e-06,
      "loss": 1.2516,
      "step": 5650
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.051358772845925,
      "learning_rate": 7.97714769360982e-06,
      "loss": 1.2117,
      "step": 5655
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.070032008142176,
      "learning_rate": 7.984200874594443e-06,
      "loss": 1.2234,
      "step": 5660
    },
    {
      "epoch": 0.24,
      "grad_norm": 26.233579984973666,
      "learning_rate": 7.991254055579066e-06,
      "loss": 1.235,
      "step": 5665
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.280753878087175,
      "learning_rate": 7.99830723656369e-06,
      "loss": 1.2167,
      "step": 5670
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.019445593955554,
      "learning_rate": 8.005360417548315e-06,
      "loss": 1.2258,
      "step": 5675
    },
    {
      "epoch": 0.24,
      "grad_norm": 22.261336225693217,
      "learning_rate": 8.012413598532939e-06,
      "loss": 1.272,
      "step": 5680
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.891613852460367,
      "learning_rate": 8.019466779517562e-06,
      "loss": 1.2047,
      "step": 5685
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.411480507776263,
      "learning_rate": 8.026519960502188e-06,
      "loss": 1.2797,
      "step": 5690
    },
    {
      "epoch": 0.24,
      "grad_norm": 35.70797685198074,
      "learning_rate": 8.033573141486811e-06,
      "loss": 1.2431,
      "step": 5695
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.769754151648415,
      "learning_rate": 8.040626322471435e-06,
      "loss": 1.2125,
      "step": 5700
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.950516163892479,
      "learning_rate": 8.04767950345606e-06,
      "loss": 1.2535,
      "step": 5705
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.365075035102075,
      "learning_rate": 8.054732684440684e-06,
      "loss": 1.2078,
      "step": 5710
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.740482274832386,
      "learning_rate": 8.061785865425307e-06,
      "loss": 1.2612,
      "step": 5715
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.74218161219491,
      "learning_rate": 8.068839046409933e-06,
      "loss": 1.2607,
      "step": 5720
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.590193525435973,
      "learning_rate": 8.075892227394556e-06,
      "loss": 1.2207,
      "step": 5725
    },
    {
      "epoch": 0.24,
      "grad_norm": 10.94878125209523,
      "learning_rate": 8.08294540837918e-06,
      "loss": 1.2702,
      "step": 5730
    },
    {
      "epoch": 0.24,
      "grad_norm": 14.180769665936394,
      "learning_rate": 8.089998589363804e-06,
      "loss": 1.2464,
      "step": 5735
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.462500006829307,
      "learning_rate": 8.097051770348427e-06,
      "loss": 1.2221,
      "step": 5740
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.413536579040953,
      "learning_rate": 8.104104951333052e-06,
      "loss": 1.2738,
      "step": 5745
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.131398238975681,
      "learning_rate": 8.111158132317676e-06,
      "loss": 1.2675,
      "step": 5750
    },
    {
      "epoch": 0.24,
      "grad_norm": 9.714857970314563,
      "learning_rate": 8.1182113133023e-06,
      "loss": 1.2363,
      "step": 5755
    },
    {
      "epoch": 0.24,
      "grad_norm": 18.14180250016712,
      "learning_rate": 8.125264494286925e-06,
      "loss": 1.221,
      "step": 5760
    },
    {
      "epoch": 0.24,
      "grad_norm": 13.905825663762378,
      "learning_rate": 8.132317675271548e-06,
      "loss": 1.1805,
      "step": 5765
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.844830230607571,
      "learning_rate": 8.139370856256172e-06,
      "loss": 1.2156,
      "step": 5770
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.475546565266745,
      "learning_rate": 8.146424037240797e-06,
      "loss": 1.253,
      "step": 5775
    },
    {
      "epoch": 0.24,
      "grad_norm": 11.619127623301695,
      "learning_rate": 8.153477218225421e-06,
      "loss": 1.2056,
      "step": 5780
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.777988691308142,
      "learning_rate": 8.160530399210045e-06,
      "loss": 1.2465,
      "step": 5785
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.56102143403454,
      "learning_rate": 8.167583580194668e-06,
      "loss": 1.2206,
      "step": 5790
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.398107937329991,
      "learning_rate": 8.174636761179292e-06,
      "loss": 1.297,
      "step": 5795
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.610804066433388,
      "learning_rate": 8.181689942163915e-06,
      "loss": 1.2474,
      "step": 5800
    },
    {
      "epoch": 0.25,
      "grad_norm": 12.748263168704009,
      "learning_rate": 8.18874312314854e-06,
      "loss": 1.1945,
      "step": 5805
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.716600253053812,
      "learning_rate": 8.195796304133164e-06,
      "loss": 1.2332,
      "step": 5810
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.54240303361471,
      "learning_rate": 8.202849485117788e-06,
      "loss": 1.2739,
      "step": 5815
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.835208832050242,
      "learning_rate": 8.209902666102413e-06,
      "loss": 1.2132,
      "step": 5820
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.55848121281398,
      "learning_rate": 8.216955847087037e-06,
      "loss": 1.254,
      "step": 5825
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.473766456855103,
      "learning_rate": 8.22400902807166e-06,
      "loss": 1.2187,
      "step": 5830
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.971680188358672,
      "learning_rate": 8.231062209056286e-06,
      "loss": 1.2146,
      "step": 5835
    },
    {
      "epoch": 0.25,
      "grad_norm": 29.133127126204222,
      "learning_rate": 8.23811539004091e-06,
      "loss": 1.2497,
      "step": 5840
    },
    {
      "epoch": 0.25,
      "grad_norm": 28.868645770345797,
      "learning_rate": 8.245168571025533e-06,
      "loss": 1.2448,
      "step": 5845
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.03794233105637,
      "learning_rate": 8.252221752010158e-06,
      "loss": 1.2285,
      "step": 5850
    },
    {
      "epoch": 0.25,
      "grad_norm": 29.153293390917938,
      "learning_rate": 8.25927493299478e-06,
      "loss": 1.2408,
      "step": 5855
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.821871459501633,
      "learning_rate": 8.266328113979405e-06,
      "loss": 1.251,
      "step": 5860
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.098621548804456,
      "learning_rate": 8.273381294964029e-06,
      "loss": 1.2557,
      "step": 5865
    },
    {
      "epoch": 0.25,
      "grad_norm": 33.754388048352,
      "learning_rate": 8.280434475948653e-06,
      "loss": 1.2415,
      "step": 5870
    },
    {
      "epoch": 0.25,
      "grad_norm": 17.433105404568604,
      "learning_rate": 8.287487656933278e-06,
      "loss": 1.2526,
      "step": 5875
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.112171234106306,
      "learning_rate": 8.294540837917901e-06,
      "loss": 1.2705,
      "step": 5880
    },
    {
      "epoch": 0.25,
      "grad_norm": 27.981100316425948,
      "learning_rate": 8.301594018902525e-06,
      "loss": 1.2355,
      "step": 5885
    },
    {
      "epoch": 0.25,
      "grad_norm": 16.9682915748727,
      "learning_rate": 8.30864719988715e-06,
      "loss": 1.2916,
      "step": 5890
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.561662484568016,
      "learning_rate": 8.315700380871774e-06,
      "loss": 1.2448,
      "step": 5895
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.83498002391689,
      "learning_rate": 8.322753561856397e-06,
      "loss": 1.2385,
      "step": 5900
    },
    {
      "epoch": 0.25,
      "grad_norm": 41.10703286431987,
      "learning_rate": 8.329806742841023e-06,
      "loss": 1.2532,
      "step": 5905
    },
    {
      "epoch": 0.25,
      "grad_norm": 41.24506108790113,
      "learning_rate": 8.336859923825646e-06,
      "loss": 1.2591,
      "step": 5910
    },
    {
      "epoch": 0.25,
      "grad_norm": 11.094347359313005,
      "learning_rate": 8.34391310481027e-06,
      "loss": 1.1914,
      "step": 5915
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.222070668505308,
      "learning_rate": 8.350966285794894e-06,
      "loss": 1.2535,
      "step": 5920
    },
    {
      "epoch": 0.25,
      "grad_norm": 40.26337514147059,
      "learning_rate": 8.358019466779517e-06,
      "loss": 1.2681,
      "step": 5925
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.333942163534006,
      "learning_rate": 8.365072647764142e-06,
      "loss": 1.2403,
      "step": 5930
    },
    {
      "epoch": 0.25,
      "grad_norm": 32.37180920346308,
      "learning_rate": 8.372125828748766e-06,
      "loss": 1.2256,
      "step": 5935
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.203347877396613,
      "learning_rate": 8.37917900973339e-06,
      "loss": 1.2219,
      "step": 5940
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.855367688307293,
      "learning_rate": 8.386232190718015e-06,
      "loss": 1.2552,
      "step": 5945
    },
    {
      "epoch": 0.25,
      "grad_norm": 31.131371072859213,
      "learning_rate": 8.393285371702639e-06,
      "loss": 1.233,
      "step": 5950
    },
    {
      "epoch": 0.25,
      "grad_norm": 21.566081614895563,
      "learning_rate": 8.400338552687262e-06,
      "loss": 1.217,
      "step": 5955
    },
    {
      "epoch": 0.25,
      "grad_norm": 24.159453225969166,
      "learning_rate": 8.407391733671887e-06,
      "loss": 1.2386,
      "step": 5960
    },
    {
      "epoch": 0.25,
      "grad_norm": 33.781014459154264,
      "learning_rate": 8.414444914656511e-06,
      "loss": 1.2681,
      "step": 5965
    },
    {
      "epoch": 0.25,
      "grad_norm": 51.31754001607082,
      "learning_rate": 8.421498095641135e-06,
      "loss": 1.2287,
      "step": 5970
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.985695551876963,
      "learning_rate": 8.42855127662576e-06,
      "loss": 1.2531,
      "step": 5975
    },
    {
      "epoch": 0.25,
      "grad_norm": 9.859082281561756,
      "learning_rate": 8.435604457610384e-06,
      "loss": 1.2726,
      "step": 5980
    },
    {
      "epoch": 0.25,
      "grad_norm": 23.738347826502846,
      "learning_rate": 8.442657638595007e-06,
      "loss": 1.3014,
      "step": 5985
    },
    {
      "epoch": 0.25,
      "grad_norm": 80.91375473200746,
      "learning_rate": 8.44971081957963e-06,
      "loss": 1.2805,
      "step": 5990
    },
    {
      "epoch": 0.25,
      "grad_norm": 75.49345948735171,
      "learning_rate": 8.456764000564254e-06,
      "loss": 1.2187,
      "step": 5995
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.493516296256855,
      "learning_rate": 8.46381718154888e-06,
      "loss": 1.2316,
      "step": 6000
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.678058973313023,
      "learning_rate": 8.470870362533503e-06,
      "loss": 1.2581,
      "step": 6005
    },
    {
      "epoch": 0.25,
      "grad_norm": 36.77265560264806,
      "learning_rate": 8.477923543518127e-06,
      "loss": 1.2469,
      "step": 6010
    },
    {
      "epoch": 0.25,
      "grad_norm": 40.824198800378696,
      "learning_rate": 8.484976724502752e-06,
      "loss": 1.2639,
      "step": 6015
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.046645380727732,
      "learning_rate": 8.492029905487376e-06,
      "loss": 1.2586,
      "step": 6020
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.701584889580545,
      "learning_rate": 8.499083086472e-06,
      "loss": 1.2471,
      "step": 6025
    },
    {
      "epoch": 0.26,
      "grad_norm": 38.999248208946106,
      "learning_rate": 8.506136267456625e-06,
      "loss": 1.2623,
      "step": 6030
    },
    {
      "epoch": 0.26,
      "grad_norm": 27.593393111353738,
      "learning_rate": 8.513189448441248e-06,
      "loss": 1.2096,
      "step": 6035
    },
    {
      "epoch": 0.26,
      "grad_norm": 24.224788728771827,
      "learning_rate": 8.520242629425872e-06,
      "loss": 1.2188,
      "step": 6040
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.235996778586854,
      "learning_rate": 8.527295810410497e-06,
      "loss": 1.2542,
      "step": 6045
    },
    {
      "epoch": 0.26,
      "grad_norm": 29.151698753961035,
      "learning_rate": 8.534348991395119e-06,
      "loss": 1.2384,
      "step": 6050
    },
    {
      "epoch": 0.26,
      "grad_norm": 34.17689264705203,
      "learning_rate": 8.541402172379744e-06,
      "loss": 1.2338,
      "step": 6055
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.745395446213094,
      "learning_rate": 8.548455353364368e-06,
      "loss": 1.2153,
      "step": 6060
    },
    {
      "epoch": 0.26,
      "grad_norm": 32.48340329821858,
      "learning_rate": 8.555508534348991e-06,
      "loss": 1.2329,
      "step": 6065
    },
    {
      "epoch": 0.26,
      "grad_norm": 51.215089575102965,
      "learning_rate": 8.562561715333617e-06,
      "loss": 1.2319,
      "step": 6070
    },
    {
      "epoch": 0.26,
      "grad_norm": 24.584438140553477,
      "learning_rate": 8.56961489631824e-06,
      "loss": 1.2332,
      "step": 6075
    },
    {
      "epoch": 0.26,
      "grad_norm": 11.967200680729418,
      "learning_rate": 8.576668077302864e-06,
      "loss": 1.254,
      "step": 6080
    },
    {
      "epoch": 0.26,
      "grad_norm": 38.11061044214363,
      "learning_rate": 8.58372125828749e-06,
      "loss": 1.2552,
      "step": 6085
    },
    {
      "epoch": 0.26,
      "grad_norm": 54.682670316095404,
      "learning_rate": 8.590774439272113e-06,
      "loss": 1.2329,
      "step": 6090
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.26925850042772,
      "learning_rate": 8.597827620256736e-06,
      "loss": 1.2312,
      "step": 6095
    },
    {
      "epoch": 0.26,
      "grad_norm": 18.7128259255319,
      "learning_rate": 8.604880801241362e-06,
      "loss": 1.2597,
      "step": 6100
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.500827664986156,
      "learning_rate": 8.611933982225985e-06,
      "loss": 1.2203,
      "step": 6105
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.919881059791322,
      "learning_rate": 8.618987163210609e-06,
      "loss": 1.2393,
      "step": 6110
    },
    {
      "epoch": 0.26,
      "grad_norm": 21.491534830197576,
      "learning_rate": 8.626040344195233e-06,
      "loss": 1.223,
      "step": 6115
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.02363809501699,
      "learning_rate": 8.633093525179856e-06,
      "loss": 1.2377,
      "step": 6120
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.420006385311563,
      "learning_rate": 8.640146706164481e-06,
      "loss": 1.2609,
      "step": 6125
    },
    {
      "epoch": 0.26,
      "grad_norm": 20.282643488172756,
      "learning_rate": 8.647199887149105e-06,
      "loss": 1.2317,
      "step": 6130
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.959388297459531,
      "learning_rate": 8.654253068133729e-06,
      "loss": 1.1877,
      "step": 6135
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.17308093008255,
      "learning_rate": 8.661306249118354e-06,
      "loss": 1.2313,
      "step": 6140
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.94757800224602,
      "learning_rate": 8.668359430102977e-06,
      "loss": 1.282,
      "step": 6145
    },
    {
      "epoch": 0.26,
      "grad_norm": 21.881167665155324,
      "learning_rate": 8.675412611087601e-06,
      "loss": 1.213,
      "step": 6150
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.648062439059323,
      "learning_rate": 8.682465792072225e-06,
      "loss": 1.2579,
      "step": 6155
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.385133852783163,
      "learning_rate": 8.68951897305685e-06,
      "loss": 1.1988,
      "step": 6160
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.878996002885085,
      "learning_rate": 8.696572154041474e-06,
      "loss": 1.2769,
      "step": 6165
    },
    {
      "epoch": 0.26,
      "grad_norm": 24.433149853698712,
      "learning_rate": 8.703625335026097e-06,
      "loss": 1.2139,
      "step": 6170
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.331256218984516,
      "learning_rate": 8.71067851601072e-06,
      "loss": 1.2681,
      "step": 6175
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.511879388422308,
      "learning_rate": 8.717731696995344e-06,
      "loss": 1.2452,
      "step": 6180
    },
    {
      "epoch": 0.26,
      "grad_norm": 17.812720243099903,
      "learning_rate": 8.72478487797997e-06,
      "loss": 1.229,
      "step": 6185
    },
    {
      "epoch": 0.26,
      "grad_norm": 16.770009248289686,
      "learning_rate": 8.731838058964593e-06,
      "loss": 1.247,
      "step": 6190
    },
    {
      "epoch": 0.26,
      "grad_norm": 57.829209743218975,
      "learning_rate": 8.738891239949217e-06,
      "loss": 1.3041,
      "step": 6195
    },
    {
      "epoch": 0.26,
      "grad_norm": 99.97774675141999,
      "learning_rate": 8.745944420933842e-06,
      "loss": 1.1974,
      "step": 6200
    },
    {
      "epoch": 0.26,
      "grad_norm": 54.45871800251744,
      "learning_rate": 8.752997601918466e-06,
      "loss": 1.2489,
      "step": 6205
    },
    {
      "epoch": 0.26,
      "grad_norm": 25.189597602922817,
      "learning_rate": 8.76005078290309e-06,
      "loss": 1.2836,
      "step": 6210
    },
    {
      "epoch": 0.26,
      "grad_norm": 20.13169430046746,
      "learning_rate": 8.767103963887715e-06,
      "loss": 1.2246,
      "step": 6215
    },
    {
      "epoch": 0.26,
      "grad_norm": 73.49461140527075,
      "learning_rate": 8.774157144872338e-06,
      "loss": 1.2681,
      "step": 6220
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.636753707659049,
      "learning_rate": 8.781210325856962e-06,
      "loss": 1.2503,
      "step": 6225
    },
    {
      "epoch": 0.26,
      "grad_norm": 13.662445957353112,
      "learning_rate": 8.788263506841587e-06,
      "loss": 1.1943,
      "step": 6230
    },
    {
      "epoch": 0.26,
      "grad_norm": 12.197452669599823,
      "learning_rate": 8.79531668782621e-06,
      "loss": 1.2352,
      "step": 6235
    },
    {
      "epoch": 0.26,
      "grad_norm": 19.297528218532758,
      "learning_rate": 8.802369868810834e-06,
      "loss": 1.2723,
      "step": 6240
    },
    {
      "epoch": 0.26,
      "grad_norm": 20.29766624538617,
      "learning_rate": 8.809423049795458e-06,
      "loss": 1.287,
      "step": 6245
    },
    {
      "epoch": 0.26,
      "grad_norm": 9.211322098722844,
      "learning_rate": 8.816476230780082e-06,
      "loss": 1.2602,
      "step": 6250
    },
    {
      "epoch": 0.26,
      "grad_norm": 41.46066578954416,
      "learning_rate": 8.823529411764707e-06,
      "loss": 1.2413,
      "step": 6255
    },
    {
      "epoch": 0.26,
      "grad_norm": 8.084909680587566,
      "learning_rate": 8.83058259274933e-06,
      "loss": 1.2291,
      "step": 6260
    },
    {
      "epoch": 0.27,
      "grad_norm": 18.44768238199645,
      "learning_rate": 8.837635773733954e-06,
      "loss": 1.2437,
      "step": 6265
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.24746135596397,
      "learning_rate": 8.84468895471858e-06,
      "loss": 1.2853,
      "step": 6270
    },
    {
      "epoch": 0.27,
      "grad_norm": 14.996544223196262,
      "learning_rate": 8.851742135703203e-06,
      "loss": 1.2332,
      "step": 6275
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.337923811698634,
      "learning_rate": 8.858795316687826e-06,
      "loss": 1.2634,
      "step": 6280
    },
    {
      "epoch": 0.27,
      "grad_norm": 26.86281771116737,
      "learning_rate": 8.865848497672452e-06,
      "loss": 1.2536,
      "step": 6285
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.650889930117474,
      "learning_rate": 8.872901678657075e-06,
      "loss": 1.237,
      "step": 6290
    },
    {
      "epoch": 0.27,
      "grad_norm": 15.607836824309675,
      "learning_rate": 8.879954859641699e-06,
      "loss": 1.2362,
      "step": 6295
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.340547701232726,
      "learning_rate": 8.887008040626324e-06,
      "loss": 1.231,
      "step": 6300
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.975445501193548,
      "learning_rate": 8.894061221610946e-06,
      "loss": 1.2056,
      "step": 6305
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.705302544141029,
      "learning_rate": 8.901114402595571e-06,
      "loss": 1.2842,
      "step": 6310
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.47465157592828,
      "learning_rate": 8.908167583580195e-06,
      "loss": 1.1873,
      "step": 6315
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.636478594612388,
      "learning_rate": 8.915220764564819e-06,
      "loss": 1.3166,
      "step": 6320
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.77143050385025,
      "learning_rate": 8.922273945549444e-06,
      "loss": 1.2491,
      "step": 6325
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.06744959236745,
      "learning_rate": 8.929327126534068e-06,
      "loss": 1.2689,
      "step": 6330
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.405857344209824,
      "learning_rate": 8.936380307518691e-06,
      "loss": 1.2329,
      "step": 6335
    },
    {
      "epoch": 0.27,
      "grad_norm": 37.69327845881934,
      "learning_rate": 8.943433488503316e-06,
      "loss": 1.2392,
      "step": 6340
    },
    {
      "epoch": 0.27,
      "grad_norm": 24.45051235208879,
      "learning_rate": 8.95048666948794e-06,
      "loss": 1.3396,
      "step": 6345
    },
    {
      "epoch": 0.27,
      "grad_norm": 13.613444741263459,
      "learning_rate": 8.957539850472564e-06,
      "loss": 1.2743,
      "step": 6350
    },
    {
      "epoch": 0.27,
      "grad_norm": 22.628575415059263,
      "learning_rate": 8.964593031457189e-06,
      "loss": 1.2392,
      "step": 6355
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.993092754354002,
      "learning_rate": 8.971646212441813e-06,
      "loss": 1.2434,
      "step": 6360
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.008685196231513,
      "learning_rate": 8.978699393426436e-06,
      "loss": 1.2424,
      "step": 6365
    },
    {
      "epoch": 0.27,
      "grad_norm": 22.11892486800647,
      "learning_rate": 8.98575257441106e-06,
      "loss": 1.2522,
      "step": 6370
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.695346758875903,
      "learning_rate": 8.992805755395683e-06,
      "loss": 1.2533,
      "step": 6375
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.153682954502635,
      "learning_rate": 8.999858936380309e-06,
      "loss": 1.2763,
      "step": 6380
    },
    {
      "epoch": 0.27,
      "grad_norm": 29.32547480642583,
      "learning_rate": 9.006912117364932e-06,
      "loss": 1.2429,
      "step": 6385
    },
    {
      "epoch": 0.27,
      "grad_norm": 56.28021316586544,
      "learning_rate": 9.013965298349556e-06,
      "loss": 1.2663,
      "step": 6390
    },
    {
      "epoch": 0.27,
      "grad_norm": 16.2498082014171,
      "learning_rate": 9.021018479334181e-06,
      "loss": 1.2698,
      "step": 6395
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.103811062616357,
      "learning_rate": 9.028071660318805e-06,
      "loss": 1.2241,
      "step": 6400
    },
    {
      "epoch": 0.27,
      "grad_norm": 22.002297844588433,
      "learning_rate": 9.035124841303428e-06,
      "loss": 1.2484,
      "step": 6405
    },
    {
      "epoch": 0.27,
      "grad_norm": 45.690699292059094,
      "learning_rate": 9.042178022288054e-06,
      "loss": 1.2834,
      "step": 6410
    },
    {
      "epoch": 0.27,
      "grad_norm": 24.57463607698267,
      "learning_rate": 9.049231203272677e-06,
      "loss": 1.2319,
      "step": 6415
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.18168321404906,
      "learning_rate": 9.0562843842573e-06,
      "loss": 1.2147,
      "step": 6420
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.25467353498429,
      "learning_rate": 9.063337565241926e-06,
      "loss": 1.2572,
      "step": 6425
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.442951622927744,
      "learning_rate": 9.07039074622655e-06,
      "loss": 1.2408,
      "step": 6430
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.571936240030306,
      "learning_rate": 9.077443927211173e-06,
      "loss": 1.228,
      "step": 6435
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.258119211497204,
      "learning_rate": 9.084497108195797e-06,
      "loss": 1.2615,
      "step": 6440
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.547620224390553,
      "learning_rate": 9.09155028918042e-06,
      "loss": 1.2115,
      "step": 6445
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.738589655965189,
      "learning_rate": 9.098603470165046e-06,
      "loss": 1.2314,
      "step": 6450
    },
    {
      "epoch": 0.27,
      "grad_norm": 12.209345287249919,
      "learning_rate": 9.10565665114967e-06,
      "loss": 1.2512,
      "step": 6455
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.23917268726962,
      "learning_rate": 9.112709832134293e-06,
      "loss": 1.2604,
      "step": 6460
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.885138688466684,
      "learning_rate": 9.119763013118918e-06,
      "loss": 1.1988,
      "step": 6465
    },
    {
      "epoch": 0.27,
      "grad_norm": 20.387843405529964,
      "learning_rate": 9.126816194103542e-06,
      "loss": 1.2722,
      "step": 6470
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.078784511863649,
      "learning_rate": 9.133869375088165e-06,
      "loss": 1.2903,
      "step": 6475
    },
    {
      "epoch": 0.27,
      "grad_norm": 40.183169101161546,
      "learning_rate": 9.14092255607279e-06,
      "loss": 1.2208,
      "step": 6480
    },
    {
      "epoch": 0.27,
      "grad_norm": 19.525544665540576,
      "learning_rate": 9.147975737057414e-06,
      "loss": 1.2695,
      "step": 6485
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.910153420949271,
      "learning_rate": 9.155028918042038e-06,
      "loss": 1.21,
      "step": 6490
    },
    {
      "epoch": 0.27,
      "grad_norm": 39.21260531852875,
      "learning_rate": 9.162082099026662e-06,
      "loss": 1.2776,
      "step": 6495
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.92815337795953,
      "learning_rate": 9.169135280011285e-06,
      "loss": 1.2124,
      "step": 6500
    },
    {
      "epoch": 0.28,
      "grad_norm": 26.03417346214362,
      "learning_rate": 9.176188460995909e-06,
      "loss": 1.2389,
      "step": 6505
    },
    {
      "epoch": 0.28,
      "grad_norm": 99.72050497541237,
      "learning_rate": 9.183241641980534e-06,
      "loss": 1.243,
      "step": 6510
    },
    {
      "epoch": 0.28,
      "grad_norm": 46.92661924822619,
      "learning_rate": 9.190294822965158e-06,
      "loss": 1.2116,
      "step": 6515
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.603076940015615,
      "learning_rate": 9.197348003949781e-06,
      "loss": 1.2536,
      "step": 6520
    },
    {
      "epoch": 0.28,
      "grad_norm": 40.0576205211251,
      "learning_rate": 9.204401184934406e-06,
      "loss": 1.2216,
      "step": 6525
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.887788606086115,
      "learning_rate": 9.21145436591903e-06,
      "loss": 1.2492,
      "step": 6530
    },
    {
      "epoch": 0.28,
      "grad_norm": 20.05273742866196,
      "learning_rate": 9.218507546903654e-06,
      "loss": 1.2406,
      "step": 6535
    },
    {
      "epoch": 0.28,
      "grad_norm": 22.823627519107824,
      "learning_rate": 9.225560727888279e-06,
      "loss": 1.2376,
      "step": 6540
    },
    {
      "epoch": 0.28,
      "grad_norm": 13.041483215640545,
      "learning_rate": 9.232613908872903e-06,
      "loss": 1.269,
      "step": 6545
    },
    {
      "epoch": 0.28,
      "grad_norm": 34.13540556399942,
      "learning_rate": 9.239667089857526e-06,
      "loss": 1.2383,
      "step": 6550
    },
    {
      "epoch": 0.28,
      "grad_norm": 18.320458508306157,
      "learning_rate": 9.246720270842151e-06,
      "loss": 1.2276,
      "step": 6555
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.5330072443733656,
      "learning_rate": 9.253773451826773e-06,
      "loss": 1.2875,
      "step": 6560
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.27703167706361,
      "learning_rate": 9.260826632811399e-06,
      "loss": 1.2494,
      "step": 6565
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.531046672259434,
      "learning_rate": 9.267879813796022e-06,
      "loss": 1.2304,
      "step": 6570
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.248786361276295,
      "learning_rate": 9.274932994780646e-06,
      "loss": 1.2394,
      "step": 6575
    },
    {
      "epoch": 0.28,
      "grad_norm": 27.714492410526507,
      "learning_rate": 9.281986175765271e-06,
      "loss": 1.2613,
      "step": 6580
    },
    {
      "epoch": 0.28,
      "grad_norm": 19.51501165108399,
      "learning_rate": 9.289039356749895e-06,
      "loss": 1.2815,
      "step": 6585
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.436359038800738,
      "learning_rate": 9.296092537734518e-06,
      "loss": 1.2282,
      "step": 6590
    },
    {
      "epoch": 0.28,
      "grad_norm": 43.23269701575712,
      "learning_rate": 9.303145718719144e-06,
      "loss": 1.2262,
      "step": 6595
    },
    {
      "epoch": 0.28,
      "grad_norm": 21.101540281939712,
      "learning_rate": 9.310198899703767e-06,
      "loss": 1.2463,
      "step": 6600
    },
    {
      "epoch": 0.28,
      "grad_norm": 34.49422984799425,
      "learning_rate": 9.31725208068839e-06,
      "loss": 1.2238,
      "step": 6605
    },
    {
      "epoch": 0.28,
      "grad_norm": 33.850133561648185,
      "learning_rate": 9.324305261673016e-06,
      "loss": 1.2259,
      "step": 6610
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.220414096605742,
      "learning_rate": 9.33135844265764e-06,
      "loss": 1.26,
      "step": 6615
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.9514579009943,
      "learning_rate": 9.338411623642263e-06,
      "loss": 1.2266,
      "step": 6620
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.419869255961212,
      "learning_rate": 9.345464804626887e-06,
      "loss": 1.2271,
      "step": 6625
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.086626967310414,
      "learning_rate": 9.35251798561151e-06,
      "loss": 1.226,
      "step": 6630
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.506884833581381,
      "learning_rate": 9.359571166596136e-06,
      "loss": 1.2283,
      "step": 6635
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.91163949574279,
      "learning_rate": 9.36662434758076e-06,
      "loss": 1.2172,
      "step": 6640
    },
    {
      "epoch": 0.28,
      "grad_norm": 12.803823993791173,
      "learning_rate": 9.373677528565383e-06,
      "loss": 1.2227,
      "step": 6645
    },
    {
      "epoch": 0.28,
      "grad_norm": 32.0430165125269,
      "learning_rate": 9.380730709550008e-06,
      "loss": 1.2504,
      "step": 6650
    },
    {
      "epoch": 0.28,
      "grad_norm": 44.66390322780334,
      "learning_rate": 9.387783890534632e-06,
      "loss": 1.252,
      "step": 6655
    },
    {
      "epoch": 0.28,
      "grad_norm": 97.49006407157172,
      "learning_rate": 9.394837071519255e-06,
      "loss": 1.2577,
      "step": 6660
    },
    {
      "epoch": 0.28,
      "grad_norm": 62.61288477401787,
      "learning_rate": 9.40189025250388e-06,
      "loss": 1.2364,
      "step": 6665
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.024011085526986,
      "learning_rate": 9.408943433488504e-06,
      "loss": 1.3067,
      "step": 6670
    },
    {
      "epoch": 0.28,
      "grad_norm": 69.01029904561533,
      "learning_rate": 9.415996614473128e-06,
      "loss": 1.3036,
      "step": 6675
    },
    {
      "epoch": 0.28,
      "grad_norm": 25.084797032255793,
      "learning_rate": 9.423049795457753e-06,
      "loss": 1.2838,
      "step": 6680
    },
    {
      "epoch": 0.28,
      "grad_norm": 19.60869619954398,
      "learning_rate": 9.430102976442377e-06,
      "loss": 1.2856,
      "step": 6685
    },
    {
      "epoch": 0.28,
      "grad_norm": 11.290648170649316,
      "learning_rate": 9.437156157427e-06,
      "loss": 1.2291,
      "step": 6690
    },
    {
      "epoch": 0.28,
      "grad_norm": 15.222827943551566,
      "learning_rate": 9.444209338411624e-06,
      "loss": 1.2397,
      "step": 6695
    },
    {
      "epoch": 0.28,
      "grad_norm": 9.133795897181006,
      "learning_rate": 9.451262519396248e-06,
      "loss": 1.2603,
      "step": 6700
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.681778696999336,
      "learning_rate": 9.458315700380873e-06,
      "loss": 1.2379,
      "step": 6705
    },
    {
      "epoch": 0.28,
      "grad_norm": 10.5031107852674,
      "learning_rate": 9.465368881365497e-06,
      "loss": 1.2541,
      "step": 6710
    },
    {
      "epoch": 0.28,
      "grad_norm": 14.839698285967577,
      "learning_rate": 9.47242206235012e-06,
      "loss": 1.2514,
      "step": 6715
    },
    {
      "epoch": 0.28,
      "grad_norm": 8.061437442300123,
      "learning_rate": 9.479475243334745e-06,
      "loss": 1.2607,
      "step": 6720
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.432480326890641,
      "learning_rate": 9.486528424319369e-06,
      "loss": 1.2898,
      "step": 6725
    },
    {
      "epoch": 0.28,
      "grad_norm": 26.456548063118635,
      "learning_rate": 9.493581605303993e-06,
      "loss": 1.2432,
      "step": 6730
    },
    {
      "epoch": 0.29,
      "grad_norm": 36.9215479602442,
      "learning_rate": 9.500634786288618e-06,
      "loss": 1.2933,
      "step": 6735
    },
    {
      "epoch": 0.29,
      "grad_norm": 27.30788433252015,
      "learning_rate": 9.507687967273242e-06,
      "loss": 1.2425,
      "step": 6740
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.314027848854106,
      "learning_rate": 9.514741148257865e-06,
      "loss": 1.2306,
      "step": 6745
    },
    {
      "epoch": 0.29,
      "grad_norm": 21.18878334837518,
      "learning_rate": 9.521794329242489e-06,
      "loss": 1.2251,
      "step": 6750
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.69161743226239,
      "learning_rate": 9.528847510227112e-06,
      "loss": 1.2971,
      "step": 6755
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.398504422452305,
      "learning_rate": 9.535900691211738e-06,
      "loss": 1.2514,
      "step": 6760
    },
    {
      "epoch": 0.29,
      "grad_norm": 25.03698505170366,
      "learning_rate": 9.542953872196361e-06,
      "loss": 1.2264,
      "step": 6765
    },
    {
      "epoch": 0.29,
      "grad_norm": 24.830659858829385,
      "learning_rate": 9.550007053180985e-06,
      "loss": 1.2423,
      "step": 6770
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.721940880879972,
      "learning_rate": 9.55706023416561e-06,
      "loss": 1.2323,
      "step": 6775
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.333025291327758,
      "learning_rate": 9.564113415150234e-06,
      "loss": 1.2237,
      "step": 6780
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.070519039905475,
      "learning_rate": 9.571166596134857e-06,
      "loss": 1.2337,
      "step": 6785
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.17048396112688,
      "learning_rate": 9.578219777119483e-06,
      "loss": 1.2148,
      "step": 6790
    },
    {
      "epoch": 0.29,
      "grad_norm": 33.68604246681523,
      "learning_rate": 9.585272958104106e-06,
      "loss": 1.2104,
      "step": 6795
    },
    {
      "epoch": 0.29,
      "grad_norm": 27.06048167618181,
      "learning_rate": 9.59232613908873e-06,
      "loss": 1.2669,
      "step": 6800
    },
    {
      "epoch": 0.29,
      "grad_norm": 51.53117728762783,
      "learning_rate": 9.599379320073355e-06,
      "loss": 1.2563,
      "step": 6805
    },
    {
      "epoch": 0.29,
      "grad_norm": 25.84955784474549,
      "learning_rate": 9.606432501057979e-06,
      "loss": 1.2569,
      "step": 6810
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.361937831104246,
      "learning_rate": 9.613485682042602e-06,
      "loss": 1.2629,
      "step": 6815
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.58280208568006,
      "learning_rate": 9.620538863027226e-06,
      "loss": 1.2569,
      "step": 6820
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.147141882700138,
      "learning_rate": 9.62759204401185e-06,
      "loss": 1.2467,
      "step": 6825
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.276443836964162,
      "learning_rate": 9.634645224996475e-06,
      "loss": 1.227,
      "step": 6830
    },
    {
      "epoch": 0.29,
      "grad_norm": 38.07613603311421,
      "learning_rate": 9.641698405981098e-06,
      "loss": 1.2474,
      "step": 6835
    },
    {
      "epoch": 0.29,
      "grad_norm": 17.687385407167028,
      "learning_rate": 9.648751586965722e-06,
      "loss": 1.2672,
      "step": 6840
    },
    {
      "epoch": 0.29,
      "grad_norm": 10.298038992959794,
      "learning_rate": 9.655804767950346e-06,
      "loss": 1.2562,
      "step": 6845
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.810686984475629,
      "learning_rate": 9.66285794893497e-06,
      "loss": 1.2536,
      "step": 6850
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.555461508206447,
      "learning_rate": 9.669911129919594e-06,
      "loss": 1.2054,
      "step": 6855
    },
    {
      "epoch": 0.29,
      "grad_norm": 25.506303203520943,
      "learning_rate": 9.676964310904218e-06,
      "loss": 1.2778,
      "step": 6860
    },
    {
      "epoch": 0.29,
      "grad_norm": 50.132857462998636,
      "learning_rate": 9.684017491888843e-06,
      "loss": 1.2548,
      "step": 6865
    },
    {
      "epoch": 0.29,
      "grad_norm": 84.18993846114763,
      "learning_rate": 9.691070672873467e-06,
      "loss": 1.2724,
      "step": 6870
    },
    {
      "epoch": 0.29,
      "grad_norm": 48.67721445772831,
      "learning_rate": 9.69812385385809e-06,
      "loss": 1.2141,
      "step": 6875
    },
    {
      "epoch": 0.29,
      "grad_norm": 15.120591634256032,
      "learning_rate": 9.705177034842714e-06,
      "loss": 1.2837,
      "step": 6880
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.91538361254449,
      "learning_rate": 9.712230215827338e-06,
      "loss": 1.2584,
      "step": 6885
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.264430773104124,
      "learning_rate": 9.719283396811963e-06,
      "loss": 1.226,
      "step": 6890
    },
    {
      "epoch": 0.29,
      "grad_norm": 37.88382063845113,
      "learning_rate": 9.726336577796587e-06,
      "loss": 1.2702,
      "step": 6895
    },
    {
      "epoch": 0.29,
      "grad_norm": 19.170282497488508,
      "learning_rate": 9.73338975878121e-06,
      "loss": 1.3045,
      "step": 6900
    },
    {
      "epoch": 0.29,
      "grad_norm": 29.493663417607124,
      "learning_rate": 9.740442939765835e-06,
      "loss": 1.2242,
      "step": 6905
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.667086924220484,
      "learning_rate": 9.747496120750459e-06,
      "loss": 1.2449,
      "step": 6910
    },
    {
      "epoch": 0.29,
      "grad_norm": 68.56140151164675,
      "learning_rate": 9.754549301735083e-06,
      "loss": 1.2189,
      "step": 6915
    },
    {
      "epoch": 0.29,
      "grad_norm": 29.96503198593971,
      "learning_rate": 9.761602482719708e-06,
      "loss": 1.2171,
      "step": 6920
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.155145918203711,
      "learning_rate": 9.768655663704332e-06,
      "loss": 1.2007,
      "step": 6925
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.511859808137694,
      "learning_rate": 9.775708844688955e-06,
      "loss": 1.2025,
      "step": 6930
    },
    {
      "epoch": 0.29,
      "grad_norm": 20.271716498027313,
      "learning_rate": 9.78276202567358e-06,
      "loss": 1.2113,
      "step": 6935
    },
    {
      "epoch": 0.29,
      "grad_norm": 8.139895279217205,
      "learning_rate": 9.789815206658204e-06,
      "loss": 1.2641,
      "step": 6940
    },
    {
      "epoch": 0.29,
      "grad_norm": 40.850804477581036,
      "learning_rate": 9.796868387642828e-06,
      "loss": 1.2336,
      "step": 6945
    },
    {
      "epoch": 0.29,
      "grad_norm": 30.815248966199405,
      "learning_rate": 9.803921568627451e-06,
      "loss": 1.2198,
      "step": 6950
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.216597164118205,
      "learning_rate": 9.810974749612075e-06,
      "loss": 1.2363,
      "step": 6955
    },
    {
      "epoch": 0.29,
      "grad_norm": 32.232897698220725,
      "learning_rate": 9.8180279305967e-06,
      "loss": 1.2591,
      "step": 6960
    },
    {
      "epoch": 0.29,
      "grad_norm": 13.372140403537237,
      "learning_rate": 9.825081111581324e-06,
      "loss": 1.2617,
      "step": 6965
    },
    {
      "epoch": 0.29,
      "grad_norm": 27.69909019104548,
      "learning_rate": 9.832134292565947e-06,
      "loss": 1.2243,
      "step": 6970
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.394667773664086,
      "learning_rate": 9.839187473550573e-06,
      "loss": 1.2555,
      "step": 6975
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.234862274671976,
      "learning_rate": 9.846240654535196e-06,
      "loss": 1.2151,
      "step": 6980
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.849116326056897,
      "learning_rate": 9.85329383551982e-06,
      "loss": 1.2567,
      "step": 6985
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.259288433082045,
      "learning_rate": 9.860347016504445e-06,
      "loss": 1.262,
      "step": 6990
    },
    {
      "epoch": 0.3,
      "grad_norm": 18.285717134687246,
      "learning_rate": 9.867400197489069e-06,
      "loss": 1.2321,
      "step": 6995
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.882948609293603,
      "learning_rate": 9.874453378473692e-06,
      "loss": 1.244,
      "step": 7000
    },
    {
      "epoch": 0.3,
      "grad_norm": 18.44038198421586,
      "learning_rate": 9.881506559458318e-06,
      "loss": 1.2736,
      "step": 7005
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.523441913220164,
      "learning_rate": 9.88855974044294e-06,
      "loss": 1.2524,
      "step": 7010
    },
    {
      "epoch": 0.3,
      "grad_norm": 13.735473518728714,
      "learning_rate": 9.895612921427565e-06,
      "loss": 1.2293,
      "step": 7015
    },
    {
      "epoch": 0.3,
      "grad_norm": 26.071146100547974,
      "learning_rate": 9.902666102412188e-06,
      "loss": 1.2245,
      "step": 7020
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.670824286415508,
      "learning_rate": 9.909719283396812e-06,
      "loss": 1.2643,
      "step": 7025
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.696896219798152,
      "learning_rate": 9.916772464381437e-06,
      "loss": 1.2103,
      "step": 7030
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.637212164904719,
      "learning_rate": 9.923825645366061e-06,
      "loss": 1.2176,
      "step": 7035
    },
    {
      "epoch": 0.3,
      "grad_norm": 23.71924414263371,
      "learning_rate": 9.930878826350684e-06,
      "loss": 1.2756,
      "step": 7040
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.989907121354511,
      "learning_rate": 9.93793200733531e-06,
      "loss": 1.2984,
      "step": 7045
    },
    {
      "epoch": 0.3,
      "grad_norm": 16.661405452546745,
      "learning_rate": 9.944985188319933e-06,
      "loss": 1.2339,
      "step": 7050
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.547809511768321,
      "learning_rate": 9.952038369304557e-06,
      "loss": 1.2349,
      "step": 7055
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.34676097748933,
      "learning_rate": 9.959091550289182e-06,
      "loss": 1.2088,
      "step": 7060
    },
    {
      "epoch": 0.3,
      "grad_norm": 14.030020061469806,
      "learning_rate": 9.966144731273806e-06,
      "loss": 1.2487,
      "step": 7065
    },
    {
      "epoch": 0.3,
      "grad_norm": 33.3367943143273,
      "learning_rate": 9.97319791225843e-06,
      "loss": 1.1999,
      "step": 7070
    },
    {
      "epoch": 0.3,
      "grad_norm": 27.162683049854188,
      "learning_rate": 9.980251093243053e-06,
      "loss": 1.2396,
      "step": 7075
    },
    {
      "epoch": 0.3,
      "grad_norm": 12.318519946093282,
      "learning_rate": 9.987304274227677e-06,
      "loss": 1.2475,
      "step": 7080
    },
    {
      "epoch": 0.3,
      "grad_norm": 55.24829758647621,
      "learning_rate": 9.994357455212302e-06,
      "loss": 1.2372,
      "step": 7085
    },
    {
      "epoch": 0.3,
      "grad_norm": 23.118377040536743,
      "learning_rate": 9.999999993937295e-06,
      "loss": 1.2352,
      "step": 7090
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.567172442993417,
      "learning_rate": 9.999999781742606e-06,
      "loss": 1.2776,
      "step": 7095
    },
    {
      "epoch": 0.3,
      "grad_norm": 28.114851435665756,
      "learning_rate": 9.999999266412656e-06,
      "loss": 1.2927,
      "step": 7100
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.013488713382067,
      "learning_rate": 9.99999844794748e-06,
      "loss": 1.2322,
      "step": 7105
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.806948014231962,
      "learning_rate": 9.999997326347127e-06,
      "loss": 1.2195,
      "step": 7110
    },
    {
      "epoch": 0.3,
      "grad_norm": 11.482976873118554,
      "learning_rate": 9.999995901611661e-06,
      "loss": 1.2068,
      "step": 7115
    },
    {
      "epoch": 0.3,
      "grad_norm": 31.26450098534542,
      "learning_rate": 9.999994173741175e-06,
      "loss": 1.251,
      "step": 7120
    },
    {
      "epoch": 0.3,
      "grad_norm": 14.927637286752303,
      "learning_rate": 9.999992142735769e-06,
      "loss": 1.1983,
      "step": 7125
    },
    {
      "epoch": 0.3,
      "grad_norm": 21.77653752243219,
      "learning_rate": 9.999989808595565e-06,
      "loss": 1.2704,
      "step": 7130
    },
    {
      "epoch": 0.3,
      "grad_norm": 29.279723439352683,
      "learning_rate": 9.999987171320711e-06,
      "loss": 1.2205,
      "step": 7135
    },
    {
      "epoch": 0.3,
      "grad_norm": 29.929199323968046,
      "learning_rate": 9.99998423091136e-06,
      "loss": 1.2348,
      "step": 7140
    },
    {
      "epoch": 0.3,
      "grad_norm": 70.22256448811632,
      "learning_rate": 9.999980987367693e-06,
      "loss": 1.2314,
      "step": 7145
    },
    {
      "epoch": 0.3,
      "grad_norm": 95.94222776095704,
      "learning_rate": 9.999977440689907e-06,
      "loss": 1.2624,
      "step": 7150
    },
    {
      "epoch": 0.3,
      "grad_norm": 87.03008705004844,
      "learning_rate": 9.999973590878217e-06,
      "loss": 1.304,
      "step": 7155
    },
    {
      "epoch": 0.3,
      "grad_norm": 65.70503905454055,
      "learning_rate": 9.999969437932857e-06,
      "loss": 1.2935,
      "step": 7160
    },
    {
      "epoch": 0.3,
      "grad_norm": 58.053975816463925,
      "learning_rate": 9.999964981854076e-06,
      "loss": 1.5134,
      "step": 7165
    },
    {
      "epoch": 0.3,
      "grad_norm": 39.86573643074013,
      "learning_rate": 9.999960222642147e-06,
      "loss": 1.3608,
      "step": 7170
    },
    {
      "epoch": 0.3,
      "grad_norm": 30.920617348566125,
      "learning_rate": 9.999955160297358e-06,
      "loss": 1.347,
      "step": 7175
    },
    {
      "epoch": 0.3,
      "grad_norm": 24.510902413567575,
      "learning_rate": 9.999949794820015e-06,
      "loss": 1.2376,
      "step": 7180
    },
    {
      "epoch": 0.3,
      "grad_norm": 45.94314202948226,
      "learning_rate": 9.999944126210443e-06,
      "loss": 1.2942,
      "step": 7185
    },
    {
      "epoch": 0.3,
      "grad_norm": 40.64924461179615,
      "learning_rate": 9.999938154468988e-06,
      "loss": 1.3315,
      "step": 7190
    },
    {
      "epoch": 0.3,
      "grad_norm": 17.441152486298893,
      "learning_rate": 9.99993187959601e-06,
      "loss": 1.3502,
      "step": 7195
    },
    {
      "epoch": 0.3,
      "grad_norm": 70.72250142919358,
      "learning_rate": 9.999925301591889e-06,
      "loss": 1.2887,
      "step": 7200
    },
    {
      "epoch": 0.3,
      "grad_norm": 85.83398730521338,
      "learning_rate": 9.999918420457026e-06,
      "loss": 1.3308,
      "step": 7205
    },
    {
      "epoch": 0.31,
      "grad_norm": 30.257110268519835,
      "learning_rate": 9.999911236191835e-06,
      "loss": 1.265,
      "step": 7210
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.828095931794895,
      "learning_rate": 9.999903748796755e-06,
      "loss": 1.2408,
      "step": 7215
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.21232869967946,
      "learning_rate": 9.99989595827224e-06,
      "loss": 1.2553,
      "step": 7220
    },
    {
      "epoch": 0.31,
      "grad_norm": 31.846923253910173,
      "learning_rate": 9.999887864618757e-06,
      "loss": 1.266,
      "step": 7225
    },
    {
      "epoch": 0.31,
      "grad_norm": 40.571439949018696,
      "learning_rate": 9.999879467836802e-06,
      "loss": 1.2926,
      "step": 7230
    },
    {
      "epoch": 0.31,
      "grad_norm": 79.6937695566078,
      "learning_rate": 9.999870767926882e-06,
      "loss": 1.3066,
      "step": 7235
    },
    {
      "epoch": 0.31,
      "grad_norm": 57.81566886552214,
      "learning_rate": 9.999861764889527e-06,
      "loss": 1.2911,
      "step": 7240
    },
    {
      "epoch": 0.31,
      "grad_norm": 97.85933247019047,
      "learning_rate": 9.99985245872528e-06,
      "loss": 1.257,
      "step": 7245
    },
    {
      "epoch": 0.31,
      "grad_norm": 58.23713217890289,
      "learning_rate": 9.999842849434704e-06,
      "loss": 1.2629,
      "step": 7250
    },
    {
      "epoch": 0.31,
      "grad_norm": 50.65625672662297,
      "learning_rate": 9.999832937018385e-06,
      "loss": 1.2899,
      "step": 7255
    },
    {
      "epoch": 0.31,
      "grad_norm": 34.63620857818771,
      "learning_rate": 9.999822721476924e-06,
      "loss": 1.2584,
      "step": 7260
    },
    {
      "epoch": 0.31,
      "grad_norm": 59.04136237806621,
      "learning_rate": 9.999812202810936e-06,
      "loss": 1.4571,
      "step": 7265
    },
    {
      "epoch": 0.31,
      "grad_norm": 40.152131744650944,
      "learning_rate": 9.999801381021064e-06,
      "loss": 1.48,
      "step": 7270
    },
    {
      "epoch": 0.31,
      "grad_norm": 167.66359659883003,
      "learning_rate": 9.99979025610796e-06,
      "loss": 1.5042,
      "step": 7275
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.792388454188853,
      "learning_rate": 9.999778828072301e-06,
      "loss": 1.37,
      "step": 7280
    },
    {
      "epoch": 0.31,
      "grad_norm": 63.373408611411264,
      "learning_rate": 9.999767096914779e-06,
      "loss": 1.3056,
      "step": 7285
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.42701285172961,
      "learning_rate": 9.999755062636105e-06,
      "loss": 1.4225,
      "step": 7290
    },
    {
      "epoch": 0.31,
      "grad_norm": 26.463401732432942,
      "learning_rate": 9.999742725237007e-06,
      "loss": 1.3252,
      "step": 7295
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.546278624607,
      "learning_rate": 9.999730084718237e-06,
      "loss": 1.2829,
      "step": 7300
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.439375097900484,
      "learning_rate": 9.999717141080561e-06,
      "loss": 1.3155,
      "step": 7305
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.437065581949568,
      "learning_rate": 9.999703894324758e-06,
      "loss": 1.2805,
      "step": 7310
    },
    {
      "epoch": 0.31,
      "grad_norm": 19.46991340240531,
      "learning_rate": 9.999690344451636e-06,
      "loss": 1.245,
      "step": 7315
    },
    {
      "epoch": 0.31,
      "grad_norm": 23.870439760675467,
      "learning_rate": 9.999676491462017e-06,
      "loss": 1.31,
      "step": 7320
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.004614503633006,
      "learning_rate": 9.999662335356739e-06,
      "loss": 1.2967,
      "step": 7325
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.86470357855451,
      "learning_rate": 9.99964787613666e-06,
      "loss": 1.2743,
      "step": 7330
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.191662355901187,
      "learning_rate": 9.999633113802658e-06,
      "loss": 1.3193,
      "step": 7335
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.873228836838706,
      "learning_rate": 9.999618048355626e-06,
      "loss": 1.301,
      "step": 7340
    },
    {
      "epoch": 0.31,
      "grad_norm": 38.18032122958042,
      "learning_rate": 9.999602679796479e-06,
      "loss": 1.2095,
      "step": 7345
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.96525312990652,
      "learning_rate": 9.999587008126149e-06,
      "loss": 1.1979,
      "step": 7350
    },
    {
      "epoch": 0.31,
      "grad_norm": 40.087840048579,
      "learning_rate": 9.999571033345584e-06,
      "loss": 1.2875,
      "step": 7355
    },
    {
      "epoch": 0.31,
      "grad_norm": 17.576188453115833,
      "learning_rate": 9.999554755455754e-06,
      "loss": 1.2333,
      "step": 7360
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.562209904492508,
      "learning_rate": 9.999538174457647e-06,
      "loss": 1.2304,
      "step": 7365
    },
    {
      "epoch": 0.31,
      "grad_norm": 33.890982837458935,
      "learning_rate": 9.999521290352268e-06,
      "loss": 1.255,
      "step": 7370
    },
    {
      "epoch": 0.31,
      "grad_norm": 33.44197930525397,
      "learning_rate": 9.999504103140638e-06,
      "loss": 1.284,
      "step": 7375
    },
    {
      "epoch": 0.31,
      "grad_norm": 13.273275265151184,
      "learning_rate": 9.9994866128238e-06,
      "loss": 1.2505,
      "step": 7380
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.328347103549904,
      "learning_rate": 9.999468819402818e-06,
      "loss": 1.31,
      "step": 7385
    },
    {
      "epoch": 0.31,
      "grad_norm": 36.641863082702564,
      "learning_rate": 9.999450722878766e-06,
      "loss": 1.2526,
      "step": 7390
    },
    {
      "epoch": 0.31,
      "grad_norm": 20.00053943239536,
      "learning_rate": 9.999432323252742e-06,
      "loss": 1.2419,
      "step": 7395
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.422848996931565,
      "learning_rate": 9.999413620525865e-06,
      "loss": 1.2642,
      "step": 7400
    },
    {
      "epoch": 0.31,
      "grad_norm": 44.88626897977361,
      "learning_rate": 9.999394614699265e-06,
      "loss": 1.2653,
      "step": 7405
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.444568888006865,
      "learning_rate": 9.999375305774096e-06,
      "loss": 1.2753,
      "step": 7410
    },
    {
      "epoch": 0.31,
      "grad_norm": 22.598004780732158,
      "learning_rate": 9.999355693751529e-06,
      "loss": 1.223,
      "step": 7415
    },
    {
      "epoch": 0.31,
      "grad_norm": 15.225997803410483,
      "learning_rate": 9.99933577863275e-06,
      "loss": 1.2429,
      "step": 7420
    },
    {
      "epoch": 0.31,
      "grad_norm": 31.59763094890729,
      "learning_rate": 9.999315560418972e-06,
      "loss": 1.2213,
      "step": 7425
    },
    {
      "epoch": 0.31,
      "grad_norm": 45.27902333938972,
      "learning_rate": 9.999295039111417e-06,
      "loss": 1.2645,
      "step": 7430
    },
    {
      "epoch": 0.31,
      "grad_norm": 16.295561266548244,
      "learning_rate": 9.999274214711327e-06,
      "loss": 1.2021,
      "step": 7435
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.520466979433348,
      "learning_rate": 9.99925308721997e-06,
      "loss": 1.2561,
      "step": 7440
    },
    {
      "epoch": 0.32,
      "grad_norm": 26.513562267392324,
      "learning_rate": 9.999231656638623e-06,
      "loss": 1.2444,
      "step": 7445
    },
    {
      "epoch": 0.32,
      "grad_norm": 22.962308581453126,
      "learning_rate": 9.999209922968587e-06,
      "loss": 1.2082,
      "step": 7450
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.499960610839619,
      "learning_rate": 9.999187886211177e-06,
      "loss": 1.2607,
      "step": 7455
    },
    {
      "epoch": 0.32,
      "grad_norm": 19.160465456910075,
      "learning_rate": 9.999165546367734e-06,
      "loss": 1.2092,
      "step": 7460
    },
    {
      "epoch": 0.32,
      "grad_norm": 15.42316580287671,
      "learning_rate": 9.999142903439608e-06,
      "loss": 1.2618,
      "step": 7465
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.162399948110279,
      "learning_rate": 9.999119957428172e-06,
      "loss": 1.262,
      "step": 7470
    },
    {
      "epoch": 0.32,
      "grad_norm": 15.732472202398695,
      "learning_rate": 9.99909670833482e-06,
      "loss": 1.262,
      "step": 7475
    },
    {
      "epoch": 0.32,
      "grad_norm": 33.46859983735862,
      "learning_rate": 9.99907315616096e-06,
      "loss": 1.2333,
      "step": 7480
    },
    {
      "epoch": 0.32,
      "grad_norm": 68.6028957957856,
      "learning_rate": 9.999049300908018e-06,
      "loss": 1.3109,
      "step": 7485
    },
    {
      "epoch": 0.32,
      "grad_norm": 35.18566845440983,
      "learning_rate": 9.999025142577443e-06,
      "loss": 1.2191,
      "step": 7490
    },
    {
      "epoch": 0.32,
      "grad_norm": 22.031380148984464,
      "learning_rate": 9.999000681170701e-06,
      "loss": 1.2188,
      "step": 7495
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.583683859556693,
      "learning_rate": 9.99897591668927e-06,
      "loss": 1.2341,
      "step": 7500
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.399575555971131,
      "learning_rate": 9.998950849134655e-06,
      "loss": 1.2638,
      "step": 7505
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.681108162959296,
      "learning_rate": 9.998925478508375e-06,
      "loss": 1.2617,
      "step": 7510
    },
    {
      "epoch": 0.32,
      "grad_norm": 26.73927786581154,
      "learning_rate": 9.998899804811967e-06,
      "loss": 1.23,
      "step": 7515
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.357775940658529,
      "learning_rate": 9.99887382804699e-06,
      "loss": 1.2488,
      "step": 7520
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.655827071683177,
      "learning_rate": 9.998847548215018e-06,
      "loss": 1.2971,
      "step": 7525
    },
    {
      "epoch": 0.32,
      "grad_norm": 26.12113740009643,
      "learning_rate": 9.998820965317642e-06,
      "loss": 1.2357,
      "step": 7530
    },
    {
      "epoch": 0.32,
      "grad_norm": 34.18938653487381,
      "learning_rate": 9.998794079356476e-06,
      "loss": 1.2019,
      "step": 7535
    },
    {
      "epoch": 0.32,
      "grad_norm": 29.794034792221073,
      "learning_rate": 9.99876689033315e-06,
      "loss": 1.2942,
      "step": 7540
    },
    {
      "epoch": 0.32,
      "grad_norm": 37.92169362942798,
      "learning_rate": 9.998739398249311e-06,
      "loss": 1.2463,
      "step": 7545
    },
    {
      "epoch": 0.32,
      "grad_norm": 41.69638564952318,
      "learning_rate": 9.998711603106625e-06,
      "loss": 1.3201,
      "step": 7550
    },
    {
      "epoch": 0.32,
      "grad_norm": 21.89090509019165,
      "learning_rate": 9.998683504906782e-06,
      "loss": 1.2507,
      "step": 7555
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.62595707000461,
      "learning_rate": 9.99865510365148e-06,
      "loss": 1.2628,
      "step": 7560
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.22148233153354,
      "learning_rate": 9.998626399342444e-06,
      "loss": 1.2531,
      "step": 7565
    },
    {
      "epoch": 0.32,
      "grad_norm": 27.631231675971897,
      "learning_rate": 9.998597391981413e-06,
      "loss": 1.1924,
      "step": 7570
    },
    {
      "epoch": 0.32,
      "grad_norm": 17.810600021103443,
      "learning_rate": 9.998568081570146e-06,
      "loss": 1.2559,
      "step": 7575
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.322156600940671,
      "learning_rate": 9.99853846811042e-06,
      "loss": 1.1985,
      "step": 7580
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.057831224628954,
      "learning_rate": 9.998508551604031e-06,
      "loss": 1.2874,
      "step": 7585
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.836707416794145,
      "learning_rate": 9.998478332052793e-06,
      "loss": 1.2336,
      "step": 7590
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.913961140074016,
      "learning_rate": 9.998447809458533e-06,
      "loss": 1.2407,
      "step": 7595
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.185825281546505,
      "learning_rate": 9.99841698382311e-06,
      "loss": 1.189,
      "step": 7600
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.03689758383576,
      "learning_rate": 9.998385855148387e-06,
      "loss": 1.2225,
      "step": 7605
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.959434158162542,
      "learning_rate": 9.998354423436253e-06,
      "loss": 1.2405,
      "step": 7610
    },
    {
      "epoch": 0.32,
      "grad_norm": 9.437909546331637,
      "learning_rate": 9.998322688688614e-06,
      "loss": 1.207,
      "step": 7615
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.984877873962511,
      "learning_rate": 9.998290650907391e-06,
      "loss": 1.3317,
      "step": 7620
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.750241980292008,
      "learning_rate": 9.998258310094532e-06,
      "loss": 1.2167,
      "step": 7625
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.79780561274728,
      "learning_rate": 9.998225666251993e-06,
      "loss": 1.2334,
      "step": 7630
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.21279698053489,
      "learning_rate": 9.998192719381753e-06,
      "loss": 1.285,
      "step": 7635
    },
    {
      "epoch": 0.32,
      "grad_norm": 10.791513369959016,
      "learning_rate": 9.998159469485814e-06,
      "loss": 1.2154,
      "step": 7640
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.51909210772539,
      "learning_rate": 9.998125916566189e-06,
      "loss": 1.2389,
      "step": 7645
    },
    {
      "epoch": 0.32,
      "grad_norm": 14.751255139905355,
      "learning_rate": 9.99809206062491e-06,
      "loss": 1.2967,
      "step": 7650
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.810064123766226,
      "learning_rate": 9.998057901664032e-06,
      "loss": 1.2417,
      "step": 7655
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.120699512188413,
      "learning_rate": 9.998023439685627e-06,
      "loss": 1.277,
      "step": 7660
    },
    {
      "epoch": 0.32,
      "grad_norm": 19.53438067734915,
      "learning_rate": 9.997988674691782e-06,
      "loss": 1.2577,
      "step": 7665
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.689096741715723,
      "learning_rate": 9.997953606684605e-06,
      "loss": 1.1871,
      "step": 7670
    },
    {
      "epoch": 0.32,
      "grad_norm": 16.15410418789472,
      "learning_rate": 9.997918235666223e-06,
      "loss": 1.2454,
      "step": 7675
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.6757528872797325,
      "learning_rate": 9.997882561638782e-06,
      "loss": 1.2168,
      "step": 7680
    },
    {
      "epoch": 0.33,
      "grad_norm": 19.133715988011993,
      "learning_rate": 9.997846584604443e-06,
      "loss": 1.1902,
      "step": 7685
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.679947110622978,
      "learning_rate": 9.997810304565386e-06,
      "loss": 1.2614,
      "step": 7690
    },
    {
      "epoch": 0.33,
      "grad_norm": 23.004587892593033,
      "learning_rate": 9.997773721523812e-06,
      "loss": 1.2344,
      "step": 7695
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.67682874432702,
      "learning_rate": 9.99773683548194e-06,
      "loss": 1.2305,
      "step": 7700
    },
    {
      "epoch": 0.33,
      "grad_norm": 18.400094449302642,
      "learning_rate": 9.997699646442003e-06,
      "loss": 1.2357,
      "step": 7705
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.49392548640079,
      "learning_rate": 9.997662154406258e-06,
      "loss": 1.2657,
      "step": 7710
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.736775222487934,
      "learning_rate": 9.997624359376979e-06,
      "loss": 1.2515,
      "step": 7715
    },
    {
      "epoch": 0.33,
      "grad_norm": 41.4303926787153,
      "learning_rate": 9.997586261356453e-06,
      "loss": 1.2132,
      "step": 7720
    },
    {
      "epoch": 0.33,
      "grad_norm": 35.184748895730245,
      "learning_rate": 9.997547860346996e-06,
      "loss": 1.221,
      "step": 7725
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.457499390191062,
      "learning_rate": 9.997509156350932e-06,
      "loss": 1.2248,
      "step": 7730
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.796548136163155,
      "learning_rate": 9.997470149370608e-06,
      "loss": 1.2587,
      "step": 7735
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.546352192140306,
      "learning_rate": 9.99743083940839e-06,
      "loss": 1.2349,
      "step": 7740
    },
    {
      "epoch": 0.33,
      "grad_norm": 20.295144935838564,
      "learning_rate": 9.99739122646666e-06,
      "loss": 1.1899,
      "step": 7745
    },
    {
      "epoch": 0.33,
      "grad_norm": 53.787368747896565,
      "learning_rate": 9.997351310547821e-06,
      "loss": 1.2089,
      "step": 7750
    },
    {
      "epoch": 0.33,
      "grad_norm": 40.22827895925454,
      "learning_rate": 9.997311091654291e-06,
      "loss": 1.2261,
      "step": 7755
    },
    {
      "epoch": 0.33,
      "grad_norm": 10.179337496482024,
      "learning_rate": 9.997270569788511e-06,
      "loss": 1.228,
      "step": 7760
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.965323017882522,
      "learning_rate": 9.997229744952937e-06,
      "loss": 1.223,
      "step": 7765
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.503594652631213,
      "learning_rate": 9.99718861715004e-06,
      "loss": 1.2266,
      "step": 7770
    },
    {
      "epoch": 0.33,
      "grad_norm": 41.67889440230999,
      "learning_rate": 9.99714718638232e-06,
      "loss": 1.2406,
      "step": 7775
    },
    {
      "epoch": 0.33,
      "grad_norm": 23.892405951309872,
      "learning_rate": 9.997105452652284e-06,
      "loss": 1.2496,
      "step": 7780
    },
    {
      "epoch": 0.33,
      "grad_norm": 63.42668530241874,
      "learning_rate": 9.997063415962464e-06,
      "loss": 1.2372,
      "step": 7785
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.343590782148018,
      "learning_rate": 9.99702107631541e-06,
      "loss": 1.2494,
      "step": 7790
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.999735564065194,
      "learning_rate": 9.996978433713687e-06,
      "loss": 1.2866,
      "step": 7795
    },
    {
      "epoch": 0.33,
      "grad_norm": 19.826309999008483,
      "learning_rate": 9.99693548815988e-06,
      "loss": 1.2461,
      "step": 7800
    },
    {
      "epoch": 0.33,
      "grad_norm": 19.750264485329023,
      "learning_rate": 9.996892239656592e-06,
      "loss": 1.2406,
      "step": 7805
    },
    {
      "epoch": 0.33,
      "grad_norm": 19.688627552278987,
      "learning_rate": 9.996848688206448e-06,
      "loss": 1.23,
      "step": 7810
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.938342338364622,
      "learning_rate": 9.996804833812086e-06,
      "loss": 1.239,
      "step": 7815
    },
    {
      "epoch": 0.33,
      "grad_norm": 38.06803747943537,
      "learning_rate": 9.996760676476168e-06,
      "loss": 1.2259,
      "step": 7820
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.5285251319783,
      "learning_rate": 9.996716216201366e-06,
      "loss": 1.2307,
      "step": 7825
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.94350367479322,
      "learning_rate": 9.996671452990377e-06,
      "loss": 1.2223,
      "step": 7830
    },
    {
      "epoch": 0.33,
      "grad_norm": 8.698492823605658,
      "learning_rate": 9.996626386845918e-06,
      "loss": 1.272,
      "step": 7835
    },
    {
      "epoch": 0.33,
      "grad_norm": 30.432536493106333,
      "learning_rate": 9.996581017770719e-06,
      "loss": 1.2543,
      "step": 7840
    },
    {
      "epoch": 0.33,
      "grad_norm": 27.97645657966109,
      "learning_rate": 9.99653534576753e-06,
      "loss": 1.2245,
      "step": 7845
    },
    {
      "epoch": 0.33,
      "grad_norm": 84.9303648895969,
      "learning_rate": 9.996489370839122e-06,
      "loss": 1.2887,
      "step": 7850
    },
    {
      "epoch": 0.33,
      "grad_norm": 101.70790412364232,
      "learning_rate": 9.996443092988279e-06,
      "loss": 1.2718,
      "step": 7855
    },
    {
      "epoch": 0.33,
      "grad_norm": 74.92552930292959,
      "learning_rate": 9.99639651221781e-06,
      "loss": 1.2506,
      "step": 7860
    },
    {
      "epoch": 0.33,
      "grad_norm": 14.316343216911543,
      "learning_rate": 9.99634962853054e-06,
      "loss": 1.2369,
      "step": 7865
    },
    {
      "epoch": 0.33,
      "grad_norm": 24.512540844754735,
      "learning_rate": 9.996302441929306e-06,
      "loss": 1.235,
      "step": 7870
    },
    {
      "epoch": 0.33,
      "grad_norm": 26.647148749658335,
      "learning_rate": 9.996254952416974e-06,
      "loss": 1.2123,
      "step": 7875
    },
    {
      "epoch": 0.33,
      "grad_norm": 12.573684349444694,
      "learning_rate": 9.996207159996421e-06,
      "loss": 1.2301,
      "step": 7880
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.08187471953139,
      "learning_rate": 9.996159064670542e-06,
      "loss": 1.2543,
      "step": 7885
    },
    {
      "epoch": 0.33,
      "grad_norm": 32.892617578869405,
      "learning_rate": 9.99611066644226e-06,
      "loss": 1.2378,
      "step": 7890
    },
    {
      "epoch": 0.33,
      "grad_norm": 17.88049211720188,
      "learning_rate": 9.996061965314502e-06,
      "loss": 1.2068,
      "step": 7895
    },
    {
      "epoch": 0.33,
      "grad_norm": 21.080132073202652,
      "learning_rate": 9.996012961290223e-06,
      "loss": 1.2323,
      "step": 7900
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.198564632839893,
      "learning_rate": 9.995963654372396e-06,
      "loss": 1.1807,
      "step": 7905
    },
    {
      "epoch": 0.33,
      "grad_norm": 13.01419295448973,
      "learning_rate": 9.995914044564008e-06,
      "loss": 1.2347,
      "step": 7910
    },
    {
      "epoch": 0.33,
      "grad_norm": 29.497398275690852,
      "learning_rate": 9.995864131868066e-06,
      "loss": 1.2257,
      "step": 7915
    },
    {
      "epoch": 0.34,
      "grad_norm": 33.43508479103748,
      "learning_rate": 9.995813916287602e-06,
      "loss": 1.224,
      "step": 7920
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.52845846510623,
      "learning_rate": 9.995763397825651e-06,
      "loss": 1.2573,
      "step": 7925
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.311686554471457,
      "learning_rate": 9.995712576485284e-06,
      "loss": 1.2407,
      "step": 7930
    },
    {
      "epoch": 0.34,
      "grad_norm": 19.7741420666964,
      "learning_rate": 9.995661452269578e-06,
      "loss": 1.2345,
      "step": 7935
    },
    {
      "epoch": 0.34,
      "grad_norm": 40.743896167918365,
      "learning_rate": 9.995610025181633e-06,
      "loss": 1.2415,
      "step": 7940
    },
    {
      "epoch": 0.34,
      "grad_norm": 25.302561478764066,
      "learning_rate": 9.995558295224565e-06,
      "loss": 1.2255,
      "step": 7945
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.209111688247388,
      "learning_rate": 9.995506262401516e-06,
      "loss": 1.3095,
      "step": 7950
    },
    {
      "epoch": 0.34,
      "grad_norm": 40.61841968723452,
      "learning_rate": 9.995453926715635e-06,
      "loss": 1.2154,
      "step": 7955
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.519057812630436,
      "learning_rate": 9.995401288170097e-06,
      "loss": 1.2436,
      "step": 7960
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.248821001314777,
      "learning_rate": 9.995348346768094e-06,
      "loss": 1.1808,
      "step": 7965
    },
    {
      "epoch": 0.34,
      "grad_norm": 28.084413250987303,
      "learning_rate": 9.995295102512837e-06,
      "loss": 1.2527,
      "step": 7970
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.836964539629219,
      "learning_rate": 9.99524155540755e-06,
      "loss": 1.2256,
      "step": 7975
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.832120488861237,
      "learning_rate": 9.99518770545548e-06,
      "loss": 1.241,
      "step": 7980
    },
    {
      "epoch": 0.34,
      "grad_norm": 19.17927685882691,
      "learning_rate": 9.995133552659895e-06,
      "loss": 1.2742,
      "step": 7985
    },
    {
      "epoch": 0.34,
      "grad_norm": 15.558874917828359,
      "learning_rate": 9.995079097024076e-06,
      "loss": 1.2293,
      "step": 7990
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.357318209950067,
      "learning_rate": 9.995024338551324e-06,
      "loss": 1.2094,
      "step": 7995
    },
    {
      "epoch": 0.34,
      "grad_norm": 63.034609308669275,
      "learning_rate": 9.994969277244961e-06,
      "loss": 1.2543,
      "step": 8000
    },
    {
      "epoch": 0.34,
      "grad_norm": 20.097186013491743,
      "learning_rate": 9.994913913108324e-06,
      "loss": 1.2389,
      "step": 8005
    },
    {
      "epoch": 0.34,
      "grad_norm": 25.499347342565777,
      "learning_rate": 9.994858246144769e-06,
      "loss": 1.2486,
      "step": 8010
    },
    {
      "epoch": 0.34,
      "grad_norm": 80.8675075405855,
      "learning_rate": 9.99480227635767e-06,
      "loss": 1.2531,
      "step": 8015
    },
    {
      "epoch": 0.34,
      "grad_norm": 81.82780626138764,
      "learning_rate": 9.994746003750423e-06,
      "loss": 1.2832,
      "step": 8020
    },
    {
      "epoch": 0.34,
      "grad_norm": 71.51347532104299,
      "learning_rate": 9.994689428326438e-06,
      "loss": 1.2762,
      "step": 8025
    },
    {
      "epoch": 0.34,
      "grad_norm": 94.67499035587554,
      "learning_rate": 9.994632550089147e-06,
      "loss": 1.2469,
      "step": 8030
    },
    {
      "epoch": 0.34,
      "grad_norm": 54.31907865475567,
      "learning_rate": 9.994575369041995e-06,
      "loss": 1.3078,
      "step": 8035
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.760119873262017,
      "learning_rate": 9.99451788518845e-06,
      "loss": 1.2339,
      "step": 8040
    },
    {
      "epoch": 0.34,
      "grad_norm": 14.417887412121337,
      "learning_rate": 9.994460098531998e-06,
      "loss": 1.2521,
      "step": 8045
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.271854470515674,
      "learning_rate": 9.994402009076144e-06,
      "loss": 1.2073,
      "step": 8050
    },
    {
      "epoch": 0.34,
      "grad_norm": 9.833875269160144,
      "learning_rate": 9.994343616824404e-06,
      "loss": 1.2184,
      "step": 8055
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.445511082474821,
      "learning_rate": 9.994284921780323e-06,
      "loss": 1.1979,
      "step": 8060
    },
    {
      "epoch": 0.34,
      "grad_norm": 20.19483393932203,
      "learning_rate": 9.994225923947459e-06,
      "loss": 1.2111,
      "step": 8065
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.2307867214878,
      "learning_rate": 9.99416662332939e-06,
      "loss": 1.2233,
      "step": 8070
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.346104187003878,
      "learning_rate": 9.994107019929707e-06,
      "loss": 1.1857,
      "step": 8075
    },
    {
      "epoch": 0.34,
      "grad_norm": 16.273289567960617,
      "learning_rate": 9.994047113752025e-06,
      "loss": 1.1966,
      "step": 8080
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.741119461132842,
      "learning_rate": 9.993986904799979e-06,
      "loss": 1.2129,
      "step": 8085
    },
    {
      "epoch": 0.34,
      "grad_norm": 19.90653974587955,
      "learning_rate": 9.993926393077215e-06,
      "loss": 1.2397,
      "step": 8090
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.282798225737693,
      "learning_rate": 9.993865578587407e-06,
      "loss": 1.2103,
      "step": 8095
    },
    {
      "epoch": 0.34,
      "grad_norm": 21.24182674626366,
      "learning_rate": 9.993804461334238e-06,
      "loss": 1.1706,
      "step": 8100
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.099869396552126,
      "learning_rate": 9.993743041321413e-06,
      "loss": 1.1946,
      "step": 8105
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.8129641406443,
      "learning_rate": 9.993681318552658e-06,
      "loss": 1.244,
      "step": 8110
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.37912288019264,
      "learning_rate": 9.993619293031713e-06,
      "loss": 1.194,
      "step": 8115
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.201576798859058,
      "learning_rate": 9.99355696476234e-06,
      "loss": 1.2422,
      "step": 8120
    },
    {
      "epoch": 0.34,
      "grad_norm": 34.121740170383944,
      "learning_rate": 9.993494333748317e-06,
      "loss": 1.2311,
      "step": 8125
    },
    {
      "epoch": 0.34,
      "grad_norm": 23.12775019310754,
      "learning_rate": 9.993431399993442e-06,
      "loss": 1.2201,
      "step": 8130
    },
    {
      "epoch": 0.34,
      "grad_norm": 19.871244630607208,
      "learning_rate": 9.993368163501529e-06,
      "loss": 1.1514,
      "step": 8135
    },
    {
      "epoch": 0.34,
      "grad_norm": 37.6939611663003,
      "learning_rate": 9.993304624276413e-06,
      "loss": 1.1888,
      "step": 8140
    },
    {
      "epoch": 0.34,
      "grad_norm": 31.48436670499356,
      "learning_rate": 9.993240782321946e-06,
      "loss": 1.2134,
      "step": 8145
    },
    {
      "epoch": 0.34,
      "grad_norm": 12.000254758597286,
      "learning_rate": 9.993176637641999e-06,
      "loss": 1.2127,
      "step": 8150
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.115236506025425,
      "learning_rate": 9.99311219024046e-06,
      "loss": 1.2451,
      "step": 8155
    },
    {
      "epoch": 0.35,
      "grad_norm": 26.95413463578466,
      "learning_rate": 9.993047440121236e-06,
      "loss": 1.1916,
      "step": 8160
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.110935308606182,
      "learning_rate": 9.992982387288254e-06,
      "loss": 1.2311,
      "step": 8165
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.544790635763175,
      "learning_rate": 9.992917031745457e-06,
      "loss": 1.2719,
      "step": 8170
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.279036810480022,
      "learning_rate": 9.992851373496808e-06,
      "loss": 1.1894,
      "step": 8175
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.789545001376956,
      "learning_rate": 9.992785412546287e-06,
      "loss": 1.2384,
      "step": 8180
    },
    {
      "epoch": 0.35,
      "grad_norm": 17.02853229262546,
      "learning_rate": 9.992719148897891e-06,
      "loss": 1.1855,
      "step": 8185
    },
    {
      "epoch": 0.35,
      "grad_norm": 36.68130672151206,
      "learning_rate": 9.992652582555643e-06,
      "loss": 1.2494,
      "step": 8190
    },
    {
      "epoch": 0.35,
      "grad_norm": 40.171663041857016,
      "learning_rate": 9.992585713523572e-06,
      "loss": 1.2288,
      "step": 8195
    },
    {
      "epoch": 0.35,
      "grad_norm": 78.18049083936823,
      "learning_rate": 9.992518541805737e-06,
      "loss": 1.2246,
      "step": 8200
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.209966268972378,
      "learning_rate": 9.992451067406208e-06,
      "loss": 1.2702,
      "step": 8205
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.833679272598818,
      "learning_rate": 9.992383290329077e-06,
      "loss": 1.2214,
      "step": 8210
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.623403413698795,
      "learning_rate": 9.992315210578454e-06,
      "loss": 1.1954,
      "step": 8215
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.918535797297617,
      "learning_rate": 9.992246828158463e-06,
      "loss": 1.2429,
      "step": 8220
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.954019510201876,
      "learning_rate": 9.992178143073252e-06,
      "loss": 1.1988,
      "step": 8225
    },
    {
      "epoch": 0.35,
      "grad_norm": 24.854442925528698,
      "learning_rate": 9.992109155326986e-06,
      "loss": 1.215,
      "step": 8230
    },
    {
      "epoch": 0.35,
      "grad_norm": 22.89155681525993,
      "learning_rate": 9.992039864923849e-06,
      "loss": 1.2506,
      "step": 8235
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.781146440290884,
      "learning_rate": 9.991970271868035e-06,
      "loss": 1.1868,
      "step": 8240
    },
    {
      "epoch": 0.35,
      "grad_norm": 28.355165901493866,
      "learning_rate": 9.99190037616377e-06,
      "loss": 1.2444,
      "step": 8245
    },
    {
      "epoch": 0.35,
      "grad_norm": 30.469499153565415,
      "learning_rate": 9.991830177815289e-06,
      "loss": 1.2607,
      "step": 8250
    },
    {
      "epoch": 0.35,
      "grad_norm": 33.17248520538483,
      "learning_rate": 9.99175967682685e-06,
      "loss": 1.2455,
      "step": 8255
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.863038883572782,
      "learning_rate": 9.991688873202723e-06,
      "loss": 1.1944,
      "step": 8260
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.309650321051647,
      "learning_rate": 9.991617766947204e-06,
      "loss": 1.2785,
      "step": 8265
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.524302448390912,
      "learning_rate": 9.991546358064603e-06,
      "loss": 1.2365,
      "step": 8270
    },
    {
      "epoch": 0.35,
      "grad_norm": 19.728092271346227,
      "learning_rate": 9.991474646559249e-06,
      "loss": 1.2437,
      "step": 8275
    },
    {
      "epoch": 0.35,
      "grad_norm": 59.56060337780727,
      "learning_rate": 9.991402632435491e-06,
      "loss": 1.2216,
      "step": 8280
    },
    {
      "epoch": 0.35,
      "grad_norm": 9.849854878122546,
      "learning_rate": 9.991330315697693e-06,
      "loss": 1.2205,
      "step": 8285
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.29344131643094,
      "learning_rate": 9.991257696350242e-06,
      "loss": 1.2128,
      "step": 8290
    },
    {
      "epoch": 0.35,
      "grad_norm": 15.632113433715542,
      "learning_rate": 9.991184774397537e-06,
      "loss": 1.2315,
      "step": 8295
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.313737404668368,
      "learning_rate": 9.991111549844004e-06,
      "loss": 1.2292,
      "step": 8300
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.953074968542056,
      "learning_rate": 9.991038022694078e-06,
      "loss": 1.1862,
      "step": 8305
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.73974789227461,
      "learning_rate": 9.990964192952218e-06,
      "loss": 1.2926,
      "step": 8310
    },
    {
      "epoch": 0.35,
      "grad_norm": 10.908010535863633,
      "learning_rate": 9.990890060622899e-06,
      "loss": 1.1757,
      "step": 8315
    },
    {
      "epoch": 0.35,
      "grad_norm": 16.05993049729101,
      "learning_rate": 9.99081562571062e-06,
      "loss": 1.2032,
      "step": 8320
    },
    {
      "epoch": 0.35,
      "grad_norm": 30.869534435483125,
      "learning_rate": 9.990740888219887e-06,
      "loss": 1.278,
      "step": 8325
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.363204753364154,
      "learning_rate": 9.990665848155237e-06,
      "loss": 1.1965,
      "step": 8330
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.506123380321823,
      "learning_rate": 9.990590505521216e-06,
      "loss": 1.1889,
      "step": 8335
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.322006757393218,
      "learning_rate": 9.990514860322393e-06,
      "loss": 1.2443,
      "step": 8340
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.952070178029388,
      "learning_rate": 9.990438912563355e-06,
      "loss": 1.2335,
      "step": 8345
    },
    {
      "epoch": 0.35,
      "grad_norm": 14.990567241234745,
      "learning_rate": 9.990362662248705e-06,
      "loss": 1.2414,
      "step": 8350
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.105396826159597,
      "learning_rate": 9.990286109383067e-06,
      "loss": 1.2282,
      "step": 8355
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.70209767655266,
      "learning_rate": 9.990209253971082e-06,
      "loss": 1.2345,
      "step": 8360
    },
    {
      "epoch": 0.35,
      "grad_norm": 12.36698546637828,
      "learning_rate": 9.990132096017407e-06,
      "loss": 1.2684,
      "step": 8365
    },
    {
      "epoch": 0.35,
      "grad_norm": 13.583087022136393,
      "learning_rate": 9.990054635526722e-06,
      "loss": 1.2101,
      "step": 8370
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.895110158292534,
      "learning_rate": 9.989976872503725e-06,
      "loss": 1.2425,
      "step": 8375
    },
    {
      "epoch": 0.35,
      "grad_norm": 8.050796838688562,
      "learning_rate": 9.989898806953127e-06,
      "loss": 1.2478,
      "step": 8380
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.869059538645431,
      "learning_rate": 9.989820438879664e-06,
      "loss": 1.2392,
      "step": 8385
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.896020847728801,
      "learning_rate": 9.989741768288086e-06,
      "loss": 1.1862,
      "step": 8390
    },
    {
      "epoch": 0.36,
      "grad_norm": 45.703456980106715,
      "learning_rate": 9.989662795183162e-06,
      "loss": 1.242,
      "step": 8395
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.579828700821357,
      "learning_rate": 9.98958351956968e-06,
      "loss": 1.1891,
      "step": 8400
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.481324903792737,
      "learning_rate": 9.989503941452445e-06,
      "loss": 1.21,
      "step": 8405
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.583278438929653,
      "learning_rate": 9.989424060836283e-06,
      "loss": 1.2205,
      "step": 8410
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.376906541576698,
      "learning_rate": 9.98934387772604e-06,
      "loss": 1.1888,
      "step": 8415
    },
    {
      "epoch": 0.36,
      "grad_norm": 10.57658585583126,
      "learning_rate": 9.989263392126573e-06,
      "loss": 1.2269,
      "step": 8420
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.275333470481575,
      "learning_rate": 9.989182604042762e-06,
      "loss": 1.242,
      "step": 8425
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.11764500076583,
      "learning_rate": 9.989101513479506e-06,
      "loss": 1.2365,
      "step": 8430
    },
    {
      "epoch": 0.36,
      "grad_norm": 9.452450309467196,
      "learning_rate": 9.98902012044172e-06,
      "loss": 1.2123,
      "step": 8435
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.149069983795446,
      "learning_rate": 9.988938424934343e-06,
      "loss": 1.2663,
      "step": 8440
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.659271865325053,
      "learning_rate": 9.988856426962324e-06,
      "loss": 1.2146,
      "step": 8445
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.20109071245686,
      "learning_rate": 9.988774126530634e-06,
      "loss": 1.2251,
      "step": 8450
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.207621549082202,
      "learning_rate": 9.988691523644262e-06,
      "loss": 1.2343,
      "step": 8455
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.82954071691952,
      "learning_rate": 9.98860861830822e-06,
      "loss": 1.2121,
      "step": 8460
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.376664299695347,
      "learning_rate": 9.988525410527531e-06,
      "loss": 1.1781,
      "step": 8465
    },
    {
      "epoch": 0.36,
      "grad_norm": 49.39244173055254,
      "learning_rate": 9.988441900307242e-06,
      "loss": 1.2269,
      "step": 8470
    },
    {
      "epoch": 0.36,
      "grad_norm": 50.8504410856827,
      "learning_rate": 9.988358087652412e-06,
      "loss": 1.2337,
      "step": 8475
    },
    {
      "epoch": 0.36,
      "grad_norm": 39.10687162104513,
      "learning_rate": 9.988273972568126e-06,
      "loss": 1.1752,
      "step": 8480
    },
    {
      "epoch": 0.36,
      "grad_norm": 67.31066961645065,
      "learning_rate": 9.988189555059483e-06,
      "loss": 1.2588,
      "step": 8485
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.69441032988168,
      "learning_rate": 9.988104835131598e-06,
      "loss": 1.1815,
      "step": 8490
    },
    {
      "epoch": 0.36,
      "grad_norm": 33.467110310779006,
      "learning_rate": 9.988019812789613e-06,
      "loss": 1.2216,
      "step": 8495
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.671270848207165,
      "learning_rate": 9.987934488038678e-06,
      "loss": 1.1768,
      "step": 8500
    },
    {
      "epoch": 0.36,
      "grad_norm": 28.150762478072988,
      "learning_rate": 9.987848860883968e-06,
      "loss": 1.2599,
      "step": 8505
    },
    {
      "epoch": 0.36,
      "grad_norm": 74.5275694909008,
      "learning_rate": 9.987762931330673e-06,
      "loss": 1.2056,
      "step": 8510
    },
    {
      "epoch": 0.36,
      "grad_norm": 30.507403480430057,
      "learning_rate": 9.987676699384004e-06,
      "loss": 1.2353,
      "step": 8515
    },
    {
      "epoch": 0.36,
      "grad_norm": 23.543567868742304,
      "learning_rate": 9.987590165049188e-06,
      "loss": 1.2098,
      "step": 8520
    },
    {
      "epoch": 0.36,
      "grad_norm": 25.656673732893854,
      "learning_rate": 9.987503328331472e-06,
      "loss": 1.2859,
      "step": 8525
    },
    {
      "epoch": 0.36,
      "grad_norm": 28.959458074427264,
      "learning_rate": 9.987416189236121e-06,
      "loss": 1.1597,
      "step": 8530
    },
    {
      "epoch": 0.36,
      "grad_norm": 43.86085421227326,
      "learning_rate": 9.987328747768416e-06,
      "loss": 1.2405,
      "step": 8535
    },
    {
      "epoch": 0.36,
      "grad_norm": 32.72468333360079,
      "learning_rate": 9.98724100393366e-06,
      "loss": 1.2069,
      "step": 8540
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.995011286255298,
      "learning_rate": 9.987152957737172e-06,
      "loss": 1.2345,
      "step": 8545
    },
    {
      "epoch": 0.36,
      "grad_norm": 73.06658047429889,
      "learning_rate": 9.987064609184291e-06,
      "loss": 1.2452,
      "step": 8550
    },
    {
      "epoch": 0.36,
      "grad_norm": 24.817118783827976,
      "learning_rate": 9.986975958280373e-06,
      "loss": 1.2164,
      "step": 8555
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.990043689200686,
      "learning_rate": 9.986887005030792e-06,
      "loss": 1.2518,
      "step": 8560
    },
    {
      "epoch": 0.36,
      "grad_norm": 31.572278664978654,
      "learning_rate": 9.986797749440943e-06,
      "loss": 1.2592,
      "step": 8565
    },
    {
      "epoch": 0.36,
      "grad_norm": 19.435482013691885,
      "learning_rate": 9.986708191516232e-06,
      "loss": 1.2424,
      "step": 8570
    },
    {
      "epoch": 0.36,
      "grad_norm": 63.693461961295036,
      "learning_rate": 9.986618331262095e-06,
      "loss": 1.2578,
      "step": 8575
    },
    {
      "epoch": 0.36,
      "grad_norm": 31.446843658547774,
      "learning_rate": 9.986528168683975e-06,
      "loss": 1.1853,
      "step": 8580
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.301533764234394,
      "learning_rate": 9.986437703787342e-06,
      "loss": 1.1933,
      "step": 8585
    },
    {
      "epoch": 0.36,
      "grad_norm": 19.8460881160074,
      "learning_rate": 9.986346936577678e-06,
      "loss": 1.2295,
      "step": 8590
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.55608879962847,
      "learning_rate": 9.986255867060489e-06,
      "loss": 1.1702,
      "step": 8595
    },
    {
      "epoch": 0.36,
      "grad_norm": 44.75645688198931,
      "learning_rate": 9.98616449524129e-06,
      "loss": 1.2348,
      "step": 8600
    },
    {
      "epoch": 0.36,
      "grad_norm": 26.859487581688636,
      "learning_rate": 9.986072821125629e-06,
      "loss": 1.2364,
      "step": 8605
    },
    {
      "epoch": 0.36,
      "grad_norm": 16.91839599023379,
      "learning_rate": 9.985980844719059e-06,
      "loss": 1.2198,
      "step": 8610
    },
    {
      "epoch": 0.36,
      "grad_norm": 26.15389996024574,
      "learning_rate": 9.985888566027156e-06,
      "loss": 1.2076,
      "step": 8615
    },
    {
      "epoch": 0.36,
      "grad_norm": 14.085079912611075,
      "learning_rate": 9.985795985055516e-06,
      "loss": 1.1763,
      "step": 8620
    },
    {
      "epoch": 0.37,
      "grad_norm": 33.792171021963675,
      "learning_rate": 9.985703101809752e-06,
      "loss": 1.2248,
      "step": 8625
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.3260248214026396,
      "learning_rate": 9.985609916295493e-06,
      "loss": 1.2355,
      "step": 8630
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.311401163652134,
      "learning_rate": 9.985516428518393e-06,
      "loss": 1.2196,
      "step": 8635
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.378826713087722,
      "learning_rate": 9.985422638484116e-06,
      "loss": 1.2499,
      "step": 8640
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.541024734387541,
      "learning_rate": 9.98532854619835e-06,
      "loss": 1.2503,
      "step": 8645
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.773454926388277,
      "learning_rate": 9.985234151666797e-06,
      "loss": 1.2345,
      "step": 8650
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.599773351956539,
      "learning_rate": 9.985139454895186e-06,
      "loss": 1.2246,
      "step": 8655
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.615899718378288,
      "learning_rate": 9.98504445588925e-06,
      "loss": 1.1834,
      "step": 8660
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.15254060666548,
      "learning_rate": 9.984949154654756e-06,
      "loss": 1.2356,
      "step": 8665
    },
    {
      "epoch": 0.37,
      "grad_norm": 14.55878618843663,
      "learning_rate": 9.984853551197476e-06,
      "loss": 1.2019,
      "step": 8670
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.53312672145288,
      "learning_rate": 9.984757645523208e-06,
      "loss": 1.2075,
      "step": 8675
    },
    {
      "epoch": 0.37,
      "grad_norm": 14.028101438814463,
      "learning_rate": 9.984661437637768e-06,
      "loss": 1.1971,
      "step": 8680
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.962681134293541,
      "learning_rate": 9.984564927546989e-06,
      "loss": 1.2292,
      "step": 8685
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.105873883293662,
      "learning_rate": 9.98446811525672e-06,
      "loss": 1.2133,
      "step": 8690
    },
    {
      "epoch": 0.37,
      "grad_norm": 34.975382958953205,
      "learning_rate": 9.98437100077283e-06,
      "loss": 1.2186,
      "step": 8695
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.37087709348676,
      "learning_rate": 9.984273584101211e-06,
      "loss": 1.1987,
      "step": 8700
    },
    {
      "epoch": 0.37,
      "grad_norm": 71.08804182456454,
      "learning_rate": 9.984175865247766e-06,
      "loss": 1.1986,
      "step": 8705
    },
    {
      "epoch": 0.37,
      "grad_norm": 52.10398569690377,
      "learning_rate": 9.984077844218416e-06,
      "loss": 1.2244,
      "step": 8710
    },
    {
      "epoch": 0.37,
      "grad_norm": 37.81548927949398,
      "learning_rate": 9.983979521019111e-06,
      "loss": 1.2653,
      "step": 8715
    },
    {
      "epoch": 0.37,
      "grad_norm": 12.85074773890059,
      "learning_rate": 9.983880895655806e-06,
      "loss": 1.2603,
      "step": 8720
    },
    {
      "epoch": 0.37,
      "grad_norm": 32.64581515056232,
      "learning_rate": 9.983781968134485e-06,
      "loss": 1.242,
      "step": 8725
    },
    {
      "epoch": 0.37,
      "grad_norm": 17.099230723476186,
      "learning_rate": 9.983682738461143e-06,
      "loss": 1.2303,
      "step": 8730
    },
    {
      "epoch": 0.37,
      "grad_norm": 36.264254645909595,
      "learning_rate": 9.983583206641796e-06,
      "loss": 1.2389,
      "step": 8735
    },
    {
      "epoch": 0.37,
      "grad_norm": 45.86121139741487,
      "learning_rate": 9.98348337268248e-06,
      "loss": 1.2036,
      "step": 8740
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.538096106752206,
      "learning_rate": 9.983383236589243e-06,
      "loss": 1.2565,
      "step": 8745
    },
    {
      "epoch": 0.37,
      "grad_norm": 33.679354719636336,
      "learning_rate": 9.983282798368162e-06,
      "loss": 1.2806,
      "step": 8750
    },
    {
      "epoch": 0.37,
      "grad_norm": 20.910192795381583,
      "learning_rate": 9.983182058025323e-06,
      "loss": 1.2322,
      "step": 8755
    },
    {
      "epoch": 0.37,
      "grad_norm": 30.370012964051725,
      "learning_rate": 9.983081015566835e-06,
      "loss": 1.2055,
      "step": 8760
    },
    {
      "epoch": 0.37,
      "grad_norm": 29.424674522331955,
      "learning_rate": 9.982979670998823e-06,
      "loss": 1.2074,
      "step": 8765
    },
    {
      "epoch": 0.37,
      "grad_norm": 32.41331231991534,
      "learning_rate": 9.982878024327432e-06,
      "loss": 1.2511,
      "step": 8770
    },
    {
      "epoch": 0.37,
      "grad_norm": 46.05780744909376,
      "learning_rate": 9.982776075558822e-06,
      "loss": 1.2172,
      "step": 8775
    },
    {
      "epoch": 0.37,
      "grad_norm": 74.16296669242696,
      "learning_rate": 9.982673824699178e-06,
      "loss": 1.245,
      "step": 8780
    },
    {
      "epoch": 0.37,
      "grad_norm": 45.764843173008884,
      "learning_rate": 9.982571271754696e-06,
      "loss": 1.2429,
      "step": 8785
    },
    {
      "epoch": 0.37,
      "grad_norm": 53.24053470535527,
      "learning_rate": 9.982468416731595e-06,
      "loss": 1.2552,
      "step": 8790
    },
    {
      "epoch": 0.37,
      "grad_norm": 27.331446935935848,
      "learning_rate": 9.982365259636109e-06,
      "loss": 1.2329,
      "step": 8795
    },
    {
      "epoch": 0.37,
      "grad_norm": 21.411223037335787,
      "learning_rate": 9.982261800474493e-06,
      "loss": 1.2223,
      "step": 8800
    },
    {
      "epoch": 0.37,
      "grad_norm": 17.64017648011196,
      "learning_rate": 9.98215803925302e-06,
      "loss": 1.2085,
      "step": 8805
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.656967648553069,
      "learning_rate": 9.982053975977983e-06,
      "loss": 1.1984,
      "step": 8810
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.252972648379149,
      "learning_rate": 9.981949610655685e-06,
      "loss": 1.2275,
      "step": 8815
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.496446996490132,
      "learning_rate": 9.98184494329246e-06,
      "loss": 1.1854,
      "step": 8820
    },
    {
      "epoch": 0.37,
      "grad_norm": 11.380020725748448,
      "learning_rate": 9.981739973894649e-06,
      "loss": 1.2188,
      "step": 8825
    },
    {
      "epoch": 0.37,
      "grad_norm": 13.801606564301279,
      "learning_rate": 9.981634702468619e-06,
      "loss": 1.179,
      "step": 8830
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.634426159659236,
      "learning_rate": 9.981529129020748e-06,
      "loss": 1.2104,
      "step": 8835
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.448842060217998,
      "learning_rate": 9.981423253557441e-06,
      "loss": 1.2178,
      "step": 8840
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.91483111537622,
      "learning_rate": 9.981317076085116e-06,
      "loss": 1.1853,
      "step": 8845
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.894164928940903,
      "learning_rate": 9.981210596610209e-06,
      "loss": 1.2524,
      "step": 8850
    },
    {
      "epoch": 0.37,
      "grad_norm": 14.56533418155501,
      "learning_rate": 9.981103815139177e-06,
      "loss": 1.2328,
      "step": 8855
    },
    {
      "epoch": 0.37,
      "grad_norm": 23.480236730572294,
      "learning_rate": 9.980996731678492e-06,
      "loss": 1.2229,
      "step": 8860
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.096746190639573,
      "learning_rate": 9.980889346234647e-06,
      "loss": 1.2436,
      "step": 8865
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.167888811625206,
      "learning_rate": 9.980781658814153e-06,
      "loss": 1.1952,
      "step": 8870
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.15643863428365,
      "learning_rate": 9.98067366942354e-06,
      "loss": 1.2394,
      "step": 8875
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.635171481534623,
      "learning_rate": 9.980565378069351e-06,
      "loss": 1.2384,
      "step": 8880
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.9089943342825455,
      "learning_rate": 9.980456784758155e-06,
      "loss": 1.1857,
      "step": 8885
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.2258435060369575,
      "learning_rate": 9.980347889496535e-06,
      "loss": 1.227,
      "step": 8890
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.606807741332121,
      "learning_rate": 9.980238692291092e-06,
      "loss": 1.2276,
      "step": 8895
    },
    {
      "epoch": 0.38,
      "grad_norm": 24.85900410879738,
      "learning_rate": 9.980129193148447e-06,
      "loss": 1.2489,
      "step": 8900
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.428713981368675,
      "learning_rate": 9.980019392075237e-06,
      "loss": 1.1592,
      "step": 8905
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.980839871296275,
      "learning_rate": 9.979909289078122e-06,
      "loss": 1.1661,
      "step": 8910
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.313186641896175,
      "learning_rate": 9.979798884163777e-06,
      "loss": 1.2035,
      "step": 8915
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.548352876330076,
      "learning_rate": 9.979688177338891e-06,
      "loss": 1.2039,
      "step": 8920
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.12293553689391,
      "learning_rate": 9.97957716861018e-06,
      "loss": 1.2454,
      "step": 8925
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.097587309196077,
      "learning_rate": 9.979465857984374e-06,
      "loss": 1.1947,
      "step": 8930
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.829882468003256,
      "learning_rate": 9.97935424546822e-06,
      "loss": 1.2133,
      "step": 8935
    },
    {
      "epoch": 0.38,
      "grad_norm": 23.637873643294103,
      "learning_rate": 9.979242331068486e-06,
      "loss": 1.2483,
      "step": 8940
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.449244133554773,
      "learning_rate": 9.979130114791956e-06,
      "loss": 1.2092,
      "step": 8945
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.49330264158639,
      "learning_rate": 9.979017596645433e-06,
      "loss": 1.2071,
      "step": 8950
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.026036396204606,
      "learning_rate": 9.97890477663574e-06,
      "loss": 1.1998,
      "step": 8955
    },
    {
      "epoch": 0.38,
      "grad_norm": 38.581175690273106,
      "learning_rate": 9.978791654769715e-06,
      "loss": 1.202,
      "step": 8960
    },
    {
      "epoch": 0.38,
      "grad_norm": 16.265219455304024,
      "learning_rate": 9.97867823105422e-06,
      "loss": 1.2264,
      "step": 8965
    },
    {
      "epoch": 0.38,
      "grad_norm": 34.11472475520768,
      "learning_rate": 9.978564505496127e-06,
      "loss": 1.2313,
      "step": 8970
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.337839150110696,
      "learning_rate": 9.978450478102332e-06,
      "loss": 1.2064,
      "step": 8975
    },
    {
      "epoch": 0.38,
      "grad_norm": 18.227860649309974,
      "learning_rate": 9.97833614887975e-06,
      "loss": 1.2129,
      "step": 8980
    },
    {
      "epoch": 0.38,
      "grad_norm": 38.56136195651849,
      "learning_rate": 9.978221517835313e-06,
      "loss": 1.2209,
      "step": 8985
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.615944511583525,
      "learning_rate": 9.978106584975968e-06,
      "loss": 1.2103,
      "step": 8990
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.0190235325719375,
      "learning_rate": 9.977991350308685e-06,
      "loss": 1.1965,
      "step": 8995
    },
    {
      "epoch": 0.38,
      "grad_norm": 17.377226855422318,
      "learning_rate": 9.977875813840448e-06,
      "loss": 1.2246,
      "step": 9000
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.4866811491150065,
      "learning_rate": 9.977759975578263e-06,
      "loss": 1.2314,
      "step": 9005
    },
    {
      "epoch": 0.38,
      "grad_norm": 30.561186684380058,
      "learning_rate": 9.977643835529154e-06,
      "loss": 1.1982,
      "step": 9010
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.614377286199833,
      "learning_rate": 9.977527393700161e-06,
      "loss": 1.1769,
      "step": 9015
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.648626162771638,
      "learning_rate": 9.977410650098345e-06,
      "loss": 1.2375,
      "step": 9020
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.88166927988688,
      "learning_rate": 9.977293604730783e-06,
      "loss": 1.1966,
      "step": 9025
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.355959328176522,
      "learning_rate": 9.977176257604569e-06,
      "loss": 1.2331,
      "step": 9030
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.856032834593801,
      "learning_rate": 9.97705860872682e-06,
      "loss": 1.1486,
      "step": 9035
    },
    {
      "epoch": 0.38,
      "grad_norm": 35.17041070558963,
      "learning_rate": 9.97694065810467e-06,
      "loss": 1.1925,
      "step": 9040
    },
    {
      "epoch": 0.38,
      "grad_norm": 51.191372284353925,
      "learning_rate": 9.976822405745266e-06,
      "loss": 1.2622,
      "step": 9045
    },
    {
      "epoch": 0.38,
      "grad_norm": 44.731476925687744,
      "learning_rate": 9.976703851655777e-06,
      "loss": 1.2073,
      "step": 9050
    },
    {
      "epoch": 0.38,
      "grad_norm": 51.585680710726216,
      "learning_rate": 9.976584995843397e-06,
      "loss": 1.1665,
      "step": 9055
    },
    {
      "epoch": 0.38,
      "grad_norm": 74.17018898861281,
      "learning_rate": 9.976465838315326e-06,
      "loss": 1.2279,
      "step": 9060
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.071347112116781,
      "learning_rate": 9.97634637907879e-06,
      "loss": 1.2556,
      "step": 9065
    },
    {
      "epoch": 0.38,
      "grad_norm": 15.740017956468256,
      "learning_rate": 9.976226618141032e-06,
      "loss": 1.216,
      "step": 9070
    },
    {
      "epoch": 0.38,
      "grad_norm": 24.379319275867136,
      "learning_rate": 9.97610655550931e-06,
      "loss": 1.2112,
      "step": 9075
    },
    {
      "epoch": 0.38,
      "grad_norm": 18.377575647372183,
      "learning_rate": 9.975986191190907e-06,
      "loss": 1.2016,
      "step": 9080
    },
    {
      "epoch": 0.38,
      "grad_norm": 12.094704476883008,
      "learning_rate": 9.975865525193117e-06,
      "loss": 1.2667,
      "step": 9085
    },
    {
      "epoch": 0.38,
      "grad_norm": 21.242836312515045,
      "learning_rate": 9.97574455752326e-06,
      "loss": 1.2677,
      "step": 9090
    },
    {
      "epoch": 0.38,
      "grad_norm": 17.88504325575899,
      "learning_rate": 9.975623288188663e-06,
      "loss": 1.2623,
      "step": 9095
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.534679209993314,
      "learning_rate": 9.975501717196685e-06,
      "loss": 1.2321,
      "step": 9100
    },
    {
      "epoch": 0.39,
      "grad_norm": 30.52300065728163,
      "learning_rate": 9.975379844554693e-06,
      "loss": 1.233,
      "step": 9105
    },
    {
      "epoch": 0.39,
      "grad_norm": 38.665283899913156,
      "learning_rate": 9.975257670270075e-06,
      "loss": 1.2354,
      "step": 9110
    },
    {
      "epoch": 0.39,
      "grad_norm": 36.6414471320579,
      "learning_rate": 9.975135194350242e-06,
      "loss": 1.2214,
      "step": 9115
    },
    {
      "epoch": 0.39,
      "grad_norm": 32.84228839088985,
      "learning_rate": 9.975012416802616e-06,
      "loss": 1.2103,
      "step": 9120
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.323335299056941,
      "learning_rate": 9.974889337634641e-06,
      "loss": 1.245,
      "step": 9125
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.934028539857001,
      "learning_rate": 9.97476595685378e-06,
      "loss": 1.1905,
      "step": 9130
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.674077430134009,
      "learning_rate": 9.974642274467514e-06,
      "loss": 1.2332,
      "step": 9135
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.956130481943227,
      "learning_rate": 9.974518290483337e-06,
      "loss": 1.2256,
      "step": 9140
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.718214599864837,
      "learning_rate": 9.974394004908772e-06,
      "loss": 1.1681,
      "step": 9145
    },
    {
      "epoch": 0.39,
      "grad_norm": 22.08091334930836,
      "learning_rate": 9.97426941775135e-06,
      "loss": 1.2137,
      "step": 9150
    },
    {
      "epoch": 0.39,
      "grad_norm": 25.003739736335447,
      "learning_rate": 9.974144529018624e-06,
      "loss": 1.2542,
      "step": 9155
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.06498327456631,
      "learning_rate": 9.974019338718169e-06,
      "loss": 1.1996,
      "step": 9160
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.305698985940414,
      "learning_rate": 9.973893846857571e-06,
      "loss": 1.2505,
      "step": 9165
    },
    {
      "epoch": 0.39,
      "grad_norm": 32.528895746257916,
      "learning_rate": 9.973768053444442e-06,
      "loss": 1.2005,
      "step": 9170
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.681747307156694,
      "learning_rate": 9.973641958486406e-06,
      "loss": 1.2146,
      "step": 9175
    },
    {
      "epoch": 0.39,
      "grad_norm": 43.466986428302825,
      "learning_rate": 9.97351556199111e-06,
      "loss": 1.244,
      "step": 9180
    },
    {
      "epoch": 0.39,
      "grad_norm": 24.838146302228044,
      "learning_rate": 9.973388863966212e-06,
      "loss": 1.1969,
      "step": 9185
    },
    {
      "epoch": 0.39,
      "grad_norm": 18.117779674843764,
      "learning_rate": 9.9732618644194e-06,
      "loss": 1.2609,
      "step": 9190
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.313913530776297,
      "learning_rate": 9.973134563358368e-06,
      "loss": 1.1797,
      "step": 9195
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.817488959664177,
      "learning_rate": 9.973006960790838e-06,
      "loss": 1.1973,
      "step": 9200
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.319552373347777,
      "learning_rate": 9.972879056724543e-06,
      "loss": 1.2126,
      "step": 9205
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.863014477728541,
      "learning_rate": 9.97275085116724e-06,
      "loss": 1.2109,
      "step": 9210
    },
    {
      "epoch": 0.39,
      "grad_norm": 9.692998717379215,
      "learning_rate": 9.972622344126698e-06,
      "loss": 1.201,
      "step": 9215
    },
    {
      "epoch": 0.39,
      "grad_norm": 13.32763675096636,
      "learning_rate": 9.972493535610714e-06,
      "loss": 1.2042,
      "step": 9220
    },
    {
      "epoch": 0.39,
      "grad_norm": 16.761512287758755,
      "learning_rate": 9.972364425627093e-06,
      "loss": 1.2137,
      "step": 9225
    },
    {
      "epoch": 0.39,
      "grad_norm": 24.78386503786613,
      "learning_rate": 9.972235014183663e-06,
      "loss": 1.2106,
      "step": 9230
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.916391574474108,
      "learning_rate": 9.97210530128827e-06,
      "loss": 1.2147,
      "step": 9235
    },
    {
      "epoch": 0.39,
      "grad_norm": 17.10123306781741,
      "learning_rate": 9.97197528694878e-06,
      "loss": 1.244,
      "step": 9240
    },
    {
      "epoch": 0.39,
      "grad_norm": 19.835980329009985,
      "learning_rate": 9.971844971173073e-06,
      "loss": 1.2327,
      "step": 9245
    },
    {
      "epoch": 0.39,
      "grad_norm": 36.644335544164925,
      "learning_rate": 9.97171435396905e-06,
      "loss": 1.1874,
      "step": 9250
    },
    {
      "epoch": 0.39,
      "grad_norm": 34.14895463364265,
      "learning_rate": 9.97158343534463e-06,
      "loss": 1.1941,
      "step": 9255
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.07805329527793,
      "learning_rate": 9.97145221530775e-06,
      "loss": 1.1773,
      "step": 9260
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.123960321358524,
      "learning_rate": 9.971320693866368e-06,
      "loss": 1.244,
      "step": 9265
    },
    {
      "epoch": 0.39,
      "grad_norm": 36.284677104994806,
      "learning_rate": 9.971188871028455e-06,
      "loss": 1.1978,
      "step": 9270
    },
    {
      "epoch": 0.39,
      "grad_norm": 23.821124025631523,
      "learning_rate": 9.971056746802004e-06,
      "loss": 1.2352,
      "step": 9275
    },
    {
      "epoch": 0.39,
      "grad_norm": 36.58726253370013,
      "learning_rate": 9.970924321195024e-06,
      "loss": 1.2086,
      "step": 9280
    },
    {
      "epoch": 0.39,
      "grad_norm": 27.682805783908606,
      "learning_rate": 9.970791594215547e-06,
      "loss": 1.2619,
      "step": 9285
    },
    {
      "epoch": 0.39,
      "grad_norm": 10.743290347808301,
      "learning_rate": 9.970658565871615e-06,
      "loss": 1.2008,
      "step": 9290
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.412098045778697,
      "learning_rate": 9.970525236171297e-06,
      "loss": 1.2233,
      "step": 9295
    },
    {
      "epoch": 0.39,
      "grad_norm": 25.903932901089096,
      "learning_rate": 9.970391605122676e-06,
      "loss": 1.2391,
      "step": 9300
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.922382831878737,
      "learning_rate": 9.970257672733852e-06,
      "loss": 1.216,
      "step": 9305
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.014446775158502,
      "learning_rate": 9.970123439012946e-06,
      "loss": 1.1972,
      "step": 9310
    },
    {
      "epoch": 0.39,
      "grad_norm": 18.875799486524393,
      "learning_rate": 9.969988903968095e-06,
      "loss": 1.2416,
      "step": 9315
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.636377736334097,
      "learning_rate": 9.969854067607456e-06,
      "loss": 1.226,
      "step": 9320
    },
    {
      "epoch": 0.39,
      "grad_norm": 12.341771794662598,
      "learning_rate": 9.969718929939205e-06,
      "loss": 1.2059,
      "step": 9325
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.616163269366373,
      "learning_rate": 9.969583490971535e-06,
      "loss": 1.2015,
      "step": 9330
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.245377853271005,
      "learning_rate": 9.969447750712655e-06,
      "loss": 1.2215,
      "step": 9335
    },
    {
      "epoch": 0.4,
      "grad_norm": 16.460400561065228,
      "learning_rate": 9.969311709170796e-06,
      "loss": 1.2048,
      "step": 9340
    },
    {
      "epoch": 0.4,
      "grad_norm": 13.400660646223276,
      "learning_rate": 9.969175366354204e-06,
      "loss": 1.2148,
      "step": 9345
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.307860109074186,
      "learning_rate": 9.96903872227115e-06,
      "loss": 1.2117,
      "step": 9350
    },
    {
      "epoch": 0.4,
      "grad_norm": 11.560237957166326,
      "learning_rate": 9.968901776929913e-06,
      "loss": 1.2413,
      "step": 9355
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.224681080050976,
      "learning_rate": 9.968764530338799e-06,
      "loss": 1.1956,
      "step": 9360
    },
    {
      "epoch": 0.4,
      "grad_norm": 92.6125976287972,
      "learning_rate": 9.968626982506127e-06,
      "loss": 1.2209,
      "step": 9365
    },
    {
      "epoch": 0.4,
      "grad_norm": 48.374082869780395,
      "learning_rate": 9.968489133440235e-06,
      "loss": 1.2498,
      "step": 9370
    },
    {
      "epoch": 0.4,
      "grad_norm": 51.47521895669246,
      "learning_rate": 9.968350983149484e-06,
      "loss": 1.228,
      "step": 9375
    },
    {
      "epoch": 0.4,
      "grad_norm": 39.49127084087763,
      "learning_rate": 9.968212531642247e-06,
      "loss": 1.1937,
      "step": 9380
    },
    {
      "epoch": 0.4,
      "grad_norm": 71.86835692271305,
      "learning_rate": 9.968073778926919e-06,
      "loss": 1.2229,
      "step": 9385
    },
    {
      "epoch": 0.4,
      "grad_norm": 47.10702971907258,
      "learning_rate": 9.967934725011911e-06,
      "loss": 1.2573,
      "step": 9390
    },
    {
      "epoch": 0.4,
      "grad_norm": 61.89525407565561,
      "learning_rate": 9.967795369905654e-06,
      "loss": 1.2283,
      "step": 9395
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.992972884755579,
      "learning_rate": 9.967655713616598e-06,
      "loss": 1.2561,
      "step": 9400
    },
    {
      "epoch": 0.4,
      "grad_norm": 16.67991763705181,
      "learning_rate": 9.967515756153207e-06,
      "loss": 1.2726,
      "step": 9405
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.497996936958751,
      "learning_rate": 9.967375497523969e-06,
      "loss": 1.2193,
      "step": 9410
    },
    {
      "epoch": 0.4,
      "grad_norm": 25.499726306243453,
      "learning_rate": 9.967234937737387e-06,
      "loss": 1.2312,
      "step": 9415
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.283297652763751,
      "learning_rate": 9.967094076801982e-06,
      "loss": 1.1984,
      "step": 9420
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.781712508453821,
      "learning_rate": 9.966952914726294e-06,
      "loss": 1.2253,
      "step": 9425
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.504400914115953,
      "learning_rate": 9.96681145151888e-06,
      "loss": 1.2019,
      "step": 9430
    },
    {
      "epoch": 0.4,
      "grad_norm": 17.823813006269976,
      "learning_rate": 9.966669687188319e-06,
      "loss": 1.187,
      "step": 9435
    },
    {
      "epoch": 0.4,
      "grad_norm": 59.731765636649634,
      "learning_rate": 9.966527621743206e-06,
      "loss": 1.2216,
      "step": 9440
    },
    {
      "epoch": 0.4,
      "grad_norm": 23.48687006658772,
      "learning_rate": 9.96638525519215e-06,
      "loss": 1.222,
      "step": 9445
    },
    {
      "epoch": 0.4,
      "grad_norm": 36.304092430688456,
      "learning_rate": 9.966242587543787e-06,
      "loss": 1.2238,
      "step": 9450
    },
    {
      "epoch": 0.4,
      "grad_norm": 16.035660789309322,
      "learning_rate": 9.966099618806764e-06,
      "loss": 1.2561,
      "step": 9455
    },
    {
      "epoch": 0.4,
      "grad_norm": 133.60400532452636,
      "learning_rate": 9.96595634898975e-06,
      "loss": 1.5928,
      "step": 9460
    },
    {
      "epoch": 0.4,
      "grad_norm": 1246.3639950308245,
      "learning_rate": 9.96581277810143e-06,
      "loss": 3.5181,
      "step": 9465
    },
    {
      "epoch": 0.4,
      "grad_norm": 1564.101183261985,
      "learning_rate": 9.965668906150508e-06,
      "loss": 13.1298,
      "step": 9470
    },
    {
      "epoch": 0.4,
      "grad_norm": 648.1757424480604,
      "learning_rate": 9.965524733145705e-06,
      "loss": 31.2339,
      "step": 9475
    },
    {
      "epoch": 0.4,
      "grad_norm": 435.7401949488054,
      "learning_rate": 9.965380259095767e-06,
      "loss": 22.4813,
      "step": 9480
    },
    {
      "epoch": 0.4,
      "grad_norm": 207.38204420943876,
      "learning_rate": 9.965235484009449e-06,
      "loss": 15.5626,
      "step": 9485
    },
    {
      "epoch": 0.4,
      "grad_norm": 131.22620060772448,
      "learning_rate": 9.965090407895529e-06,
      "loss": 13.1922,
      "step": 9490
    },
    {
      "epoch": 0.4,
      "grad_norm": 170.61189718333307,
      "learning_rate": 9.964945030762803e-06,
      "loss": 10.9917,
      "step": 9495
    },
    {
      "epoch": 0.4,
      "grad_norm": 154.81543956380625,
      "learning_rate": 9.964799352620084e-06,
      "loss": 9.926,
      "step": 9500
    },
    {
      "epoch": 0.4,
      "grad_norm": 41.39714851341611,
      "learning_rate": 9.964653373476206e-06,
      "loss": 9.0876,
      "step": 9505
    },
    {
      "epoch": 0.4,
      "grad_norm": 45.46295761377743,
      "learning_rate": 9.964507093340015e-06,
      "loss": 8.4916,
      "step": 9510
    },
    {
      "epoch": 0.4,
      "grad_norm": 45.630639893960875,
      "learning_rate": 9.964360512220384e-06,
      "loss": 8.2392,
      "step": 9515
    },
    {
      "epoch": 0.4,
      "grad_norm": 28.34539297347392,
      "learning_rate": 9.964213630126198e-06,
      "loss": 7.8186,
      "step": 9520
    },
    {
      "epoch": 0.4,
      "grad_norm": 51.916435761534224,
      "learning_rate": 9.964066447066363e-06,
      "loss": 7.7061,
      "step": 9525
    },
    {
      "epoch": 0.4,
      "grad_norm": 37.494201471517435,
      "learning_rate": 9.963918963049802e-06,
      "loss": 7.4171,
      "step": 9530
    },
    {
      "epoch": 0.4,
      "grad_norm": 23.851659564101183,
      "learning_rate": 9.963771178085453e-06,
      "loss": 7.3364,
      "step": 9535
    },
    {
      "epoch": 0.4,
      "grad_norm": 40.157847317787926,
      "learning_rate": 9.963623092182282e-06,
      "loss": 7.1963,
      "step": 9540
    },
    {
      "epoch": 0.4,
      "grad_norm": 25.138830161265012,
      "learning_rate": 9.963474705349261e-06,
      "loss": 6.9885,
      "step": 9545
    },
    {
      "epoch": 0.4,
      "grad_norm": 89.29539425932496,
      "learning_rate": 9.963326017595391e-06,
      "loss": 6.9163,
      "step": 9550
    },
    {
      "epoch": 0.4,
      "grad_norm": 42.64818735240843,
      "learning_rate": 9.963177028929684e-06,
      "loss": 6.9437,
      "step": 9555
    },
    {
      "epoch": 0.4,
      "grad_norm": 35.26096380157238,
      "learning_rate": 9.963027739361173e-06,
      "loss": 6.8422,
      "step": 9560
    },
    {
      "epoch": 0.4,
      "grad_norm": 251.2796965724704,
      "learning_rate": 9.96287814889891e-06,
      "loss": 6.8281,
      "step": 9565
    },
    {
      "epoch": 0.41,
      "grad_norm": 130.01734608880602,
      "learning_rate": 9.962728257551962e-06,
      "loss": 6.8942,
      "step": 9570
    },
    {
      "epoch": 0.41,
      "grad_norm": 54.406652980621814,
      "learning_rate": 9.962578065329418e-06,
      "loss": 6.6734,
      "step": 9575
    },
    {
      "epoch": 0.41,
      "grad_norm": 31.478822197942467,
      "learning_rate": 9.962427572240386e-06,
      "loss": 6.6376,
      "step": 9580
    },
    {
      "epoch": 0.41,
      "grad_norm": 19.134982938702933,
      "learning_rate": 9.962276778293985e-06,
      "loss": 6.5207,
      "step": 9585
    },
    {
      "epoch": 0.41,
      "grad_norm": 75.0329054930933,
      "learning_rate": 9.96212568349936e-06,
      "loss": 6.5194,
      "step": 9590
    },
    {
      "epoch": 0.41,
      "grad_norm": 69.17552242907516,
      "learning_rate": 9.961974287865673e-06,
      "loss": 6.4638,
      "step": 9595
    },
    {
      "epoch": 0.41,
      "grad_norm": 26.344162018491446,
      "learning_rate": 9.961822591402099e-06,
      "loss": 6.4512,
      "step": 9600
    },
    {
      "epoch": 0.41,
      "grad_norm": 64.74865605741786,
      "learning_rate": 9.961670594117838e-06,
      "loss": 6.4404,
      "step": 9605
    },
    {
      "epoch": 0.41,
      "grad_norm": 133.72320086340795,
      "learning_rate": 9.961518296022103e-06,
      "loss": 6.367,
      "step": 9610
    },
    {
      "epoch": 0.41,
      "grad_norm": 85.25682159143611,
      "learning_rate": 9.96136569712413e-06,
      "loss": 6.3148,
      "step": 9615
    },
    {
      "epoch": 0.41,
      "grad_norm": 76.04932912541877,
      "learning_rate": 9.961212797433168e-06,
      "loss": 6.3086,
      "step": 9620
    },
    {
      "epoch": 0.41,
      "grad_norm": 47.48107198192422,
      "learning_rate": 9.961059596958487e-06,
      "loss": 6.0329,
      "step": 9625
    },
    {
      "epoch": 0.41,
      "grad_norm": 560.7905988781308,
      "learning_rate": 9.960906095709378e-06,
      "loss": 7.3474,
      "step": 9630
    },
    {
      "epoch": 0.41,
      "grad_norm": 107.42656906943576,
      "learning_rate": 9.960752293695142e-06,
      "loss": 8.1501,
      "step": 9635
    },
    {
      "epoch": 0.41,
      "grad_norm": 140.85335787901693,
      "learning_rate": 9.96059819092511e-06,
      "loss": 7.6273,
      "step": 9640
    },
    {
      "epoch": 0.41,
      "grad_norm": 76.80491095184355,
      "learning_rate": 9.96044378740862e-06,
      "loss": 7.2934,
      "step": 9645
    },
    {
      "epoch": 0.41,
      "grad_norm": 88.51053018950759,
      "learning_rate": 9.960289083155035e-06,
      "loss": 7.1166,
      "step": 9650
    },
    {
      "epoch": 0.41,
      "grad_norm": 86.82587076796945,
      "learning_rate": 9.960134078173734e-06,
      "loss": 6.9439,
      "step": 9655
    },
    {
      "epoch": 0.41,
      "grad_norm": 80.49749753416499,
      "learning_rate": 9.959978772474113e-06,
      "loss": 6.7618,
      "step": 9660
    },
    {
      "epoch": 0.41,
      "grad_norm": 82.71371750955271,
      "learning_rate": 9.959823166065592e-06,
      "loss": 6.6756,
      "step": 9665
    },
    {
      "epoch": 0.41,
      "grad_norm": 44.079068344126675,
      "learning_rate": 9.9596672589576e-06,
      "loss": 6.5916,
      "step": 9670
    },
    {
      "epoch": 0.41,
      "grad_norm": 34.73313355427132,
      "learning_rate": 9.95951105115959e-06,
      "loss": 6.5831,
      "step": 9675
    },
    {
      "epoch": 0.41,
      "grad_norm": 34.90671996650152,
      "learning_rate": 9.959354542681037e-06,
      "loss": 6.3289,
      "step": 9680
    },
    {
      "epoch": 0.41,
      "grad_norm": 32.08593072464212,
      "learning_rate": 9.959197733531425e-06,
      "loss": 6.5721,
      "step": 9685
    },
    {
      "epoch": 0.41,
      "grad_norm": 81.5408683228983,
      "learning_rate": 9.959040623720261e-06,
      "loss": 6.4341,
      "step": 9690
    },
    {
      "epoch": 0.41,
      "grad_norm": 18.9408354844983,
      "learning_rate": 9.958883213257074e-06,
      "loss": 6.3347,
      "step": 9695
    },
    {
      "epoch": 0.41,
      "grad_norm": 18.890746090474316,
      "learning_rate": 9.958725502151402e-06,
      "loss": 6.2115,
      "step": 9700
    },
    {
      "epoch": 0.41,
      "grad_norm": 18.216768757550874,
      "learning_rate": 9.95856749041281e-06,
      "loss": 6.2049,
      "step": 9705
    },
    {
      "epoch": 0.41,
      "grad_norm": 26.134445645789967,
      "learning_rate": 9.958409178050878e-06,
      "loss": 6.134,
      "step": 9710
    },
    {
      "epoch": 0.41,
      "grad_norm": 27.01987844611143,
      "learning_rate": 9.958250565075202e-06,
      "loss": 6.1552,
      "step": 9715
    },
    {
      "epoch": 0.41,
      "grad_norm": 29.999863123436292,
      "learning_rate": 9.958091651495399e-06,
      "loss": 6.0328,
      "step": 9720
    },
    {
      "epoch": 0.41,
      "grad_norm": 31.27345738732662,
      "learning_rate": 9.957932437321104e-06,
      "loss": 5.9652,
      "step": 9725
    },
    {
      "epoch": 0.41,
      "grad_norm": 42.5058427296302,
      "learning_rate": 9.95777292256197e-06,
      "loss": 5.8758,
      "step": 9730
    },
    {
      "epoch": 0.41,
      "grad_norm": 57.62779698359081,
      "learning_rate": 9.957613107227666e-06,
      "loss": 5.8712,
      "step": 9735
    },
    {
      "epoch": 0.41,
      "grad_norm": 52.777051997228845,
      "learning_rate": 9.957452991327882e-06,
      "loss": 5.7757,
      "step": 9740
    },
    {
      "epoch": 0.41,
      "grad_norm": 166.2712541342479,
      "learning_rate": 9.957292574872329e-06,
      "loss": 5.8184,
      "step": 9745
    },
    {
      "epoch": 0.41,
      "grad_norm": 82.29460628475253,
      "learning_rate": 9.957131857870728e-06,
      "loss": 6.0833,
      "step": 9750
    },
    {
      "epoch": 0.41,
      "grad_norm": 38.91824140406657,
      "learning_rate": 9.956970840332823e-06,
      "loss": 5.8787,
      "step": 9755
    },
    {
      "epoch": 0.41,
      "grad_norm": 72.19171063211087,
      "learning_rate": 9.956809522268378e-06,
      "loss": 5.7328,
      "step": 9760
    },
    {
      "epoch": 0.41,
      "grad_norm": 125.33977216827597,
      "learning_rate": 9.95664790368717e-06,
      "loss": 5.5346,
      "step": 9765
    },
    {
      "epoch": 0.41,
      "grad_norm": 42.511155139928945,
      "learning_rate": 9.956485984599003e-06,
      "loss": 5.4775,
      "step": 9770
    },
    {
      "epoch": 0.41,
      "grad_norm": 92.39804911136169,
      "learning_rate": 9.95632376501369e-06,
      "loss": 5.3655,
      "step": 9775
    },
    {
      "epoch": 0.41,
      "grad_norm": 270.82508199954304,
      "learning_rate": 9.956161244941067e-06,
      "loss": 5.1565,
      "step": 9780
    },
    {
      "epoch": 0.41,
      "grad_norm": 105.35039052278564,
      "learning_rate": 9.955998424390985e-06,
      "loss": 5.1373,
      "step": 9785
    },
    {
      "epoch": 0.41,
      "grad_norm": 230.6278558606637,
      "learning_rate": 9.955835303373318e-06,
      "loss": 4.9518,
      "step": 9790
    },
    {
      "epoch": 0.41,
      "grad_norm": 57.15879498637197,
      "learning_rate": 9.955671881897954e-06,
      "loss": 4.8838,
      "step": 9795
    },
    {
      "epoch": 0.41,
      "grad_norm": 62.03105504704029,
      "learning_rate": 9.9555081599748e-06,
      "loss": 4.4759,
      "step": 9800
    },
    {
      "epoch": 0.41,
      "grad_norm": 164.88866486764255,
      "learning_rate": 9.955344137613785e-06,
      "loss": 4.0557,
      "step": 9805
    },
    {
      "epoch": 0.42,
      "grad_norm": 76.10018504744276,
      "learning_rate": 9.955179814824852e-06,
      "loss": 3.7505,
      "step": 9810
    },
    {
      "epoch": 0.42,
      "grad_norm": 123.94440081408953,
      "learning_rate": 9.95501519161796e-06,
      "loss": 3.4826,
      "step": 9815
    },
    {
      "epoch": 0.42,
      "grad_norm": 23.852795754122074,
      "learning_rate": 9.954850268003095e-06,
      "loss": 3.0169,
      "step": 9820
    },
    {
      "epoch": 0.42,
      "grad_norm": 58.00109644180814,
      "learning_rate": 9.954685043990252e-06,
      "loss": 2.7229,
      "step": 9825
    },
    {
      "epoch": 0.42,
      "grad_norm": 45.564132396663126,
      "learning_rate": 9.954519519589449e-06,
      "loss": 2.4982,
      "step": 9830
    },
    {
      "epoch": 0.42,
      "grad_norm": 89.10698141972962,
      "learning_rate": 9.954353694810722e-06,
      "loss": 2.4265,
      "step": 9835
    },
    {
      "epoch": 0.42,
      "grad_norm": 13.89724085652203,
      "learning_rate": 9.954187569664124e-06,
      "loss": 2.2177,
      "step": 9840
    },
    {
      "epoch": 0.42,
      "grad_norm": 16.38689386273016,
      "learning_rate": 9.954021144159726e-06,
      "loss": 2.1623,
      "step": 9845
    },
    {
      "epoch": 0.42,
      "grad_norm": 64.54936960834726,
      "learning_rate": 9.95385441830762e-06,
      "loss": 1.8974,
      "step": 9850
    },
    {
      "epoch": 0.42,
      "grad_norm": 26.561861795726873,
      "learning_rate": 9.953687392117913e-06,
      "loss": 1.7854,
      "step": 9855
    },
    {
      "epoch": 0.42,
      "grad_norm": 32.04597088306472,
      "learning_rate": 9.95352006560073e-06,
      "loss": 1.9033,
      "step": 9860
    },
    {
      "epoch": 0.42,
      "grad_norm": 33.74923644329989,
      "learning_rate": 9.953352438766217e-06,
      "loss": 1.9048,
      "step": 9865
    },
    {
      "epoch": 0.42,
      "grad_norm": 88.22758671179918,
      "learning_rate": 9.953184511624535e-06,
      "loss": 1.8914,
      "step": 9870
    },
    {
      "epoch": 0.42,
      "grad_norm": 48.07428025503431,
      "learning_rate": 9.953016284185867e-06,
      "loss": 1.847,
      "step": 9875
    },
    {
      "epoch": 0.42,
      "grad_norm": 27.518829521750558,
      "learning_rate": 9.952847756460411e-06,
      "loss": 1.6421,
      "step": 9880
    },
    {
      "epoch": 0.42,
      "grad_norm": 70.77074742148697,
      "learning_rate": 9.952678928458384e-06,
      "loss": 1.6483,
      "step": 9885
    },
    {
      "epoch": 0.42,
      "grad_norm": 22.329833525195394,
      "learning_rate": 9.952509800190023e-06,
      "loss": 1.5452,
      "step": 9890
    },
    {
      "epoch": 0.42,
      "grad_norm": 22.527894812873893,
      "learning_rate": 9.952340371665581e-06,
      "loss": 1.5635,
      "step": 9895
    },
    {
      "epoch": 0.42,
      "grad_norm": 30.679732319240273,
      "learning_rate": 9.952170642895331e-06,
      "loss": 1.5152,
      "step": 9900
    },
    {
      "epoch": 0.42,
      "grad_norm": 15.910460973885947,
      "learning_rate": 9.95200061388956e-06,
      "loss": 1.4791,
      "step": 9905
    },
    {
      "epoch": 0.42,
      "grad_norm": 15.673746860152601,
      "learning_rate": 9.951830284658579e-06,
      "loss": 1.4556,
      "step": 9910
    },
    {
      "epoch": 0.42,
      "grad_norm": 20.242012716298134,
      "learning_rate": 9.951659655212713e-06,
      "loss": 1.5105,
      "step": 9915
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.390724594801506,
      "learning_rate": 9.951488725562309e-06,
      "loss": 1.4504,
      "step": 9920
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.077366181486358,
      "learning_rate": 9.951317495717729e-06,
      "loss": 1.4336,
      "step": 9925
    },
    {
      "epoch": 0.42,
      "grad_norm": 8.71516293169589,
      "learning_rate": 9.951145965689353e-06,
      "loss": 1.3927,
      "step": 9930
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.01714917542507,
      "learning_rate": 9.950974135487581e-06,
      "loss": 1.3778,
      "step": 9935
    },
    {
      "epoch": 0.42,
      "grad_norm": 18.000653886509415,
      "learning_rate": 9.950802005122832e-06,
      "loss": 1.3947,
      "step": 9940
    },
    {
      "epoch": 0.42,
      "grad_norm": 13.401114901020838,
      "learning_rate": 9.950629574605539e-06,
      "loss": 1.365,
      "step": 9945
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.594615964471209,
      "learning_rate": 9.95045684394616e-06,
      "loss": 1.2829,
      "step": 9950
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.144439761709952,
      "learning_rate": 9.950283813155162e-06,
      "loss": 1.3727,
      "step": 9955
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.608225881956869,
      "learning_rate": 9.950110482243037e-06,
      "loss": 1.3639,
      "step": 9960
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.918362963941835,
      "learning_rate": 9.949936851220297e-06,
      "loss": 1.4023,
      "step": 9965
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.838873323810942,
      "learning_rate": 9.949762920097466e-06,
      "loss": 1.2959,
      "step": 9970
    },
    {
      "epoch": 0.42,
      "grad_norm": 7.2778845793516975,
      "learning_rate": 9.949588688885088e-06,
      "loss": 1.3927,
      "step": 9975
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.547618869449657,
      "learning_rate": 9.949414157593726e-06,
      "loss": 1.3341,
      "step": 9980
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.497332697543262,
      "learning_rate": 9.949239326233967e-06,
      "loss": 1.3462,
      "step": 9985
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.97726774216965,
      "learning_rate": 9.949064194816403e-06,
      "loss": 1.3584,
      "step": 9990
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.34222557050974,
      "learning_rate": 9.948888763351655e-06,
      "loss": 1.3885,
      "step": 9995
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.622687670045263,
      "learning_rate": 9.948713031850359e-06,
      "loss": 1.363,
      "step": 10000
    },
    {
      "epoch": 0.42,
      "grad_norm": 23.099097828415495,
      "learning_rate": 9.948537000323168e-06,
      "loss": 1.2767,
      "step": 10005
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.171415356852316,
      "learning_rate": 9.948360668780757e-06,
      "loss": 1.3293,
      "step": 10010
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.526871210026702,
      "learning_rate": 9.948184037233812e-06,
      "loss": 1.3387,
      "step": 10015
    },
    {
      "epoch": 0.42,
      "grad_norm": 10.064163367669915,
      "learning_rate": 9.948007105693047e-06,
      "loss": 1.31,
      "step": 10020
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.252667667761068,
      "learning_rate": 9.947829874169183e-06,
      "loss": 1.3373,
      "step": 10025
    },
    {
      "epoch": 0.42,
      "grad_norm": 18.30752951354965,
      "learning_rate": 9.947652342672969e-06,
      "loss": 1.2854,
      "step": 10030
    },
    {
      "epoch": 0.42,
      "grad_norm": 16.352958613582345,
      "learning_rate": 9.94747451121517e-06,
      "loss": 1.3213,
      "step": 10035
    },
    {
      "epoch": 0.42,
      "grad_norm": 12.29479058175808,
      "learning_rate": 9.947296379806561e-06,
      "loss": 1.2825,
      "step": 10040
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.01481681442109,
      "learning_rate": 9.947117948457948e-06,
      "loss": 1.2957,
      "step": 10045
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.300700150619039,
      "learning_rate": 9.946939217180143e-06,
      "loss": 1.2922,
      "step": 10050
    },
    {
      "epoch": 0.43,
      "grad_norm": 20.127238513405484,
      "learning_rate": 9.946760185983988e-06,
      "loss": 1.2786,
      "step": 10055
    },
    {
      "epoch": 0.43,
      "grad_norm": 16.126504988198615,
      "learning_rate": 9.946580854880332e-06,
      "loss": 1.2778,
      "step": 10060
    },
    {
      "epoch": 0.43,
      "grad_norm": 23.495416520452515,
      "learning_rate": 9.94640122388005e-06,
      "loss": 1.2576,
      "step": 10065
    },
    {
      "epoch": 0.43,
      "grad_norm": 26.79450746953665,
      "learning_rate": 9.946221292994033e-06,
      "loss": 1.2895,
      "step": 10070
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.657243521737724,
      "learning_rate": 9.946041062233188e-06,
      "loss": 1.2628,
      "step": 10075
    },
    {
      "epoch": 0.43,
      "grad_norm": 18.45985948516765,
      "learning_rate": 9.945860531608442e-06,
      "loss": 1.2894,
      "step": 10080
    },
    {
      "epoch": 0.43,
      "grad_norm": 23.609942320006592,
      "learning_rate": 9.945679701130739e-06,
      "loss": 1.2995,
      "step": 10085
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.477147046333467,
      "learning_rate": 9.945498570811045e-06,
      "loss": 1.2799,
      "step": 10090
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.371002370930244,
      "learning_rate": 9.945317140660342e-06,
      "loss": 1.33,
      "step": 10095
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.251357492721981,
      "learning_rate": 9.945135410689626e-06,
      "loss": 1.2775,
      "step": 10100
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.784609324227422,
      "learning_rate": 9.944953380909917e-06,
      "loss": 1.2835,
      "step": 10105
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.5387534223853185,
      "learning_rate": 9.944771051332249e-06,
      "loss": 1.2862,
      "step": 10110
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.798370850984581,
      "learning_rate": 9.944588421967677e-06,
      "loss": 1.2451,
      "step": 10115
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.34992074620122,
      "learning_rate": 9.944405492827277e-06,
      "loss": 1.2479,
      "step": 10120
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.835843580870858,
      "learning_rate": 9.944222263922135e-06,
      "loss": 1.2436,
      "step": 10125
    },
    {
      "epoch": 0.43,
      "grad_norm": 20.53673234569956,
      "learning_rate": 9.944038735263359e-06,
      "loss": 1.2684,
      "step": 10130
    },
    {
      "epoch": 0.43,
      "grad_norm": 18.354472401129364,
      "learning_rate": 9.943854906862078e-06,
      "loss": 1.293,
      "step": 10135
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.749836491121126,
      "learning_rate": 9.943670778729438e-06,
      "loss": 1.2302,
      "step": 10140
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.777341395366744,
      "learning_rate": 9.943486350876599e-06,
      "loss": 1.249,
      "step": 10145
    },
    {
      "epoch": 0.43,
      "grad_norm": 11.779285606737107,
      "learning_rate": 9.943301623314745e-06,
      "loss": 1.2554,
      "step": 10150
    },
    {
      "epoch": 0.43,
      "grad_norm": 25.19685136713185,
      "learning_rate": 9.943116596055076e-06,
      "loss": 1.2424,
      "step": 10155
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.531134343319577,
      "learning_rate": 9.942931269108807e-06,
      "loss": 1.2701,
      "step": 10160
    },
    {
      "epoch": 0.43,
      "grad_norm": 22.493845189413534,
      "learning_rate": 9.942745642487174e-06,
      "loss": 1.2611,
      "step": 10165
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.022938427879689,
      "learning_rate": 9.942559716201432e-06,
      "loss": 1.2111,
      "step": 10170
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.996480938568168,
      "learning_rate": 9.942373490262856e-06,
      "loss": 1.2027,
      "step": 10175
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.126233348452894,
      "learning_rate": 9.94218696468273e-06,
      "loss": 1.196,
      "step": 10180
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.269249382801538,
      "learning_rate": 9.942000139472368e-06,
      "loss": 1.2359,
      "step": 10185
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.743907405383896,
      "learning_rate": 9.941813014643093e-06,
      "loss": 1.2303,
      "step": 10190
    },
    {
      "epoch": 0.43,
      "grad_norm": 20.409102476243174,
      "learning_rate": 9.941625590206254e-06,
      "loss": 1.2467,
      "step": 10195
    },
    {
      "epoch": 0.43,
      "grad_norm": 18.379766589318194,
      "learning_rate": 9.94143786617321e-06,
      "loss": 1.2728,
      "step": 10200
    },
    {
      "epoch": 0.43,
      "grad_norm": 29.223060156588254,
      "learning_rate": 9.941249842555344e-06,
      "loss": 1.2467,
      "step": 10205
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.571588802146547,
      "learning_rate": 9.941061519364054e-06,
      "loss": 1.2114,
      "step": 10210
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.506940018892727,
      "learning_rate": 9.94087289661076e-06,
      "loss": 1.2424,
      "step": 10215
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.040622817211133,
      "learning_rate": 9.940683974306896e-06,
      "loss": 1.2287,
      "step": 10220
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.128572303196038,
      "learning_rate": 9.940494752463917e-06,
      "loss": 1.2209,
      "step": 10225
    },
    {
      "epoch": 0.43,
      "grad_norm": 12.880363564274072,
      "learning_rate": 9.940305231093291e-06,
      "loss": 1.2293,
      "step": 10230
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.2771416985494914,
      "learning_rate": 9.940115410206512e-06,
      "loss": 1.2518,
      "step": 10235
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.486797815515412,
      "learning_rate": 9.939925289815087e-06,
      "loss": 1.2335,
      "step": 10240
    },
    {
      "epoch": 0.43,
      "grad_norm": 14.069423242264826,
      "learning_rate": 9.939734869930544e-06,
      "loss": 1.2441,
      "step": 10245
    },
    {
      "epoch": 0.43,
      "grad_norm": 13.42878969194725,
      "learning_rate": 9.939544150564425e-06,
      "loss": 1.2234,
      "step": 10250
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.924472816865137,
      "learning_rate": 9.939353131728295e-06,
      "loss": 1.2169,
      "step": 10255
    },
    {
      "epoch": 0.43,
      "grad_norm": 31.972436499585463,
      "learning_rate": 9.939161813433733e-06,
      "loss": 1.2647,
      "step": 10260
    },
    {
      "epoch": 0.43,
      "grad_norm": 31.627258030667768,
      "learning_rate": 9.93897019569234e-06,
      "loss": 1.2107,
      "step": 10265
    },
    {
      "epoch": 0.43,
      "grad_norm": 54.57097622714503,
      "learning_rate": 9.938778278515733e-06,
      "loss": 1.2562,
      "step": 10270
    },
    {
      "epoch": 0.43,
      "grad_norm": 21.665738941858674,
      "learning_rate": 9.938586061915545e-06,
      "loss": 1.211,
      "step": 10275
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.0504571677781,
      "learning_rate": 9.938393545903431e-06,
      "loss": 1.1989,
      "step": 10280
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.359550985591298,
      "learning_rate": 9.938200730491064e-06,
      "loss": 1.212,
      "step": 10285
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.139990522598348,
      "learning_rate": 9.938007615690133e-06,
      "loss": 1.2356,
      "step": 10290
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.328557260145226,
      "learning_rate": 9.937814201512343e-06,
      "loss": 1.2207,
      "step": 10295
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.353832565514606,
      "learning_rate": 9.937620487969426e-06,
      "loss": 1.2234,
      "step": 10300
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.055326277381491,
      "learning_rate": 9.93742647507312e-06,
      "loss": 1.1901,
      "step": 10305
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.51592292175932,
      "learning_rate": 9.937232162835194e-06,
      "loss": 1.2267,
      "step": 10310
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.588113392157487,
      "learning_rate": 9.93703755126742e-06,
      "loss": 1.1771,
      "step": 10315
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.226208415684464,
      "learning_rate": 9.936842640381607e-06,
      "loss": 1.1875,
      "step": 10320
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.603391632253025,
      "learning_rate": 9.936647430189563e-06,
      "loss": 1.2562,
      "step": 10325
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.950184690178073,
      "learning_rate": 9.93645192070313e-06,
      "loss": 1.1953,
      "step": 10330
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.765367393192978,
      "learning_rate": 9.936256111934155e-06,
      "loss": 1.2267,
      "step": 10335
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.4761024012374735,
      "learning_rate": 9.936060003894514e-06,
      "loss": 1.2096,
      "step": 10340
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.081180503907223,
      "learning_rate": 9.935863596596094e-06,
      "loss": 1.2242,
      "step": 10345
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.787545341346834,
      "learning_rate": 9.935666890050802e-06,
      "loss": 1.2399,
      "step": 10350
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.032648729682437,
      "learning_rate": 9.935469884270566e-06,
      "loss": 1.2055,
      "step": 10355
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.62949883737289,
      "learning_rate": 9.935272579267328e-06,
      "loss": 1.1796,
      "step": 10360
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.5222261312219745,
      "learning_rate": 9.935074975053053e-06,
      "loss": 1.2745,
      "step": 10365
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.163099772861981,
      "learning_rate": 9.934877071639716e-06,
      "loss": 1.2093,
      "step": 10370
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.613481196334059,
      "learning_rate": 9.93467886903932e-06,
      "loss": 1.2463,
      "step": 10375
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.931685390895252,
      "learning_rate": 9.93448036726388e-06,
      "loss": 1.2018,
      "step": 10380
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.975094818793018,
      "learning_rate": 9.934281566325431e-06,
      "loss": 1.19,
      "step": 10385
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.298937810576603,
      "learning_rate": 9.934082466236024e-06,
      "loss": 1.2056,
      "step": 10390
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.161477414177748,
      "learning_rate": 9.93388306700773e-06,
      "loss": 1.2265,
      "step": 10395
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.388915051137957,
      "learning_rate": 9.93368336865264e-06,
      "loss": 1.1905,
      "step": 10400
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.427422972855142,
      "learning_rate": 9.93348337118286e-06,
      "loss": 1.1469,
      "step": 10405
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.127322135473031,
      "learning_rate": 9.933283074610516e-06,
      "loss": 1.1783,
      "step": 10410
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.070302630294536,
      "learning_rate": 9.933082478947749e-06,
      "loss": 1.2091,
      "step": 10415
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.167061128334721,
      "learning_rate": 9.932881584206722e-06,
      "loss": 1.2058,
      "step": 10420
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.671128985805976,
      "learning_rate": 9.932680390399616e-06,
      "loss": 1.2503,
      "step": 10425
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.428469323767652,
      "learning_rate": 9.932478897538628e-06,
      "loss": 1.2091,
      "step": 10430
    },
    {
      "epoch": 0.44,
      "grad_norm": 9.027737668250325,
      "learning_rate": 9.932277105635973e-06,
      "loss": 1.2004,
      "step": 10435
    },
    {
      "epoch": 0.44,
      "grad_norm": 17.639585985612662,
      "learning_rate": 9.932075014703886e-06,
      "loss": 1.2108,
      "step": 10440
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.846812907766918,
      "learning_rate": 9.931872624754618e-06,
      "loss": 1.1871,
      "step": 10445
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.769758325038678,
      "learning_rate": 9.93166993580044e-06,
      "loss": 1.2511,
      "step": 10450
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.353791888476142,
      "learning_rate": 9.93146694785364e-06,
      "loss": 1.215,
      "step": 10455
    },
    {
      "epoch": 0.44,
      "grad_norm": 16.6760296090444,
      "learning_rate": 9.931263660926527e-06,
      "loss": 1.2557,
      "step": 10460
    },
    {
      "epoch": 0.44,
      "grad_norm": 14.307014761297724,
      "learning_rate": 9.931060075031423e-06,
      "loss": 1.2008,
      "step": 10465
    },
    {
      "epoch": 0.44,
      "grad_norm": 12.395314637904972,
      "learning_rate": 9.93085619018067e-06,
      "loss": 1.1386,
      "step": 10470
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.765312547763097,
      "learning_rate": 9.930652006386632e-06,
      "loss": 1.1823,
      "step": 10475
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.7062486490976685,
      "learning_rate": 9.930447523661686e-06,
      "loss": 1.2302,
      "step": 10480
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.317431111849158,
      "learning_rate": 9.93024274201823e-06,
      "loss": 1.2074,
      "step": 10485
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.294735669461464,
      "learning_rate": 9.930037661468679e-06,
      "loss": 1.2016,
      "step": 10490
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.501356204117338,
      "learning_rate": 9.929832282025466e-06,
      "loss": 1.225,
      "step": 10495
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.941532626605468,
      "learning_rate": 9.929626603701044e-06,
      "loss": 1.2284,
      "step": 10500
    },
    {
      "epoch": 0.44,
      "grad_norm": 17.715653674861617,
      "learning_rate": 9.929420626507882e-06,
      "loss": 1.1705,
      "step": 10505
    },
    {
      "epoch": 0.44,
      "grad_norm": 17.15148177217384,
      "learning_rate": 9.929214350458465e-06,
      "loss": 1.2225,
      "step": 10510
    },
    {
      "epoch": 0.45,
      "grad_norm": 23.90702080577724,
      "learning_rate": 9.929007775565305e-06,
      "loss": 1.1634,
      "step": 10515
    },
    {
      "epoch": 0.45,
      "grad_norm": 30.177090621342753,
      "learning_rate": 9.928800901840921e-06,
      "loss": 1.1539,
      "step": 10520
    },
    {
      "epoch": 0.45,
      "grad_norm": 17.44135285936784,
      "learning_rate": 9.928593729297855e-06,
      "loss": 1.1366,
      "step": 10525
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.270926953351559,
      "learning_rate": 9.92838625794867e-06,
      "loss": 1.1712,
      "step": 10530
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.348831852629494,
      "learning_rate": 9.928178487805945e-06,
      "loss": 1.2401,
      "step": 10535
    },
    {
      "epoch": 0.45,
      "grad_norm": 22.577816872892846,
      "learning_rate": 9.927970418882272e-06,
      "loss": 1.2159,
      "step": 10540
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.183281260052668,
      "learning_rate": 9.927762051190268e-06,
      "loss": 1.2373,
      "step": 10545
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.7125380455308825,
      "learning_rate": 9.927553384742567e-06,
      "loss": 1.2083,
      "step": 10550
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.94323068571043,
      "learning_rate": 9.927344419551822e-06,
      "loss": 1.1979,
      "step": 10555
    },
    {
      "epoch": 0.45,
      "grad_norm": 10.586780828445535,
      "learning_rate": 9.927135155630694e-06,
      "loss": 1.1675,
      "step": 10560
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.978746246588358,
      "learning_rate": 9.926925592991876e-06,
      "loss": 1.2019,
      "step": 10565
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.632834658864674,
      "learning_rate": 9.926715731648073e-06,
      "loss": 1.2366,
      "step": 10570
    },
    {
      "epoch": 0.45,
      "grad_norm": 14.174405240556075,
      "learning_rate": 9.926505571612008e-06,
      "loss": 1.1868,
      "step": 10575
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.3040202484705175,
      "learning_rate": 9.92629511289642e-06,
      "loss": 1.2072,
      "step": 10580
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.341495925088784,
      "learning_rate": 9.926084355514073e-06,
      "loss": 1.202,
      "step": 10585
    },
    {
      "epoch": 0.45,
      "grad_norm": 28.351874817946406,
      "learning_rate": 9.92587329947774e-06,
      "loss": 1.1906,
      "step": 10590
    },
    {
      "epoch": 0.45,
      "grad_norm": 25.798690096809228,
      "learning_rate": 9.925661944800218e-06,
      "loss": 1.1903,
      "step": 10595
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.5458717369179285,
      "learning_rate": 9.92545029149432e-06,
      "loss": 1.2071,
      "step": 10600
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.424439651233213,
      "learning_rate": 9.925238339572883e-06,
      "loss": 1.1945,
      "step": 10605
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.717217661607357,
      "learning_rate": 9.92502608904875e-06,
      "loss": 1.1704,
      "step": 10610
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.751766054459633,
      "learning_rate": 9.924813539934794e-06,
      "loss": 1.1939,
      "step": 10615
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.462968047747415,
      "learning_rate": 9.924600692243898e-06,
      "loss": 1.1689,
      "step": 10620
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.607152682053208,
      "learning_rate": 9.92438754598897e-06,
      "loss": 1.2046,
      "step": 10625
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.2681757986446645,
      "learning_rate": 9.924174101182928e-06,
      "loss": 1.1635,
      "step": 10630
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.920607135261278,
      "learning_rate": 9.923960357838717e-06,
      "loss": 1.1499,
      "step": 10635
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.156130633446434,
      "learning_rate": 9.923746315969292e-06,
      "loss": 1.194,
      "step": 10640
    },
    {
      "epoch": 0.45,
      "grad_norm": 16.3057165190141,
      "learning_rate": 9.923531975587633e-06,
      "loss": 1.1468,
      "step": 10645
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.44207288541131,
      "learning_rate": 9.923317336706732e-06,
      "loss": 1.1572,
      "step": 10650
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.491729437834684,
      "learning_rate": 9.923102399339602e-06,
      "loss": 1.1538,
      "step": 10655
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.769953007994715,
      "learning_rate": 9.922887163499276e-06,
      "loss": 1.2177,
      "step": 10660
    },
    {
      "epoch": 0.45,
      "grad_norm": 19.14152021932703,
      "learning_rate": 9.922671629198802e-06,
      "loss": 1.2153,
      "step": 10665
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.978992801364809,
      "learning_rate": 9.922455796451249e-06,
      "loss": 1.2067,
      "step": 10670
    },
    {
      "epoch": 0.45,
      "grad_norm": 12.661404526341018,
      "learning_rate": 9.922239665269697e-06,
      "loss": 1.1566,
      "step": 10675
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.327352607288402,
      "learning_rate": 9.922023235667255e-06,
      "loss": 1.1975,
      "step": 10680
    },
    {
      "epoch": 0.45,
      "grad_norm": 11.745619312070527,
      "learning_rate": 9.921806507657044e-06,
      "loss": 1.2059,
      "step": 10685
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.176168882922202,
      "learning_rate": 9.921589481252201e-06,
      "loss": 1.1482,
      "step": 10690
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.78473586808955,
      "learning_rate": 9.921372156465884e-06,
      "loss": 1.1564,
      "step": 10695
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.819308712859828,
      "learning_rate": 9.92115453331127e-06,
      "loss": 1.1899,
      "step": 10700
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.4444198557490315,
      "learning_rate": 9.920936611801552e-06,
      "loss": 1.1829,
      "step": 10705
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.0103124452863135,
      "learning_rate": 9.920718391949943e-06,
      "loss": 1.1114,
      "step": 10710
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.218072493475162,
      "learning_rate": 9.920499873769671e-06,
      "loss": 1.1903,
      "step": 10715
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.811183074814596,
      "learning_rate": 9.920281057273988e-06,
      "loss": 1.1242,
      "step": 10720
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.6195642267455765,
      "learning_rate": 9.920061942476156e-06,
      "loss": 1.1823,
      "step": 10725
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.882686024666247,
      "learning_rate": 9.91984252938946e-06,
      "loss": 1.1829,
      "step": 10730
    },
    {
      "epoch": 0.45,
      "grad_norm": 7.687098526835065,
      "learning_rate": 9.919622818027204e-06,
      "loss": 1.1504,
      "step": 10735
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.901209944181845,
      "learning_rate": 9.91940280840271e-06,
      "loss": 1.1802,
      "step": 10740
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.741967202730921,
      "learning_rate": 9.919182500529311e-06,
      "loss": 1.1603,
      "step": 10745
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.49314460818403,
      "learning_rate": 9.91896189442037e-06,
      "loss": 1.1385,
      "step": 10750
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.846613960624861,
      "learning_rate": 9.918740990089256e-06,
      "loss": 1.1914,
      "step": 10755
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.628152978231483,
      "learning_rate": 9.918519787549366e-06,
      "loss": 1.1483,
      "step": 10760
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.360298756997522,
      "learning_rate": 9.918298286814108e-06,
      "loss": 1.1382,
      "step": 10765
    },
    {
      "epoch": 0.46,
      "grad_norm": 17.051419942281832,
      "learning_rate": 9.918076487896912e-06,
      "loss": 1.1592,
      "step": 10770
    },
    {
      "epoch": 0.46,
      "grad_norm": 17.184741407663875,
      "learning_rate": 9.917854390811227e-06,
      "loss": 1.1841,
      "step": 10775
    },
    {
      "epoch": 0.46,
      "grad_norm": 20.95540204270835,
      "learning_rate": 9.917631995570516e-06,
      "loss": 1.1757,
      "step": 10780
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.5651707387832205,
      "learning_rate": 9.917409302188261e-06,
      "loss": 1.1859,
      "step": 10785
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.32757135428432,
      "learning_rate": 9.917186310677967e-06,
      "loss": 1.1413,
      "step": 10790
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.345522140302892,
      "learning_rate": 9.916963021053148e-06,
      "loss": 1.2138,
      "step": 10795
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.546032560435066,
      "learning_rate": 9.916739433327347e-06,
      "loss": 1.1614,
      "step": 10800
    },
    {
      "epoch": 0.46,
      "grad_norm": 15.037771854499969,
      "learning_rate": 9.916515547514116e-06,
      "loss": 1.1657,
      "step": 10805
    },
    {
      "epoch": 0.46,
      "grad_norm": 12.962530625932569,
      "learning_rate": 9.916291363627028e-06,
      "loss": 1.1548,
      "step": 10810
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.733858126523737,
      "learning_rate": 9.916066881679678e-06,
      "loss": 1.2221,
      "step": 10815
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.4197775228783485,
      "learning_rate": 9.915842101685672e-06,
      "loss": 1.2215,
      "step": 10820
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.446367708194145,
      "learning_rate": 9.915617023658642e-06,
      "loss": 1.1865,
      "step": 10825
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.909688736237768,
      "learning_rate": 9.915391647612228e-06,
      "loss": 1.1873,
      "step": 10830
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.560801320774795,
      "learning_rate": 9.9151659735601e-06,
      "loss": 1.116,
      "step": 10835
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.92035753231308,
      "learning_rate": 9.914940001515937e-06,
      "loss": 1.131,
      "step": 10840
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.47640382614778,
      "learning_rate": 9.914713731493437e-06,
      "loss": 1.1332,
      "step": 10845
    },
    {
      "epoch": 0.46,
      "grad_norm": 13.442824328039446,
      "learning_rate": 9.914487163506323e-06,
      "loss": 1.1487,
      "step": 10850
    },
    {
      "epoch": 0.46,
      "grad_norm": 10.219963387438971,
      "learning_rate": 9.914260297568326e-06,
      "loss": 1.1919,
      "step": 10855
    },
    {
      "epoch": 0.46,
      "grad_norm": 17.830512400940798,
      "learning_rate": 9.914033133693204e-06,
      "loss": 1.1713,
      "step": 10860
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.115427526870187,
      "learning_rate": 9.913805671894727e-06,
      "loss": 1.1565,
      "step": 10865
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.363191331256874,
      "learning_rate": 9.913577912186688e-06,
      "loss": 1.1689,
      "step": 10870
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.708693374997952,
      "learning_rate": 9.91334985458289e-06,
      "loss": 1.1921,
      "step": 10875
    },
    {
      "epoch": 0.46,
      "grad_norm": 11.043603686528792,
      "learning_rate": 9.913121499097165e-06,
      "loss": 1.1896,
      "step": 10880
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.127183021152288,
      "learning_rate": 9.912892845743356e-06,
      "loss": 1.156,
      "step": 10885
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.784181488898342,
      "learning_rate": 9.912663894535325e-06,
      "loss": 1.1585,
      "step": 10890
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.946782241883635,
      "learning_rate": 9.912434645486953e-06,
      "loss": 1.1534,
      "step": 10895
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.684903220320171,
      "learning_rate": 9.912205098612138e-06,
      "loss": 1.1408,
      "step": 10900
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.512343867443584,
      "learning_rate": 9.911975253924797e-06,
      "loss": 1.1632,
      "step": 10905
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.717541036271599,
      "learning_rate": 9.911745111438867e-06,
      "loss": 1.1869,
      "step": 10910
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.509526494615323,
      "learning_rate": 9.911514671168296e-06,
      "loss": 1.148,
      "step": 10915
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.182359008080275,
      "learning_rate": 9.911283933127061e-06,
      "loss": 1.1491,
      "step": 10920
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.7066101472575665,
      "learning_rate": 9.911052897329145e-06,
      "loss": 1.1815,
      "step": 10925
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.857949064757787,
      "learning_rate": 9.91082156378856e-06,
      "loss": 1.1518,
      "step": 10930
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.724963982439529,
      "learning_rate": 9.910589932519327e-06,
      "loss": 1.1351,
      "step": 10935
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.159897620900517,
      "learning_rate": 9.910358003535491e-06,
      "loss": 1.1423,
      "step": 10940
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.45715717713818,
      "learning_rate": 9.910125776851114e-06,
      "loss": 1.131,
      "step": 10945
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.8028978521572725,
      "learning_rate": 9.909893252480273e-06,
      "loss": 1.1664,
      "step": 10950
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.87835802927879,
      "learning_rate": 9.90966043043707e-06,
      "loss": 1.204,
      "step": 10955
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.652741016209589,
      "learning_rate": 9.909427310735615e-06,
      "loss": 1.159,
      "step": 10960
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.679135125939183,
      "learning_rate": 9.909193893390043e-06,
      "loss": 1.1371,
      "step": 10965
    },
    {
      "epoch": 0.46,
      "grad_norm": 9.593510965900318,
      "learning_rate": 9.908960178414505e-06,
      "loss": 1.1884,
      "step": 10970
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.342203549577611,
      "learning_rate": 9.908726165823172e-06,
      "loss": 1.1862,
      "step": 10975
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.65191266709501,
      "learning_rate": 9.908491855630232e-06,
      "loss": 1.133,
      "step": 10980
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.986985423965624,
      "learning_rate": 9.90825724784989e-06,
      "loss": 1.1273,
      "step": 10985
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.8708933815736755,
      "learning_rate": 9.908022342496365e-06,
      "loss": 1.1642,
      "step": 10990
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.22685427868934,
      "learning_rate": 9.907787139583904e-06,
      "loss": 1.1549,
      "step": 10995
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.662857809878343,
      "learning_rate": 9.907551639126766e-06,
      "loss": 1.186,
      "step": 11000
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.706587813035455,
      "learning_rate": 9.907315841139226e-06,
      "loss": 1.2093,
      "step": 11005
    },
    {
      "epoch": 0.47,
      "grad_norm": 23.324541045987278,
      "learning_rate": 9.907079745635583e-06,
      "loss": 1.1803,
      "step": 11010
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.8620193130054945,
      "learning_rate": 9.90684335263015e-06,
      "loss": 1.1758,
      "step": 11015
    },
    {
      "epoch": 0.47,
      "grad_norm": 28.110493558181762,
      "learning_rate": 9.906606662137257e-06,
      "loss": 1.1587,
      "step": 11020
    },
    {
      "epoch": 0.47,
      "grad_norm": 20.75697424936364,
      "learning_rate": 9.906369674171257e-06,
      "loss": 1.1616,
      "step": 11025
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.74740157103678,
      "learning_rate": 9.906132388746512e-06,
      "loss": 1.1667,
      "step": 11030
    },
    {
      "epoch": 0.47,
      "grad_norm": 16.221646397570733,
      "learning_rate": 9.905894805877413e-06,
      "loss": 1.1574,
      "step": 11035
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.315429426999328,
      "learning_rate": 9.905656925578365e-06,
      "loss": 1.1335,
      "step": 11040
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.289812467695272,
      "learning_rate": 9.905418747863786e-06,
      "loss": 1.1973,
      "step": 11045
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.691144253246272,
      "learning_rate": 9.905180272748119e-06,
      "loss": 1.2163,
      "step": 11050
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.125900649933367,
      "learning_rate": 9.90494150024582e-06,
      "loss": 1.1528,
      "step": 11055
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.967571187709722,
      "learning_rate": 9.904702430371365e-06,
      "loss": 1.129,
      "step": 11060
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.232701180420777,
      "learning_rate": 9.904463063139249e-06,
      "loss": 1.1288,
      "step": 11065
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.9823375151125235,
      "learning_rate": 9.904223398563983e-06,
      "loss": 1.1561,
      "step": 11070
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.328654473530816,
      "learning_rate": 9.903983436660101e-06,
      "loss": 1.1787,
      "step": 11075
    },
    {
      "epoch": 0.47,
      "grad_norm": 21.760572758974128,
      "learning_rate": 9.903743177442146e-06,
      "loss": 1.2078,
      "step": 11080
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.31790366894579,
      "learning_rate": 9.903502620924688e-06,
      "loss": 1.2201,
      "step": 11085
    },
    {
      "epoch": 0.47,
      "grad_norm": 24.48203400676599,
      "learning_rate": 9.90326176712231e-06,
      "loss": 1.1926,
      "step": 11090
    },
    {
      "epoch": 0.47,
      "grad_norm": 17.618875344588776,
      "learning_rate": 9.903020616049613e-06,
      "loss": 1.2167,
      "step": 11095
    },
    {
      "epoch": 0.47,
      "grad_norm": 17.325643244351472,
      "learning_rate": 9.90277916772122e-06,
      "loss": 1.1709,
      "step": 11100
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.934175497898546,
      "learning_rate": 9.902537422151766e-06,
      "loss": 1.2131,
      "step": 11105
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.16973069180265,
      "learning_rate": 9.90229537935591e-06,
      "loss": 1.1607,
      "step": 11110
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.288594806946907,
      "learning_rate": 9.902053039348325e-06,
      "loss": 1.1726,
      "step": 11115
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.854571230860627,
      "learning_rate": 9.901810402143703e-06,
      "loss": 1.1653,
      "step": 11120
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.151482995214919,
      "learning_rate": 9.901567467756757e-06,
      "loss": 1.1681,
      "step": 11125
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.632971060096517,
      "learning_rate": 9.90132423620221e-06,
      "loss": 1.1442,
      "step": 11130
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.215110069184859,
      "learning_rate": 9.901080707494814e-06,
      "loss": 1.0976,
      "step": 11135
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.286433333904665,
      "learning_rate": 9.900836881649332e-06,
      "loss": 1.1856,
      "step": 11140
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.377274914922964,
      "learning_rate": 9.900592758680544e-06,
      "loss": 1.2035,
      "step": 11145
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.379698830333417,
      "learning_rate": 9.900348338603256e-06,
      "loss": 1.1318,
      "step": 11150
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.058571170997721,
      "learning_rate": 9.900103621432278e-06,
      "loss": 1.1988,
      "step": 11155
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.483186687397655,
      "learning_rate": 9.899858607182454e-06,
      "loss": 1.1307,
      "step": 11160
    },
    {
      "epoch": 0.47,
      "grad_norm": 15.372143287190866,
      "learning_rate": 9.899613295868635e-06,
      "loss": 1.1436,
      "step": 11165
    },
    {
      "epoch": 0.47,
      "grad_norm": 15.441760553757844,
      "learning_rate": 9.899367687505695e-06,
      "loss": 1.1663,
      "step": 11170
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.067452984345042,
      "learning_rate": 9.899121782108521e-06,
      "loss": 1.1389,
      "step": 11175
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.493878741153034,
      "learning_rate": 9.898875579692027e-06,
      "loss": 1.1334,
      "step": 11180
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.877351187180319,
      "learning_rate": 9.898629080271135e-06,
      "loss": 1.1579,
      "step": 11185
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.306766372634094,
      "learning_rate": 9.898382283860792e-06,
      "loss": 1.1591,
      "step": 11190
    },
    {
      "epoch": 0.47,
      "grad_norm": 13.997116383859714,
      "learning_rate": 9.89813519047596e-06,
      "loss": 1.1434,
      "step": 11195
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.456764815120626,
      "learning_rate": 9.897887800131618e-06,
      "loss": 1.2191,
      "step": 11200
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.380024381576704,
      "learning_rate": 9.897640112842769e-06,
      "loss": 1.1698,
      "step": 11205
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.062180917583801,
      "learning_rate": 9.897392128624422e-06,
      "loss": 1.1533,
      "step": 11210
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.372322311934546,
      "learning_rate": 9.897143847491619e-06,
      "loss": 1.1595,
      "step": 11215
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.137540645494184,
      "learning_rate": 9.896895269459408e-06,
      "loss": 1.1745,
      "step": 11220
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.488009026628987,
      "learning_rate": 9.896646394542861e-06,
      "loss": 1.1979,
      "step": 11225
    },
    {
      "epoch": 0.48,
      "grad_norm": 21.479331352304104,
      "learning_rate": 9.896397222757069e-06,
      "loss": 1.1748,
      "step": 11230
    },
    {
      "epoch": 0.48,
      "grad_norm": 32.17705539395102,
      "learning_rate": 9.896147754117133e-06,
      "loss": 1.1763,
      "step": 11235
    },
    {
      "epoch": 0.48,
      "grad_norm": 29.641023561579832,
      "learning_rate": 9.895897988638184e-06,
      "loss": 1.1612,
      "step": 11240
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.618612499955207,
      "learning_rate": 9.895647926335359e-06,
      "loss": 1.1143,
      "step": 11245
    },
    {
      "epoch": 0.48,
      "grad_norm": 24.552763013483638,
      "learning_rate": 9.895397567223821e-06,
      "loss": 1.1828,
      "step": 11250
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.827877276813723,
      "learning_rate": 9.895146911318749e-06,
      "loss": 1.1868,
      "step": 11255
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.126715833569445,
      "learning_rate": 9.89489595863534e-06,
      "loss": 1.1495,
      "step": 11260
    },
    {
      "epoch": 0.48,
      "grad_norm": 20.218207112675735,
      "learning_rate": 9.894644709188805e-06,
      "loss": 1.1768,
      "step": 11265
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.921473798859457,
      "learning_rate": 9.894393162994381e-06,
      "loss": 1.199,
      "step": 11270
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.33166151881928,
      "learning_rate": 9.894141320067316e-06,
      "loss": 1.1707,
      "step": 11275
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.565354823955448,
      "learning_rate": 9.893889180422878e-06,
      "loss": 1.164,
      "step": 11280
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.776125898509952,
      "learning_rate": 9.893636744076357e-06,
      "loss": 1.198,
      "step": 11285
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.152371316351982,
      "learning_rate": 9.89338401104305e-06,
      "loss": 1.1801,
      "step": 11290
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.126109015757814,
      "learning_rate": 9.893130981338289e-06,
      "loss": 1.1661,
      "step": 11295
    },
    {
      "epoch": 0.48,
      "grad_norm": 14.083493515860624,
      "learning_rate": 9.892877654977407e-06,
      "loss": 1.1731,
      "step": 11300
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.998715081522729,
      "learning_rate": 9.892624031975766e-06,
      "loss": 1.1692,
      "step": 11305
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.411370045012572,
      "learning_rate": 9.89237011234874e-06,
      "loss": 1.1651,
      "step": 11310
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.578768938585467,
      "learning_rate": 9.89211589611173e-06,
      "loss": 1.1566,
      "step": 11315
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.316568316732369,
      "learning_rate": 9.89186138328014e-06,
      "loss": 1.105,
      "step": 11320
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.548768655491145,
      "learning_rate": 9.891606573869403e-06,
      "loss": 1.1269,
      "step": 11325
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.0706108911607455,
      "learning_rate": 9.89135146789497e-06,
      "loss": 1.1464,
      "step": 11330
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.419681955203033,
      "learning_rate": 9.891096065372304e-06,
      "loss": 1.1455,
      "step": 11335
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.676930979505991,
      "learning_rate": 9.890840366316892e-06,
      "loss": 1.2126,
      "step": 11340
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.52047652324343,
      "learning_rate": 9.890584370744233e-06,
      "loss": 1.1998,
      "step": 11345
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.455724561809516,
      "learning_rate": 9.890328078669851e-06,
      "loss": 1.1378,
      "step": 11350
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.426712602811074,
      "learning_rate": 9.890071490109283e-06,
      "loss": 1.1505,
      "step": 11355
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.674302587106533,
      "learning_rate": 9.889814605078084e-06,
      "loss": 1.1526,
      "step": 11360
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.134097221101347,
      "learning_rate": 9.889557423591829e-06,
      "loss": 1.1458,
      "step": 11365
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.898009761601114,
      "learning_rate": 9.889299945666108e-06,
      "loss": 1.2127,
      "step": 11370
    },
    {
      "epoch": 0.48,
      "grad_norm": 9.632360731331874,
      "learning_rate": 9.889042171316537e-06,
      "loss": 1.1626,
      "step": 11375
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.307125513946716,
      "learning_rate": 9.888784100558739e-06,
      "loss": 1.1575,
      "step": 11380
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.228412416269242,
      "learning_rate": 9.88852573340836e-06,
      "loss": 1.1368,
      "step": 11385
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.802860959891278,
      "learning_rate": 9.888267069881066e-06,
      "loss": 1.1398,
      "step": 11390
    },
    {
      "epoch": 0.48,
      "grad_norm": 12.361479226155447,
      "learning_rate": 9.88800810999254e-06,
      "loss": 1.1186,
      "step": 11395
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.43506792761793,
      "learning_rate": 9.887748853758477e-06,
      "loss": 1.1355,
      "step": 11400
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.477860747686588,
      "learning_rate": 9.887489301194602e-06,
      "loss": 1.1583,
      "step": 11405
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.60647595615046,
      "learning_rate": 9.887229452316644e-06,
      "loss": 1.1852,
      "step": 11410
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.555330057575714,
      "learning_rate": 9.886969307140363e-06,
      "loss": 1.1814,
      "step": 11415
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.367462869352004,
      "learning_rate": 9.886708865681526e-06,
      "loss": 1.1296,
      "step": 11420
    },
    {
      "epoch": 0.48,
      "grad_norm": 11.8952387272428,
      "learning_rate": 9.886448127955925e-06,
      "loss": 1.1599,
      "step": 11425
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.796066164176459,
      "learning_rate": 9.886187093979369e-06,
      "loss": 1.1442,
      "step": 11430
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.384124771531093,
      "learning_rate": 9.88592576376768e-06,
      "loss": 1.1124,
      "step": 11435
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.365682983872836,
      "learning_rate": 9.885664137336706e-06,
      "loss": 1.1528,
      "step": 11440
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.466847423576245,
      "learning_rate": 9.885402214702306e-06,
      "loss": 1.1681,
      "step": 11445
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.85172158396155,
      "learning_rate": 9.88513999588036e-06,
      "loss": 1.1054,
      "step": 11450
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.954694471587122,
      "learning_rate": 9.884877480886765e-06,
      "loss": 1.1429,
      "step": 11455
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.109127539501378,
      "learning_rate": 9.884614669737439e-06,
      "loss": 1.2066,
      "step": 11460
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.70820384291903,
      "learning_rate": 9.88435156244831e-06,
      "loss": 1.1445,
      "step": 11465
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.487129277691837,
      "learning_rate": 9.884088159035336e-06,
      "loss": 1.1642,
      "step": 11470
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.258581288924794,
      "learning_rate": 9.883824459514484e-06,
      "loss": 1.1408,
      "step": 11475
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.318640774823393,
      "learning_rate": 9.883560463901738e-06,
      "loss": 1.1729,
      "step": 11480
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.708089259801228,
      "learning_rate": 9.883296172213108e-06,
      "loss": 1.1813,
      "step": 11485
    },
    {
      "epoch": 0.49,
      "grad_norm": 11.455045503263648,
      "learning_rate": 9.883031584464616e-06,
      "loss": 1.174,
      "step": 11490
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.396411284344936,
      "learning_rate": 9.882766700672301e-06,
      "loss": 1.1069,
      "step": 11495
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.535704924614947,
      "learning_rate": 9.882501520852223e-06,
      "loss": 1.2267,
      "step": 11500
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.242712743571305,
      "learning_rate": 9.882236045020462e-06,
      "loss": 1.1192,
      "step": 11505
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.20033342471653,
      "learning_rate": 9.881970273193109e-06,
      "loss": 1.1573,
      "step": 11510
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.817606417347065,
      "learning_rate": 9.88170420538628e-06,
      "loss": 1.1386,
      "step": 11515
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.848578061698391,
      "learning_rate": 9.881437841616103e-06,
      "loss": 1.1566,
      "step": 11520
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.89601129169758,
      "learning_rate": 9.88117118189873e-06,
      "loss": 1.1164,
      "step": 11525
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.32008606488712,
      "learning_rate": 9.880904226250325e-06,
      "loss": 1.1142,
      "step": 11530
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.772024052164713,
      "learning_rate": 9.880636974687074e-06,
      "loss": 1.1134,
      "step": 11535
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.853995128338482,
      "learning_rate": 9.88036942722518e-06,
      "loss": 1.133,
      "step": 11540
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.69844691158747,
      "learning_rate": 9.880101583880863e-06,
      "loss": 1.1452,
      "step": 11545
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.587696727294008,
      "learning_rate": 9.879833444670361e-06,
      "loss": 1.192,
      "step": 11550
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.417324970434418,
      "learning_rate": 9.879565009609933e-06,
      "loss": 1.1693,
      "step": 11555
    },
    {
      "epoch": 0.49,
      "grad_norm": 19.74729653004002,
      "learning_rate": 9.87929627871585e-06,
      "loss": 1.1348,
      "step": 11560
    },
    {
      "epoch": 0.49,
      "grad_norm": 17.75267370519572,
      "learning_rate": 9.879027252004408e-06,
      "loss": 1.1499,
      "step": 11565
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.246657708584769,
      "learning_rate": 9.878757929491915e-06,
      "loss": 1.1304,
      "step": 11570
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.807281544727722,
      "learning_rate": 9.878488311194698e-06,
      "loss": 1.1753,
      "step": 11575
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.60747522287729,
      "learning_rate": 9.878218397129106e-06,
      "loss": 1.1186,
      "step": 11580
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.474045394785346,
      "learning_rate": 9.877948187311501e-06,
      "loss": 1.162,
      "step": 11585
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.7293988831557625,
      "learning_rate": 9.877677681758267e-06,
      "loss": 1.1007,
      "step": 11590
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.72239460734512,
      "learning_rate": 9.877406880485803e-06,
      "loss": 1.1875,
      "step": 11595
    },
    {
      "epoch": 0.49,
      "grad_norm": 10.430059083077545,
      "learning_rate": 9.877135783510526e-06,
      "loss": 1.1778,
      "step": 11600
    },
    {
      "epoch": 0.49,
      "grad_norm": 15.11322880660984,
      "learning_rate": 9.876864390848871e-06,
      "loss": 1.1018,
      "step": 11605
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.953724958813943,
      "learning_rate": 9.876592702517294e-06,
      "loss": 1.1063,
      "step": 11610
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.047768168085878,
      "learning_rate": 9.876320718532268e-06,
      "loss": 1.1484,
      "step": 11615
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.633786295620178,
      "learning_rate": 9.876048438910279e-06,
      "loss": 1.128,
      "step": 11620
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.325124078791975,
      "learning_rate": 9.875775863667835e-06,
      "loss": 1.1114,
      "step": 11625
    },
    {
      "epoch": 0.49,
      "grad_norm": 9.767935984740749,
      "learning_rate": 9.875502992821464e-06,
      "loss": 1.145,
      "step": 11630
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.112306088496393,
      "learning_rate": 9.875229826387706e-06,
      "loss": 1.1438,
      "step": 11635
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.120903752262135,
      "learning_rate": 9.874956364383127e-06,
      "loss": 1.1587,
      "step": 11640
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.986091808075715,
      "learning_rate": 9.874682606824301e-06,
      "loss": 1.131,
      "step": 11645
    },
    {
      "epoch": 0.49,
      "grad_norm": 22.406463114043778,
      "learning_rate": 9.874408553727828e-06,
      "loss": 1.1337,
      "step": 11650
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.712704924972094,
      "learning_rate": 9.874134205110325e-06,
      "loss": 1.1075,
      "step": 11655
    },
    {
      "epoch": 0.49,
      "grad_norm": 25.079555102890428,
      "learning_rate": 9.87385956098842e-06,
      "loss": 1.166,
      "step": 11660
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.954483214737107,
      "learning_rate": 9.873584621378765e-06,
      "loss": 1.1655,
      "step": 11665
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.097962197693542,
      "learning_rate": 9.873309386298031e-06,
      "loss": 1.1765,
      "step": 11670
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.677980010760047,
      "learning_rate": 9.873033855762903e-06,
      "loss": 1.1326,
      "step": 11675
    },
    {
      "epoch": 0.49,
      "grad_norm": 19.441939212176987,
      "learning_rate": 9.872758029790088e-06,
      "loss": 1.1481,
      "step": 11680
    },
    {
      "epoch": 0.49,
      "grad_norm": 20.724179910793424,
      "learning_rate": 9.872481908396305e-06,
      "loss": 1.1623,
      "step": 11685
    },
    {
      "epoch": 0.49,
      "grad_norm": 27.182337162688494,
      "learning_rate": 9.872205491598299e-06,
      "loss": 1.1807,
      "step": 11690
    },
    {
      "epoch": 0.49,
      "grad_norm": 18.04984628427394,
      "learning_rate": 9.871928779412822e-06,
      "loss": 1.1557,
      "step": 11695
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.077281823470901,
      "learning_rate": 9.871651771856656e-06,
      "loss": 1.1259,
      "step": 11700
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.989097544793694,
      "learning_rate": 9.871374468946593e-06,
      "loss": 1.1845,
      "step": 11705
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.182163676999648,
      "learning_rate": 9.871096870699446e-06,
      "loss": 1.2276,
      "step": 11710
    },
    {
      "epoch": 0.5,
      "grad_norm": 15.305742794562596,
      "learning_rate": 9.870818977132042e-06,
      "loss": 1.1558,
      "step": 11715
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.286994580972955,
      "learning_rate": 9.870540788261232e-06,
      "loss": 1.1256,
      "step": 11720
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.976057077609181,
      "learning_rate": 9.87026230410388e-06,
      "loss": 1.151,
      "step": 11725
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.290155160380621,
      "learning_rate": 9.869983524676872e-06,
      "loss": 1.112,
      "step": 11730
    },
    {
      "epoch": 0.5,
      "grad_norm": 15.640665605558757,
      "learning_rate": 9.869704449997106e-06,
      "loss": 1.1689,
      "step": 11735
    },
    {
      "epoch": 0.5,
      "grad_norm": 19.726793316037796,
      "learning_rate": 9.869425080081504e-06,
      "loss": 1.1257,
      "step": 11740
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.918565976829093,
      "learning_rate": 9.869145414947002e-06,
      "loss": 1.1265,
      "step": 11745
    },
    {
      "epoch": 0.5,
      "grad_norm": 24.131782175561387,
      "learning_rate": 9.868865454610557e-06,
      "loss": 1.1579,
      "step": 11750
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.8976802694399035,
      "learning_rate": 9.868585199089141e-06,
      "loss": 1.1422,
      "step": 11755
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.921775189693692,
      "learning_rate": 9.868304648399744e-06,
      "loss": 1.172,
      "step": 11760
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.36851777741143,
      "learning_rate": 9.86802380255938e-06,
      "loss": 1.1871,
      "step": 11765
    },
    {
      "epoch": 0.5,
      "grad_norm": 16.40339172531344,
      "learning_rate": 9.867742661585069e-06,
      "loss": 1.1623,
      "step": 11770
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.176582100990586,
      "learning_rate": 9.86746122549386e-06,
      "loss": 1.1564,
      "step": 11775
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.080872944075164,
      "learning_rate": 9.867179494302812e-06,
      "loss": 1.1331,
      "step": 11780
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.506834876776587,
      "learning_rate": 9.866897468029011e-06,
      "loss": 1.1192,
      "step": 11785
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.992554052031743,
      "learning_rate": 9.866615146689551e-06,
      "loss": 1.1307,
      "step": 11790
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.137574257507623,
      "learning_rate": 9.86633253030155e-06,
      "loss": 1.1357,
      "step": 11795
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.130393457328972,
      "learning_rate": 9.866049618882142e-06,
      "loss": 1.1271,
      "step": 11800
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.689287447938648,
      "learning_rate": 9.865766412448479e-06,
      "loss": 1.1208,
      "step": 11805
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.993566818309528,
      "learning_rate": 9.865482911017731e-06,
      "loss": 1.1143,
      "step": 11810
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.610987656025143,
      "learning_rate": 9.865199114607085e-06,
      "loss": 1.1449,
      "step": 11815
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.86202202993207,
      "learning_rate": 9.86491502323375e-06,
      "loss": 1.1436,
      "step": 11820
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.824515678162103,
      "learning_rate": 9.864630636914945e-06,
      "loss": 1.1821,
      "step": 11825
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.742485790665893,
      "learning_rate": 9.864345955667914e-06,
      "loss": 1.135,
      "step": 11830
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.738242649379986,
      "learning_rate": 9.864060979509917e-06,
      "loss": 1.1031,
      "step": 11835
    },
    {
      "epoch": 0.5,
      "grad_norm": 22.016315223740655,
      "learning_rate": 9.86377570845823e-06,
      "loss": 1.1167,
      "step": 11840
    },
    {
      "epoch": 0.5,
      "grad_norm": 20.104340436977523,
      "learning_rate": 9.863490142530148e-06,
      "loss": 1.1509,
      "step": 11845
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.793524747206087,
      "learning_rate": 9.863204281742986e-06,
      "loss": 1.1273,
      "step": 11850
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.055343078022386,
      "learning_rate": 9.862918126114072e-06,
      "loss": 1.1877,
      "step": 11855
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.114627805232248,
      "learning_rate": 9.862631675660756e-06,
      "loss": 1.1714,
      "step": 11860
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.36176240385175,
      "learning_rate": 9.862344930400404e-06,
      "loss": 1.1365,
      "step": 11865
    },
    {
      "epoch": 0.5,
      "grad_norm": 9.119053552631945,
      "learning_rate": 9.862057890350404e-06,
      "loss": 1.1911,
      "step": 11870
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.5555333501647866,
      "learning_rate": 9.861770555528155e-06,
      "loss": 1.112,
      "step": 11875
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.8162057066325525,
      "learning_rate": 9.861482925951078e-06,
      "loss": 1.1905,
      "step": 11880
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.992003210854348,
      "learning_rate": 9.86119500163661e-06,
      "loss": 1.1283,
      "step": 11885
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.889743370342024,
      "learning_rate": 9.86090678260221e-06,
      "loss": 1.1337,
      "step": 11890
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.803355038794472,
      "learning_rate": 9.860618268865348e-06,
      "loss": 1.1648,
      "step": 11895
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.987428176412822,
      "learning_rate": 9.860329460443519e-06,
      "loss": 1.1388,
      "step": 11900
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.44307176779172,
      "learning_rate": 9.860040357354232e-06,
      "loss": 1.1044,
      "step": 11905
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.738452427336835,
      "learning_rate": 9.859750959615012e-06,
      "loss": 1.1741,
      "step": 11910
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.93914151574783,
      "learning_rate": 9.859461267243407e-06,
      "loss": 1.2008,
      "step": 11915
    },
    {
      "epoch": 0.5,
      "grad_norm": 14.973737260555566,
      "learning_rate": 9.859171280256981e-06,
      "loss": 1.1892,
      "step": 11920
    },
    {
      "epoch": 0.5,
      "grad_norm": 15.26786668503323,
      "learning_rate": 9.85888099867331e-06,
      "loss": 1.1432,
      "step": 11925
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.213308210695427,
      "learning_rate": 9.858590422509999e-06,
      "loss": 1.1755,
      "step": 11930
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.4987460503540895,
      "learning_rate": 9.858299551784662e-06,
      "loss": 1.1261,
      "step": 11935
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.231781553337773,
      "learning_rate": 9.858008386514933e-06,
      "loss": 1.1421,
      "step": 11940
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.494777253475009,
      "learning_rate": 9.857716926718465e-06,
      "loss": 1.0886,
      "step": 11945
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.847364984317209,
      "learning_rate": 9.85742517241293e-06,
      "loss": 1.154,
      "step": 11950
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.527729858172997,
      "learning_rate": 9.857133123616013e-06,
      "loss": 1.1518,
      "step": 11955
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.682313498314514,
      "learning_rate": 9.856840780345423e-06,
      "loss": 1.197,
      "step": 11960
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.54415687857973,
      "learning_rate": 9.856548142618882e-06,
      "loss": 1.1779,
      "step": 11965
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.561235190352997,
      "learning_rate": 9.856255210454134e-06,
      "loss": 1.1911,
      "step": 11970
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.241505704266924,
      "learning_rate": 9.855961983868936e-06,
      "loss": 1.1173,
      "step": 11975
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.585499314172027,
      "learning_rate": 9.855668462881068e-06,
      "loss": 1.1293,
      "step": 11980
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.90281683007015,
      "learning_rate": 9.855374647508322e-06,
      "loss": 1.1305,
      "step": 11985
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.18070927433015,
      "learning_rate": 9.855080537768515e-06,
      "loss": 1.1195,
      "step": 11990
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.334953659829334,
      "learning_rate": 9.854786133679474e-06,
      "loss": 1.1617,
      "step": 11995
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.425613823444337,
      "learning_rate": 9.854491435259051e-06,
      "loss": 1.1605,
      "step": 12000
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.043819698721772,
      "learning_rate": 9.854196442525113e-06,
      "loss": 1.1403,
      "step": 12005
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.840243726585245,
      "learning_rate": 9.85390115549554e-06,
      "loss": 1.1572,
      "step": 12010
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.902078776899268,
      "learning_rate": 9.853605574188241e-06,
      "loss": 1.1378,
      "step": 12015
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.4409833323953505,
      "learning_rate": 9.85330969862113e-06,
      "loss": 1.1054,
      "step": 12020
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.569106348711084,
      "learning_rate": 9.85301352881215e-06,
      "loss": 1.1288,
      "step": 12025
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.641174835617154,
      "learning_rate": 9.852717064779253e-06,
      "loss": 1.1175,
      "step": 12030
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.205088887888452,
      "learning_rate": 9.852420306540416e-06,
      "loss": 1.1288,
      "step": 12035
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.499418937192942,
      "learning_rate": 9.852123254113628e-06,
      "loss": 1.1452,
      "step": 12040
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.598682355484569,
      "learning_rate": 9.851825907516901e-06,
      "loss": 1.1074,
      "step": 12045
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.358570349103616,
      "learning_rate": 9.85152826676826e-06,
      "loss": 1.1532,
      "step": 12050
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.60412566758258,
      "learning_rate": 9.85123033188575e-06,
      "loss": 1.1361,
      "step": 12055
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.030084181433548,
      "learning_rate": 9.850932102887435e-06,
      "loss": 1.1439,
      "step": 12060
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.751578427274406,
      "learning_rate": 9.850633579791395e-06,
      "loss": 1.1975,
      "step": 12065
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.70896980875338,
      "learning_rate": 9.85033476261573e-06,
      "loss": 1.134,
      "step": 12070
    },
    {
      "epoch": 0.51,
      "grad_norm": 14.877170969086212,
      "learning_rate": 9.850035651378556e-06,
      "loss": 1.1283,
      "step": 12075
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.75355538586889,
      "learning_rate": 9.849736246098006e-06,
      "loss": 1.1375,
      "step": 12080
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.401902740850932,
      "learning_rate": 9.849436546792232e-06,
      "loss": 1.112,
      "step": 12085
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.490671748975146,
      "learning_rate": 9.849136553479406e-06,
      "loss": 1.2094,
      "step": 12090
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.213197699095515,
      "learning_rate": 9.848836266177715e-06,
      "loss": 1.159,
      "step": 12095
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.130490153881233,
      "learning_rate": 9.848535684905362e-06,
      "loss": 1.1072,
      "step": 12100
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.175600104181329,
      "learning_rate": 9.848234809680573e-06,
      "loss": 1.1511,
      "step": 12105
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.363976292832451,
      "learning_rate": 9.847933640521588e-06,
      "loss": 1.1702,
      "step": 12110
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.123969685033667,
      "learning_rate": 9.847632177446666e-06,
      "loss": 1.1188,
      "step": 12115
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.306763945286356,
      "learning_rate": 9.847330420474086e-06,
      "loss": 1.1382,
      "step": 12120
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.023152029184388,
      "learning_rate": 9.84702836962214e-06,
      "loss": 1.1379,
      "step": 12125
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.908879001583405,
      "learning_rate": 9.846726024909141e-06,
      "loss": 1.0826,
      "step": 12130
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.461184144598489,
      "learning_rate": 9.84642338635342e-06,
      "loss": 1.14,
      "step": 12135
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.656567574368935,
      "learning_rate": 9.846120453973324e-06,
      "loss": 1.16,
      "step": 12140
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.545006166182492,
      "learning_rate": 9.84581722778722e-06,
      "loss": 1.1671,
      "step": 12145
    },
    {
      "epoch": 0.51,
      "grad_norm": 6.954054299452032,
      "learning_rate": 9.845513707813492e-06,
      "loss": 1.1496,
      "step": 12150
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.675309374480587,
      "learning_rate": 9.845209894070542e-06,
      "loss": 1.1485,
      "step": 12155
    },
    {
      "epoch": 0.51,
      "grad_norm": 11.551780199834928,
      "learning_rate": 9.844905786576785e-06,
      "loss": 1.1599,
      "step": 12160
    },
    {
      "epoch": 0.51,
      "grad_norm": 10.382406762670646,
      "learning_rate": 9.844601385350663e-06,
      "loss": 1.141,
      "step": 12165
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.548914499956982,
      "learning_rate": 9.84429669041063e-06,
      "loss": 1.1339,
      "step": 12170
    },
    {
      "epoch": 0.52,
      "grad_norm": 12.219673744661028,
      "learning_rate": 9.843991701775157e-06,
      "loss": 1.1126,
      "step": 12175
    },
    {
      "epoch": 0.52,
      "grad_norm": 18.084839847096443,
      "learning_rate": 9.843686419462735e-06,
      "loss": 1.1076,
      "step": 12180
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.595470266242617,
      "learning_rate": 9.843380843491873e-06,
      "loss": 1.2094,
      "step": 12185
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.273411631819693,
      "learning_rate": 9.843074973881098e-06,
      "loss": 1.0957,
      "step": 12190
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.8993901549616,
      "learning_rate": 9.842768810648953e-06,
      "loss": 1.1489,
      "step": 12195
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.101486783709928,
      "learning_rate": 9.842462353813999e-06,
      "loss": 1.1784,
      "step": 12200
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.08392377893216,
      "learning_rate": 9.842155603394817e-06,
      "loss": 1.1808,
      "step": 12205
    },
    {
      "epoch": 0.52,
      "grad_norm": 17.361090798023238,
      "learning_rate": 9.841848559410004e-06,
      "loss": 1.1442,
      "step": 12210
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.627036436334981,
      "learning_rate": 9.841541221878174e-06,
      "loss": 1.1487,
      "step": 12215
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.14019284124057,
      "learning_rate": 9.841233590817963e-06,
      "loss": 1.1025,
      "step": 12220
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.612791421299756,
      "learning_rate": 9.840925666248018e-06,
      "loss": 1.1266,
      "step": 12225
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.726548685526764,
      "learning_rate": 9.84061744818701e-06,
      "loss": 1.1427,
      "step": 12230
    },
    {
      "epoch": 0.52,
      "grad_norm": 9.314003498907114,
      "learning_rate": 9.840308936653624e-06,
      "loss": 1.145,
      "step": 12235
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.567299277357848,
      "learning_rate": 9.840000131666564e-06,
      "loss": 1.1142,
      "step": 12240
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.518454294916243,
      "learning_rate": 9.839691033244556e-06,
      "loss": 1.1624,
      "step": 12245
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.855018387313464,
      "learning_rate": 9.839381641406334e-06,
      "loss": 1.1292,
      "step": 12250
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.742207974064404,
      "learning_rate": 9.839071956170658e-06,
      "loss": 1.1543,
      "step": 12255
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.838812325933963,
      "learning_rate": 9.838761977556304e-06,
      "loss": 1.1783,
      "step": 12260
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.890962090327382,
      "learning_rate": 9.838451705582062e-06,
      "loss": 1.1565,
      "step": 12265
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.19375103386591,
      "learning_rate": 9.838141140266749e-06,
      "loss": 1.137,
      "step": 12270
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.889923065665591,
      "learning_rate": 9.837830281629186e-06,
      "loss": 1.0944,
      "step": 12275
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.170740026527037,
      "learning_rate": 9.837519129688227e-06,
      "loss": 1.1337,
      "step": 12280
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.290042217647155,
      "learning_rate": 9.837207684462729e-06,
      "loss": 1.1765,
      "step": 12285
    },
    {
      "epoch": 0.52,
      "grad_norm": 15.132686526221493,
      "learning_rate": 9.836895945971581e-06,
      "loss": 1.157,
      "step": 12290
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.92011040772267,
      "learning_rate": 9.836583914233677e-06,
      "loss": 1.1129,
      "step": 12295
    },
    {
      "epoch": 0.52,
      "grad_norm": 16.96459814968213,
      "learning_rate": 9.836271589267938e-06,
      "loss": 1.1892,
      "step": 12300
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.478207855501797,
      "learning_rate": 9.835958971093296e-06,
      "loss": 1.1556,
      "step": 12305
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.210182340022875,
      "learning_rate": 9.83564605972871e-06,
      "loss": 1.1311,
      "step": 12310
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.717631978597978,
      "learning_rate": 9.835332855193145e-06,
      "loss": 1.1124,
      "step": 12315
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.960093758316839,
      "learning_rate": 9.835019357505592e-06,
      "loss": 1.146,
      "step": 12320
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.015175594719165,
      "learning_rate": 9.834705566685058e-06,
      "loss": 1.1082,
      "step": 12325
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.36279854848031,
      "learning_rate": 9.834391482750566e-06,
      "loss": 1.1079,
      "step": 12330
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.009252765865015,
      "learning_rate": 9.834077105721159e-06,
      "loss": 1.1276,
      "step": 12335
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.048365631873782,
      "learning_rate": 9.833762435615897e-06,
      "loss": 1.1155,
      "step": 12340
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.785986992471564,
      "learning_rate": 9.833447472453855e-06,
      "loss": 1.1139,
      "step": 12345
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.911923101542478,
      "learning_rate": 9.83313221625413e-06,
      "loss": 1.1512,
      "step": 12350
    },
    {
      "epoch": 0.52,
      "grad_norm": 23.79428146410804,
      "learning_rate": 9.832816667035838e-06,
      "loss": 1.1757,
      "step": 12355
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.239342441896353,
      "learning_rate": 9.832500824818105e-06,
      "loss": 1.1321,
      "step": 12360
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.58435822384286,
      "learning_rate": 9.832184689620082e-06,
      "loss": 1.1473,
      "step": 12365
    },
    {
      "epoch": 0.52,
      "grad_norm": 13.56254222195042,
      "learning_rate": 9.831868261460935e-06,
      "loss": 1.1171,
      "step": 12370
    },
    {
      "epoch": 0.52,
      "grad_norm": 8.202953054801172,
      "learning_rate": 9.831551540359848e-06,
      "loss": 1.0989,
      "step": 12375
    },
    {
      "epoch": 0.52,
      "grad_norm": 14.997296884312815,
      "learning_rate": 9.831234526336022e-06,
      "loss": 1.161,
      "step": 12380
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.967783318989335,
      "learning_rate": 9.830917219408678e-06,
      "loss": 1.1303,
      "step": 12385
    },
    {
      "epoch": 0.52,
      "grad_norm": 11.072364252202307,
      "learning_rate": 9.830599619597055e-06,
      "loss": 1.1275,
      "step": 12390
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.005212657771845,
      "learning_rate": 9.830281726920403e-06,
      "loss": 1.1175,
      "step": 12395
    },
    {
      "epoch": 0.52,
      "grad_norm": 12.244484428979382,
      "learning_rate": 9.829963541398e-06,
      "loss": 1.156,
      "step": 12400
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.140462559704638,
      "learning_rate": 9.829645063049134e-06,
      "loss": 1.116,
      "step": 12405
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.5892189571569455,
      "learning_rate": 9.829326291893114e-06,
      "loss": 1.1308,
      "step": 12410
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.395369945832596,
      "learning_rate": 9.829007227949266e-06,
      "loss": 1.1662,
      "step": 12415
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.342506794748691,
      "learning_rate": 9.828687871236935e-06,
      "loss": 1.1115,
      "step": 12420
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.434145580962522,
      "learning_rate": 9.82836822177548e-06,
      "loss": 1.1159,
      "step": 12425
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.50773849951274,
      "learning_rate": 9.828048279584284e-06,
      "loss": 1.1084,
      "step": 12430
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.79929312415913,
      "learning_rate": 9.82772804468274e-06,
      "loss": 1.1001,
      "step": 12435
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.916878234370095,
      "learning_rate": 9.827407517090268e-06,
      "loss": 1.1864,
      "step": 12440
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.624640186750703,
      "learning_rate": 9.827086696826295e-06,
      "loss": 1.1754,
      "step": 12445
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.983018743222077,
      "learning_rate": 9.826765583910277e-06,
      "loss": 1.1326,
      "step": 12450
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.773415605407802,
      "learning_rate": 9.826444178361678e-06,
      "loss": 1.1252,
      "step": 12455
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.64346583503369,
      "learning_rate": 9.826122480199984e-06,
      "loss": 1.1121,
      "step": 12460
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.806255887153926,
      "learning_rate": 9.825800489444702e-06,
      "loss": 1.1332,
      "step": 12465
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.946886018001433,
      "learning_rate": 9.825478206115348e-06,
      "loss": 1.1263,
      "step": 12470
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.7102504435850525,
      "learning_rate": 9.825155630231468e-06,
      "loss": 1.1914,
      "step": 12475
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.187177547860948,
      "learning_rate": 9.824832761812612e-06,
      "loss": 1.1005,
      "step": 12480
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.932640658112496,
      "learning_rate": 9.82450960087836e-06,
      "loss": 1.0953,
      "step": 12485
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.713554701617836,
      "learning_rate": 9.824186147448302e-06,
      "loss": 1.1101,
      "step": 12490
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.940642765514276,
      "learning_rate": 9.823862401542047e-06,
      "loss": 1.1848,
      "step": 12495
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.374947051131372,
      "learning_rate": 9.823538363179224e-06,
      "loss": 1.125,
      "step": 12500
    },
    {
      "epoch": 0.53,
      "grad_norm": 25.063050467263594,
      "learning_rate": 9.823214032379477e-06,
      "loss": 1.1641,
      "step": 12505
    },
    {
      "epoch": 0.53,
      "grad_norm": 31.946356147034052,
      "learning_rate": 9.822889409162472e-06,
      "loss": 1.1359,
      "step": 12510
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.352240288543752,
      "learning_rate": 9.822564493547888e-06,
      "loss": 1.1469,
      "step": 12515
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.341985165460384,
      "learning_rate": 9.822239285555426e-06,
      "loss": 1.1558,
      "step": 12520
    },
    {
      "epoch": 0.53,
      "grad_norm": 13.90649560814533,
      "learning_rate": 9.821913785204798e-06,
      "loss": 1.2014,
      "step": 12525
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.377792204778581,
      "learning_rate": 9.82158799251574e-06,
      "loss": 1.097,
      "step": 12530
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.663508185237124,
      "learning_rate": 9.821261907508007e-06,
      "loss": 1.1191,
      "step": 12535
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.410816308626841,
      "learning_rate": 9.820935530201364e-06,
      "loss": 1.1672,
      "step": 12540
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.703330757894308,
      "learning_rate": 9.8206088606156e-06,
      "loss": 1.1126,
      "step": 12545
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.376797202098381,
      "learning_rate": 9.820281898770522e-06,
      "loss": 1.0941,
      "step": 12550
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.020811120181008,
      "learning_rate": 9.819954644685951e-06,
      "loss": 1.1362,
      "step": 12555
    },
    {
      "epoch": 0.53,
      "grad_norm": 14.561177397605544,
      "learning_rate": 9.819627098381727e-06,
      "loss": 1.1461,
      "step": 12560
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.375564514096008,
      "learning_rate": 9.819299259877708e-06,
      "loss": 1.1366,
      "step": 12565
    },
    {
      "epoch": 0.53,
      "grad_norm": 9.46221781201574,
      "learning_rate": 9.818971129193772e-06,
      "loss": 1.1309,
      "step": 12570
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.227433319273252,
      "learning_rate": 9.81864270634981e-06,
      "loss": 1.1424,
      "step": 12575
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.015066391850717,
      "learning_rate": 9.818313991365736e-06,
      "loss": 1.1272,
      "step": 12580
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.431547869025431,
      "learning_rate": 9.817984984261476e-06,
      "loss": 1.1126,
      "step": 12585
    },
    {
      "epoch": 0.53,
      "grad_norm": 14.452244129505589,
      "learning_rate": 9.81765568505698e-06,
      "loss": 1.1295,
      "step": 12590
    },
    {
      "epoch": 0.53,
      "grad_norm": 16.044731059430934,
      "learning_rate": 9.817326093772208e-06,
      "loss": 1.1428,
      "step": 12595
    },
    {
      "epoch": 0.53,
      "grad_norm": 12.009431959670444,
      "learning_rate": 9.816996210427147e-06,
      "loss": 1.0916,
      "step": 12600
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.532064237578303,
      "learning_rate": 9.816666035041793e-06,
      "loss": 1.1325,
      "step": 12605
    },
    {
      "epoch": 0.53,
      "grad_norm": 8.229898720306615,
      "learning_rate": 9.816335567636165e-06,
      "loss": 1.1359,
      "step": 12610
    },
    {
      "epoch": 0.53,
      "grad_norm": 10.480896011148356,
      "learning_rate": 9.816004808230302e-06,
      "loss": 1.1097,
      "step": 12615
    },
    {
      "epoch": 0.53,
      "grad_norm": 11.08466030790317,
      "learning_rate": 9.815673756844251e-06,
      "loss": 1.1308,
      "step": 12620
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.434118290289443,
      "learning_rate": 9.815342413498085e-06,
      "loss": 1.1649,
      "step": 12625
    },
    {
      "epoch": 0.53,
      "grad_norm": 14.253357449063092,
      "learning_rate": 9.815010778211894e-06,
      "loss": 1.1472,
      "step": 12630
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.003635887936105,
      "learning_rate": 9.814678851005782e-06,
      "loss": 1.1347,
      "step": 12635
    },
    {
      "epoch": 0.53,
      "grad_norm": 22.983628003413013,
      "learning_rate": 9.814346631899871e-06,
      "loss": 1.1425,
      "step": 12640
    },
    {
      "epoch": 0.54,
      "grad_norm": 30.646669162736565,
      "learning_rate": 9.814014120914307e-06,
      "loss": 1.1457,
      "step": 12645
    },
    {
      "epoch": 0.54,
      "grad_norm": 21.95596516313099,
      "learning_rate": 9.813681318069247e-06,
      "loss": 1.1085,
      "step": 12650
    },
    {
      "epoch": 0.54,
      "grad_norm": 23.921706235280478,
      "learning_rate": 9.813348223384866e-06,
      "loss": 1.1443,
      "step": 12655
    },
    {
      "epoch": 0.54,
      "grad_norm": 24.963542697151166,
      "learning_rate": 9.813014836881363e-06,
      "loss": 1.1096,
      "step": 12660
    },
    {
      "epoch": 0.54,
      "grad_norm": 55.33834627454908,
      "learning_rate": 9.812681158578947e-06,
      "loss": 1.1325,
      "step": 12665
    },
    {
      "epoch": 0.54,
      "grad_norm": 34.10999318071417,
      "learning_rate": 9.812347188497848e-06,
      "loss": 1.1356,
      "step": 12670
    },
    {
      "epoch": 0.54,
      "grad_norm": 44.40637943552744,
      "learning_rate": 9.812012926658315e-06,
      "loss": 1.183,
      "step": 12675
    },
    {
      "epoch": 0.54,
      "grad_norm": 32.96691171243745,
      "learning_rate": 9.811678373080611e-06,
      "loss": 1.1082,
      "step": 12680
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.224975241150997,
      "learning_rate": 9.811343527785023e-06,
      "loss": 1.1141,
      "step": 12685
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.394199433038377,
      "learning_rate": 9.811008390791845e-06,
      "loss": 1.1469,
      "step": 12690
    },
    {
      "epoch": 0.54,
      "grad_norm": 17.29075550459264,
      "learning_rate": 9.810672962121403e-06,
      "loss": 1.145,
      "step": 12695
    },
    {
      "epoch": 0.54,
      "grad_norm": 22.42269663005383,
      "learning_rate": 9.81033724179403e-06,
      "loss": 1.1455,
      "step": 12700
    },
    {
      "epoch": 0.54,
      "grad_norm": 21.35283491140994,
      "learning_rate": 9.810001229830078e-06,
      "loss": 1.1254,
      "step": 12705
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.17781485465181,
      "learning_rate": 9.809664926249918e-06,
      "loss": 1.145,
      "step": 12710
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.730954399864765,
      "learning_rate": 9.809328331073943e-06,
      "loss": 1.1819,
      "step": 12715
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.30549736617908,
      "learning_rate": 9.808991444322558e-06,
      "loss": 1.1581,
      "step": 12720
    },
    {
      "epoch": 0.54,
      "grad_norm": 9.926665877801275,
      "learning_rate": 9.808654266016186e-06,
      "loss": 1.1038,
      "step": 12725
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.135075584557665,
      "learning_rate": 9.80831679617527e-06,
      "loss": 1.1031,
      "step": 12730
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.136098592509309,
      "learning_rate": 9.80797903482027e-06,
      "loss": 1.1215,
      "step": 12735
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.901920819566608,
      "learning_rate": 9.807640981971664e-06,
      "loss": 1.1156,
      "step": 12740
    },
    {
      "epoch": 0.54,
      "grad_norm": 8.920824264615367,
      "learning_rate": 9.807302637649947e-06,
      "loss": 1.1256,
      "step": 12745
    },
    {
      "epoch": 0.54,
      "grad_norm": 16.526178949933264,
      "learning_rate": 9.806964001875632e-06,
      "loss": 1.1082,
      "step": 12750
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.221653671846322,
      "learning_rate": 9.806625074669248e-06,
      "loss": 1.1521,
      "step": 12755
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.624469505400643,
      "learning_rate": 9.806285856051345e-06,
      "loss": 1.1303,
      "step": 12760
    },
    {
      "epoch": 0.54,
      "grad_norm": 23.123082479810204,
      "learning_rate": 9.805946346042486e-06,
      "loss": 1.1037,
      "step": 12765
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.680466429914798,
      "learning_rate": 9.805606544663257e-06,
      "loss": 1.1262,
      "step": 12770
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.556054906504793,
      "learning_rate": 9.80526645193426e-06,
      "loss": 1.1625,
      "step": 12775
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.657764965405082,
      "learning_rate": 9.804926067876112e-06,
      "loss": 1.1915,
      "step": 12780
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.42309815476824,
      "learning_rate": 9.804585392509448e-06,
      "loss": 1.1129,
      "step": 12785
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.267310833703389,
      "learning_rate": 9.804244425854927e-06,
      "loss": 1.1317,
      "step": 12790
    },
    {
      "epoch": 0.54,
      "grad_norm": 14.011457840467049,
      "learning_rate": 9.803903167933216e-06,
      "loss": 1.1128,
      "step": 12795
    },
    {
      "epoch": 0.54,
      "grad_norm": 16.280056218264157,
      "learning_rate": 9.803561618765007e-06,
      "loss": 1.1294,
      "step": 12800
    },
    {
      "epoch": 0.54,
      "grad_norm": 12.124777958454295,
      "learning_rate": 9.803219778371005e-06,
      "loss": 1.1371,
      "step": 12805
    },
    {
      "epoch": 0.54,
      "grad_norm": 30.772978134808493,
      "learning_rate": 9.802877646771939e-06,
      "loss": 1.1246,
      "step": 12810
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.177471773456658,
      "learning_rate": 9.802535223988544e-06,
      "loss": 1.1362,
      "step": 12815
    },
    {
      "epoch": 0.54,
      "grad_norm": 15.673892195663568,
      "learning_rate": 9.802192510041588e-06,
      "loss": 1.1163,
      "step": 12820
    },
    {
      "epoch": 0.54,
      "grad_norm": 19.536658912570903,
      "learning_rate": 9.801849504951846e-06,
      "loss": 1.1069,
      "step": 12825
    },
    {
      "epoch": 0.54,
      "grad_norm": 17.10915844199704,
      "learning_rate": 9.80150620874011e-06,
      "loss": 1.1453,
      "step": 12830
    },
    {
      "epoch": 0.54,
      "grad_norm": 13.435559174247235,
      "learning_rate": 9.801162621427199e-06,
      "loss": 1.1161,
      "step": 12835
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.029011371702906,
      "learning_rate": 9.800818743033939e-06,
      "loss": 1.1586,
      "step": 12840
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.447254382637652,
      "learning_rate": 9.80047457358118e-06,
      "loss": 1.1156,
      "step": 12845
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.16029474577938,
      "learning_rate": 9.800130113089788e-06,
      "loss": 1.2196,
      "step": 12850
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.7279993904063184,
      "learning_rate": 9.799785361580646e-06,
      "loss": 1.1116,
      "step": 12855
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.524184367358572,
      "learning_rate": 9.799440319074656e-06,
      "loss": 1.1308,
      "step": 12860
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.511117821800435,
      "learning_rate": 9.799094985592736e-06,
      "loss": 1.118,
      "step": 12865
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.719451076107806,
      "learning_rate": 9.798749361155825e-06,
      "loss": 1.1574,
      "step": 12870
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.147397647563752,
      "learning_rate": 9.798403445784874e-06,
      "loss": 1.0919,
      "step": 12875
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.02630171898831,
      "learning_rate": 9.798057239500857e-06,
      "loss": 1.1158,
      "step": 12880
    },
    {
      "epoch": 0.55,
      "grad_norm": 21.51259520835566,
      "learning_rate": 9.797710742324762e-06,
      "loss": 1.1634,
      "step": 12885
    },
    {
      "epoch": 0.55,
      "grad_norm": 20.52457904564015,
      "learning_rate": 9.797363954277597e-06,
      "loss": 1.1102,
      "step": 12890
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.405605907789512,
      "learning_rate": 9.797016875380387e-06,
      "loss": 1.0867,
      "step": 12895
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.751986469586543,
      "learning_rate": 9.796669505654174e-06,
      "loss": 1.1125,
      "step": 12900
    },
    {
      "epoch": 0.55,
      "grad_norm": 14.914563903998776,
      "learning_rate": 9.796321845120019e-06,
      "loss": 1.1791,
      "step": 12905
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.078160319987566,
      "learning_rate": 9.795973893798995e-06,
      "loss": 1.1318,
      "step": 12910
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.398663014267725,
      "learning_rate": 9.795625651712205e-06,
      "loss": 1.1231,
      "step": 12915
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.264465644945457,
      "learning_rate": 9.795277118880755e-06,
      "loss": 1.1441,
      "step": 12920
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.928179448521267,
      "learning_rate": 9.794928295325779e-06,
      "loss": 1.1839,
      "step": 12925
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.980306187907086,
      "learning_rate": 9.794579181068424e-06,
      "loss": 1.1098,
      "step": 12930
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.425153845735667,
      "learning_rate": 9.794229776129855e-06,
      "loss": 1.1161,
      "step": 12935
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.590078288626902,
      "learning_rate": 9.79388008053126e-06,
      "loss": 1.1725,
      "step": 12940
    },
    {
      "epoch": 0.55,
      "grad_norm": 23.779202647418575,
      "learning_rate": 9.793530094293834e-06,
      "loss": 1.1069,
      "step": 12945
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.318809585200905,
      "learning_rate": 9.793179817438797e-06,
      "loss": 1.15,
      "step": 12950
    },
    {
      "epoch": 0.55,
      "grad_norm": 24.039171889246116,
      "learning_rate": 9.79282924998739e-06,
      "loss": 1.0864,
      "step": 12955
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.701881120168887,
      "learning_rate": 9.79247839196086e-06,
      "loss": 1.1382,
      "step": 12960
    },
    {
      "epoch": 0.55,
      "grad_norm": 20.868832837071682,
      "learning_rate": 9.792127243380485e-06,
      "loss": 1.1006,
      "step": 12965
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.805720085328945,
      "learning_rate": 9.79177580426755e-06,
      "loss": 1.114,
      "step": 12970
    },
    {
      "epoch": 0.55,
      "grad_norm": 20.943371067699374,
      "learning_rate": 9.79142407464336e-06,
      "loss": 1.1025,
      "step": 12975
    },
    {
      "epoch": 0.55,
      "grad_norm": 25.874461585041974,
      "learning_rate": 9.791072054529245e-06,
      "loss": 1.1405,
      "step": 12980
    },
    {
      "epoch": 0.55,
      "grad_norm": 20.662401985754244,
      "learning_rate": 9.790719743946542e-06,
      "loss": 1.1386,
      "step": 12985
    },
    {
      "epoch": 0.55,
      "grad_norm": 18.911195988444344,
      "learning_rate": 9.790367142916613e-06,
      "loss": 1.1325,
      "step": 12990
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.043732067918676,
      "learning_rate": 9.790014251460835e-06,
      "loss": 1.1395,
      "step": 12995
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.912730063709543,
      "learning_rate": 9.789661069600603e-06,
      "loss": 1.1412,
      "step": 13000
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.143892789975729,
      "learning_rate": 9.789307597357327e-06,
      "loss": 1.1014,
      "step": 13005
    },
    {
      "epoch": 0.55,
      "grad_norm": 24.767448383009672,
      "learning_rate": 9.788953834752439e-06,
      "loss": 1.1729,
      "step": 13010
    },
    {
      "epoch": 0.55,
      "grad_norm": 17.37375828521619,
      "learning_rate": 9.788599781807387e-06,
      "loss": 1.1142,
      "step": 13015
    },
    {
      "epoch": 0.55,
      "grad_norm": 14.446506230426834,
      "learning_rate": 9.788245438543635e-06,
      "loss": 1.1723,
      "step": 13020
    },
    {
      "epoch": 0.55,
      "grad_norm": 12.718297894103685,
      "learning_rate": 9.787890804982667e-06,
      "loss": 1.1274,
      "step": 13025
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.281808194011177,
      "learning_rate": 9.787535881145983e-06,
      "loss": 1.1229,
      "step": 13030
    },
    {
      "epoch": 0.55,
      "grad_norm": 16.231563175521178,
      "learning_rate": 9.7871806670551e-06,
      "loss": 1.1112,
      "step": 13035
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.332706893444458,
      "learning_rate": 9.786825162731554e-06,
      "loss": 1.133,
      "step": 13040
    },
    {
      "epoch": 0.55,
      "grad_norm": 13.073314528986915,
      "learning_rate": 9.7864693681969e-06,
      "loss": 1.1103,
      "step": 13045
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.191234904647198,
      "learning_rate": 9.786113283472706e-06,
      "loss": 1.1365,
      "step": 13050
    },
    {
      "epoch": 0.55,
      "grad_norm": 9.688970311284944,
      "learning_rate": 9.785756908580562e-06,
      "loss": 1.1674,
      "step": 13055
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.66700676991886,
      "learning_rate": 9.785400243542072e-06,
      "loss": 1.1107,
      "step": 13060
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.482845963252206,
      "learning_rate": 9.785043288378863e-06,
      "loss": 1.1549,
      "step": 13065
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.2982069199988775,
      "learning_rate": 9.784686043112574e-06,
      "loss": 1.1173,
      "step": 13070
    },
    {
      "epoch": 0.55,
      "grad_norm": 20.93924534870786,
      "learning_rate": 9.784328507764861e-06,
      "loss": 1.1022,
      "step": 13075
    },
    {
      "epoch": 0.55,
      "grad_norm": 17.482783700983667,
      "learning_rate": 9.783970682357407e-06,
      "loss": 1.107,
      "step": 13080
    },
    {
      "epoch": 0.55,
      "grad_norm": 26.29068808237692,
      "learning_rate": 9.783612566911901e-06,
      "loss": 1.1432,
      "step": 13085
    },
    {
      "epoch": 0.55,
      "grad_norm": 18.0251665510324,
      "learning_rate": 9.783254161450056e-06,
      "loss": 1.2032,
      "step": 13090
    },
    {
      "epoch": 0.55,
      "grad_norm": 13.233875236490901,
      "learning_rate": 9.782895465993601e-06,
      "loss": 1.1393,
      "step": 13095
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.454995385907694,
      "learning_rate": 9.782536480564282e-06,
      "loss": 1.1571,
      "step": 13100
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.030320811867446,
      "learning_rate": 9.782177205183863e-06,
      "loss": 1.0999,
      "step": 13105
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.734900954082516,
      "learning_rate": 9.781817639874126e-06,
      "loss": 1.1176,
      "step": 13110
    },
    {
      "epoch": 0.56,
      "grad_norm": 22.133119933221828,
      "learning_rate": 9.781457784656872e-06,
      "loss": 1.0956,
      "step": 13115
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.938707646888012,
      "learning_rate": 9.781097639553916e-06,
      "loss": 1.1223,
      "step": 13120
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.564486410330966,
      "learning_rate": 9.780737204587094e-06,
      "loss": 1.1954,
      "step": 13125
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.985861891688922,
      "learning_rate": 9.780376479778257e-06,
      "loss": 1.1281,
      "step": 13130
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.705715360507454,
      "learning_rate": 9.780015465149275e-06,
      "loss": 1.0931,
      "step": 13135
    },
    {
      "epoch": 0.56,
      "grad_norm": 16.620647171367157,
      "learning_rate": 9.779654160722035e-06,
      "loss": 1.1835,
      "step": 13140
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.299144146136616,
      "learning_rate": 9.779292566518442e-06,
      "loss": 1.1564,
      "step": 13145
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.521385037124467,
      "learning_rate": 9.778930682560419e-06,
      "loss": 1.1982,
      "step": 13150
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.840960460428551,
      "learning_rate": 9.778568508869906e-06,
      "loss": 1.1115,
      "step": 13155
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.6274360915454,
      "learning_rate": 9.77820604546886e-06,
      "loss": 1.1553,
      "step": 13160
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.768737399090917,
      "learning_rate": 9.777843292379255e-06,
      "loss": 1.1328,
      "step": 13165
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.133797486111874,
      "learning_rate": 9.777480249623086e-06,
      "loss": 1.1522,
      "step": 13170
    },
    {
      "epoch": 0.56,
      "grad_norm": 13.654277072752937,
      "learning_rate": 9.777116917222362e-06,
      "loss": 1.1358,
      "step": 13175
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.167302637129885,
      "learning_rate": 9.776753295199111e-06,
      "loss": 1.1703,
      "step": 13180
    },
    {
      "epoch": 0.56,
      "grad_norm": 21.932925745156684,
      "learning_rate": 9.776389383575377e-06,
      "loss": 1.1098,
      "step": 13185
    },
    {
      "epoch": 0.56,
      "grad_norm": 25.755926387523015,
      "learning_rate": 9.776025182373225e-06,
      "loss": 1.1554,
      "step": 13190
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.18752070214331,
      "learning_rate": 9.775660691614733e-06,
      "loss": 1.1828,
      "step": 13195
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.4710004089662485,
      "learning_rate": 9.775295911322002e-06,
      "loss": 1.1657,
      "step": 13200
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.907298713227695,
      "learning_rate": 9.774930841517146e-06,
      "loss": 1.1235,
      "step": 13205
    },
    {
      "epoch": 0.56,
      "grad_norm": 27.11503848014546,
      "learning_rate": 9.774565482222297e-06,
      "loss": 1.117,
      "step": 13210
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.551029589771035,
      "learning_rate": 9.774199833459606e-06,
      "loss": 1.1581,
      "step": 13215
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.828455747367782,
      "learning_rate": 9.773833895251243e-06,
      "loss": 1.1336,
      "step": 13220
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.552269523706661,
      "learning_rate": 9.773467667619394e-06,
      "loss": 1.1383,
      "step": 13225
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.5264486516209255,
      "learning_rate": 9.77310115058626e-06,
      "loss": 1.1657,
      "step": 13230
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.143082185385055,
      "learning_rate": 9.772734344174062e-06,
      "loss": 1.0869,
      "step": 13235
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.501204261538233,
      "learning_rate": 9.77236724840504e-06,
      "loss": 1.1671,
      "step": 13240
    },
    {
      "epoch": 0.56,
      "grad_norm": 17.787340675653407,
      "learning_rate": 9.77199986330145e-06,
      "loss": 1.0893,
      "step": 13245
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.627285149178578,
      "learning_rate": 9.771632188885563e-06,
      "loss": 1.1145,
      "step": 13250
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.613557491982418,
      "learning_rate": 9.771264225179673e-06,
      "loss": 1.0989,
      "step": 13255
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.458179462400088,
      "learning_rate": 9.770895972206087e-06,
      "loss": 1.1087,
      "step": 13260
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.599539106617368,
      "learning_rate": 9.77052742998713e-06,
      "loss": 1.1729,
      "step": 13265
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.427568675308853,
      "learning_rate": 9.77015859854515e-06,
      "loss": 1.1244,
      "step": 13270
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.338757436962799,
      "learning_rate": 9.769789477902503e-06,
      "loss": 1.1314,
      "step": 13275
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.127939920679577,
      "learning_rate": 9.76942006808157e-06,
      "loss": 1.1432,
      "step": 13280
    },
    {
      "epoch": 0.56,
      "grad_norm": 17.482971661011053,
      "learning_rate": 9.769050369104747e-06,
      "loss": 1.1435,
      "step": 13285
    },
    {
      "epoch": 0.56,
      "grad_norm": 18.968380951063157,
      "learning_rate": 9.76868038099445e-06,
      "loss": 1.1223,
      "step": 13290
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.851121365538491,
      "learning_rate": 9.768310103773105e-06,
      "loss": 1.1565,
      "step": 13295
    },
    {
      "epoch": 0.56,
      "grad_norm": 18.52619269070551,
      "learning_rate": 9.767939537463164e-06,
      "loss": 1.0886,
      "step": 13300
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.914681046230499,
      "learning_rate": 9.767568682087094e-06,
      "loss": 1.1477,
      "step": 13305
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.957620866819147,
      "learning_rate": 9.767197537667379e-06,
      "loss": 1.1418,
      "step": 13310
    },
    {
      "epoch": 0.56,
      "grad_norm": 15.458989591754985,
      "learning_rate": 9.766826104226519e-06,
      "loss": 1.1442,
      "step": 13315
    },
    {
      "epoch": 0.56,
      "grad_norm": 12.85429724644223,
      "learning_rate": 9.766454381787034e-06,
      "loss": 1.1263,
      "step": 13320
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.9052956788645385,
      "learning_rate": 9.766082370371458e-06,
      "loss": 1.088,
      "step": 13325
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.644506708146974,
      "learning_rate": 9.76571007000235e-06,
      "loss": 1.1216,
      "step": 13330
    },
    {
      "epoch": 0.56,
      "grad_norm": 11.936463125365135,
      "learning_rate": 9.765337480702276e-06,
      "loss": 1.152,
      "step": 13335
    },
    {
      "epoch": 0.56,
      "grad_norm": 10.375067115709802,
      "learning_rate": 9.764964602493828e-06,
      "loss": 1.1596,
      "step": 13340
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.902950180571504,
      "learning_rate": 9.76459143539961e-06,
      "loss": 1.1224,
      "step": 13345
    },
    {
      "epoch": 0.56,
      "grad_norm": 21.837736280134347,
      "learning_rate": 9.76421797944225e-06,
      "loss": 1.1504,
      "step": 13350
    },
    {
      "epoch": 0.57,
      "grad_norm": 16.842568521064024,
      "learning_rate": 9.763844234644386e-06,
      "loss": 1.1049,
      "step": 13355
    },
    {
      "epoch": 0.57,
      "grad_norm": 18.12746969766832,
      "learning_rate": 9.76347020102868e-06,
      "loss": 1.1119,
      "step": 13360
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.935129818403512,
      "learning_rate": 9.763095878617806e-06,
      "loss": 1.1298,
      "step": 13365
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.539806839753402,
      "learning_rate": 9.76272126743446e-06,
      "loss": 1.1265,
      "step": 13370
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.547612250773259,
      "learning_rate": 9.76234636750135e-06,
      "loss": 1.1188,
      "step": 13375
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.68767556630829,
      "learning_rate": 9.76197117884121e-06,
      "loss": 1.0998,
      "step": 13380
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.448630061504575,
      "learning_rate": 9.761595701476784e-06,
      "loss": 1.1226,
      "step": 13385
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.38702777729219,
      "learning_rate": 9.761219935430836e-06,
      "loss": 1.1193,
      "step": 13390
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.590565912070996,
      "learning_rate": 9.760843880726148e-06,
      "loss": 1.1309,
      "step": 13395
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.055441084633275,
      "learning_rate": 9.760467537385519e-06,
      "loss": 1.1085,
      "step": 13400
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.93073082795177,
      "learning_rate": 9.760090905431767e-06,
      "loss": 1.1238,
      "step": 13405
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.027074855740675,
      "learning_rate": 9.759713984887723e-06,
      "loss": 1.0785,
      "step": 13410
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.404823610158473,
      "learning_rate": 9.75933677577624e-06,
      "loss": 1.1467,
      "step": 13415
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.184722880962163,
      "learning_rate": 9.758959278120189e-06,
      "loss": 1.0982,
      "step": 13420
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.452705144229121,
      "learning_rate": 9.758581491942455e-06,
      "loss": 1.1098,
      "step": 13425
    },
    {
      "epoch": 0.57,
      "grad_norm": 15.831042783525902,
      "learning_rate": 9.75820341726594e-06,
      "loss": 1.1453,
      "step": 13430
    },
    {
      "epoch": 0.57,
      "grad_norm": 4.9233019978285775,
      "learning_rate": 9.757825054113569e-06,
      "loss": 1.1134,
      "step": 13435
    },
    {
      "epoch": 0.57,
      "grad_norm": 15.175120159721672,
      "learning_rate": 9.75744640250828e-06,
      "loss": 1.1236,
      "step": 13440
    },
    {
      "epoch": 0.57,
      "grad_norm": 18.157207065832747,
      "learning_rate": 9.757067462473026e-06,
      "loss": 1.1399,
      "step": 13445
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.618263639784244,
      "learning_rate": 9.756688234030788e-06,
      "loss": 1.1042,
      "step": 13450
    },
    {
      "epoch": 0.57,
      "grad_norm": 15.872953737147169,
      "learning_rate": 9.756308717204552e-06,
      "loss": 1.1474,
      "step": 13455
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.4409305621497275,
      "learning_rate": 9.75592891201733e-06,
      "loss": 1.154,
      "step": 13460
    },
    {
      "epoch": 0.57,
      "grad_norm": 37.656149382181226,
      "learning_rate": 9.755548818492145e-06,
      "loss": 1.1332,
      "step": 13465
    },
    {
      "epoch": 0.57,
      "grad_norm": 39.410070890251866,
      "learning_rate": 9.755168436652044e-06,
      "loss": 1.1329,
      "step": 13470
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.494051807223958,
      "learning_rate": 9.754787766520087e-06,
      "loss": 1.1539,
      "step": 13475
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.13011391022796,
      "learning_rate": 9.754406808119355e-06,
      "loss": 1.1663,
      "step": 13480
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.419113045490374,
      "learning_rate": 9.754025561472942e-06,
      "loss": 1.1087,
      "step": 13485
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.516755129262156,
      "learning_rate": 9.753644026603961e-06,
      "loss": 1.189,
      "step": 13490
    },
    {
      "epoch": 0.57,
      "grad_norm": 12.978766271770928,
      "learning_rate": 9.753262203535547e-06,
      "loss": 1.1421,
      "step": 13495
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.117541691549466,
      "learning_rate": 9.752880092290846e-06,
      "loss": 1.1256,
      "step": 13500
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.252353779560716,
      "learning_rate": 9.752497692893024e-06,
      "loss": 1.1152,
      "step": 13505
    },
    {
      "epoch": 0.57,
      "grad_norm": 11.319861335587225,
      "learning_rate": 9.752115005365268e-06,
      "loss": 1.1269,
      "step": 13510
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.993503207907033,
      "learning_rate": 9.751732029730776e-06,
      "loss": 1.0822,
      "step": 13515
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.9736893549726755,
      "learning_rate": 9.751348766012768e-06,
      "loss": 1.1072,
      "step": 13520
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.278543704694994,
      "learning_rate": 9.75096521423448e-06,
      "loss": 1.0968,
      "step": 13525
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.823957093385041,
      "learning_rate": 9.750581374419165e-06,
      "loss": 1.1128,
      "step": 13530
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.354038431585671,
      "learning_rate": 9.750197246590094e-06,
      "loss": 1.1535,
      "step": 13535
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.392417723197982,
      "learning_rate": 9.749812830770557e-06,
      "loss": 1.0951,
      "step": 13540
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.583628493567195,
      "learning_rate": 9.74942812698386e-06,
      "loss": 1.1295,
      "step": 13545
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.242990049423247,
      "learning_rate": 9.749043135253325e-06,
      "loss": 1.1074,
      "step": 13550
    },
    {
      "epoch": 0.57,
      "grad_norm": 18.01123885900048,
      "learning_rate": 9.748657855602291e-06,
      "loss": 1.1773,
      "step": 13555
    },
    {
      "epoch": 0.57,
      "grad_norm": 22.042223974575627,
      "learning_rate": 9.748272288054121e-06,
      "loss": 1.192,
      "step": 13560
    },
    {
      "epoch": 0.57,
      "grad_norm": 14.879311750750942,
      "learning_rate": 9.747886432632188e-06,
      "loss": 1.1903,
      "step": 13565
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.711097179570518,
      "learning_rate": 9.747500289359888e-06,
      "loss": 1.144,
      "step": 13570
    },
    {
      "epoch": 0.57,
      "grad_norm": 23.60807442866413,
      "learning_rate": 9.747113858260627e-06,
      "loss": 1.1923,
      "step": 13575
    },
    {
      "epoch": 0.57,
      "grad_norm": 32.66858165124962,
      "learning_rate": 9.746727139357836e-06,
      "loss": 1.1754,
      "step": 13580
    },
    {
      "epoch": 0.57,
      "grad_norm": 20.49191445348783,
      "learning_rate": 9.746340132674963e-06,
      "loss": 1.197,
      "step": 13585
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.596808091652,
      "learning_rate": 9.745952838235466e-06,
      "loss": 1.1538,
      "step": 13590
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.961680188320575,
      "learning_rate": 9.74556525606283e-06,
      "loss": 1.1269,
      "step": 13595
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.699277531139078,
      "learning_rate": 9.74517738618055e-06,
      "loss": 1.1891,
      "step": 13600
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.596301814481997,
      "learning_rate": 9.744789228612143e-06,
      "loss": 1.1343,
      "step": 13605
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.07379404370688,
      "learning_rate": 9.74440078338114e-06,
      "loss": 1.1586,
      "step": 13610
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.745282290035206,
      "learning_rate": 9.744012050511093e-06,
      "loss": 1.0977,
      "step": 13615
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.093819211165586,
      "learning_rate": 9.74362303002557e-06,
      "loss": 1.1253,
      "step": 13620
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.86902471573825,
      "learning_rate": 9.743233721948155e-06,
      "loss": 1.1165,
      "step": 13625
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.982300055530532,
      "learning_rate": 9.742844126302451e-06,
      "loss": 1.1519,
      "step": 13630
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.238724259179179,
      "learning_rate": 9.74245424311208e-06,
      "loss": 1.1186,
      "step": 13635
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.663496552845421,
      "learning_rate": 9.742064072400676e-06,
      "loss": 1.0901,
      "step": 13640
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.083149295452875,
      "learning_rate": 9.741673614191896e-06,
      "loss": 1.1641,
      "step": 13645
    },
    {
      "epoch": 0.58,
      "grad_norm": 8.079652347188938,
      "learning_rate": 9.74128286850941e-06,
      "loss": 1.1024,
      "step": 13650
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.683752446150179,
      "learning_rate": 9.74089183537691e-06,
      "loss": 1.1076,
      "step": 13655
    },
    {
      "epoch": 0.58,
      "grad_norm": 21.919906620793522,
      "learning_rate": 9.740500514818105e-06,
      "loss": 1.1159,
      "step": 13660
    },
    {
      "epoch": 0.58,
      "grad_norm": 16.774229118180383,
      "learning_rate": 9.740108906856716e-06,
      "loss": 1.1621,
      "step": 13665
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.083800645690784,
      "learning_rate": 9.739717011516488e-06,
      "loss": 1.1314,
      "step": 13670
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.218406244385575,
      "learning_rate": 9.739324828821177e-06,
      "loss": 1.1417,
      "step": 13675
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.438724003076803,
      "learning_rate": 9.738932358794561e-06,
      "loss": 1.1099,
      "step": 13680
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.825192852736801,
      "learning_rate": 9.738539601460437e-06,
      "loss": 1.1449,
      "step": 13685
    },
    {
      "epoch": 0.58,
      "grad_norm": 13.254767362012407,
      "learning_rate": 9.738146556842616e-06,
      "loss": 1.1387,
      "step": 13690
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.341334327589971,
      "learning_rate": 9.737753224964924e-06,
      "loss": 1.1537,
      "step": 13695
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.57147756649199,
      "learning_rate": 9.737359605851209e-06,
      "loss": 1.1073,
      "step": 13700
    },
    {
      "epoch": 0.58,
      "grad_norm": 30.182589884308637,
      "learning_rate": 9.736965699525335e-06,
      "loss": 1.1537,
      "step": 13705
    },
    {
      "epoch": 0.58,
      "grad_norm": 84.36950066110069,
      "learning_rate": 9.736571506011186e-06,
      "loss": 1.1251,
      "step": 13710
    },
    {
      "epoch": 0.58,
      "grad_norm": 85.08634303108919,
      "learning_rate": 9.736177025332655e-06,
      "loss": 1.1921,
      "step": 13715
    },
    {
      "epoch": 0.58,
      "grad_norm": 56.43510058129571,
      "learning_rate": 9.735782257513666e-06,
      "loss": 1.1423,
      "step": 13720
    },
    {
      "epoch": 0.58,
      "grad_norm": 46.89654876156769,
      "learning_rate": 9.735387202578146e-06,
      "loss": 1.1496,
      "step": 13725
    },
    {
      "epoch": 0.58,
      "grad_norm": 12.981475427424796,
      "learning_rate": 9.73499186055005e-06,
      "loss": 1.1316,
      "step": 13730
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.249150020614929,
      "learning_rate": 9.734596231453344e-06,
      "loss": 1.1715,
      "step": 13735
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.96237764776874,
      "learning_rate": 9.734200315312015e-06,
      "loss": 1.0843,
      "step": 13740
    },
    {
      "epoch": 0.58,
      "grad_norm": 16.453782232823215,
      "learning_rate": 9.733804112150065e-06,
      "loss": 1.1192,
      "step": 13745
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.457717433038631,
      "learning_rate": 9.733407621991516e-06,
      "loss": 1.1513,
      "step": 13750
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.563850450972106,
      "learning_rate": 9.733010844860405e-06,
      "loss": 1.1506,
      "step": 13755
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.7193049880769715,
      "learning_rate": 9.73261378078079e-06,
      "loss": 1.1181,
      "step": 13760
    },
    {
      "epoch": 0.58,
      "grad_norm": 11.184904037318445,
      "learning_rate": 9.73221642977674e-06,
      "loss": 1.1236,
      "step": 13765
    },
    {
      "epoch": 0.58,
      "grad_norm": 15.912619313513849,
      "learning_rate": 9.731818791872347e-06,
      "loss": 1.1639,
      "step": 13770
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.133080421518732,
      "learning_rate": 9.73142086709172e-06,
      "loss": 1.1199,
      "step": 13775
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.340388594522803,
      "learning_rate": 9.731022655458981e-06,
      "loss": 1.0936,
      "step": 13780
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.278513456854574,
      "learning_rate": 9.730624156998277e-06,
      "loss": 1.1234,
      "step": 13785
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.500570873388791,
      "learning_rate": 9.730225371733762e-06,
      "loss": 1.1168,
      "step": 13790
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.250782373989602,
      "learning_rate": 9.729826299689617e-06,
      "loss": 1.1311,
      "step": 13795
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.741684382842332,
      "learning_rate": 9.729426940890037e-06,
      "loss": 1.1329,
      "step": 13800
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.4175202822844435,
      "learning_rate": 9.72902729535923e-06,
      "loss": 1.1001,
      "step": 13805
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.272043537203692,
      "learning_rate": 9.72862736312143e-06,
      "loss": 1.1515,
      "step": 13810
    },
    {
      "epoch": 0.58,
      "grad_norm": 14.496071095936655,
      "learning_rate": 9.72822714420088e-06,
      "loss": 1.1328,
      "step": 13815
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.016957578389495,
      "learning_rate": 9.727826638621847e-06,
      "loss": 1.0869,
      "step": 13820
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.763331240721669,
      "learning_rate": 9.72742584640861e-06,
      "loss": 1.1273,
      "step": 13825
    },
    {
      "epoch": 0.59,
      "grad_norm": 20.149762669917695,
      "learning_rate": 9.727024767585469e-06,
      "loss": 1.1247,
      "step": 13830
    },
    {
      "epoch": 0.59,
      "grad_norm": 15.028645592929125,
      "learning_rate": 9.726623402176741e-06,
      "loss": 1.1096,
      "step": 13835
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.254216583524284,
      "learning_rate": 9.726221750206759e-06,
      "loss": 1.1012,
      "step": 13840
    },
    {
      "epoch": 0.59,
      "grad_norm": 12.932555915952687,
      "learning_rate": 9.725819811699874e-06,
      "loss": 1.0829,
      "step": 13845
    },
    {
      "epoch": 0.59,
      "grad_norm": 18.495150180801897,
      "learning_rate": 9.725417586680454e-06,
      "loss": 1.1774,
      "step": 13850
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.736696370952199,
      "learning_rate": 9.725015075172883e-06,
      "loss": 1.0882,
      "step": 13855
    },
    {
      "epoch": 0.59,
      "grad_norm": 14.387044403058038,
      "learning_rate": 9.724612277201569e-06,
      "loss": 1.1382,
      "step": 13860
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.636671771889029,
      "learning_rate": 9.724209192790928e-06,
      "loss": 1.1354,
      "step": 13865
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.586172667041027,
      "learning_rate": 9.7238058219654e-06,
      "loss": 1.1096,
      "step": 13870
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.586335870123596,
      "learning_rate": 9.723402164749438e-06,
      "loss": 1.1567,
      "step": 13875
    },
    {
      "epoch": 0.59,
      "grad_norm": 11.408673890467808,
      "learning_rate": 9.722998221167517e-06,
      "loss": 1.1554,
      "step": 13880
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.564909108022746,
      "learning_rate": 9.722593991244125e-06,
      "loss": 1.1875,
      "step": 13885
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.454590478153151,
      "learning_rate": 9.722189475003771e-06,
      "loss": 1.0963,
      "step": 13890
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.029927090426701,
      "learning_rate": 9.721784672470978e-06,
      "loss": 1.1357,
      "step": 13895
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.891036632495453,
      "learning_rate": 9.72137958367029e-06,
      "loss": 1.1459,
      "step": 13900
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.365981627250617,
      "learning_rate": 9.720974208626264e-06,
      "loss": 1.1083,
      "step": 13905
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.8215572344329685,
      "learning_rate": 9.720568547363479e-06,
      "loss": 1.1109,
      "step": 13910
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.248447010576974,
      "learning_rate": 9.720162599906526e-06,
      "loss": 1.1341,
      "step": 13915
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.525695339655307,
      "learning_rate": 9.71975636628002e-06,
      "loss": 1.0999,
      "step": 13920
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.3392416167837915,
      "learning_rate": 9.719349846508587e-06,
      "loss": 1.1087,
      "step": 13925
    },
    {
      "epoch": 0.59,
      "grad_norm": 21.205709958927134,
      "learning_rate": 9.718943040616875e-06,
      "loss": 1.1688,
      "step": 13930
    },
    {
      "epoch": 0.59,
      "grad_norm": 29.453578570488713,
      "learning_rate": 9.718535948629546e-06,
      "loss": 1.1688,
      "step": 13935
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.778574829443633,
      "learning_rate": 9.71812857057128e-06,
      "loss": 1.1433,
      "step": 13940
    },
    {
      "epoch": 0.59,
      "grad_norm": 12.220735777567683,
      "learning_rate": 9.717720906466779e-06,
      "loss": 1.1192,
      "step": 13945
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.563226486213903,
      "learning_rate": 9.717312956340756e-06,
      "loss": 1.1463,
      "step": 13950
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.515330457974046,
      "learning_rate": 9.716904720217942e-06,
      "loss": 1.1598,
      "step": 13955
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.854047314048814,
      "learning_rate": 9.716496198123088e-06,
      "loss": 1.126,
      "step": 13960
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.019166990613669,
      "learning_rate": 9.716087390080966e-06,
      "loss": 1.0897,
      "step": 13965
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.789855726925458,
      "learning_rate": 9.715678296116354e-06,
      "loss": 1.1387,
      "step": 13970
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.132125621767042,
      "learning_rate": 9.71526891625406e-06,
      "loss": 1.1022,
      "step": 13975
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.427187535674997,
      "learning_rate": 9.714859250518898e-06,
      "loss": 1.1444,
      "step": 13980
    },
    {
      "epoch": 0.59,
      "grad_norm": 35.65294469608465,
      "learning_rate": 9.714449298935708e-06,
      "loss": 1.1232,
      "step": 13985
    },
    {
      "epoch": 0.59,
      "grad_norm": 18.61637138231272,
      "learning_rate": 9.714039061529346e-06,
      "loss": 1.1059,
      "step": 13990
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.714080450971223,
      "learning_rate": 9.713628538324681e-06,
      "loss": 1.1204,
      "step": 13995
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.32495006374814,
      "learning_rate": 9.713217729346602e-06,
      "loss": 1.1245,
      "step": 14000
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.8906853219804525,
      "learning_rate": 9.712806634620013e-06,
      "loss": 1.119,
      "step": 14005
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.8371994359884,
      "learning_rate": 9.712395254169842e-06,
      "loss": 1.1242,
      "step": 14010
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.6348799615958685,
      "learning_rate": 9.711983588021027e-06,
      "loss": 1.0972,
      "step": 14015
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.464835712508369,
      "learning_rate": 9.711571636198528e-06,
      "loss": 1.1374,
      "step": 14020
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.8760804537559626,
      "learning_rate": 9.711159398727318e-06,
      "loss": 1.111,
      "step": 14025
    },
    {
      "epoch": 0.59,
      "grad_norm": 16.603816509164826,
      "learning_rate": 9.71074687563239e-06,
      "loss": 1.1311,
      "step": 14030
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.885629283636852,
      "learning_rate": 9.710334066938756e-06,
      "loss": 1.0825,
      "step": 14035
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.597870646280369,
      "learning_rate": 9.709920972671442e-06,
      "loss": 1.0968,
      "step": 14040
    },
    {
      "epoch": 0.59,
      "grad_norm": 10.385967033254175,
      "learning_rate": 9.709507592855494e-06,
      "loss": 1.1209,
      "step": 14045
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.284912747176475,
      "learning_rate": 9.70909392751597e-06,
      "loss": 1.1417,
      "step": 14050
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.030024940957162,
      "learning_rate": 9.708679976677955e-06,
      "loss": 1.1091,
      "step": 14055
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.692313777138527,
      "learning_rate": 9.708265740366541e-06,
      "loss": 1.184,
      "step": 14060
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.096693910857474,
      "learning_rate": 9.707851218606845e-06,
      "loss": 1.1261,
      "step": 14065
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.063844375083768,
      "learning_rate": 9.707436411423997e-06,
      "loss": 1.1421,
      "step": 14070
    },
    {
      "epoch": 0.6,
      "grad_norm": 17.11067654018697,
      "learning_rate": 9.707021318843147e-06,
      "loss": 1.1354,
      "step": 14075
    },
    {
      "epoch": 0.6,
      "grad_norm": 20.637254197598722,
      "learning_rate": 9.706605940889456e-06,
      "loss": 1.1227,
      "step": 14080
    },
    {
      "epoch": 0.6,
      "grad_norm": 18.57758018515164,
      "learning_rate": 9.706190277588112e-06,
      "loss": 1.1679,
      "step": 14085
    },
    {
      "epoch": 0.6,
      "grad_norm": 17.011518395593868,
      "learning_rate": 9.705774328964315e-06,
      "loss": 1.1249,
      "step": 14090
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.259547528540654,
      "learning_rate": 9.705358095043282e-06,
      "loss": 1.1685,
      "step": 14095
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.804697857419432,
      "learning_rate": 9.704941575850246e-06,
      "loss": 1.0849,
      "step": 14100
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.311651499996418,
      "learning_rate": 9.704524771410461e-06,
      "loss": 1.1242,
      "step": 14105
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.630908308786798,
      "learning_rate": 9.704107681749199e-06,
      "loss": 1.1246,
      "step": 14110
    },
    {
      "epoch": 0.6,
      "grad_norm": 19.456640518105218,
      "learning_rate": 9.703690306891744e-06,
      "loss": 1.1094,
      "step": 14115
    },
    {
      "epoch": 0.6,
      "grad_norm": 16.315665765220558,
      "learning_rate": 9.7032726468634e-06,
      "loss": 1.1668,
      "step": 14120
    },
    {
      "epoch": 0.6,
      "grad_norm": 17.123573267526094,
      "learning_rate": 9.70285470168949e-06,
      "loss": 1.1391,
      "step": 14125
    },
    {
      "epoch": 0.6,
      "grad_norm": 28.912282344415782,
      "learning_rate": 9.70243647139535e-06,
      "loss": 1.1259,
      "step": 14130
    },
    {
      "epoch": 0.6,
      "grad_norm": 43.18446506558985,
      "learning_rate": 9.702017956006343e-06,
      "loss": 1.1615,
      "step": 14135
    },
    {
      "epoch": 0.6,
      "grad_norm": 17.102065114051854,
      "learning_rate": 9.701599155547832e-06,
      "loss": 1.124,
      "step": 14140
    },
    {
      "epoch": 0.6,
      "grad_norm": 20.82897825568837,
      "learning_rate": 9.701180070045217e-06,
      "loss": 1.1466,
      "step": 14145
    },
    {
      "epoch": 0.6,
      "grad_norm": 14.43507501127827,
      "learning_rate": 9.700760699523902e-06,
      "loss": 1.1053,
      "step": 14150
    },
    {
      "epoch": 0.6,
      "grad_norm": 23.000105978468245,
      "learning_rate": 9.700341044009311e-06,
      "loss": 1.1295,
      "step": 14155
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.197616835755062,
      "learning_rate": 9.699921103526888e-06,
      "loss": 1.1048,
      "step": 14160
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.870948441850969,
      "learning_rate": 9.699500878102093e-06,
      "loss": 1.092,
      "step": 14165
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.392302297617999,
      "learning_rate": 9.699080367760401e-06,
      "loss": 1.1945,
      "step": 14170
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.33064846082759,
      "learning_rate": 9.69865957252731e-06,
      "loss": 1.0801,
      "step": 14175
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.406694537179667,
      "learning_rate": 9.698238492428329e-06,
      "loss": 1.1227,
      "step": 14180
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.656183095272076,
      "learning_rate": 9.697817127488986e-06,
      "loss": 1.1013,
      "step": 14185
    },
    {
      "epoch": 0.6,
      "grad_norm": 17.706762423709147,
      "learning_rate": 9.69739547773483e-06,
      "loss": 1.1546,
      "step": 14190
    },
    {
      "epoch": 0.6,
      "grad_norm": 12.15549954656361,
      "learning_rate": 9.696973543191421e-06,
      "loss": 1.092,
      "step": 14195
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.291328539456801,
      "learning_rate": 9.696551323884344e-06,
      "loss": 1.131,
      "step": 14200
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.09770951178777,
      "learning_rate": 9.696128819839192e-06,
      "loss": 1.1339,
      "step": 14205
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.401555679099213,
      "learning_rate": 9.695706031081585e-06,
      "loss": 1.1072,
      "step": 14210
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.154129426934878,
      "learning_rate": 9.695282957637152e-06,
      "loss": 1.1523,
      "step": 14215
    },
    {
      "epoch": 0.6,
      "grad_norm": 12.555466433147776,
      "learning_rate": 9.694859599531542e-06,
      "loss": 1.1253,
      "step": 14220
    },
    {
      "epoch": 0.6,
      "grad_norm": 16.532846358518643,
      "learning_rate": 9.694435956790425e-06,
      "loss": 1.1669,
      "step": 14225
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.787662137130075,
      "learning_rate": 9.694012029439484e-06,
      "loss": 1.1046,
      "step": 14230
    },
    {
      "epoch": 0.6,
      "grad_norm": 12.321373558046599,
      "learning_rate": 9.693587817504419e-06,
      "loss": 1.1241,
      "step": 14235
    },
    {
      "epoch": 0.6,
      "grad_norm": 14.177500473313764,
      "learning_rate": 9.693163321010951e-06,
      "loss": 1.1157,
      "step": 14240
    },
    {
      "epoch": 0.6,
      "grad_norm": 17.424638170835365,
      "learning_rate": 9.692738539984817e-06,
      "loss": 1.1354,
      "step": 14245
    },
    {
      "epoch": 0.6,
      "grad_norm": 16.129220300041492,
      "learning_rate": 9.692313474451766e-06,
      "loss": 1.1508,
      "step": 14250
    },
    {
      "epoch": 0.6,
      "grad_norm": 29.534047927514592,
      "learning_rate": 9.691888124437571e-06,
      "loss": 1.1256,
      "step": 14255
    },
    {
      "epoch": 0.6,
      "grad_norm": 35.34617874476829,
      "learning_rate": 9.69146248996802e-06,
      "loss": 1.1364,
      "step": 14260
    },
    {
      "epoch": 0.6,
      "grad_norm": 25.769318004468367,
      "learning_rate": 9.691036571068916e-06,
      "loss": 1.1262,
      "step": 14265
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.156003738279631,
      "learning_rate": 9.690610367766084e-06,
      "loss": 1.141,
      "step": 14270
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.55976996682998,
      "learning_rate": 9.690183880085361e-06,
      "loss": 1.1031,
      "step": 14275
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.2408477756647365,
      "learning_rate": 9.689757108052606e-06,
      "loss": 1.1307,
      "step": 14280
    },
    {
      "epoch": 0.6,
      "grad_norm": 10.524267369930584,
      "learning_rate": 9.68933005169369e-06,
      "loss": 1.1655,
      "step": 14285
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.195124963742375,
      "learning_rate": 9.688902711034506e-06,
      "loss": 1.1158,
      "step": 14290
    },
    {
      "epoch": 0.6,
      "grad_norm": 24.639025098053352,
      "learning_rate": 9.688475086100964e-06,
      "loss": 1.1024,
      "step": 14295
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.106642013701407,
      "learning_rate": 9.688047176918987e-06,
      "loss": 1.1598,
      "step": 14300
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.017831339967914,
      "learning_rate": 9.68761898351452e-06,
      "loss": 1.1444,
      "step": 14305
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.450636171679976,
      "learning_rate": 9.68719050591352e-06,
      "loss": 1.0963,
      "step": 14310
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.592425950734231,
      "learning_rate": 9.686761744141966e-06,
      "loss": 1.1545,
      "step": 14315
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.609845210528615,
      "learning_rate": 9.686332698225853e-06,
      "loss": 1.1346,
      "step": 14320
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.691612394219117,
      "learning_rate": 9.685903368191193e-06,
      "loss": 1.1315,
      "step": 14325
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.80123532064526,
      "learning_rate": 9.685473754064017e-06,
      "loss": 1.1288,
      "step": 14330
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.765324860390736,
      "learning_rate": 9.685043855870364e-06,
      "loss": 1.1452,
      "step": 14335
    },
    {
      "epoch": 0.61,
      "grad_norm": 18.75024884075784,
      "learning_rate": 9.684613673636307e-06,
      "loss": 1.1537,
      "step": 14340
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.879513503633115,
      "learning_rate": 9.684183207387918e-06,
      "loss": 1.1086,
      "step": 14345
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.055767729742575,
      "learning_rate": 9.6837524571513e-06,
      "loss": 1.1712,
      "step": 14350
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.522534053725067,
      "learning_rate": 9.683321422952568e-06,
      "loss": 1.0963,
      "step": 14355
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.96653752963125,
      "learning_rate": 9.682890104817853e-06,
      "loss": 1.0918,
      "step": 14360
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.786773311136603,
      "learning_rate": 9.682458502773305e-06,
      "loss": 1.1482,
      "step": 14365
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.271167665895201,
      "learning_rate": 9.68202661684509e-06,
      "loss": 1.0533,
      "step": 14370
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.127188815553549,
      "learning_rate": 9.681594447059393e-06,
      "loss": 1.1342,
      "step": 14375
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.124285529409141,
      "learning_rate": 9.681161993442416e-06,
      "loss": 1.1517,
      "step": 14380
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.347945330177689,
      "learning_rate": 9.680729256020373e-06,
      "loss": 1.1027,
      "step": 14385
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.423364445638174,
      "learning_rate": 9.680296234819505e-06,
      "loss": 1.0798,
      "step": 14390
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.457764193921415,
      "learning_rate": 9.679862929866062e-06,
      "loss": 1.1257,
      "step": 14395
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.505127413413769,
      "learning_rate": 9.679429341186315e-06,
      "loss": 1.1142,
      "step": 14400
    },
    {
      "epoch": 0.61,
      "grad_norm": 16.20710828996411,
      "learning_rate": 9.67899546880655e-06,
      "loss": 1.0509,
      "step": 14405
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.762825368671312,
      "learning_rate": 9.678561312753073e-06,
      "loss": 1.1079,
      "step": 14410
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.586218987889025,
      "learning_rate": 9.678126873052204e-06,
      "loss": 1.122,
      "step": 14415
    },
    {
      "epoch": 0.61,
      "grad_norm": 10.992598271387472,
      "learning_rate": 9.677692149730282e-06,
      "loss": 1.1441,
      "step": 14420
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.249406138669678,
      "learning_rate": 9.677257142813665e-06,
      "loss": 1.0739,
      "step": 14425
    },
    {
      "epoch": 0.61,
      "grad_norm": 4.975653255007732,
      "learning_rate": 9.676821852328724e-06,
      "loss": 1.1518,
      "step": 14430
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.296017980394742,
      "learning_rate": 9.676386278301851e-06,
      "loss": 1.1016,
      "step": 14435
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.978794255748017,
      "learning_rate": 9.675950420759453e-06,
      "loss": 1.1133,
      "step": 14440
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.093122422684749,
      "learning_rate": 9.675514279727953e-06,
      "loss": 1.117,
      "step": 14445
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.0733795065361,
      "learning_rate": 9.675077855233794e-06,
      "loss": 1.1169,
      "step": 14450
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.453056572996434,
      "learning_rate": 9.674641147303437e-06,
      "loss": 1.1268,
      "step": 14455
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.266373519269233,
      "learning_rate": 9.674204155963356e-06,
      "loss": 1.1457,
      "step": 14460
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.2647054288458115,
      "learning_rate": 9.673766881240045e-06,
      "loss": 1.1204,
      "step": 14465
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.759027051071511,
      "learning_rate": 9.673329323160015e-06,
      "loss": 1.1407,
      "step": 14470
    },
    {
      "epoch": 0.61,
      "grad_norm": 15.897602252911188,
      "learning_rate": 9.672891481749796e-06,
      "loss": 1.1194,
      "step": 14475
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.052546497909548,
      "learning_rate": 9.67245335703593e-06,
      "loss": 1.1268,
      "step": 14480
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.411893677328415,
      "learning_rate": 9.672014949044977e-06,
      "loss": 1.1185,
      "step": 14485
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.5893558155293,
      "learning_rate": 9.671576257803524e-06,
      "loss": 1.0789,
      "step": 14490
    },
    {
      "epoch": 0.61,
      "grad_norm": 12.26452835241954,
      "learning_rate": 9.67113728333816e-06,
      "loss": 1.158,
      "step": 14495
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.822106493314099,
      "learning_rate": 9.670698025675502e-06,
      "loss": 1.1699,
      "step": 14500
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.284019134420179,
      "learning_rate": 9.670258484842182e-06,
      "loss": 1.1387,
      "step": 14505
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.3509592926517655,
      "learning_rate": 9.669818660864847e-06,
      "loss": 1.0986,
      "step": 14510
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.471410795592737,
      "learning_rate": 9.669378553770162e-06,
      "loss": 1.108,
      "step": 14515
    },
    {
      "epoch": 0.61,
      "grad_norm": 13.913658595388489,
      "learning_rate": 9.668938163584807e-06,
      "loss": 1.1193,
      "step": 14520
    },
    {
      "epoch": 0.61,
      "grad_norm": 9.916414957990833,
      "learning_rate": 9.668497490335487e-06,
      "loss": 1.1139,
      "step": 14525
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.577167360695154,
      "learning_rate": 9.668056534048913e-06,
      "loss": 1.1515,
      "step": 14530
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.922769047859126,
      "learning_rate": 9.667615294751824e-06,
      "loss": 1.14,
      "step": 14535
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.535203927215703,
      "learning_rate": 9.667173772470967e-06,
      "loss": 1.2139,
      "step": 14540
    },
    {
      "epoch": 0.62,
      "grad_norm": 36.48363028442636,
      "learning_rate": 9.666731967233113e-06,
      "loss": 1.1349,
      "step": 14545
    },
    {
      "epoch": 0.62,
      "grad_norm": 21.0262985066971,
      "learning_rate": 9.666289879065046e-06,
      "loss": 1.1071,
      "step": 14550
    },
    {
      "epoch": 0.62,
      "grad_norm": 58.844424374469526,
      "learning_rate": 9.665847507993567e-06,
      "loss": 1.0956,
      "step": 14555
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.52443592324907,
      "learning_rate": 9.665404854045497e-06,
      "loss": 1.1442,
      "step": 14560
    },
    {
      "epoch": 0.62,
      "grad_norm": 34.869846029476804,
      "learning_rate": 9.664961917247675e-06,
      "loss": 1.154,
      "step": 14565
    },
    {
      "epoch": 0.62,
      "grad_norm": 27.194417909613236,
      "learning_rate": 9.664518697626952e-06,
      "loss": 1.171,
      "step": 14570
    },
    {
      "epoch": 0.62,
      "grad_norm": 13.706891969274936,
      "learning_rate": 9.664075195210203e-06,
      "loss": 1.1137,
      "step": 14575
    },
    {
      "epoch": 0.62,
      "grad_norm": 36.71351842453812,
      "learning_rate": 9.66363141002431e-06,
      "loss": 1.0981,
      "step": 14580
    },
    {
      "epoch": 0.62,
      "grad_norm": 18.19588738492687,
      "learning_rate": 9.663187342096182e-06,
      "loss": 1.1632,
      "step": 14585
    },
    {
      "epoch": 0.62,
      "grad_norm": 25.530544913332296,
      "learning_rate": 9.662742991452742e-06,
      "loss": 1.1066,
      "step": 14590
    },
    {
      "epoch": 0.62,
      "grad_norm": 21.76289279151669,
      "learning_rate": 9.66229835812093e-06,
      "loss": 1.1201,
      "step": 14595
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.227374121904603,
      "learning_rate": 9.6618534421277e-06,
      "loss": 1.1448,
      "step": 14600
    },
    {
      "epoch": 0.62,
      "grad_norm": 12.131144471095478,
      "learning_rate": 9.66140824350003e-06,
      "loss": 1.1151,
      "step": 14605
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.706943321549794,
      "learning_rate": 9.660962762264908e-06,
      "loss": 1.1375,
      "step": 14610
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.292501325497955,
      "learning_rate": 9.66051699844934e-06,
      "loss": 1.1493,
      "step": 14615
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.38464317476439,
      "learning_rate": 9.660070952080358e-06,
      "loss": 1.1564,
      "step": 14620
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.85675488204554,
      "learning_rate": 9.659624623185e-06,
      "loss": 1.1196,
      "step": 14625
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.2106366247174645,
      "learning_rate": 9.659178011790326e-06,
      "loss": 1.1316,
      "step": 14630
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.121924784323083,
      "learning_rate": 9.658731117923415e-06,
      "loss": 1.1276,
      "step": 14635
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.295109594330357,
      "learning_rate": 9.658283941611356e-06,
      "loss": 1.1206,
      "step": 14640
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.992896623594316,
      "learning_rate": 9.657836482881265e-06,
      "loss": 1.1453,
      "step": 14645
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.104412657694432,
      "learning_rate": 9.657388741760268e-06,
      "loss": 1.1414,
      "step": 14650
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.853921095894423,
      "learning_rate": 9.65694071827551e-06,
      "loss": 1.1506,
      "step": 14655
    },
    {
      "epoch": 0.62,
      "grad_norm": 9.312774107907364,
      "learning_rate": 9.656492412454153e-06,
      "loss": 1.1465,
      "step": 14660
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.436309054999819,
      "learning_rate": 9.656043824323379e-06,
      "loss": 1.1072,
      "step": 14665
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.624628936860415,
      "learning_rate": 9.655594953910382e-06,
      "loss": 1.1126,
      "step": 14670
    },
    {
      "epoch": 0.62,
      "grad_norm": 14.378255020995566,
      "learning_rate": 9.655145801242376e-06,
      "loss": 1.1495,
      "step": 14675
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.236898550021181,
      "learning_rate": 9.654696366346591e-06,
      "loss": 1.0908,
      "step": 14680
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.240160620415107,
      "learning_rate": 9.654246649250279e-06,
      "loss": 1.1509,
      "step": 14685
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.412423079542824,
      "learning_rate": 9.6537966499807e-06,
      "loss": 1.1125,
      "step": 14690
    },
    {
      "epoch": 0.62,
      "grad_norm": 15.974520412331332,
      "learning_rate": 9.65334636856514e-06,
      "loss": 1.1504,
      "step": 14695
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.404924957271204,
      "learning_rate": 9.652895805030896e-06,
      "loss": 1.1567,
      "step": 14700
    },
    {
      "epoch": 0.62,
      "grad_norm": 31.70569891795151,
      "learning_rate": 9.652444959405284e-06,
      "loss": 1.1181,
      "step": 14705
    },
    {
      "epoch": 0.62,
      "grad_norm": 33.13321061852332,
      "learning_rate": 9.651993831715638e-06,
      "loss": 1.1329,
      "step": 14710
    },
    {
      "epoch": 0.62,
      "grad_norm": 22.37197901400863,
      "learning_rate": 9.651542421989311e-06,
      "loss": 1.1608,
      "step": 14715
    },
    {
      "epoch": 0.62,
      "grad_norm": 28.934620271939934,
      "learning_rate": 9.651090730253668e-06,
      "loss": 1.1543,
      "step": 14720
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.170921356817367,
      "learning_rate": 9.650638756536092e-06,
      "loss": 1.1518,
      "step": 14725
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.659704770912452,
      "learning_rate": 9.65018650086399e-06,
      "loss": 1.1423,
      "step": 14730
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.627436737501537,
      "learning_rate": 9.649733963264775e-06,
      "loss": 1.1088,
      "step": 14735
    },
    {
      "epoch": 0.62,
      "grad_norm": 20.15416503207731,
      "learning_rate": 9.649281143765888e-06,
      "loss": 1.1543,
      "step": 14740
    },
    {
      "epoch": 0.62,
      "grad_norm": 11.099105736261631,
      "learning_rate": 9.64882804239478e-06,
      "loss": 1.1466,
      "step": 14745
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.821298000599661,
      "learning_rate": 9.648374659178922e-06,
      "loss": 1.1475,
      "step": 14750
    },
    {
      "epoch": 0.62,
      "grad_norm": 18.5733303838369,
      "learning_rate": 9.6479209941458e-06,
      "loss": 1.1471,
      "step": 14755
    },
    {
      "epoch": 0.62,
      "grad_norm": 18.973824251446448,
      "learning_rate": 9.647467047322918e-06,
      "loss": 1.1101,
      "step": 14760
    },
    {
      "epoch": 0.62,
      "grad_norm": 16.425630115141452,
      "learning_rate": 9.6470128187378e-06,
      "loss": 1.1206,
      "step": 14765
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.660184417370282,
      "learning_rate": 9.646558308417983e-06,
      "loss": 1.127,
      "step": 14770
    },
    {
      "epoch": 0.63,
      "grad_norm": 18.32203882018408,
      "learning_rate": 9.646103516391021e-06,
      "loss": 1.0729,
      "step": 14775
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.301171455064672,
      "learning_rate": 9.64564844268449e-06,
      "loss": 1.1493,
      "step": 14780
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.48203082744321,
      "learning_rate": 9.645193087325978e-06,
      "loss": 1.0993,
      "step": 14785
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.764486995440448,
      "learning_rate": 9.644737450343092e-06,
      "loss": 1.1028,
      "step": 14790
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.638575846552063,
      "learning_rate": 9.644281531763456e-06,
      "loss": 1.097,
      "step": 14795
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.766919340157793,
      "learning_rate": 9.64382533161471e-06,
      "loss": 1.1222,
      "step": 14800
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.07352528372201,
      "learning_rate": 9.643368849924513e-06,
      "loss": 1.1223,
      "step": 14805
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.41311685997666,
      "learning_rate": 9.64291208672054e-06,
      "loss": 1.0925,
      "step": 14810
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.52337127878568,
      "learning_rate": 9.642455042030483e-06,
      "loss": 1.1027,
      "step": 14815
    },
    {
      "epoch": 0.63,
      "grad_norm": 20.750230322413955,
      "learning_rate": 9.641997715882054e-06,
      "loss": 1.1207,
      "step": 14820
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.045843276737358,
      "learning_rate": 9.641540108302975e-06,
      "loss": 1.0781,
      "step": 14825
    },
    {
      "epoch": 0.63,
      "grad_norm": 15.980347740082388,
      "learning_rate": 9.641082219320991e-06,
      "loss": 1.1319,
      "step": 14830
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.18362011388887,
      "learning_rate": 9.640624048963864e-06,
      "loss": 1.1315,
      "step": 14835
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.814234861088137,
      "learning_rate": 9.64016559725937e-06,
      "loss": 1.1083,
      "step": 14840
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.05604326331034,
      "learning_rate": 9.639706864235304e-06,
      "loss": 1.1311,
      "step": 14845
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.546098093676746,
      "learning_rate": 9.639247849919476e-06,
      "loss": 1.1095,
      "step": 14850
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.727428382748387,
      "learning_rate": 9.638788554339717e-06,
      "loss": 1.0946,
      "step": 14855
    },
    {
      "epoch": 0.63,
      "grad_norm": 21.40400777595905,
      "learning_rate": 9.638328977523874e-06,
      "loss": 1.149,
      "step": 14860
    },
    {
      "epoch": 0.63,
      "grad_norm": 25.56823675908026,
      "learning_rate": 9.637869119499804e-06,
      "loss": 1.1223,
      "step": 14865
    },
    {
      "epoch": 0.63,
      "grad_norm": 20.095678010055522,
      "learning_rate": 9.637408980295394e-06,
      "loss": 1.1234,
      "step": 14870
    },
    {
      "epoch": 0.63,
      "grad_norm": 46.0051044355793,
      "learning_rate": 9.636948559938537e-06,
      "loss": 1.1208,
      "step": 14875
    },
    {
      "epoch": 0.63,
      "grad_norm": 14.386593610310534,
      "learning_rate": 9.636487858457146e-06,
      "loss": 1.1485,
      "step": 14880
    },
    {
      "epoch": 0.63,
      "grad_norm": 10.780819395625295,
      "learning_rate": 9.636026875879154e-06,
      "loss": 1.1785,
      "step": 14885
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.371711799097483,
      "learning_rate": 9.635565612232508e-06,
      "loss": 1.1254,
      "step": 14890
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.981367267676593,
      "learning_rate": 9.635104067545173e-06,
      "loss": 1.1234,
      "step": 14895
    },
    {
      "epoch": 0.63,
      "grad_norm": 13.77327259153108,
      "learning_rate": 9.634642241845134e-06,
      "loss": 1.1011,
      "step": 14900
    },
    {
      "epoch": 0.63,
      "grad_norm": 21.445395018488565,
      "learning_rate": 9.634180135160385e-06,
      "loss": 1.1342,
      "step": 14905
    },
    {
      "epoch": 0.63,
      "grad_norm": 26.339476639671517,
      "learning_rate": 9.633717747518946e-06,
      "loss": 1.1879,
      "step": 14910
    },
    {
      "epoch": 0.63,
      "grad_norm": 46.58196426876102,
      "learning_rate": 9.63325507894885e-06,
      "loss": 1.2517,
      "step": 14915
    },
    {
      "epoch": 0.63,
      "grad_norm": 42.15122592569484,
      "learning_rate": 9.632792129478144e-06,
      "loss": 1.1565,
      "step": 14920
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.894451350487213,
      "learning_rate": 9.632328899134897e-06,
      "loss": 1.1536,
      "step": 14925
    },
    {
      "epoch": 0.63,
      "grad_norm": 33.16456657300665,
      "learning_rate": 9.631865387947196e-06,
      "loss": 1.1525,
      "step": 14930
    },
    {
      "epoch": 0.63,
      "grad_norm": 21.98475192767441,
      "learning_rate": 9.631401595943139e-06,
      "loss": 1.1515,
      "step": 14935
    },
    {
      "epoch": 0.63,
      "grad_norm": 26.819245939364354,
      "learning_rate": 9.630937523150848e-06,
      "loss": 1.1326,
      "step": 14940
    },
    {
      "epoch": 0.63,
      "grad_norm": 25.454359725082117,
      "learning_rate": 9.630473169598452e-06,
      "loss": 1.1537,
      "step": 14945
    },
    {
      "epoch": 0.63,
      "grad_norm": 30.190462371778626,
      "learning_rate": 9.63000853531411e-06,
      "loss": 1.1479,
      "step": 14950
    },
    {
      "epoch": 0.63,
      "grad_norm": 44.79817296059923,
      "learning_rate": 9.629543620325988e-06,
      "loss": 1.1309,
      "step": 14955
    },
    {
      "epoch": 0.63,
      "grad_norm": 36.18920488552786,
      "learning_rate": 9.629078424662273e-06,
      "loss": 1.1651,
      "step": 14960
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.481515417629619,
      "learning_rate": 9.628612948351168e-06,
      "loss": 1.1416,
      "step": 14965
    },
    {
      "epoch": 0.63,
      "grad_norm": 34.70285757368889,
      "learning_rate": 9.628147191420893e-06,
      "loss": 1.1099,
      "step": 14970
    },
    {
      "epoch": 0.63,
      "grad_norm": 44.07771473792773,
      "learning_rate": 9.627681153899689e-06,
      "loss": 1.1651,
      "step": 14975
    },
    {
      "epoch": 0.63,
      "grad_norm": 21.236827398173254,
      "learning_rate": 9.627214835815805e-06,
      "loss": 1.0993,
      "step": 14980
    },
    {
      "epoch": 0.63,
      "grad_norm": 16.80799620798502,
      "learning_rate": 9.626748237197518e-06,
      "loss": 1.1662,
      "step": 14985
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.047128242105549,
      "learning_rate": 9.626281358073113e-06,
      "loss": 1.1341,
      "step": 14990
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.836468761643587,
      "learning_rate": 9.625814198470897e-06,
      "loss": 1.1315,
      "step": 14995
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.627893008953913,
      "learning_rate": 9.625346758419191e-06,
      "loss": 1.114,
      "step": 15000
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.330496846937193,
      "learning_rate": 9.624879037946335e-06,
      "loss": 1.1373,
      "step": 15005
    },
    {
      "epoch": 0.64,
      "grad_norm": 21.802550959539268,
      "learning_rate": 9.624411037080687e-06,
      "loss": 1.1129,
      "step": 15010
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.928081298992451,
      "learning_rate": 9.623942755850619e-06,
      "loss": 1.0807,
      "step": 15015
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.989717071840893,
      "learning_rate": 9.623474194284524e-06,
      "loss": 1.1188,
      "step": 15020
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.530098397527423,
      "learning_rate": 9.623005352410805e-06,
      "loss": 1.1367,
      "step": 15025
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.7048373067148725,
      "learning_rate": 9.62253623025789e-06,
      "loss": 1.1486,
      "step": 15030
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.246290567489709,
      "learning_rate": 9.622066827854218e-06,
      "loss": 1.1514,
      "step": 15035
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.862912887518295,
      "learning_rate": 9.62159714522825e-06,
      "loss": 1.127,
      "step": 15040
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.246636272270521,
      "learning_rate": 9.621127182408461e-06,
      "loss": 1.1616,
      "step": 15045
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.3660907840974446,
      "learning_rate": 9.620656939423343e-06,
      "loss": 1.1167,
      "step": 15050
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.1388773365299265,
      "learning_rate": 9.620186416301404e-06,
      "loss": 1.1218,
      "step": 15055
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.368396681648354,
      "learning_rate": 9.619715613071173e-06,
      "loss": 1.1014,
      "step": 15060
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.622130786971702,
      "learning_rate": 9.619244529761191e-06,
      "loss": 1.1116,
      "step": 15065
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.700741596223363,
      "learning_rate": 9.618773166400022e-06,
      "loss": 1.1046,
      "step": 15070
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.334937953106167,
      "learning_rate": 9.618301523016238e-06,
      "loss": 1.1224,
      "step": 15075
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.412167580151939,
      "learning_rate": 9.61782959963844e-06,
      "loss": 1.1225,
      "step": 15080
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.381815950702849,
      "learning_rate": 9.617357396295231e-06,
      "loss": 1.1475,
      "step": 15085
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.607805687189267,
      "learning_rate": 9.616884913015247e-06,
      "loss": 1.1468,
      "step": 15090
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.31491405956735,
      "learning_rate": 9.616412149827131e-06,
      "loss": 1.1441,
      "step": 15095
    },
    {
      "epoch": 0.64,
      "grad_norm": 21.56853357776938,
      "learning_rate": 9.615939106759542e-06,
      "loss": 1.1621,
      "step": 15100
    },
    {
      "epoch": 0.64,
      "grad_norm": 35.003379998015944,
      "learning_rate": 9.615465783841163e-06,
      "loss": 1.0939,
      "step": 15105
    },
    {
      "epoch": 0.64,
      "grad_norm": 14.105672351267819,
      "learning_rate": 9.614992181100686e-06,
      "loss": 1.1313,
      "step": 15110
    },
    {
      "epoch": 0.64,
      "grad_norm": 17.003286142133955,
      "learning_rate": 9.61451829856683e-06,
      "loss": 1.1361,
      "step": 15115
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.321819577415538,
      "learning_rate": 9.614044136268321e-06,
      "loss": 1.1386,
      "step": 15120
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.925335012066343,
      "learning_rate": 9.613569694233908e-06,
      "loss": 1.118,
      "step": 15125
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.741195814966175,
      "learning_rate": 9.613094972492353e-06,
      "loss": 1.1181,
      "step": 15130
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.450759668474747,
      "learning_rate": 9.61261997107244e-06,
      "loss": 1.1353,
      "step": 15135
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.709871208971862,
      "learning_rate": 9.612144690002965e-06,
      "loss": 1.1289,
      "step": 15140
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.668294653667959,
      "learning_rate": 9.61166912931274e-06,
      "loss": 1.0972,
      "step": 15145
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.458141394285696,
      "learning_rate": 9.611193289030603e-06,
      "loss": 1.0752,
      "step": 15150
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.871885733259331,
      "learning_rate": 9.610717169185398e-06,
      "loss": 1.1394,
      "step": 15155
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.356954333625659,
      "learning_rate": 9.610240769805992e-06,
      "loss": 1.122,
      "step": 15160
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.593192755699136,
      "learning_rate": 9.60976409092127e-06,
      "loss": 1.0903,
      "step": 15165
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.608453637198863,
      "learning_rate": 9.609287132560128e-06,
      "loss": 1.1088,
      "step": 15170
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.969575739114598,
      "learning_rate": 9.608809894751483e-06,
      "loss": 1.0988,
      "step": 15175
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.107995819643573,
      "learning_rate": 9.608332377524272e-06,
      "loss": 1.098,
      "step": 15180
    },
    {
      "epoch": 0.64,
      "grad_norm": 10.229953142990972,
      "learning_rate": 9.607854580907442e-06,
      "loss": 1.1688,
      "step": 15185
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.552428134196232,
      "learning_rate": 9.607376504929963e-06,
      "loss": 1.088,
      "step": 15190
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.564121504278818,
      "learning_rate": 9.606898149620819e-06,
      "loss": 1.1601,
      "step": 15195
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.294144000050321,
      "learning_rate": 9.606419515009007e-06,
      "loss": 1.1437,
      "step": 15200
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.855905131825997,
      "learning_rate": 9.605940601123548e-06,
      "loss": 1.0986,
      "step": 15205
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.739582811248398,
      "learning_rate": 9.605461407993482e-06,
      "loss": 1.0751,
      "step": 15210
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.631573384150668,
      "learning_rate": 9.604981935647854e-06,
      "loss": 1.1277,
      "step": 15215
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.480553947910073,
      "learning_rate": 9.604502184115735e-06,
      "loss": 1.1576,
      "step": 15220
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.3106903851644764,
      "learning_rate": 9.604022153426212e-06,
      "loss": 1.1743,
      "step": 15225
    },
    {
      "epoch": 0.64,
      "grad_norm": 26.499021769743585,
      "learning_rate": 9.603541843608389e-06,
      "loss": 1.1499,
      "step": 15230
    },
    {
      "epoch": 0.64,
      "grad_norm": 24.156117228150563,
      "learning_rate": 9.603061254691382e-06,
      "loss": 1.1301,
      "step": 15235
    },
    {
      "epoch": 0.64,
      "grad_norm": 27.54954903843597,
      "learning_rate": 9.602580386704332e-06,
      "loss": 1.1537,
      "step": 15240
    },
    {
      "epoch": 0.65,
      "grad_norm": 41.92336775416203,
      "learning_rate": 9.602099239676388e-06,
      "loss": 1.1213,
      "step": 15245
    },
    {
      "epoch": 0.65,
      "grad_norm": 39.66492236584202,
      "learning_rate": 9.601617813636726e-06,
      "loss": 1.1499,
      "step": 15250
    },
    {
      "epoch": 0.65,
      "grad_norm": 20.84220989989977,
      "learning_rate": 9.601136108614529e-06,
      "loss": 1.0993,
      "step": 15255
    },
    {
      "epoch": 0.65,
      "grad_norm": 10.38712579463707,
      "learning_rate": 9.600654124639001e-06,
      "loss": 1.1428,
      "step": 15260
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.110201153788826,
      "learning_rate": 9.600171861739368e-06,
      "loss": 1.1241,
      "step": 15265
    },
    {
      "epoch": 0.65,
      "grad_norm": 16.55737513511684,
      "learning_rate": 9.599689319944863e-06,
      "loss": 1.1512,
      "step": 15270
    },
    {
      "epoch": 0.65,
      "grad_norm": 19.05739551041232,
      "learning_rate": 9.599206499284746e-06,
      "loss": 1.1165,
      "step": 15275
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.975587719188491,
      "learning_rate": 9.598723399788285e-06,
      "loss": 1.146,
      "step": 15280
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.90430246967029,
      "learning_rate": 9.598240021484769e-06,
      "loss": 1.1598,
      "step": 15285
    },
    {
      "epoch": 0.65,
      "grad_norm": 14.496261326306803,
      "learning_rate": 9.597756364403507e-06,
      "loss": 1.1223,
      "step": 15290
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.361462360200242,
      "learning_rate": 9.597272428573818e-06,
      "loss": 1.1687,
      "step": 15295
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.951477644937146,
      "learning_rate": 9.596788214025046e-06,
      "loss": 1.1478,
      "step": 15300
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.952930515222706,
      "learning_rate": 9.596303720786542e-06,
      "loss": 1.1217,
      "step": 15305
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.302537352729694,
      "learning_rate": 9.595818948887683e-06,
      "loss": 1.13,
      "step": 15310
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.75155589663738,
      "learning_rate": 9.59533389835786e-06,
      "loss": 1.1172,
      "step": 15315
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.409933249584224,
      "learning_rate": 9.594848569226477e-06,
      "loss": 1.163,
      "step": 15320
    },
    {
      "epoch": 0.65,
      "grad_norm": 32.95505167114163,
      "learning_rate": 9.59436296152296e-06,
      "loss": 1.1477,
      "step": 15325
    },
    {
      "epoch": 0.65,
      "grad_norm": 30.137321109503546,
      "learning_rate": 9.59387707527675e-06,
      "loss": 1.1104,
      "step": 15330
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.139967014584997,
      "learning_rate": 9.593390910517306e-06,
      "loss": 1.1372,
      "step": 15335
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.578959372276609,
      "learning_rate": 9.5929044672741e-06,
      "loss": 1.134,
      "step": 15340
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.620349952741133,
      "learning_rate": 9.592417745576625e-06,
      "loss": 1.0904,
      "step": 15345
    },
    {
      "epoch": 0.65,
      "grad_norm": 15.005242597791108,
      "learning_rate": 9.59193074545439e-06,
      "loss": 1.1307,
      "step": 15350
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.910151880964161,
      "learning_rate": 9.591443466936921e-06,
      "loss": 1.1156,
      "step": 15355
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.237347235567157,
      "learning_rate": 9.590955910053756e-06,
      "loss": 1.1365,
      "step": 15360
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.942465973419977,
      "learning_rate": 9.59046807483446e-06,
      "loss": 1.0889,
      "step": 15365
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.094992492945449,
      "learning_rate": 9.589979961308606e-06,
      "loss": 1.0906,
      "step": 15370
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.519965628845473,
      "learning_rate": 9.589491569505786e-06,
      "loss": 1.1386,
      "step": 15375
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.826222461543118,
      "learning_rate": 9.589002899455611e-06,
      "loss": 1.117,
      "step": 15380
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.421505559972395,
      "learning_rate": 9.588513951187709e-06,
      "loss": 1.1214,
      "step": 15385
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.681384302708578,
      "learning_rate": 9.588024724731721e-06,
      "loss": 1.1148,
      "step": 15390
    },
    {
      "epoch": 0.65,
      "grad_norm": 11.7074944230244,
      "learning_rate": 9.587535220117309e-06,
      "loss": 1.1289,
      "step": 15395
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.640409633458371,
      "learning_rate": 9.587045437374149e-06,
      "loss": 1.0997,
      "step": 15400
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.4491849647084525,
      "learning_rate": 9.586555376531935e-06,
      "loss": 1.1295,
      "step": 15405
    },
    {
      "epoch": 0.65,
      "grad_norm": 30.828365701952105,
      "learning_rate": 9.58606503762038e-06,
      "loss": 1.1218,
      "step": 15410
    },
    {
      "epoch": 0.65,
      "grad_norm": 28.877471025102423,
      "learning_rate": 9.58557442066921e-06,
      "loss": 1.1131,
      "step": 15415
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.32310077638982,
      "learning_rate": 9.58508352570817e-06,
      "loss": 1.1105,
      "step": 15420
    },
    {
      "epoch": 0.65,
      "grad_norm": 26.048816125512946,
      "learning_rate": 9.584592352767021e-06,
      "loss": 1.1031,
      "step": 15425
    },
    {
      "epoch": 0.65,
      "grad_norm": 22.648669070359446,
      "learning_rate": 9.584100901875544e-06,
      "loss": 1.1388,
      "step": 15430
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.709365412817161,
      "learning_rate": 9.583609173063531e-06,
      "loss": 1.0632,
      "step": 15435
    },
    {
      "epoch": 0.65,
      "grad_norm": 14.504446695258476,
      "learning_rate": 9.583117166360796e-06,
      "loss": 1.1326,
      "step": 15440
    },
    {
      "epoch": 0.65,
      "grad_norm": 23.930724401242085,
      "learning_rate": 9.582624881797166e-06,
      "loss": 1.1092,
      "step": 15445
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.572371547572622,
      "learning_rate": 9.582132319402489e-06,
      "loss": 1.107,
      "step": 15450
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.015536040581603,
      "learning_rate": 9.581639479206627e-06,
      "loss": 1.1566,
      "step": 15455
    },
    {
      "epoch": 0.65,
      "grad_norm": 13.490313232809362,
      "learning_rate": 9.581146361239457e-06,
      "loss": 1.119,
      "step": 15460
    },
    {
      "epoch": 0.65,
      "grad_norm": 38.448176265161806,
      "learning_rate": 9.580652965530878e-06,
      "loss": 1.1208,
      "step": 15465
    },
    {
      "epoch": 0.65,
      "grad_norm": 47.74830792797482,
      "learning_rate": 9.580159292110802e-06,
      "loss": 1.1239,
      "step": 15470
    },
    {
      "epoch": 0.65,
      "grad_norm": 26.011696593388926,
      "learning_rate": 9.579665341009161e-06,
      "loss": 1.1341,
      "step": 15475
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.986201059912118,
      "learning_rate": 9.5791711122559e-06,
      "loss": 1.0977,
      "step": 15480
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.02722953640949,
      "learning_rate": 9.578676605880983e-06,
      "loss": 1.1241,
      "step": 15485
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.694007453047524,
      "learning_rate": 9.578181821914389e-06,
      "loss": 1.129,
      "step": 15490
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.395570151734314,
      "learning_rate": 9.577686760386118e-06,
      "loss": 1.1279,
      "step": 15495
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.100312053589473,
      "learning_rate": 9.577191421326181e-06,
      "loss": 1.0861,
      "step": 15500
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.341903044806589,
      "learning_rate": 9.576695804764611e-06,
      "loss": 1.0621,
      "step": 15505
    },
    {
      "epoch": 0.66,
      "grad_norm": 23.190702477762677,
      "learning_rate": 9.576199910731457e-06,
      "loss": 1.1588,
      "step": 15510
    },
    {
      "epoch": 0.66,
      "grad_norm": 23.639729053832877,
      "learning_rate": 9.57570373925678e-06,
      "loss": 1.1282,
      "step": 15515
    },
    {
      "epoch": 0.66,
      "grad_norm": 17.305840873138052,
      "learning_rate": 9.575207290370667e-06,
      "loss": 1.1465,
      "step": 15520
    },
    {
      "epoch": 0.66,
      "grad_norm": 32.69940676134057,
      "learning_rate": 9.574710564103208e-06,
      "loss": 1.1625,
      "step": 15525
    },
    {
      "epoch": 0.66,
      "grad_norm": 37.39367118463734,
      "learning_rate": 9.574213560484526e-06,
      "loss": 1.1548,
      "step": 15530
    },
    {
      "epoch": 0.66,
      "grad_norm": 29.88323507569578,
      "learning_rate": 9.573716279544749e-06,
      "loss": 1.0891,
      "step": 15535
    },
    {
      "epoch": 0.66,
      "grad_norm": 30.463372908170502,
      "learning_rate": 9.573218721314028e-06,
      "loss": 1.1628,
      "step": 15540
    },
    {
      "epoch": 0.66,
      "grad_norm": 31.169487999370382,
      "learning_rate": 9.572720885822523e-06,
      "loss": 1.1217,
      "step": 15545
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.6138819420453805,
      "learning_rate": 9.572222773100423e-06,
      "loss": 1.1429,
      "step": 15550
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.1102214560028,
      "learning_rate": 9.571724383177923e-06,
      "loss": 1.1188,
      "step": 15555
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.633020345327878,
      "learning_rate": 9.571225716085239e-06,
      "loss": 1.1449,
      "step": 15560
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.6029530390775735,
      "learning_rate": 9.570726771852608e-06,
      "loss": 1.0976,
      "step": 15565
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.368538144798845,
      "learning_rate": 9.570227550510274e-06,
      "loss": 1.1194,
      "step": 15570
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.955013860094212,
      "learning_rate": 9.569728052088505e-06,
      "loss": 1.1142,
      "step": 15575
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.799660654824613,
      "learning_rate": 9.569228276617587e-06,
      "loss": 1.1305,
      "step": 15580
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.6834533740776925,
      "learning_rate": 9.568728224127817e-06,
      "loss": 1.1182,
      "step": 15585
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.783924326567748,
      "learning_rate": 9.56822789464951e-06,
      "loss": 1.1301,
      "step": 15590
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.054448725862445,
      "learning_rate": 9.567727288213005e-06,
      "loss": 1.0734,
      "step": 15595
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.443051879102491,
      "learning_rate": 9.567226404848648e-06,
      "loss": 1.1024,
      "step": 15600
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.629438032018662,
      "learning_rate": 9.566725244586807e-06,
      "loss": 1.0917,
      "step": 15605
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.231888706468046,
      "learning_rate": 9.566223807457867e-06,
      "loss": 1.0969,
      "step": 15610
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.790613630056143,
      "learning_rate": 9.565722093492226e-06,
      "loss": 1.1092,
      "step": 15615
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.649892906466696,
      "learning_rate": 9.565220102720306e-06,
      "loss": 1.1045,
      "step": 15620
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.83614325693871,
      "learning_rate": 9.564717835172536e-06,
      "loss": 1.0735,
      "step": 15625
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.741316253963543,
      "learning_rate": 9.564215290879371e-06,
      "loss": 1.15,
      "step": 15630
    },
    {
      "epoch": 0.66,
      "grad_norm": 10.543531557768679,
      "learning_rate": 9.563712469871277e-06,
      "loss": 1.0852,
      "step": 15635
    },
    {
      "epoch": 0.66,
      "grad_norm": 16.545885733273415,
      "learning_rate": 9.563209372178738e-06,
      "loss": 1.1104,
      "step": 15640
    },
    {
      "epoch": 0.66,
      "grad_norm": 16.379386872751233,
      "learning_rate": 9.562705997832259e-06,
      "loss": 1.1113,
      "step": 15645
    },
    {
      "epoch": 0.66,
      "grad_norm": 14.40211055238993,
      "learning_rate": 9.562202346862352e-06,
      "loss": 1.1412,
      "step": 15650
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.807773616994584,
      "learning_rate": 9.561698419299555e-06,
      "loss": 1.1149,
      "step": 15655
    },
    {
      "epoch": 0.66,
      "grad_norm": 14.39871623794586,
      "learning_rate": 9.56119421517442e-06,
      "loss": 1.1371,
      "step": 15660
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.291224197745935,
      "learning_rate": 9.560689734517517e-06,
      "loss": 1.0982,
      "step": 15665
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.707853349257634,
      "learning_rate": 9.560184977359429e-06,
      "loss": 1.1654,
      "step": 15670
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.95705431277963,
      "learning_rate": 9.559679943730757e-06,
      "loss": 1.108,
      "step": 15675
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.318265773283196,
      "learning_rate": 9.559174633662123e-06,
      "loss": 1.0886,
      "step": 15680
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.238269887472416,
      "learning_rate": 9.558669047184157e-06,
      "loss": 1.0963,
      "step": 15685
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.118318181593873,
      "learning_rate": 9.558163184327519e-06,
      "loss": 1.1661,
      "step": 15690
    },
    {
      "epoch": 0.66,
      "grad_norm": 7.576816564619521,
      "learning_rate": 9.55765704512287e-06,
      "loss": 1.097,
      "step": 15695
    },
    {
      "epoch": 0.66,
      "grad_norm": 18.722824564847656,
      "learning_rate": 9.557150629600901e-06,
      "loss": 1.1214,
      "step": 15700
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.425301804278783,
      "learning_rate": 9.556643937792313e-06,
      "loss": 1.097,
      "step": 15705
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.88683107008038,
      "learning_rate": 9.556136969727825e-06,
      "loss": 1.1574,
      "step": 15710
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.418389644282192,
      "learning_rate": 9.555629725438172e-06,
      "loss": 1.1009,
      "step": 15715
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.832660233585669,
      "learning_rate": 9.555122204954109e-06,
      "loss": 1.1345,
      "step": 15720
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.003568247307738,
      "learning_rate": 9.554614408306405e-06,
      "loss": 1.0998,
      "step": 15725
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.71699216321345,
      "learning_rate": 9.554106335525843e-06,
      "loss": 1.1264,
      "step": 15730
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.291658488293862,
      "learning_rate": 9.55359798664323e-06,
      "loss": 1.125,
      "step": 15735
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.86309031595583,
      "learning_rate": 9.553089361689382e-06,
      "loss": 1.102,
      "step": 15740
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.669922414982565,
      "learning_rate": 9.552580460695141e-06,
      "loss": 1.13,
      "step": 15745
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.947133682212815,
      "learning_rate": 9.552071283691355e-06,
      "loss": 1.0998,
      "step": 15750
    },
    {
      "epoch": 0.67,
      "grad_norm": 12.20076178027709,
      "learning_rate": 9.551561830708895e-06,
      "loss": 1.1301,
      "step": 15755
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.68342801957456,
      "learning_rate": 9.55105210177865e-06,
      "loss": 1.1394,
      "step": 15760
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.222756752599814,
      "learning_rate": 9.55054209693152e-06,
      "loss": 1.0735,
      "step": 15765
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.186772023727826,
      "learning_rate": 9.550031816198429e-06,
      "loss": 1.0959,
      "step": 15770
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.3105167317237685,
      "learning_rate": 9.54952125961031e-06,
      "loss": 1.1537,
      "step": 15775
    },
    {
      "epoch": 0.67,
      "grad_norm": 14.938936955446723,
      "learning_rate": 9.549010427198118e-06,
      "loss": 1.1429,
      "step": 15780
    },
    {
      "epoch": 0.67,
      "grad_norm": 16.576650004438832,
      "learning_rate": 9.548499318992824e-06,
      "loss": 1.1108,
      "step": 15785
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.816720702593366,
      "learning_rate": 9.547987935025412e-06,
      "loss": 1.0916,
      "step": 15790
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.261444769111539,
      "learning_rate": 9.54747627532689e-06,
      "loss": 1.1293,
      "step": 15795
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.129132085123604,
      "learning_rate": 9.546964339928276e-06,
      "loss": 1.1087,
      "step": 15800
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.167944209636138,
      "learning_rate": 9.546452128860607e-06,
      "loss": 1.1325,
      "step": 15805
    },
    {
      "epoch": 0.67,
      "grad_norm": 14.30170678593782,
      "learning_rate": 9.54593964215494e-06,
      "loss": 1.1487,
      "step": 15810
    },
    {
      "epoch": 0.67,
      "grad_norm": 15.31473977237239,
      "learning_rate": 9.545426879842341e-06,
      "loss": 1.1125,
      "step": 15815
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.344238010450255,
      "learning_rate": 9.544913841953898e-06,
      "loss": 1.1277,
      "step": 15820
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.851476086789551,
      "learning_rate": 9.544400528520718e-06,
      "loss": 1.1238,
      "step": 15825
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.344716489051912,
      "learning_rate": 9.543886939573921e-06,
      "loss": 1.1089,
      "step": 15830
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.706495578276909,
      "learning_rate": 9.543373075144642e-06,
      "loss": 1.1363,
      "step": 15835
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.833524880990271,
      "learning_rate": 9.542858935264036e-06,
      "loss": 1.103,
      "step": 15840
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.888900647368413,
      "learning_rate": 9.542344519963274e-06,
      "loss": 1.1431,
      "step": 15845
    },
    {
      "epoch": 0.67,
      "grad_norm": 24.72540705356045,
      "learning_rate": 9.541829829273546e-06,
      "loss": 1.1353,
      "step": 15850
    },
    {
      "epoch": 0.67,
      "grad_norm": 10.60610919904562,
      "learning_rate": 9.541314863226052e-06,
      "loss": 1.1313,
      "step": 15855
    },
    {
      "epoch": 0.67,
      "grad_norm": 16.89472331801454,
      "learning_rate": 9.540799621852014e-06,
      "loss": 1.1158,
      "step": 15860
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.72988600930111,
      "learning_rate": 9.540284105182671e-06,
      "loss": 1.1197,
      "step": 15865
    },
    {
      "epoch": 0.67,
      "grad_norm": 19.191752374280185,
      "learning_rate": 9.539768313249276e-06,
      "loss": 1.1337,
      "step": 15870
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.856539691834326,
      "learning_rate": 9.5392522460831e-06,
      "loss": 1.0902,
      "step": 15875
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.920787210253751,
      "learning_rate": 9.538735903715435e-06,
      "loss": 1.1644,
      "step": 15880
    },
    {
      "epoch": 0.67,
      "grad_norm": 19.515779275107036,
      "learning_rate": 9.538219286177579e-06,
      "loss": 1.0936,
      "step": 15885
    },
    {
      "epoch": 0.67,
      "grad_norm": 8.302226564565773,
      "learning_rate": 9.537702393500854e-06,
      "loss": 1.1388,
      "step": 15890
    },
    {
      "epoch": 0.67,
      "grad_norm": 43.789120777383566,
      "learning_rate": 9.537185225716601e-06,
      "loss": 1.1483,
      "step": 15895
    },
    {
      "epoch": 0.67,
      "grad_norm": 35.37208805690864,
      "learning_rate": 9.536667782856172e-06,
      "loss": 1.1654,
      "step": 15900
    },
    {
      "epoch": 0.67,
      "grad_norm": 14.85696555679927,
      "learning_rate": 9.536150064950937e-06,
      "loss": 1.184,
      "step": 15905
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.0762966819038065,
      "learning_rate": 9.535632072032288e-06,
      "loss": 1.1384,
      "step": 15910
    },
    {
      "epoch": 0.67,
      "grad_norm": 20.23928215278144,
      "learning_rate": 9.535113804131625e-06,
      "loss": 1.1511,
      "step": 15915
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.104557315697825,
      "learning_rate": 9.53459526128037e-06,
      "loss": 1.1009,
      "step": 15920
    },
    {
      "epoch": 0.67,
      "grad_norm": 15.795062846814455,
      "learning_rate": 9.534076443509963e-06,
      "loss": 1.1241,
      "step": 15925
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.636920453176176,
      "learning_rate": 9.533557350851857e-06,
      "loss": 1.1081,
      "step": 15930
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.241516908927179,
      "learning_rate": 9.533037983337521e-06,
      "loss": 1.0959,
      "step": 15935
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.778451139155578,
      "learning_rate": 9.532518340998444e-06,
      "loss": 1.1441,
      "step": 15940
    },
    {
      "epoch": 0.67,
      "grad_norm": 15.62020461296476,
      "learning_rate": 9.531998423866133e-06,
      "loss": 1.0973,
      "step": 15945
    },
    {
      "epoch": 0.68,
      "grad_norm": 20.267337819458096,
      "learning_rate": 9.531478231972107e-06,
      "loss": 1.1004,
      "step": 15950
    },
    {
      "epoch": 0.68,
      "grad_norm": 21.857553320454397,
      "learning_rate": 9.530957765347903e-06,
      "loss": 1.1235,
      "step": 15955
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.091841783402426,
      "learning_rate": 9.530437024025077e-06,
      "loss": 1.1269,
      "step": 15960
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.705531629037003,
      "learning_rate": 9.529916008035199e-06,
      "loss": 1.1122,
      "step": 15965
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.588748077856957,
      "learning_rate": 9.529394717409855e-06,
      "loss": 1.1093,
      "step": 15970
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.094558818256664,
      "learning_rate": 9.52887315218065e-06,
      "loss": 1.0989,
      "step": 15975
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.787451083155929,
      "learning_rate": 9.52835131237921e-06,
      "loss": 1.1398,
      "step": 15980
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.1548537208422305,
      "learning_rate": 9.527829198037167e-06,
      "loss": 1.1331,
      "step": 15985
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.652170694507138,
      "learning_rate": 9.527306809186176e-06,
      "loss": 1.0993,
      "step": 15990
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.655357942185252,
      "learning_rate": 9.52678414585791e-06,
      "loss": 1.1247,
      "step": 15995
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.386303646992126,
      "learning_rate": 9.526261208084054e-06,
      "loss": 1.1226,
      "step": 16000
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.6778458086669845,
      "learning_rate": 9.525737995896313e-06,
      "loss": 1.0901,
      "step": 16005
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.39788114004018,
      "learning_rate": 9.52521450932641e-06,
      "loss": 1.1147,
      "step": 16010
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.652087015538783,
      "learning_rate": 9.52469074840608e-06,
      "loss": 1.1446,
      "step": 16015
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.40592830103076,
      "learning_rate": 9.524166713167077e-06,
      "loss": 1.1256,
      "step": 16020
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.01522461655715,
      "learning_rate": 9.523642403641176e-06,
      "loss": 1.1371,
      "step": 16025
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.246350472239362,
      "learning_rate": 9.523117819860158e-06,
      "loss": 1.1202,
      "step": 16030
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.297322699968756,
      "learning_rate": 9.52259296185583e-06,
      "loss": 1.0879,
      "step": 16035
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.013068269723783,
      "learning_rate": 9.522067829660013e-06,
      "loss": 1.125,
      "step": 16040
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.293790337648549,
      "learning_rate": 9.521542423304544e-06,
      "loss": 1.159,
      "step": 16045
    },
    {
      "epoch": 0.68,
      "grad_norm": 4.835919261424686,
      "learning_rate": 9.521016742821275e-06,
      "loss": 1.1087,
      "step": 16050
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.975019618669092,
      "learning_rate": 9.52049078824208e-06,
      "loss": 1.0601,
      "step": 16055
    },
    {
      "epoch": 0.68,
      "grad_norm": 26.809077836318586,
      "learning_rate": 9.519964559598842e-06,
      "loss": 1.1079,
      "step": 16060
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.223704884661332,
      "learning_rate": 9.519438056923468e-06,
      "loss": 1.0978,
      "step": 16065
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.007418574741625,
      "learning_rate": 9.518911280247877e-06,
      "loss": 1.0972,
      "step": 16070
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.307951209339337,
      "learning_rate": 9.518384229604004e-06,
      "loss": 1.1494,
      "step": 16075
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.847877619887234,
      "learning_rate": 9.517856905023808e-06,
      "loss": 1.094,
      "step": 16080
    },
    {
      "epoch": 0.68,
      "grad_norm": 21.596597576292055,
      "learning_rate": 9.517329306539254e-06,
      "loss": 1.1117,
      "step": 16085
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.248205504346232,
      "learning_rate": 9.51680143418233e-06,
      "loss": 1.1208,
      "step": 16090
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.96613183549509,
      "learning_rate": 9.516273287985041e-06,
      "loss": 1.1585,
      "step": 16095
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.801891804621334,
      "learning_rate": 9.515744867979404e-06,
      "loss": 1.0901,
      "step": 16100
    },
    {
      "epoch": 0.68,
      "grad_norm": 11.049608898358061,
      "learning_rate": 9.515216174197457e-06,
      "loss": 1.0984,
      "step": 16105
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.075499506984404,
      "learning_rate": 9.514687206671255e-06,
      "loss": 1.139,
      "step": 16110
    },
    {
      "epoch": 0.68,
      "grad_norm": 10.316105228669224,
      "learning_rate": 9.514157965432866e-06,
      "loss": 1.0923,
      "step": 16115
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.499657914771534,
      "learning_rate": 9.513628450514376e-06,
      "loss": 1.1219,
      "step": 16120
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.68355911771093,
      "learning_rate": 9.513098661947888e-06,
      "loss": 1.1414,
      "step": 16125
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.640527817201807,
      "learning_rate": 9.51256859976552e-06,
      "loss": 1.13,
      "step": 16130
    },
    {
      "epoch": 0.68,
      "grad_norm": 16.602723941317617,
      "learning_rate": 9.512038263999414e-06,
      "loss": 1.1081,
      "step": 16135
    },
    {
      "epoch": 0.68,
      "grad_norm": 19.779127002067398,
      "learning_rate": 9.511507654681717e-06,
      "loss": 1.1269,
      "step": 16140
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.327055552687504,
      "learning_rate": 9.5109767718446e-06,
      "loss": 1.1417,
      "step": 16145
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.872922848971763,
      "learning_rate": 9.510445615520248e-06,
      "loss": 1.1774,
      "step": 16150
    },
    {
      "epoch": 0.68,
      "grad_norm": 28.237574261276997,
      "learning_rate": 9.509914185740864e-06,
      "loss": 1.1181,
      "step": 16155
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.998291608488835,
      "learning_rate": 9.50938248253867e-06,
      "loss": 1.1581,
      "step": 16160
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.765806747822108,
      "learning_rate": 9.508850505945896e-06,
      "loss": 1.1223,
      "step": 16165
    },
    {
      "epoch": 0.68,
      "grad_norm": 12.429989143848161,
      "learning_rate": 9.508318255994798e-06,
      "loss": 1.1074,
      "step": 16170
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.564872981137462,
      "learning_rate": 9.507785732717645e-06,
      "loss": 1.1088,
      "step": 16175
    },
    {
      "epoch": 0.68,
      "grad_norm": 17.90562616962865,
      "learning_rate": 9.50725293614672e-06,
      "loss": 1.1405,
      "step": 16180
    },
    {
      "epoch": 0.68,
      "grad_norm": 16.709698958734737,
      "learning_rate": 9.506719866314326e-06,
      "loss": 1.0864,
      "step": 16185
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.107245754363326,
      "learning_rate": 9.506186523252781e-06,
      "loss": 1.0789,
      "step": 16190
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.172998808660411,
      "learning_rate": 9.505652906994421e-06,
      "loss": 1.225,
      "step": 16195
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.204090149073378,
      "learning_rate": 9.505119017571599e-06,
      "loss": 1.1731,
      "step": 16200
    },
    {
      "epoch": 0.69,
      "grad_norm": 13.789001715457545,
      "learning_rate": 9.504584855016678e-06,
      "loss": 1.0912,
      "step": 16205
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.282588972993546,
      "learning_rate": 9.504050419362047e-06,
      "loss": 1.1168,
      "step": 16210
    },
    {
      "epoch": 0.69,
      "grad_norm": 26.93422669138498,
      "learning_rate": 9.503515710640108e-06,
      "loss": 1.1653,
      "step": 16215
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.774993639459181,
      "learning_rate": 9.502980728883275e-06,
      "loss": 1.0825,
      "step": 16220
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.722513696576831,
      "learning_rate": 9.502445474123984e-06,
      "loss": 1.1287,
      "step": 16225
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.518843001585743,
      "learning_rate": 9.501909946394688e-06,
      "loss": 1.1302,
      "step": 16230
    },
    {
      "epoch": 0.69,
      "grad_norm": 15.861134296712128,
      "learning_rate": 9.501374145727854e-06,
      "loss": 1.113,
      "step": 16235
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.522280441281781,
      "learning_rate": 9.500838072155961e-06,
      "loss": 1.1006,
      "step": 16240
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.687416219228599,
      "learning_rate": 9.500301725711516e-06,
      "loss": 1.1507,
      "step": 16245
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.217268767273105,
      "learning_rate": 9.499765106427033e-06,
      "loss": 1.1127,
      "step": 16250
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.184309992267938,
      "learning_rate": 9.499228214335045e-06,
      "loss": 1.1159,
      "step": 16255
    },
    {
      "epoch": 0.69,
      "grad_norm": 19.033486016390004,
      "learning_rate": 9.498691049468103e-06,
      "loss": 1.0839,
      "step": 16260
    },
    {
      "epoch": 0.69,
      "grad_norm": 16.854178720427843,
      "learning_rate": 9.498153611858776e-06,
      "loss": 1.074,
      "step": 16265
    },
    {
      "epoch": 0.69,
      "grad_norm": 23.058516645518235,
      "learning_rate": 9.497615901539642e-06,
      "loss": 1.1786,
      "step": 16270
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.234404255321616,
      "learning_rate": 9.497077918543309e-06,
      "loss": 1.1098,
      "step": 16275
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.644911172226839,
      "learning_rate": 9.496539662902384e-06,
      "loss": 1.1383,
      "step": 16280
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.895257087064357,
      "learning_rate": 9.496001134649507e-06,
      "loss": 1.1289,
      "step": 16285
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.890853218819522,
      "learning_rate": 9.495462333817323e-06,
      "loss": 1.1314,
      "step": 16290
    },
    {
      "epoch": 0.69,
      "grad_norm": 14.88983867536514,
      "learning_rate": 9.4949232604385e-06,
      "loss": 1.1254,
      "step": 16295
    },
    {
      "epoch": 0.69,
      "grad_norm": 45.02490898857594,
      "learning_rate": 9.494383914545719e-06,
      "loss": 1.1536,
      "step": 16300
    },
    {
      "epoch": 0.69,
      "grad_norm": 15.21717036655647,
      "learning_rate": 9.493844296171681e-06,
      "loss": 1.1455,
      "step": 16305
    },
    {
      "epoch": 0.69,
      "grad_norm": 29.87505299491644,
      "learning_rate": 9.493304405349101e-06,
      "loss": 1.1207,
      "step": 16310
    },
    {
      "epoch": 0.69,
      "grad_norm": 29.983671380398928,
      "learning_rate": 9.49276424211071e-06,
      "loss": 1.133,
      "step": 16315
    },
    {
      "epoch": 0.69,
      "grad_norm": 20.735292581448892,
      "learning_rate": 9.492223806489256e-06,
      "loss": 1.1071,
      "step": 16320
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.83409718003557,
      "learning_rate": 9.491683098517505e-06,
      "loss": 1.1569,
      "step": 16325
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.172079042977844,
      "learning_rate": 9.49114211822824e-06,
      "loss": 1.09,
      "step": 16330
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.763372494395667,
      "learning_rate": 9.490600865654255e-06,
      "loss": 1.1123,
      "step": 16335
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.56089879060707,
      "learning_rate": 9.490059340828369e-06,
      "loss": 1.1648,
      "step": 16340
    },
    {
      "epoch": 0.69,
      "grad_norm": 14.45202987057126,
      "learning_rate": 9.489517543783412e-06,
      "loss": 1.103,
      "step": 16345
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.019185491297037,
      "learning_rate": 9.488975474552226e-06,
      "loss": 1.081,
      "step": 16350
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.742327704659386,
      "learning_rate": 9.488433133167683e-06,
      "loss": 1.1387,
      "step": 16355
    },
    {
      "epoch": 0.69,
      "grad_norm": 11.401143502026688,
      "learning_rate": 9.487890519662662e-06,
      "loss": 1.1229,
      "step": 16360
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.89927105312118,
      "learning_rate": 9.487347634070055e-06,
      "loss": 1.0955,
      "step": 16365
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.008315941857413,
      "learning_rate": 9.48680447642278e-06,
      "loss": 1.1261,
      "step": 16370
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.744313717021116,
      "learning_rate": 9.486261046753767e-06,
      "loss": 1.122,
      "step": 16375
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.758601921147918,
      "learning_rate": 9.485717345095959e-06,
      "loss": 1.1122,
      "step": 16380
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.707997637719105,
      "learning_rate": 9.485173371482323e-06,
      "loss": 1.1099,
      "step": 16385
    },
    {
      "epoch": 0.69,
      "grad_norm": 13.043130417480645,
      "learning_rate": 9.484629125945837e-06,
      "loss": 1.1429,
      "step": 16390
    },
    {
      "epoch": 0.69,
      "grad_norm": 10.65680356110051,
      "learning_rate": 9.484084608519497e-06,
      "loss": 1.1278,
      "step": 16395
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.398807630176025,
      "learning_rate": 9.483539819236316e-06,
      "loss": 1.1294,
      "step": 16400
    },
    {
      "epoch": 0.69,
      "grad_norm": 20.79442772954482,
      "learning_rate": 9.482994758129322e-06,
      "loss": 1.111,
      "step": 16405
    },
    {
      "epoch": 0.69,
      "grad_norm": 16.319576719853313,
      "learning_rate": 9.48244942523156e-06,
      "loss": 1.1385,
      "step": 16410
    },
    {
      "epoch": 0.69,
      "grad_norm": 26.400663066036593,
      "learning_rate": 9.481903820576093e-06,
      "loss": 1.1272,
      "step": 16415
    },
    {
      "epoch": 0.69,
      "grad_norm": 13.146974735838572,
      "learning_rate": 9.481357944196001e-06,
      "loss": 1.1244,
      "step": 16420
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.770150240753477,
      "learning_rate": 9.480811796124376e-06,
      "loss": 1.1168,
      "step": 16425
    },
    {
      "epoch": 0.7,
      "grad_norm": 21.510290694212912,
      "learning_rate": 9.48026537639433e-06,
      "loss": 1.1411,
      "step": 16430
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.124776639888739,
      "learning_rate": 9.479718685038991e-06,
      "loss": 1.1448,
      "step": 16435
    },
    {
      "epoch": 0.7,
      "grad_norm": 19.330327500848504,
      "learning_rate": 9.479171722091504e-06,
      "loss": 1.1232,
      "step": 16440
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.162637567554073,
      "learning_rate": 9.478624487585031e-06,
      "loss": 1.1045,
      "step": 16445
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.40534662028943,
      "learning_rate": 9.478076981552746e-06,
      "loss": 1.1079,
      "step": 16450
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.273655272335528,
      "learning_rate": 9.477529204027844e-06,
      "loss": 1.1047,
      "step": 16455
    },
    {
      "epoch": 0.7,
      "grad_norm": 19.68001402078473,
      "learning_rate": 9.476981155043537e-06,
      "loss": 1.1248,
      "step": 16460
    },
    {
      "epoch": 0.7,
      "grad_norm": 21.380072068327515,
      "learning_rate": 9.476432834633048e-06,
      "loss": 1.1198,
      "step": 16465
    },
    {
      "epoch": 0.7,
      "grad_norm": 34.85628614751562,
      "learning_rate": 9.475884242829623e-06,
      "loss": 1.1098,
      "step": 16470
    },
    {
      "epoch": 0.7,
      "grad_norm": 47.94743515292792,
      "learning_rate": 9.47533537966652e-06,
      "loss": 1.1085,
      "step": 16475
    },
    {
      "epoch": 0.7,
      "grad_norm": 20.81681702072176,
      "learning_rate": 9.474786245177015e-06,
      "loss": 1.1517,
      "step": 16480
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.068269343963848,
      "learning_rate": 9.474236839394402e-06,
      "loss": 1.1808,
      "step": 16485
    },
    {
      "epoch": 0.7,
      "grad_norm": 21.19243581628494,
      "learning_rate": 9.473687162351987e-06,
      "loss": 1.1239,
      "step": 16490
    },
    {
      "epoch": 0.7,
      "grad_norm": 37.21512502656769,
      "learning_rate": 9.473137214083099e-06,
      "loss": 1.1444,
      "step": 16495
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.356721661213098,
      "learning_rate": 9.472586994621077e-06,
      "loss": 1.1335,
      "step": 16500
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.130062442997964,
      "learning_rate": 9.47203650399928e-06,
      "loss": 1.1571,
      "step": 16505
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.896186773646387,
      "learning_rate": 9.471485742251083e-06,
      "loss": 1.1137,
      "step": 16510
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.551283986598412,
      "learning_rate": 9.470934709409876e-06,
      "loss": 1.1337,
      "step": 16515
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.7003160427665,
      "learning_rate": 9.470383405509067e-06,
      "loss": 1.1181,
      "step": 16520
    },
    {
      "epoch": 0.7,
      "grad_norm": 16.27370780684043,
      "learning_rate": 9.469831830582082e-06,
      "loss": 1.1203,
      "step": 16525
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.450951092994064,
      "learning_rate": 9.469279984662356e-06,
      "loss": 1.1223,
      "step": 16530
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.826104983490803,
      "learning_rate": 9.468727867783352e-06,
      "loss": 1.1338,
      "step": 16535
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.548589140974116,
      "learning_rate": 9.468175479978538e-06,
      "loss": 1.0937,
      "step": 16540
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.04772186126436,
      "learning_rate": 9.467622821281408e-06,
      "loss": 1.0864,
      "step": 16545
    },
    {
      "epoch": 0.7,
      "grad_norm": 13.61626104485777,
      "learning_rate": 9.467069891725466e-06,
      "loss": 1.1257,
      "step": 16550
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.4844180416513435,
      "learning_rate": 9.466516691344235e-06,
      "loss": 1.1204,
      "step": 16555
    },
    {
      "epoch": 0.7,
      "grad_norm": 28.168245753658315,
      "learning_rate": 9.465963220171252e-06,
      "loss": 1.1052,
      "step": 16560
    },
    {
      "epoch": 0.7,
      "grad_norm": 32.393992640015774,
      "learning_rate": 9.465409478240074e-06,
      "loss": 1.1046,
      "step": 16565
    },
    {
      "epoch": 0.7,
      "grad_norm": 11.053725853513852,
      "learning_rate": 9.464855465584275e-06,
      "loss": 1.1247,
      "step": 16570
    },
    {
      "epoch": 0.7,
      "grad_norm": 16.38971028809713,
      "learning_rate": 9.464301182237439e-06,
      "loss": 1.1117,
      "step": 16575
    },
    {
      "epoch": 0.7,
      "grad_norm": 17.61020730135982,
      "learning_rate": 9.463746628233173e-06,
      "loss": 1.1814,
      "step": 16580
    },
    {
      "epoch": 0.7,
      "grad_norm": 21.44109326803698,
      "learning_rate": 9.463191803605097e-06,
      "loss": 1.1135,
      "step": 16585
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.426235470967014,
      "learning_rate": 9.462636708386849e-06,
      "loss": 1.119,
      "step": 16590
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.947325330805191,
      "learning_rate": 9.462081342612081e-06,
      "loss": 1.0905,
      "step": 16595
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.528238906675508,
      "learning_rate": 9.461525706314467e-06,
      "loss": 1.1249,
      "step": 16600
    },
    {
      "epoch": 0.7,
      "grad_norm": 18.805288994820586,
      "learning_rate": 9.46096979952769e-06,
      "loss": 1.1099,
      "step": 16605
    },
    {
      "epoch": 0.7,
      "grad_norm": 23.33570587767011,
      "learning_rate": 9.460413622285454e-06,
      "loss": 1.1433,
      "step": 16610
    },
    {
      "epoch": 0.7,
      "grad_norm": 14.004866042143698,
      "learning_rate": 9.45985717462148e-06,
      "loss": 1.0779,
      "step": 16615
    },
    {
      "epoch": 0.7,
      "grad_norm": 14.242876809876437,
      "learning_rate": 9.4593004565695e-06,
      "loss": 1.1287,
      "step": 16620
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.65946635603587,
      "learning_rate": 9.458743468163271e-06,
      "loss": 1.1391,
      "step": 16625
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.191679028924941,
      "learning_rate": 9.458186209436559e-06,
      "loss": 1.1183,
      "step": 16630
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.594517614370647,
      "learning_rate": 9.457628680423148e-06,
      "loss": 1.1034,
      "step": 16635
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.6666821311089075,
      "learning_rate": 9.45707088115684e-06,
      "loss": 1.1435,
      "step": 16640
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.787138239729822,
      "learning_rate": 9.456512811671454e-06,
      "loss": 1.1315,
      "step": 16645
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.824007676561451,
      "learning_rate": 9.455954472000824e-06,
      "loss": 1.1229,
      "step": 16650
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.70450859969771,
      "learning_rate": 9.4553958621788e-06,
      "loss": 1.1387,
      "step": 16655
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.183771705613105,
      "learning_rate": 9.454836982239248e-06,
      "loss": 1.0918,
      "step": 16660
    },
    {
      "epoch": 0.71,
      "grad_norm": 12.68487848217025,
      "learning_rate": 9.454277832216053e-06,
      "loss": 1.1459,
      "step": 16665
    },
    {
      "epoch": 0.71,
      "grad_norm": 12.588762678610511,
      "learning_rate": 9.453718412143111e-06,
      "loss": 1.1686,
      "step": 16670
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.95041425199994,
      "learning_rate": 9.453158722054342e-06,
      "loss": 1.1775,
      "step": 16675
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.389140281785481,
      "learning_rate": 9.452598761983676e-06,
      "loss": 1.0824,
      "step": 16680
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.457371902774636,
      "learning_rate": 9.452038531965064e-06,
      "loss": 1.1244,
      "step": 16685
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.138594175909104,
      "learning_rate": 9.45147803203247e-06,
      "loss": 1.1502,
      "step": 16690
    },
    {
      "epoch": 0.71,
      "grad_norm": 15.73782408581993,
      "learning_rate": 9.450917262219873e-06,
      "loss": 1.1096,
      "step": 16695
    },
    {
      "epoch": 0.71,
      "grad_norm": 48.15487289315489,
      "learning_rate": 9.450356222561275e-06,
      "loss": 1.1158,
      "step": 16700
    },
    {
      "epoch": 0.71,
      "grad_norm": 71.0660940825832,
      "learning_rate": 9.449794913090688e-06,
      "loss": 1.1074,
      "step": 16705
    },
    {
      "epoch": 0.71,
      "grad_norm": 37.908447686872755,
      "learning_rate": 9.449233333842142e-06,
      "loss": 1.1157,
      "step": 16710
    },
    {
      "epoch": 0.71,
      "grad_norm": 48.95623884410956,
      "learning_rate": 9.448671484849686e-06,
      "loss": 1.1215,
      "step": 16715
    },
    {
      "epoch": 0.71,
      "grad_norm": 26.511995987652156,
      "learning_rate": 9.44810936614738e-06,
      "loss": 1.1472,
      "step": 16720
    },
    {
      "epoch": 0.71,
      "grad_norm": 24.35007005317823,
      "learning_rate": 9.447546977769306e-06,
      "loss": 1.1275,
      "step": 16725
    },
    {
      "epoch": 0.71,
      "grad_norm": 19.500741273390563,
      "learning_rate": 9.44698431974956e-06,
      "loss": 1.0949,
      "step": 16730
    },
    {
      "epoch": 0.71,
      "grad_norm": 19.970507472491803,
      "learning_rate": 9.446421392122252e-06,
      "loss": 1.1443,
      "step": 16735
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.051456844397624,
      "learning_rate": 9.445858194921513e-06,
      "loss": 1.083,
      "step": 16740
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.514466356345627,
      "learning_rate": 9.445294728181488e-06,
      "loss": 1.173,
      "step": 16745
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.583482026424033,
      "learning_rate": 9.444730991936339e-06,
      "loss": 1.1785,
      "step": 16750
    },
    {
      "epoch": 0.71,
      "grad_norm": 16.056715779376827,
      "learning_rate": 9.444166986220241e-06,
      "loss": 1.105,
      "step": 16755
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.80940019974729,
      "learning_rate": 9.44360271106739e-06,
      "loss": 1.1694,
      "step": 16760
    },
    {
      "epoch": 0.71,
      "grad_norm": 22.361662859930433,
      "learning_rate": 9.443038166511995e-06,
      "loss": 1.1116,
      "step": 16765
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.027221559233777,
      "learning_rate": 9.442473352588284e-06,
      "loss": 1.103,
      "step": 16770
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.836040408668314,
      "learning_rate": 9.441908269330497e-06,
      "loss": 1.1322,
      "step": 16775
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.303328143306504,
      "learning_rate": 9.441342916772897e-06,
      "loss": 1.1345,
      "step": 16780
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.021272221926106,
      "learning_rate": 9.44077729494976e-06,
      "loss": 1.0654,
      "step": 16785
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.995398075099797,
      "learning_rate": 9.440211403895375e-06,
      "loss": 1.1068,
      "step": 16790
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.521030212433786,
      "learning_rate": 9.439645243644052e-06,
      "loss": 1.1246,
      "step": 16795
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.464672992246477,
      "learning_rate": 9.439078814230116e-06,
      "loss": 1.1399,
      "step": 16800
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.968872254572704,
      "learning_rate": 9.438512115687906e-06,
      "loss": 1.128,
      "step": 16805
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.658854728003837,
      "learning_rate": 9.43794514805178e-06,
      "loss": 1.0866,
      "step": 16810
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.91804251262899,
      "learning_rate": 9.437377911356113e-06,
      "loss": 1.154,
      "step": 16815
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.433954831086169,
      "learning_rate": 9.436810405635295e-06,
      "loss": 1.1803,
      "step": 16820
    },
    {
      "epoch": 0.71,
      "grad_norm": 17.669302545018628,
      "learning_rate": 9.436242630923729e-06,
      "loss": 1.1257,
      "step": 16825
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.6086021840869105,
      "learning_rate": 9.43567458725584e-06,
      "loss": 1.0987,
      "step": 16830
    },
    {
      "epoch": 0.71,
      "grad_norm": 28.14374126918958,
      "learning_rate": 9.435106274666068e-06,
      "loss": 1.1591,
      "step": 16835
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.222061371987685,
      "learning_rate": 9.434537693188866e-06,
      "loss": 1.1114,
      "step": 16840
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.979414582554136,
      "learning_rate": 9.433968842858707e-06,
      "loss": 1.1037,
      "step": 16845
    },
    {
      "epoch": 0.71,
      "grad_norm": 15.03706353610534,
      "learning_rate": 9.433399723710075e-06,
      "loss": 1.0911,
      "step": 16850
    },
    {
      "epoch": 0.71,
      "grad_norm": 14.901984005001061,
      "learning_rate": 9.432830335777478e-06,
      "loss": 1.1511,
      "step": 16855
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.1973574333662365,
      "learning_rate": 9.432260679095437e-06,
      "loss": 1.147,
      "step": 16860
    },
    {
      "epoch": 0.71,
      "grad_norm": 10.799997034857435,
      "learning_rate": 9.431690753698484e-06,
      "loss": 1.0888,
      "step": 16865
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.515932725361904,
      "learning_rate": 9.431120559621177e-06,
      "loss": 1.1235,
      "step": 16870
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.396228285215444,
      "learning_rate": 9.430550096898081e-06,
      "loss": 1.1252,
      "step": 16875
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.736459841290592,
      "learning_rate": 9.429979365563784e-06,
      "loss": 1.0846,
      "step": 16880
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.170423620966293,
      "learning_rate": 9.429408365652885e-06,
      "loss": 1.1327,
      "step": 16885
    },
    {
      "epoch": 0.71,
      "grad_norm": 15.019682647948079,
      "learning_rate": 9.428837097200007e-06,
      "loss": 1.1283,
      "step": 16890
    },
    {
      "epoch": 0.72,
      "grad_norm": 19.109949796023205,
      "learning_rate": 9.42826556023978e-06,
      "loss": 1.1327,
      "step": 16895
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.792807691435836,
      "learning_rate": 9.427693754806855e-06,
      "loss": 1.1405,
      "step": 16900
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.830178430312812,
      "learning_rate": 9.427121680935903e-06,
      "loss": 1.1026,
      "step": 16905
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.2772761218963415,
      "learning_rate": 9.4265493386616e-06,
      "loss": 1.102,
      "step": 16910
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.513334085513613,
      "learning_rate": 9.425976728018652e-06,
      "loss": 1.1461,
      "step": 16915
    },
    {
      "epoch": 0.72,
      "grad_norm": 15.461304275406098,
      "learning_rate": 9.425403849041772e-06,
      "loss": 1.1177,
      "step": 16920
    },
    {
      "epoch": 0.72,
      "grad_norm": 11.172519127997129,
      "learning_rate": 9.424830701765691e-06,
      "loss": 1.1218,
      "step": 16925
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.780778037333757,
      "learning_rate": 9.424257286225158e-06,
      "loss": 1.103,
      "step": 16930
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.074888927969338,
      "learning_rate": 9.42368360245494e-06,
      "loss": 1.1047,
      "step": 16935
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.343478713355483,
      "learning_rate": 9.423109650489813e-06,
      "loss": 1.112,
      "step": 16940
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.18160662537546,
      "learning_rate": 9.422535430364578e-06,
      "loss": 1.1071,
      "step": 16945
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.727614405773898,
      "learning_rate": 9.421960942114047e-06,
      "loss": 1.1492,
      "step": 16950
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.975550544799563,
      "learning_rate": 9.42138618577305e-06,
      "loss": 1.0887,
      "step": 16955
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.165521010744902,
      "learning_rate": 9.42081116137643e-06,
      "loss": 1.1371,
      "step": 16960
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.9756960727466275,
      "learning_rate": 9.420235868959053e-06,
      "loss": 1.102,
      "step": 16965
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.288050139755605,
      "learning_rate": 9.419660308555794e-06,
      "loss": 1.125,
      "step": 16970
    },
    {
      "epoch": 0.72,
      "grad_norm": 22.46363958680743,
      "learning_rate": 9.41908448020155e-06,
      "loss": 1.1041,
      "step": 16975
    },
    {
      "epoch": 0.72,
      "grad_norm": 25.92286580198727,
      "learning_rate": 9.41850838393123e-06,
      "loss": 1.0745,
      "step": 16980
    },
    {
      "epoch": 0.72,
      "grad_norm": 31.478701769149943,
      "learning_rate": 9.417932019779764e-06,
      "loss": 1.1298,
      "step": 16985
    },
    {
      "epoch": 0.72,
      "grad_norm": 49.98478516291867,
      "learning_rate": 9.417355387782093e-06,
      "loss": 1.2096,
      "step": 16990
    },
    {
      "epoch": 0.72,
      "grad_norm": 18.777311914589585,
      "learning_rate": 9.416778487973175e-06,
      "loss": 1.1411,
      "step": 16995
    },
    {
      "epoch": 0.72,
      "grad_norm": 29.088862736629796,
      "learning_rate": 9.416201320387987e-06,
      "loss": 1.1457,
      "step": 17000
    },
    {
      "epoch": 0.72,
      "grad_norm": 14.911032623941088,
      "learning_rate": 9.415623885061523e-06,
      "loss": 1.0973,
      "step": 17005
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.592416651664726,
      "learning_rate": 9.415046182028786e-06,
      "loss": 1.1031,
      "step": 17010
    },
    {
      "epoch": 0.72,
      "grad_norm": 24.22813894139801,
      "learning_rate": 9.414468211324807e-06,
      "loss": 1.1858,
      "step": 17015
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.970733707924163,
      "learning_rate": 9.413889972984624e-06,
      "loss": 1.1443,
      "step": 17020
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.495194174545278,
      "learning_rate": 9.413311467043291e-06,
      "loss": 1.1401,
      "step": 17025
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.941363346822824,
      "learning_rate": 9.412732693535886e-06,
      "loss": 1.0958,
      "step": 17030
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.263242700227079,
      "learning_rate": 9.412153652497495e-06,
      "loss": 1.1256,
      "step": 17035
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.335565848080336,
      "learning_rate": 9.411574343963224e-06,
      "loss": 1.0734,
      "step": 17040
    },
    {
      "epoch": 0.72,
      "grad_norm": 13.382657427490711,
      "learning_rate": 9.410994767968195e-06,
      "loss": 1.1308,
      "step": 17045
    },
    {
      "epoch": 0.72,
      "grad_norm": 24.033479591859525,
      "learning_rate": 9.410414924547547e-06,
      "loss": 1.1264,
      "step": 17050
    },
    {
      "epoch": 0.72,
      "grad_norm": 21.630015037907583,
      "learning_rate": 9.409834813736433e-06,
      "loss": 1.1463,
      "step": 17055
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.037300580388155,
      "learning_rate": 9.409254435570026e-06,
      "loss": 1.1246,
      "step": 17060
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.793201911770602,
      "learning_rate": 9.408673790083507e-06,
      "loss": 1.1093,
      "step": 17065
    },
    {
      "epoch": 0.72,
      "grad_norm": 10.380811723294608,
      "learning_rate": 9.408092877312083e-06,
      "loss": 1.1116,
      "step": 17070
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.33713237365468,
      "learning_rate": 9.407511697290974e-06,
      "loss": 1.1354,
      "step": 17075
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.693570574742182,
      "learning_rate": 9.406930250055412e-06,
      "loss": 1.0813,
      "step": 17080
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.41622611707326,
      "learning_rate": 9.406348535640652e-06,
      "loss": 1.0948,
      "step": 17085
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.456705795607186,
      "learning_rate": 9.405766554081958e-06,
      "loss": 1.0801,
      "step": 17090
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.4960135527114495,
      "learning_rate": 9.405184305414616e-06,
      "loss": 1.1138,
      "step": 17095
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.288026182079762,
      "learning_rate": 9.404601789673925e-06,
      "loss": 1.0832,
      "step": 17100
    },
    {
      "epoch": 0.72,
      "grad_norm": 19.3479543569613,
      "learning_rate": 9.404019006895203e-06,
      "loss": 1.107,
      "step": 17105
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.842926222082708,
      "learning_rate": 9.403435957113777e-06,
      "loss": 1.1164,
      "step": 17110
    },
    {
      "epoch": 0.72,
      "grad_norm": 13.281561361795422,
      "learning_rate": 9.402852640365004e-06,
      "loss": 1.0897,
      "step": 17115
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.077612566222394,
      "learning_rate": 9.402269056684243e-06,
      "loss": 1.1498,
      "step": 17120
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.166434646711995,
      "learning_rate": 9.401685206106876e-06,
      "loss": 1.1154,
      "step": 17125
    },
    {
      "epoch": 0.72,
      "grad_norm": 20.958865281871816,
      "learning_rate": 9.401101088668302e-06,
      "loss": 1.0815,
      "step": 17130
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.18135045122717,
      "learning_rate": 9.400516704403932e-06,
      "loss": 1.12,
      "step": 17135
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.568880727957094,
      "learning_rate": 9.399932053349197e-06,
      "loss": 1.1508,
      "step": 17140
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.885155559948981,
      "learning_rate": 9.399347135539541e-06,
      "loss": 1.1253,
      "step": 17145
    },
    {
      "epoch": 0.73,
      "grad_norm": 14.876977894790413,
      "learning_rate": 9.398761951010426e-06,
      "loss": 1.1455,
      "step": 17150
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.3827275308788245,
      "learning_rate": 9.398176499797332e-06,
      "loss": 1.1025,
      "step": 17155
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.951926088984365,
      "learning_rate": 9.397590781935752e-06,
      "loss": 1.1129,
      "step": 17160
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.206918830206302,
      "learning_rate": 9.397004797461197e-06,
      "loss": 1.0866,
      "step": 17165
    },
    {
      "epoch": 0.73,
      "grad_norm": 13.939961323895474,
      "learning_rate": 9.396418546409192e-06,
      "loss": 1.086,
      "step": 17170
    },
    {
      "epoch": 0.73,
      "grad_norm": 17.98587503767636,
      "learning_rate": 9.395832028815282e-06,
      "loss": 1.0882,
      "step": 17175
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.494169701184303,
      "learning_rate": 9.395245244715023e-06,
      "loss": 1.0921,
      "step": 17180
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.881569315129182,
      "learning_rate": 9.394658194143992e-06,
      "loss": 1.1037,
      "step": 17185
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.78607005751547,
      "learning_rate": 9.39407087713778e-06,
      "loss": 1.0849,
      "step": 17190
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.58776841781562,
      "learning_rate": 9.393483293731993e-06,
      "loss": 1.1106,
      "step": 17195
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.688776652978134,
      "learning_rate": 9.392895443962256e-06,
      "loss": 1.1113,
      "step": 17200
    },
    {
      "epoch": 0.73,
      "grad_norm": 16.680057695339766,
      "learning_rate": 9.392307327864208e-06,
      "loss": 1.0985,
      "step": 17205
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.334391613909822,
      "learning_rate": 9.391718945473503e-06,
      "loss": 1.1252,
      "step": 17210
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.2719185549215375,
      "learning_rate": 9.391130296825816e-06,
      "loss": 1.1098,
      "step": 17215
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.852469179256973,
      "learning_rate": 9.390541381956834e-06,
      "loss": 1.1391,
      "step": 17220
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.0455342614041525,
      "learning_rate": 9.38995220090226e-06,
      "loss": 1.1411,
      "step": 17225
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.0591929612376525,
      "learning_rate": 9.389362753697816e-06,
      "loss": 1.1087,
      "step": 17230
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.763499201327044,
      "learning_rate": 9.388773040379235e-06,
      "loss": 1.0955,
      "step": 17235
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.7384206899086205,
      "learning_rate": 9.388183060982275e-06,
      "loss": 1.0961,
      "step": 17240
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.225344644349165,
      "learning_rate": 9.387592815542701e-06,
      "loss": 1.1043,
      "step": 17245
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.081547305535541,
      "learning_rate": 9.387002304096299e-06,
      "loss": 1.0947,
      "step": 17250
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.5018757772530344,
      "learning_rate": 9.38641152667887e-06,
      "loss": 1.0955,
      "step": 17255
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.018645712561101,
      "learning_rate": 9.38582048332623e-06,
      "loss": 1.12,
      "step": 17260
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.032871693566749,
      "learning_rate": 9.38522917407421e-06,
      "loss": 1.1433,
      "step": 17265
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.085426516684288,
      "learning_rate": 9.384637598958665e-06,
      "loss": 1.1228,
      "step": 17270
    },
    {
      "epoch": 0.73,
      "grad_norm": 16.940379431112994,
      "learning_rate": 9.384045758015458e-06,
      "loss": 1.1136,
      "step": 17275
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.867936870634642,
      "learning_rate": 9.383453651280473e-06,
      "loss": 1.1184,
      "step": 17280
    },
    {
      "epoch": 0.73,
      "grad_norm": 15.159912773918336,
      "learning_rate": 9.382861278789601e-06,
      "loss": 1.0968,
      "step": 17285
    },
    {
      "epoch": 0.73,
      "grad_norm": 35.23415371528219,
      "learning_rate": 9.38226864057876e-06,
      "loss": 1.1321,
      "step": 17290
    },
    {
      "epoch": 0.73,
      "grad_norm": 38.455798061691674,
      "learning_rate": 9.381675736683882e-06,
      "loss": 1.1038,
      "step": 17295
    },
    {
      "epoch": 0.73,
      "grad_norm": 22.97976332771787,
      "learning_rate": 9.38108256714091e-06,
      "loss": 1.1248,
      "step": 17300
    },
    {
      "epoch": 0.73,
      "grad_norm": 16.54808353760375,
      "learning_rate": 9.380489131985806e-06,
      "loss": 1.0636,
      "step": 17305
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.500960986525851,
      "learning_rate": 9.37989543125455e-06,
      "loss": 1.1391,
      "step": 17310
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.254658336402659,
      "learning_rate": 9.379301464983132e-06,
      "loss": 1.0996,
      "step": 17315
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.8062352509720485,
      "learning_rate": 9.37870723320757e-06,
      "loss": 1.0848,
      "step": 17320
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.58053408673204,
      "learning_rate": 9.378112735963884e-06,
      "loss": 1.0942,
      "step": 17325
    },
    {
      "epoch": 0.73,
      "grad_norm": 11.5728432405758,
      "learning_rate": 9.377517973288122e-06,
      "loss": 1.1486,
      "step": 17330
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.758099755696685,
      "learning_rate": 9.376922945216336e-06,
      "loss": 1.0788,
      "step": 17335
    },
    {
      "epoch": 0.73,
      "grad_norm": 7.961547409784049,
      "learning_rate": 9.376327651784607e-06,
      "loss": 1.1187,
      "step": 17340
    },
    {
      "epoch": 0.73,
      "grad_norm": 10.382997081280067,
      "learning_rate": 9.375732093029022e-06,
      "loss": 1.1194,
      "step": 17345
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.939893616718557,
      "learning_rate": 9.37513626898569e-06,
      "loss": 1.0756,
      "step": 17350
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.144261339286091,
      "learning_rate": 9.374540179690732e-06,
      "loss": 1.1073,
      "step": 17355
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.522202673309658,
      "learning_rate": 9.373943825180291e-06,
      "loss": 1.1012,
      "step": 17360
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.270819616283949,
      "learning_rate": 9.373347205490519e-06,
      "loss": 1.1631,
      "step": 17365
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.940002302468172,
      "learning_rate": 9.372750320657586e-06,
      "loss": 1.0831,
      "step": 17370
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.116239465877265,
      "learning_rate": 9.372153170717684e-06,
      "loss": 1.093,
      "step": 17375
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.224456581099487,
      "learning_rate": 9.371555755707014e-06,
      "loss": 1.1437,
      "step": 17380
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.929234697607665,
      "learning_rate": 9.370958075661793e-06,
      "loss": 1.1664,
      "step": 17385
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.412018440039608,
      "learning_rate": 9.37036013061826e-06,
      "loss": 1.0812,
      "step": 17390
    },
    {
      "epoch": 0.74,
      "grad_norm": 27.757568484336996,
      "learning_rate": 9.369761920612666e-06,
      "loss": 1.1752,
      "step": 17395
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.993272958203807,
      "learning_rate": 9.369163445681278e-06,
      "loss": 1.1116,
      "step": 17400
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.819832161701608,
      "learning_rate": 9.368564705860382e-06,
      "loss": 1.0893,
      "step": 17405
    },
    {
      "epoch": 0.74,
      "grad_norm": 16.23664331522571,
      "learning_rate": 9.367965701186273e-06,
      "loss": 1.059,
      "step": 17410
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.786421369373763,
      "learning_rate": 9.367366431695271e-06,
      "loss": 1.1032,
      "step": 17415
    },
    {
      "epoch": 0.74,
      "grad_norm": 12.881198429582792,
      "learning_rate": 9.366766897423707e-06,
      "loss": 1.0745,
      "step": 17420
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.6695823337627775,
      "learning_rate": 9.366167098407929e-06,
      "loss": 1.1006,
      "step": 17425
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.397347096500923,
      "learning_rate": 9.365567034684299e-06,
      "loss": 1.0793,
      "step": 17430
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.504096159984868,
      "learning_rate": 9.3649667062892e-06,
      "loss": 1.1459,
      "step": 17435
    },
    {
      "epoch": 0.74,
      "grad_norm": 38.57755971082172,
      "learning_rate": 9.364366113259027e-06,
      "loss": 1.072,
      "step": 17440
    },
    {
      "epoch": 0.74,
      "grad_norm": 26.433593794628287,
      "learning_rate": 9.363765255630192e-06,
      "loss": 1.1172,
      "step": 17445
    },
    {
      "epoch": 0.74,
      "grad_norm": 27.051505783097266,
      "learning_rate": 9.363164133439125e-06,
      "loss": 1.1307,
      "step": 17450
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.177600258122624,
      "learning_rate": 9.362562746722266e-06,
      "loss": 1.0892,
      "step": 17455
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.911785509530484,
      "learning_rate": 9.36196109551608e-06,
      "loss": 1.1751,
      "step": 17460
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.142733250197846,
      "learning_rate": 9.36135917985704e-06,
      "loss": 1.1293,
      "step": 17465
    },
    {
      "epoch": 0.74,
      "grad_norm": 4.599561485892141,
      "learning_rate": 9.360756999781641e-06,
      "loss": 1.0728,
      "step": 17470
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.748536093258558,
      "learning_rate": 9.360154555326389e-06,
      "loss": 1.1313,
      "step": 17475
    },
    {
      "epoch": 0.74,
      "grad_norm": 16.18103476033407,
      "learning_rate": 9.359551846527811e-06,
      "loss": 1.0639,
      "step": 17480
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.837274662600126,
      "learning_rate": 9.358948873422446e-06,
      "loss": 1.0918,
      "step": 17485
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.6687705979683525,
      "learning_rate": 9.358345636046848e-06,
      "loss": 1.0925,
      "step": 17490
    },
    {
      "epoch": 0.74,
      "grad_norm": 21.562897690698215,
      "learning_rate": 9.357742134437595e-06,
      "loss": 1.1433,
      "step": 17495
    },
    {
      "epoch": 0.74,
      "grad_norm": 20.13977736018128,
      "learning_rate": 9.35713836863127e-06,
      "loss": 1.1734,
      "step": 17500
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.160711819186727,
      "learning_rate": 9.356534338664483e-06,
      "loss": 1.102,
      "step": 17505
    },
    {
      "epoch": 0.74,
      "grad_norm": 31.946127381726082,
      "learning_rate": 9.35593004457385e-06,
      "loss": 1.1448,
      "step": 17510
    },
    {
      "epoch": 0.74,
      "grad_norm": 24.62622025068199,
      "learning_rate": 9.35532548639601e-06,
      "loss": 1.0676,
      "step": 17515
    },
    {
      "epoch": 0.74,
      "grad_norm": 16.42863546105984,
      "learning_rate": 9.354720664167614e-06,
      "loss": 1.1279,
      "step": 17520
    },
    {
      "epoch": 0.74,
      "grad_norm": 26.444107761655193,
      "learning_rate": 9.354115577925332e-06,
      "loss": 1.1216,
      "step": 17525
    },
    {
      "epoch": 0.74,
      "grad_norm": 14.4372949897646,
      "learning_rate": 9.353510227705846e-06,
      "loss": 1.1043,
      "step": 17530
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.547363155591738,
      "learning_rate": 9.35290461354586e-06,
      "loss": 1.1228,
      "step": 17535
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.442737868382967,
      "learning_rate": 9.35229873548209e-06,
      "loss": 1.1339,
      "step": 17540
    },
    {
      "epoch": 0.74,
      "grad_norm": 15.48346025871906,
      "learning_rate": 9.351692593551265e-06,
      "loss": 1.1451,
      "step": 17545
    },
    {
      "epoch": 0.74,
      "grad_norm": 13.996469509651037,
      "learning_rate": 9.35108618779014e-06,
      "loss": 1.1276,
      "step": 17550
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.49324712392557,
      "learning_rate": 9.350479518235473e-06,
      "loss": 1.1509,
      "step": 17555
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.181695831946108,
      "learning_rate": 9.349872584924048e-06,
      "loss": 1.0905,
      "step": 17560
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.978894884206056,
      "learning_rate": 9.349265387892663e-06,
      "loss": 1.0913,
      "step": 17565
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.62675790500558,
      "learning_rate": 9.348657927178129e-06,
      "loss": 1.1052,
      "step": 17570
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.1643408333638945,
      "learning_rate": 9.348050202817273e-06,
      "loss": 1.108,
      "step": 17575
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.285029094815165,
      "learning_rate": 9.34744221484694e-06,
      "loss": 1.0962,
      "step": 17580
    },
    {
      "epoch": 0.74,
      "grad_norm": 13.100019988748052,
      "learning_rate": 9.346833963303993e-06,
      "loss": 1.1107,
      "step": 17585
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.649098406277306,
      "learning_rate": 9.346225448225306e-06,
      "loss": 1.1542,
      "step": 17590
    },
    {
      "epoch": 0.74,
      "grad_norm": 21.37435265346111,
      "learning_rate": 9.345616669647773e-06,
      "loss": 1.1085,
      "step": 17595
    },
    {
      "epoch": 0.74,
      "grad_norm": 10.18876984518517,
      "learning_rate": 9.3450076276083e-06,
      "loss": 1.0728,
      "step": 17600
    },
    {
      "epoch": 0.75,
      "grad_norm": 20.201192777929748,
      "learning_rate": 9.344398322143816e-06,
      "loss": 1.0933,
      "step": 17605
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.445424417338074,
      "learning_rate": 9.343788753291256e-06,
      "loss": 1.1315,
      "step": 17610
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.36200115962045,
      "learning_rate": 9.34317892108758e-06,
      "loss": 1.0948,
      "step": 17615
    },
    {
      "epoch": 0.75,
      "grad_norm": 18.17910100725324,
      "learning_rate": 9.342568825569758e-06,
      "loss": 1.0926,
      "step": 17620
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.473700386629618,
      "learning_rate": 9.34195846677478e-06,
      "loss": 1.0816,
      "step": 17625
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.48303123125485,
      "learning_rate": 9.34134784473965e-06,
      "loss": 1.1206,
      "step": 17630
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.097180306245573,
      "learning_rate": 9.34073695950139e-06,
      "loss": 1.1322,
      "step": 17635
    },
    {
      "epoch": 0.75,
      "grad_norm": 10.525758247727076,
      "learning_rate": 9.340125811097031e-06,
      "loss": 1.0828,
      "step": 17640
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.6009010325028505,
      "learning_rate": 9.339514399563631e-06,
      "loss": 1.0958,
      "step": 17645
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.640477077965488,
      "learning_rate": 9.338902724938254e-06,
      "loss": 1.1042,
      "step": 17650
    },
    {
      "epoch": 0.75,
      "grad_norm": 13.994414103562079,
      "learning_rate": 9.338290787257986e-06,
      "loss": 1.0807,
      "step": 17655
    },
    {
      "epoch": 0.75,
      "grad_norm": 17.560956580459454,
      "learning_rate": 9.337678586559926e-06,
      "loss": 1.1153,
      "step": 17660
    },
    {
      "epoch": 0.75,
      "grad_norm": 19.149023371553817,
      "learning_rate": 9.337066122881192e-06,
      "loss": 1.1285,
      "step": 17665
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.840414915856152,
      "learning_rate": 9.33645339625891e-06,
      "loss": 1.1475,
      "step": 17670
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.318343283167946,
      "learning_rate": 9.335840406730236e-06,
      "loss": 1.0679,
      "step": 17675
    },
    {
      "epoch": 0.75,
      "grad_norm": 15.886471567354363,
      "learning_rate": 9.33522715433233e-06,
      "loss": 1.0834,
      "step": 17680
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.1570218387264255,
      "learning_rate": 9.334613639102369e-06,
      "loss": 1.1588,
      "step": 17685
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.496101410888,
      "learning_rate": 9.333999861077551e-06,
      "loss": 1.1213,
      "step": 17690
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.4879162131471455,
      "learning_rate": 9.33338582029509e-06,
      "loss": 1.0902,
      "step": 17695
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.163005891091226,
      "learning_rate": 9.33277151679221e-06,
      "loss": 1.1084,
      "step": 17700
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.280386367844475,
      "learning_rate": 9.332156950606156e-06,
      "loss": 1.1438,
      "step": 17705
    },
    {
      "epoch": 0.75,
      "grad_norm": 21.097335957884365,
      "learning_rate": 9.331542121774185e-06,
      "loss": 1.1196,
      "step": 17710
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.706026362108982,
      "learning_rate": 9.330927030333576e-06,
      "loss": 1.0956,
      "step": 17715
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.424335749711798,
      "learning_rate": 9.330311676321617e-06,
      "loss": 1.0882,
      "step": 17720
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.46983326743464,
      "learning_rate": 9.329696059775617e-06,
      "loss": 1.1302,
      "step": 17725
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.2976372954129625,
      "learning_rate": 9.3290801807329e-06,
      "loss": 1.142,
      "step": 17730
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.858724317071061,
      "learning_rate": 9.3284640392308e-06,
      "loss": 1.1238,
      "step": 17735
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.370489362463976,
      "learning_rate": 9.32784763530668e-06,
      "loss": 1.087,
      "step": 17740
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.886489401922693,
      "learning_rate": 9.327230968997902e-06,
      "loss": 1.0841,
      "step": 17745
    },
    {
      "epoch": 0.75,
      "grad_norm": 11.61986652593081,
      "learning_rate": 9.326614040341856e-06,
      "loss": 1.1004,
      "step": 17750
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.961581315637526,
      "learning_rate": 9.32599684937595e-06,
      "loss": 1.0827,
      "step": 17755
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.36029046294683,
      "learning_rate": 9.325379396137592e-06,
      "loss": 1.1208,
      "step": 17760
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.653540400390449,
      "learning_rate": 9.324761680664225e-06,
      "loss": 1.1047,
      "step": 17765
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.167477681863815,
      "learning_rate": 9.324143702993297e-06,
      "loss": 1.1157,
      "step": 17770
    },
    {
      "epoch": 0.75,
      "grad_norm": 13.181209543313301,
      "learning_rate": 9.323525463162272e-06,
      "loss": 1.1029,
      "step": 17775
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.453272784212921,
      "learning_rate": 9.322906961208635e-06,
      "loss": 1.0982,
      "step": 17780
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.783406016091666,
      "learning_rate": 9.32228819716988e-06,
      "loss": 1.0925,
      "step": 17785
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.707581107944481,
      "learning_rate": 9.321669171083524e-06,
      "loss": 1.1077,
      "step": 17790
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.24513728558436,
      "learning_rate": 9.321049882987099e-06,
      "loss": 1.0983,
      "step": 17795
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.114739383565758,
      "learning_rate": 9.320430332918145e-06,
      "loss": 1.0697,
      "step": 17800
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.696646599122937,
      "learning_rate": 9.319810520914227e-06,
      "loss": 1.1071,
      "step": 17805
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.325697501887738,
      "learning_rate": 9.31919044701292e-06,
      "loss": 1.0609,
      "step": 17810
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.564186081772088,
      "learning_rate": 9.318570111251822e-06,
      "loss": 1.1163,
      "step": 17815
    },
    {
      "epoch": 0.75,
      "grad_norm": 9.896498195909624,
      "learning_rate": 9.317949513668535e-06,
      "loss": 1.1309,
      "step": 17820
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.935313062509824,
      "learning_rate": 9.31732865430069e-06,
      "loss": 1.0803,
      "step": 17825
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.371963286186434,
      "learning_rate": 9.316707533185926e-06,
      "loss": 1.1124,
      "step": 17830
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.1389184696570345,
      "learning_rate": 9.316086150361898e-06,
      "loss": 1.0989,
      "step": 17835
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.254101640906628,
      "learning_rate": 9.315464505866283e-06,
      "loss": 1.1019,
      "step": 17840
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.617964891678169,
      "learning_rate": 9.314842599736765e-06,
      "loss": 1.0296,
      "step": 17845
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.994548316031661,
      "learning_rate": 9.31422043201105e-06,
      "loss": 1.0999,
      "step": 17850
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.884209118718427,
      "learning_rate": 9.313598002726858e-06,
      "loss": 1.1303,
      "step": 17855
    },
    {
      "epoch": 0.76,
      "grad_norm": 43.291399748229495,
      "learning_rate": 9.312975311921925e-06,
      "loss": 1.1126,
      "step": 17860
    },
    {
      "epoch": 0.76,
      "grad_norm": 20.692757428848292,
      "learning_rate": 9.312352359634002e-06,
      "loss": 1.1431,
      "step": 17865
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.025039825363798,
      "learning_rate": 9.311729145900861e-06,
      "loss": 1.1445,
      "step": 17870
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.413295935876926,
      "learning_rate": 9.31110567076028e-06,
      "loss": 1.113,
      "step": 17875
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.707804017583332,
      "learning_rate": 9.310481934250061e-06,
      "loss": 1.0626,
      "step": 17880
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.335946387506712,
      "learning_rate": 9.309857936408021e-06,
      "loss": 1.08,
      "step": 17885
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.608756107676717,
      "learning_rate": 9.309233677271989e-06,
      "loss": 1.0948,
      "step": 17890
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.91177670711054,
      "learning_rate": 9.308609156879811e-06,
      "loss": 1.0963,
      "step": 17895
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.115951076738714,
      "learning_rate": 9.307984375269352e-06,
      "loss": 1.0847,
      "step": 17900
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.004888216317523,
      "learning_rate": 9.30735933247849e-06,
      "loss": 1.1112,
      "step": 17905
    },
    {
      "epoch": 0.76,
      "grad_norm": 22.769500361730856,
      "learning_rate": 9.306734028545119e-06,
      "loss": 1.1411,
      "step": 17910
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.246741595201495,
      "learning_rate": 9.306108463507151e-06,
      "loss": 1.165,
      "step": 17915
    },
    {
      "epoch": 0.76,
      "grad_norm": 60.33464362540984,
      "learning_rate": 9.30548263740251e-06,
      "loss": 1.0808,
      "step": 17920
    },
    {
      "epoch": 0.76,
      "grad_norm": 43.69616070366895,
      "learning_rate": 9.30485655026914e-06,
      "loss": 1.1266,
      "step": 17925
    },
    {
      "epoch": 0.76,
      "grad_norm": 10.321737392385257,
      "learning_rate": 9.304230202144994e-06,
      "loss": 1.0959,
      "step": 17930
    },
    {
      "epoch": 0.76,
      "grad_norm": 43.41819720045734,
      "learning_rate": 9.303603593068053e-06,
      "loss": 1.1409,
      "step": 17935
    },
    {
      "epoch": 0.76,
      "grad_norm": 24.649802928595093,
      "learning_rate": 9.302976723076302e-06,
      "loss": 1.1148,
      "step": 17940
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.029746468154173,
      "learning_rate": 9.302349592207749e-06,
      "loss": 1.1472,
      "step": 17945
    },
    {
      "epoch": 0.76,
      "grad_norm": 14.193913968429644,
      "learning_rate": 9.30172220050041e-06,
      "loss": 1.0816,
      "step": 17950
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.508118285996746,
      "learning_rate": 9.301094547992326e-06,
      "loss": 1.1818,
      "step": 17955
    },
    {
      "epoch": 0.76,
      "grad_norm": 20.527532773215444,
      "learning_rate": 9.30046663472155e-06,
      "loss": 1.0734,
      "step": 17960
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.413200169623018,
      "learning_rate": 9.299838460726147e-06,
      "loss": 1.0533,
      "step": 17965
    },
    {
      "epoch": 0.76,
      "grad_norm": 14.572408066102856,
      "learning_rate": 9.299210026044206e-06,
      "loss": 1.1383,
      "step": 17970
    },
    {
      "epoch": 0.76,
      "grad_norm": 27.10646904666803,
      "learning_rate": 9.298581330713825e-06,
      "loss": 1.148,
      "step": 17975
    },
    {
      "epoch": 0.76,
      "grad_norm": 15.448729598628704,
      "learning_rate": 9.297952374773117e-06,
      "loss": 1.1288,
      "step": 17980
    },
    {
      "epoch": 0.76,
      "grad_norm": 19.19352992645366,
      "learning_rate": 9.29732315826022e-06,
      "loss": 1.1217,
      "step": 17985
    },
    {
      "epoch": 0.76,
      "grad_norm": 13.857559087635702,
      "learning_rate": 9.296693681213278e-06,
      "loss": 1.1038,
      "step": 17990
    },
    {
      "epoch": 0.76,
      "grad_norm": 21.89320667055709,
      "learning_rate": 9.296063943670451e-06,
      "loss": 1.1167,
      "step": 17995
    },
    {
      "epoch": 0.76,
      "grad_norm": 12.428263262811615,
      "learning_rate": 9.295433945669925e-06,
      "loss": 1.1003,
      "step": 18000
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.786520722827463,
      "learning_rate": 9.29480368724989e-06,
      "loss": 1.1053,
      "step": 18005
    },
    {
      "epoch": 0.76,
      "grad_norm": 14.621900385283961,
      "learning_rate": 9.294173168448557e-06,
      "loss": 1.0974,
      "step": 18010
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.872821012468063,
      "learning_rate": 9.293542389304156e-06,
      "loss": 1.138,
      "step": 18015
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.62667057385184,
      "learning_rate": 9.292911349854925e-06,
      "loss": 1.0886,
      "step": 18020
    },
    {
      "epoch": 0.76,
      "grad_norm": 16.427205921395075,
      "learning_rate": 9.292280050139127e-06,
      "loss": 1.0571,
      "step": 18025
    },
    {
      "epoch": 0.76,
      "grad_norm": 26.118521970745363,
      "learning_rate": 9.29164849019503e-06,
      "loss": 1.0896,
      "step": 18030
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.594387110319337,
      "learning_rate": 9.291016670060927e-06,
      "loss": 1.064,
      "step": 18035
    },
    {
      "epoch": 0.76,
      "grad_norm": 18.444047145454256,
      "learning_rate": 9.290384589775124e-06,
      "loss": 1.0505,
      "step": 18040
    },
    {
      "epoch": 0.76,
      "grad_norm": 15.13698466135544,
      "learning_rate": 9.28975224937594e-06,
      "loss": 1.1057,
      "step": 18045
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.143583582507327,
      "learning_rate": 9.289119648901713e-06,
      "loss": 1.0773,
      "step": 18050
    },
    {
      "epoch": 0.76,
      "grad_norm": 15.324132017195472,
      "learning_rate": 9.288486788390796e-06,
      "loss": 1.1165,
      "step": 18055
    },
    {
      "epoch": 0.76,
      "grad_norm": 19.686030221150965,
      "learning_rate": 9.287853667881558e-06,
      "loss": 1.1165,
      "step": 18060
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.856307777689082,
      "learning_rate": 9.287220287412382e-06,
      "loss": 1.1115,
      "step": 18065
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.529049466086411,
      "learning_rate": 9.286586647021667e-06,
      "loss": 1.1069,
      "step": 18070
    },
    {
      "epoch": 0.76,
      "grad_norm": 21.14855222012083,
      "learning_rate": 9.285952746747832e-06,
      "loss": 1.1055,
      "step": 18075
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.488448627417618,
      "learning_rate": 9.285318586629307e-06,
      "loss": 1.1114,
      "step": 18080
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.68412109571187,
      "learning_rate": 9.284684166704536e-06,
      "loss": 1.115,
      "step": 18085
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.678861619744061,
      "learning_rate": 9.284049487011986e-06,
      "loss": 1.1153,
      "step": 18090
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.751745063264902,
      "learning_rate": 9.283414547590136e-06,
      "loss": 1.1135,
      "step": 18095
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.728169038157912,
      "learning_rate": 9.28277934847748e-06,
      "loss": 1.0642,
      "step": 18100
    },
    {
      "epoch": 0.77,
      "grad_norm": 13.563350595731961,
      "learning_rate": 9.282143889712525e-06,
      "loss": 1.1345,
      "step": 18105
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.644341445123432,
      "learning_rate": 9.281508171333799e-06,
      "loss": 1.0736,
      "step": 18110
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.201505795645917,
      "learning_rate": 9.280872193379846e-06,
      "loss": 1.1169,
      "step": 18115
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.743236804679773,
      "learning_rate": 9.280235955889223e-06,
      "loss": 1.125,
      "step": 18120
    },
    {
      "epoch": 0.77,
      "grad_norm": 17.444599243677327,
      "learning_rate": 9.2795994589005e-06,
      "loss": 1.138,
      "step": 18125
    },
    {
      "epoch": 0.77,
      "grad_norm": 23.029678620204297,
      "learning_rate": 9.278962702452268e-06,
      "loss": 1.1331,
      "step": 18130
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.891904219305596,
      "learning_rate": 9.27832568658313e-06,
      "loss": 1.1102,
      "step": 18135
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.6712913782314,
      "learning_rate": 9.27768841133171e-06,
      "loss": 1.1113,
      "step": 18140
    },
    {
      "epoch": 0.77,
      "grad_norm": 19.658027741518705,
      "learning_rate": 9.27705087673664e-06,
      "loss": 1.118,
      "step": 18145
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.549807127756542,
      "learning_rate": 9.276413082836577e-06,
      "loss": 1.118,
      "step": 18150
    },
    {
      "epoch": 0.77,
      "grad_norm": 20.480539715359487,
      "learning_rate": 9.275775029670182e-06,
      "loss": 1.1275,
      "step": 18155
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.764272951596997,
      "learning_rate": 9.275136717276143e-06,
      "loss": 1.1518,
      "step": 18160
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.756798278971463,
      "learning_rate": 9.274498145693159e-06,
      "loss": 1.0684,
      "step": 18165
    },
    {
      "epoch": 0.77,
      "grad_norm": 32.97820281601068,
      "learning_rate": 9.27385931495994e-06,
      "loss": 1.1132,
      "step": 18170
    },
    {
      "epoch": 0.77,
      "grad_norm": 28.049341545594245,
      "learning_rate": 9.273220225115222e-06,
      "loss": 1.1441,
      "step": 18175
    },
    {
      "epoch": 0.77,
      "grad_norm": 53.46882825394066,
      "learning_rate": 9.27258087619775e-06,
      "loss": 1.1038,
      "step": 18180
    },
    {
      "epoch": 0.77,
      "grad_norm": 51.659466178457414,
      "learning_rate": 9.271941268246284e-06,
      "loss": 1.1818,
      "step": 18185
    },
    {
      "epoch": 0.77,
      "grad_norm": 32.20468389648628,
      "learning_rate": 9.2713014012996e-06,
      "loss": 1.1615,
      "step": 18190
    },
    {
      "epoch": 0.77,
      "grad_norm": 25.027750831389525,
      "learning_rate": 9.270661275396496e-06,
      "loss": 1.1027,
      "step": 18195
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.19228217314939,
      "learning_rate": 9.270020890575778e-06,
      "loss": 1.182,
      "step": 18200
    },
    {
      "epoch": 0.77,
      "grad_norm": 11.256715054942326,
      "learning_rate": 9.269380246876272e-06,
      "loss": 1.0932,
      "step": 18205
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.695021259289096,
      "learning_rate": 9.268739344336816e-06,
      "loss": 1.0704,
      "step": 18210
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.134439070614352,
      "learning_rate": 9.268098182996267e-06,
      "loss": 1.0791,
      "step": 18215
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.478176467403107,
      "learning_rate": 9.267456762893499e-06,
      "loss": 1.1099,
      "step": 18220
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.418045680315594,
      "learning_rate": 9.266815084067398e-06,
      "loss": 1.1563,
      "step": 18225
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.349754864637389,
      "learning_rate": 9.266173146556865e-06,
      "loss": 1.1304,
      "step": 18230
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.6400881312963085,
      "learning_rate": 9.265530950400821e-06,
      "loss": 1.1084,
      "step": 18235
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.626035085964305,
      "learning_rate": 9.2648884956382e-06,
      "loss": 1.1198,
      "step": 18240
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.461797984128544,
      "learning_rate": 9.264245782307953e-06,
      "loss": 1.1128,
      "step": 18245
    },
    {
      "epoch": 0.77,
      "grad_norm": 12.192555207213774,
      "learning_rate": 9.263602810449046e-06,
      "loss": 1.1169,
      "step": 18250
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.466099159104695,
      "learning_rate": 9.262959580100457e-06,
      "loss": 1.1208,
      "step": 18255
    },
    {
      "epoch": 0.77,
      "grad_norm": 16.593427194263676,
      "learning_rate": 9.262316091301187e-06,
      "loss": 1.131,
      "step": 18260
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.11305565293234,
      "learning_rate": 9.261672344090247e-06,
      "loss": 1.1044,
      "step": 18265
    },
    {
      "epoch": 0.77,
      "grad_norm": 19.64817614237251,
      "learning_rate": 9.261028338506666e-06,
      "loss": 1.094,
      "step": 18270
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.723357452339682,
      "learning_rate": 9.260384074589487e-06,
      "loss": 1.153,
      "step": 18275
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.534030160676935,
      "learning_rate": 9.259739552377773e-06,
      "loss": 1.0862,
      "step": 18280
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.441695749826908,
      "learning_rate": 9.259094771910597e-06,
      "loss": 1.095,
      "step": 18285
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.146424724688408,
      "learning_rate": 9.258449733227049e-06,
      "loss": 1.0893,
      "step": 18290
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.34937160389928,
      "learning_rate": 9.257804436366239e-06,
      "loss": 1.0733,
      "step": 18295
    },
    {
      "epoch": 0.77,
      "grad_norm": 14.743914833467466,
      "learning_rate": 9.257158881367287e-06,
      "loss": 1.0894,
      "step": 18300
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.70492924811333,
      "learning_rate": 9.256513068269333e-06,
      "loss": 1.0656,
      "step": 18305
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.972679783154874,
      "learning_rate": 9.255866997111527e-06,
      "loss": 1.0959,
      "step": 18310
    },
    {
      "epoch": 0.78,
      "grad_norm": 20.10908837704645,
      "learning_rate": 9.255220667933044e-06,
      "loss": 1.1272,
      "step": 18315
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.791800090881644,
      "learning_rate": 9.254574080773065e-06,
      "loss": 1.1172,
      "step": 18320
    },
    {
      "epoch": 0.78,
      "grad_norm": 20.716794303522093,
      "learning_rate": 9.25392723567079e-06,
      "loss": 1.1157,
      "step": 18325
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.876879265665632,
      "learning_rate": 9.253280132665438e-06,
      "loss": 1.129,
      "step": 18330
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.140608566368778,
      "learning_rate": 9.252632771796241e-06,
      "loss": 1.0718,
      "step": 18335
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.32721571880933,
      "learning_rate": 9.251985153102445e-06,
      "loss": 1.0714,
      "step": 18340
    },
    {
      "epoch": 0.78,
      "grad_norm": 21.412824484047892,
      "learning_rate": 9.251337276623315e-06,
      "loss": 1.1693,
      "step": 18345
    },
    {
      "epoch": 0.78,
      "grad_norm": 13.576899680911131,
      "learning_rate": 9.250689142398128e-06,
      "loss": 1.1573,
      "step": 18350
    },
    {
      "epoch": 0.78,
      "grad_norm": 20.368065654779944,
      "learning_rate": 9.250040750466177e-06,
      "loss": 1.0656,
      "step": 18355
    },
    {
      "epoch": 0.78,
      "grad_norm": 20.024988350800516,
      "learning_rate": 9.249392100866777e-06,
      "loss": 1.1204,
      "step": 18360
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.315975053603287,
      "learning_rate": 9.24874319363925e-06,
      "loss": 1.1216,
      "step": 18365
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.061537153000435,
      "learning_rate": 9.248094028822938e-06,
      "loss": 1.0548,
      "step": 18370
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.866519384445859,
      "learning_rate": 9.247444606457199e-06,
      "loss": 1.0836,
      "step": 18375
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.840992088596085,
      "learning_rate": 9.246794926581404e-06,
      "loss": 1.1002,
      "step": 18380
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.125959813886627,
      "learning_rate": 9.246144989234942e-06,
      "loss": 1.1379,
      "step": 18385
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.093677475359922,
      "learning_rate": 9.245494794457216e-06,
      "loss": 1.1096,
      "step": 18390
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.956324046401775,
      "learning_rate": 9.244844342287647e-06,
      "loss": 1.1054,
      "step": 18395
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.923510744953704,
      "learning_rate": 9.24419363276567e-06,
      "loss": 1.0951,
      "step": 18400
    },
    {
      "epoch": 0.78,
      "grad_norm": 14.04440030568993,
      "learning_rate": 9.243542665930734e-06,
      "loss": 1.1066,
      "step": 18405
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.648409708461334,
      "learning_rate": 9.242891441822304e-06,
      "loss": 1.0764,
      "step": 18410
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.341981280631778,
      "learning_rate": 9.242239960479866e-06,
      "loss": 1.0876,
      "step": 18415
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.1259447557074305,
      "learning_rate": 9.241588221942914e-06,
      "loss": 1.069,
      "step": 18420
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.085677030674877,
      "learning_rate": 9.240936226250962e-06,
      "loss": 1.0434,
      "step": 18425
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.735618808712853,
      "learning_rate": 9.24028397344354e-06,
      "loss": 1.13,
      "step": 18430
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.742183155115284,
      "learning_rate": 9.23963146356019e-06,
      "loss": 1.1425,
      "step": 18435
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.034925119789757,
      "learning_rate": 9.238978696640473e-06,
      "loss": 1.1215,
      "step": 18440
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.957544218782891,
      "learning_rate": 9.238325672723964e-06,
      "loss": 1.1179,
      "step": 18445
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.0995740054848255,
      "learning_rate": 9.237672391850255e-06,
      "loss": 1.1248,
      "step": 18450
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.46097926388208,
      "learning_rate": 9.237018854058949e-06,
      "loss": 1.0863,
      "step": 18455
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.3683893659291035,
      "learning_rate": 9.236365059389672e-06,
      "loss": 1.1092,
      "step": 18460
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.614740256304254,
      "learning_rate": 9.235711007882057e-06,
      "loss": 1.0744,
      "step": 18465
    },
    {
      "epoch": 0.78,
      "grad_norm": 18.380904701796076,
      "learning_rate": 9.235056699575763e-06,
      "loss": 1.1049,
      "step": 18470
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.530310889229228,
      "learning_rate": 9.234402134510456e-06,
      "loss": 1.1031,
      "step": 18475
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.240197606016899,
      "learning_rate": 9.23374731272582e-06,
      "loss": 1.1016,
      "step": 18480
    },
    {
      "epoch": 0.78,
      "grad_norm": 16.227232089645792,
      "learning_rate": 9.233092234261557e-06,
      "loss": 1.1076,
      "step": 18485
    },
    {
      "epoch": 0.78,
      "grad_norm": 4.960853602955756,
      "learning_rate": 9.23243689915738e-06,
      "loss": 1.1437,
      "step": 18490
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.844782647036462,
      "learning_rate": 9.231781307453019e-06,
      "loss": 1.0723,
      "step": 18495
    },
    {
      "epoch": 0.78,
      "grad_norm": 22.168588671403622,
      "learning_rate": 9.231125459188223e-06,
      "loss": 1.104,
      "step": 18500
    },
    {
      "epoch": 0.78,
      "grad_norm": 12.297962124650041,
      "learning_rate": 9.230469354402755e-06,
      "loss": 1.1606,
      "step": 18505
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.858954595797314,
      "learning_rate": 9.229812993136392e-06,
      "loss": 1.0898,
      "step": 18510
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.039162007908019,
      "learning_rate": 9.229156375428925e-06,
      "loss": 1.0209,
      "step": 18515
    },
    {
      "epoch": 0.78,
      "grad_norm": 14.488953158080891,
      "learning_rate": 9.228499501320164e-06,
      "loss": 1.0841,
      "step": 18520
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.901604927770394,
      "learning_rate": 9.227842370849933e-06,
      "loss": 1.0845,
      "step": 18525
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.274488941514678,
      "learning_rate": 9.227184984058075e-06,
      "loss": 1.048,
      "step": 18530
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.215552861968598,
      "learning_rate": 9.226527340984442e-06,
      "loss": 1.0646,
      "step": 18535
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.669264228477738,
      "learning_rate": 9.225869441668906e-06,
      "loss": 1.1157,
      "step": 18540
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.411649102171838,
      "learning_rate": 9.225211286151353e-06,
      "loss": 1.1102,
      "step": 18545
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.654181227317375,
      "learning_rate": 9.224552874471685e-06,
      "loss": 1.098,
      "step": 18550
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.102364558992827,
      "learning_rate": 9.22389420666982e-06,
      "loss": 1.0893,
      "step": 18555
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.382254349643196,
      "learning_rate": 9.223235282785692e-06,
      "loss": 1.104,
      "step": 18560
    },
    {
      "epoch": 0.79,
      "grad_norm": 11.827504129215075,
      "learning_rate": 9.22257610285925e-06,
      "loss": 1.0865,
      "step": 18565
    },
    {
      "epoch": 0.79,
      "grad_norm": 11.825696764593685,
      "learning_rate": 9.221916666930455e-06,
      "loss": 1.1063,
      "step": 18570
    },
    {
      "epoch": 0.79,
      "grad_norm": 26.225003421671392,
      "learning_rate": 9.22125697503929e-06,
      "loss": 1.1692,
      "step": 18575
    },
    {
      "epoch": 0.79,
      "grad_norm": 23.087112157295447,
      "learning_rate": 9.220597027225746e-06,
      "loss": 1.1267,
      "step": 18580
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.137806415286787,
      "learning_rate": 9.219936823529838e-06,
      "loss": 1.1274,
      "step": 18585
    },
    {
      "epoch": 0.79,
      "grad_norm": 30.88384659808171,
      "learning_rate": 9.219276363991592e-06,
      "loss": 1.1425,
      "step": 18590
    },
    {
      "epoch": 0.79,
      "grad_norm": 18.257608901237255,
      "learning_rate": 9.218615648651046e-06,
      "loss": 1.1239,
      "step": 18595
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.841531358854471,
      "learning_rate": 9.217954677548261e-06,
      "loss": 1.1022,
      "step": 18600
    },
    {
      "epoch": 0.79,
      "grad_norm": 30.428804838114356,
      "learning_rate": 9.217293450723307e-06,
      "loss": 1.1175,
      "step": 18605
    },
    {
      "epoch": 0.79,
      "grad_norm": 22.431269080361552,
      "learning_rate": 9.216631968216275e-06,
      "loss": 1.0723,
      "step": 18610
    },
    {
      "epoch": 0.79,
      "grad_norm": 9.17281384190824,
      "learning_rate": 9.215970230067268e-06,
      "loss": 1.0997,
      "step": 18615
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.9231234344069685,
      "learning_rate": 9.215308236316402e-06,
      "loss": 1.0959,
      "step": 18620
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.33253390344492,
      "learning_rate": 9.214645987003817e-06,
      "loss": 1.1546,
      "step": 18625
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.182887200506652,
      "learning_rate": 9.213983482169658e-06,
      "loss": 1.1178,
      "step": 18630
    },
    {
      "epoch": 0.79,
      "grad_norm": 18.184168576979193,
      "learning_rate": 9.213320721854095e-06,
      "loss": 1.1034,
      "step": 18635
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.343857355602296,
      "learning_rate": 9.212657706097305e-06,
      "loss": 1.095,
      "step": 18640
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.258431794972262,
      "learning_rate": 9.21199443493949e-06,
      "loss": 1.0829,
      "step": 18645
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.923175420362524,
      "learning_rate": 9.211330908420857e-06,
      "loss": 1.0947,
      "step": 18650
    },
    {
      "epoch": 0.79,
      "grad_norm": 34.55885679530513,
      "learning_rate": 9.210667126581636e-06,
      "loss": 1.1119,
      "step": 18655
    },
    {
      "epoch": 0.79,
      "grad_norm": 26.29222103457184,
      "learning_rate": 9.210003089462071e-06,
      "loss": 1.1027,
      "step": 18660
    },
    {
      "epoch": 0.79,
      "grad_norm": 39.09497882226064,
      "learning_rate": 9.20933879710242e-06,
      "loss": 1.1229,
      "step": 18665
    },
    {
      "epoch": 0.79,
      "grad_norm": 22.170186193514663,
      "learning_rate": 9.208674249542956e-06,
      "loss": 1.0841,
      "step": 18670
    },
    {
      "epoch": 0.79,
      "grad_norm": 11.492904018642998,
      "learning_rate": 9.20800944682397e-06,
      "loss": 1.0961,
      "step": 18675
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.959205187086468,
      "learning_rate": 9.207344388985766e-06,
      "loss": 1.096,
      "step": 18680
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.470877767454973,
      "learning_rate": 9.206679076068665e-06,
      "loss": 1.1152,
      "step": 18685
    },
    {
      "epoch": 0.79,
      "grad_norm": 19.487070228600583,
      "learning_rate": 9.206013508113e-06,
      "loss": 1.1321,
      "step": 18690
    },
    {
      "epoch": 0.79,
      "grad_norm": 19.664031635949794,
      "learning_rate": 9.20534768515913e-06,
      "loss": 1.1228,
      "step": 18695
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.076270648885565,
      "learning_rate": 9.204681607247415e-06,
      "loss": 1.0941,
      "step": 18700
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.218854558319101,
      "learning_rate": 9.204015274418239e-06,
      "loss": 1.1111,
      "step": 18705
    },
    {
      "epoch": 0.79,
      "grad_norm": 13.939959121000188,
      "learning_rate": 9.203348686712e-06,
      "loss": 1.0835,
      "step": 18710
    },
    {
      "epoch": 0.79,
      "grad_norm": 28.39451363343461,
      "learning_rate": 9.202681844169112e-06,
      "loss": 1.0916,
      "step": 18715
    },
    {
      "epoch": 0.79,
      "grad_norm": 14.97537603596154,
      "learning_rate": 9.202014746830004e-06,
      "loss": 1.065,
      "step": 18720
    },
    {
      "epoch": 0.79,
      "grad_norm": 36.05035382625349,
      "learning_rate": 9.201347394735119e-06,
      "loss": 1.1081,
      "step": 18725
    },
    {
      "epoch": 0.79,
      "grad_norm": 37.27792781835876,
      "learning_rate": 9.200679787924915e-06,
      "loss": 1.0945,
      "step": 18730
    },
    {
      "epoch": 0.79,
      "grad_norm": 26.38255343026431,
      "learning_rate": 9.200011926439872e-06,
      "loss": 1.1013,
      "step": 18735
    },
    {
      "epoch": 0.79,
      "grad_norm": 31.212573738405045,
      "learning_rate": 9.199343810320474e-06,
      "loss": 1.0924,
      "step": 18740
    },
    {
      "epoch": 0.79,
      "grad_norm": 23.260054360062796,
      "learning_rate": 9.198675439607232e-06,
      "loss": 1.1118,
      "step": 18745
    },
    {
      "epoch": 0.79,
      "grad_norm": 16.96758238167588,
      "learning_rate": 9.198006814340666e-06,
      "loss": 1.0758,
      "step": 18750
    },
    {
      "epoch": 0.79,
      "grad_norm": 18.199432752815447,
      "learning_rate": 9.197337934561312e-06,
      "loss": 1.1021,
      "step": 18755
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.77019913257233,
      "learning_rate": 9.196668800309724e-06,
      "loss": 1.1559,
      "step": 18760
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.660052237914633,
      "learning_rate": 9.195999411626467e-06,
      "loss": 1.0785,
      "step": 18765
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.406555244894811,
      "learning_rate": 9.195329768552125e-06,
      "loss": 1.0345,
      "step": 18770
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.56550525952078,
      "learning_rate": 9.194659871127299e-06,
      "loss": 1.1117,
      "step": 18775
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.169526120138727,
      "learning_rate": 9.193989719392599e-06,
      "loss": 1.0913,
      "step": 18780
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.14875418458956,
      "learning_rate": 9.193319313388657e-06,
      "loss": 1.0948,
      "step": 18785
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.59929653686058,
      "learning_rate": 9.192648653156115e-06,
      "loss": 1.0959,
      "step": 18790
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.401496333026,
      "learning_rate": 9.191977738735638e-06,
      "loss": 1.1397,
      "step": 18795
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.36854464421448,
      "learning_rate": 9.191306570167897e-06,
      "loss": 1.0903,
      "step": 18800
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.888985182158882,
      "learning_rate": 9.190635147493583e-06,
      "loss": 1.1292,
      "step": 18805
    },
    {
      "epoch": 0.8,
      "grad_norm": 15.694279392670527,
      "learning_rate": 9.189963470753406e-06,
      "loss": 1.0974,
      "step": 18810
    },
    {
      "epoch": 0.8,
      "grad_norm": 12.549628981745615,
      "learning_rate": 9.189291539988085e-06,
      "loss": 1.0868,
      "step": 18815
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.924012018978287,
      "learning_rate": 9.18861935523836e-06,
      "loss": 1.064,
      "step": 18820
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.163714404107731,
      "learning_rate": 9.187946916544979e-06,
      "loss": 1.0679,
      "step": 18825
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.562522128754315,
      "learning_rate": 9.187274223948713e-06,
      "loss": 1.1086,
      "step": 18830
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.356125579253188,
      "learning_rate": 9.186601277490345e-06,
      "loss": 1.0623,
      "step": 18835
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.4770963823764545,
      "learning_rate": 9.185928077210673e-06,
      "loss": 1.0885,
      "step": 18840
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.764008627091845,
      "learning_rate": 9.185254623150512e-06,
      "loss": 1.0774,
      "step": 18845
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.253763251913796,
      "learning_rate": 9.184580915350692e-06,
      "loss": 1.092,
      "step": 18850
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.89590025114452,
      "learning_rate": 9.183906953852055e-06,
      "loss": 1.0722,
      "step": 18855
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.4778403355398195,
      "learning_rate": 9.183232738695466e-06,
      "loss": 1.0935,
      "step": 18860
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.940980094998093,
      "learning_rate": 9.182558269921798e-06,
      "loss": 1.0745,
      "step": 18865
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.440141846582979,
      "learning_rate": 9.181883547571941e-06,
      "loss": 1.0837,
      "step": 18870
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.5994434274352765,
      "learning_rate": 9.181208571686803e-06,
      "loss": 1.1118,
      "step": 18875
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.011216745860771,
      "learning_rate": 9.180533342307305e-06,
      "loss": 1.1157,
      "step": 18880
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.521502193027965,
      "learning_rate": 9.179857859474386e-06,
      "loss": 1.0823,
      "step": 18885
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.346760599258086,
      "learning_rate": 9.179182123228997e-06,
      "loss": 1.063,
      "step": 18890
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.878544302423429,
      "learning_rate": 9.178506133612105e-06,
      "loss": 1.1045,
      "step": 18895
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.178471866740171,
      "learning_rate": 9.177829890664695e-06,
      "loss": 1.0584,
      "step": 18900
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.241658516893521,
      "learning_rate": 9.177153394427765e-06,
      "loss": 1.0686,
      "step": 18905
    },
    {
      "epoch": 0.8,
      "grad_norm": 17.54081778180161,
      "learning_rate": 9.176476644942328e-06,
      "loss": 1.0617,
      "step": 18910
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.572734515834037,
      "learning_rate": 9.175799642249415e-06,
      "loss": 1.0825,
      "step": 18915
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.223257770848594,
      "learning_rate": 9.175122386390069e-06,
      "loss": 1.0758,
      "step": 18920
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.588816346488197,
      "learning_rate": 9.174444877405353e-06,
      "loss": 1.0816,
      "step": 18925
    },
    {
      "epoch": 0.8,
      "grad_norm": 13.27215080545339,
      "learning_rate": 9.173767115336338e-06,
      "loss": 1.1355,
      "step": 18930
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.464572280662654,
      "learning_rate": 9.173089100224118e-06,
      "loss": 1.1102,
      "step": 18935
    },
    {
      "epoch": 0.8,
      "grad_norm": 9.647966082272436,
      "learning_rate": 9.1724108321098e-06,
      "loss": 1.0575,
      "step": 18940
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.249108019532874,
      "learning_rate": 9.1717323110345e-06,
      "loss": 1.1121,
      "step": 18945
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.032627924416587,
      "learning_rate": 9.17105353703936e-06,
      "loss": 1.0519,
      "step": 18950
    },
    {
      "epoch": 0.8,
      "grad_norm": 14.13910179747161,
      "learning_rate": 9.17037451016553e-06,
      "loss": 1.0683,
      "step": 18955
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.344457997778898,
      "learning_rate": 9.16969523045418e-06,
      "loss": 1.0494,
      "step": 18960
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.207826443825296,
      "learning_rate": 9.169015697946486e-06,
      "loss": 1.0673,
      "step": 18965
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.292452697401298,
      "learning_rate": 9.168335912683653e-06,
      "loss": 1.1212,
      "step": 18970
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.76765548899204,
      "learning_rate": 9.167655874706891e-06,
      "loss": 1.0713,
      "step": 18975
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.915497305254457,
      "learning_rate": 9.166975584057432e-06,
      "loss": 1.0762,
      "step": 18980
    },
    {
      "epoch": 0.8,
      "grad_norm": 15.030166307519531,
      "learning_rate": 9.166295040776514e-06,
      "loss": 1.1023,
      "step": 18985
    },
    {
      "epoch": 0.8,
      "grad_norm": 8.328793277685982,
      "learning_rate": 9.165614244905402e-06,
      "loss": 1.1102,
      "step": 18990
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.164809125917204,
      "learning_rate": 9.164933196485368e-06,
      "loss": 1.0572,
      "step": 18995
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.264076581011489,
      "learning_rate": 9.164251895557704e-06,
      "loss": 1.0856,
      "step": 19000
    },
    {
      "epoch": 0.8,
      "grad_norm": 19.88653791877927,
      "learning_rate": 9.163570342163712e-06,
      "loss": 1.0841,
      "step": 19005
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.56257230491043,
      "learning_rate": 9.162888536344714e-06,
      "loss": 1.0661,
      "step": 19010
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.829204000371574,
      "learning_rate": 9.162206478142047e-06,
      "loss": 1.1083,
      "step": 19015
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.96648729986634,
      "learning_rate": 9.16152416759706e-06,
      "loss": 1.0787,
      "step": 19020
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.823120734154049,
      "learning_rate": 9.160841604751124e-06,
      "loss": 1.0879,
      "step": 19025
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.557988977111406,
      "learning_rate": 9.160158789645616e-06,
      "loss": 1.0861,
      "step": 19030
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.323859105048502,
      "learning_rate": 9.159475722321934e-06,
      "loss": 1.0493,
      "step": 19035
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.146001316945294,
      "learning_rate": 9.158792402821493e-06,
      "loss": 1.0749,
      "step": 19040
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.444302499400449,
      "learning_rate": 9.158108831185716e-06,
      "loss": 1.1309,
      "step": 19045
    },
    {
      "epoch": 0.81,
      "grad_norm": 27.666702919519214,
      "learning_rate": 9.157425007456051e-06,
      "loss": 1.0981,
      "step": 19050
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.295897843798757,
      "learning_rate": 9.156740931673954e-06,
      "loss": 1.0839,
      "step": 19055
    },
    {
      "epoch": 0.81,
      "grad_norm": 18.243577534648047,
      "learning_rate": 9.156056603880896e-06,
      "loss": 1.136,
      "step": 19060
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.791910175421773,
      "learning_rate": 9.15537202411837e-06,
      "loss": 1.0691,
      "step": 19065
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.836083884623443,
      "learning_rate": 9.154687192427878e-06,
      "loss": 1.1183,
      "step": 19070
    },
    {
      "epoch": 0.81,
      "grad_norm": 4.678385330915157,
      "learning_rate": 9.154002108850938e-06,
      "loss": 1.1134,
      "step": 19075
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.305402511879571,
      "learning_rate": 9.153316773429089e-06,
      "loss": 1.1513,
      "step": 19080
    },
    {
      "epoch": 0.81,
      "grad_norm": 18.107870763211942,
      "learning_rate": 9.152631186203877e-06,
      "loss": 1.1157,
      "step": 19085
    },
    {
      "epoch": 0.81,
      "grad_norm": 33.44419033884193,
      "learning_rate": 9.151945347216867e-06,
      "loss": 1.0885,
      "step": 19090
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.62082181138283,
      "learning_rate": 9.151259256509641e-06,
      "loss": 1.09,
      "step": 19095
    },
    {
      "epoch": 0.81,
      "grad_norm": 15.11849150910975,
      "learning_rate": 9.150572914123795e-06,
      "loss": 1.079,
      "step": 19100
    },
    {
      "epoch": 0.81,
      "grad_norm": 14.763376625110595,
      "learning_rate": 9.149886320100939e-06,
      "loss": 1.1357,
      "step": 19105
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.495754179251353,
      "learning_rate": 9.1491994744827e-06,
      "loss": 1.0966,
      "step": 19110
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.86885507172298,
      "learning_rate": 9.148512377310716e-06,
      "loss": 1.086,
      "step": 19115
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.035546965614941,
      "learning_rate": 9.14782502862665e-06,
      "loss": 1.0783,
      "step": 19120
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.639223711428465,
      "learning_rate": 9.147137428472167e-06,
      "loss": 1.0931,
      "step": 19125
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.007655885003556,
      "learning_rate": 9.146449576888958e-06,
      "loss": 1.0942,
      "step": 19130
    },
    {
      "epoch": 0.81,
      "grad_norm": 16.38938102398451,
      "learning_rate": 9.145761473918727e-06,
      "loss": 1.0962,
      "step": 19135
    },
    {
      "epoch": 0.81,
      "grad_norm": 30.007942150758492,
      "learning_rate": 9.145073119603188e-06,
      "loss": 1.0505,
      "step": 19140
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.754590507173786,
      "learning_rate": 9.144384513984076e-06,
      "loss": 1.099,
      "step": 19145
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.650281206295421,
      "learning_rate": 9.14369565710314e-06,
      "loss": 1.1044,
      "step": 19150
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.2447836248349216,
      "learning_rate": 9.14300654900214e-06,
      "loss": 1.1031,
      "step": 19155
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.459504512729552,
      "learning_rate": 9.142317189722856e-06,
      "loss": 1.0634,
      "step": 19160
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.093171162012137,
      "learning_rate": 9.141627579307084e-06,
      "loss": 1.1065,
      "step": 19165
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.419564898508516,
      "learning_rate": 9.140937717796631e-06,
      "loss": 1.0867,
      "step": 19170
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.544563631803051,
      "learning_rate": 9.140247605233321e-06,
      "loss": 1.1049,
      "step": 19175
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.777214234831718,
      "learning_rate": 9.139557241658995e-06,
      "loss": 1.1437,
      "step": 19180
    },
    {
      "epoch": 0.81,
      "grad_norm": 17.913400095080576,
      "learning_rate": 9.138866627115507e-06,
      "loss": 1.0458,
      "step": 19185
    },
    {
      "epoch": 0.81,
      "grad_norm": 22.401171282603954,
      "learning_rate": 9.138175761644726e-06,
      "loss": 1.0968,
      "step": 19190
    },
    {
      "epoch": 0.81,
      "grad_norm": 11.235794915938502,
      "learning_rate": 9.137484645288539e-06,
      "loss": 1.0726,
      "step": 19195
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.814588677429702,
      "learning_rate": 9.136793278088844e-06,
      "loss": 1.0953,
      "step": 19200
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.83755866548187,
      "learning_rate": 9.13610166008756e-06,
      "loss": 1.0906,
      "step": 19205
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.309167887529496,
      "learning_rate": 9.135409791326614e-06,
      "loss": 1.0794,
      "step": 19210
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.009610075576476,
      "learning_rate": 9.134717671847954e-06,
      "loss": 1.0954,
      "step": 19215
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.613001490424345,
      "learning_rate": 9.134025301693542e-06,
      "loss": 1.0821,
      "step": 19220
    },
    {
      "epoch": 0.81,
      "grad_norm": 14.780312253020812,
      "learning_rate": 9.13333268090535e-06,
      "loss": 1.078,
      "step": 19225
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.60593018369186,
      "learning_rate": 9.132639809525374e-06,
      "loss": 1.0939,
      "step": 19230
    },
    {
      "epoch": 0.81,
      "grad_norm": 13.688194726528058,
      "learning_rate": 9.131946687595621e-06,
      "loss": 1.0813,
      "step": 19235
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.7923322045800605,
      "learning_rate": 9.13125331515811e-06,
      "loss": 1.0552,
      "step": 19240
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.905352240938075,
      "learning_rate": 9.13055969225488e-06,
      "loss": 1.0717,
      "step": 19245
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.973108039942938,
      "learning_rate": 9.129865818927984e-06,
      "loss": 1.1372,
      "step": 19250
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.565336859234096,
      "learning_rate": 9.129171695219486e-06,
      "loss": 1.1252,
      "step": 19255
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.9793779482690805,
      "learning_rate": 9.128477321171471e-06,
      "loss": 1.0295,
      "step": 19260
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.301272087600875,
      "learning_rate": 9.127782696826037e-06,
      "loss": 1.0889,
      "step": 19265
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.770130018522191,
      "learning_rate": 9.127087822225298e-06,
      "loss": 1.0691,
      "step": 19270
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.107034374003805,
      "learning_rate": 9.126392697411382e-06,
      "loss": 1.0682,
      "step": 19275
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.8254758260181845,
      "learning_rate": 9.12569732242643e-06,
      "loss": 1.0907,
      "step": 19280
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.877748248639448,
      "learning_rate": 9.1250016973126e-06,
      "loss": 1.0949,
      "step": 19285
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.994028524515401,
      "learning_rate": 9.124305822112071e-06,
      "loss": 1.015,
      "step": 19290
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.6559920477563645,
      "learning_rate": 9.123609696867028e-06,
      "loss": 1.0665,
      "step": 19295
    },
    {
      "epoch": 0.82,
      "grad_norm": 18.045241127899516,
      "learning_rate": 9.122913321619675e-06,
      "loss": 1.0739,
      "step": 19300
    },
    {
      "epoch": 0.82,
      "grad_norm": 28.321087239813032,
      "learning_rate": 9.122216696412232e-06,
      "loss": 1.1056,
      "step": 19305
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.012864156176525,
      "learning_rate": 9.121519821286933e-06,
      "loss": 1.0769,
      "step": 19310
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.527751634950696,
      "learning_rate": 9.120822696286027e-06,
      "loss": 1.0878,
      "step": 19315
    },
    {
      "epoch": 0.82,
      "grad_norm": 24.561124908224215,
      "learning_rate": 9.12012532145178e-06,
      "loss": 1.1165,
      "step": 19320
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.484752725806603,
      "learning_rate": 9.11942769682647e-06,
      "loss": 1.118,
      "step": 19325
    },
    {
      "epoch": 0.82,
      "grad_norm": 15.799356698049024,
      "learning_rate": 9.118729822452394e-06,
      "loss": 1.0865,
      "step": 19330
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.404032264513745,
      "learning_rate": 9.11803169837186e-06,
      "loss": 1.114,
      "step": 19335
    },
    {
      "epoch": 0.82,
      "grad_norm": 17.759231298478277,
      "learning_rate": 9.117333324627195e-06,
      "loss": 1.0864,
      "step": 19340
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.872268322535767,
      "learning_rate": 9.116634701260738e-06,
      "loss": 1.0893,
      "step": 19345
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.282267041584997,
      "learning_rate": 9.115935828314847e-06,
      "loss": 1.1351,
      "step": 19350
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.09022097777567,
      "learning_rate": 9.115236705831886e-06,
      "loss": 1.1151,
      "step": 19355
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.97505691874872,
      "learning_rate": 9.11453733385425e-06,
      "loss": 1.0476,
      "step": 19360
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.117871109799094,
      "learning_rate": 9.113837712424332e-06,
      "loss": 1.1249,
      "step": 19365
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.911394135113546,
      "learning_rate": 9.113137841584552e-06,
      "loss": 1.1155,
      "step": 19370
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.6286089657656575,
      "learning_rate": 9.112437721377341e-06,
      "loss": 1.0551,
      "step": 19375
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.309863351619459,
      "learning_rate": 9.111737351845144e-06,
      "loss": 1.0989,
      "step": 19380
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.530328281457148,
      "learning_rate": 9.111036733030425e-06,
      "loss": 1.0951,
      "step": 19385
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.869816279715394,
      "learning_rate": 9.110335864975657e-06,
      "loss": 1.0407,
      "step": 19390
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.611064080496048,
      "learning_rate": 9.109634747723331e-06,
      "loss": 1.0611,
      "step": 19395
    },
    {
      "epoch": 0.82,
      "grad_norm": 12.201274680216573,
      "learning_rate": 9.108933381315958e-06,
      "loss": 1.1019,
      "step": 19400
    },
    {
      "epoch": 0.82,
      "grad_norm": 24.447852592558526,
      "learning_rate": 9.108231765796055e-06,
      "loss": 1.0933,
      "step": 19405
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.72430836714062,
      "learning_rate": 9.107529901206164e-06,
      "loss": 1.0837,
      "step": 19410
    },
    {
      "epoch": 0.82,
      "grad_norm": 8.772729618562487,
      "learning_rate": 9.106827787588832e-06,
      "loss": 1.091,
      "step": 19415
    },
    {
      "epoch": 0.82,
      "grad_norm": 20.059071865937053,
      "learning_rate": 9.106125424986629e-06,
      "loss": 1.1038,
      "step": 19420
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.652340814006722,
      "learning_rate": 9.105422813442137e-06,
      "loss": 1.0728,
      "step": 19425
    },
    {
      "epoch": 0.82,
      "grad_norm": 34.18818336171334,
      "learning_rate": 9.104719952997953e-06,
      "loss": 1.1001,
      "step": 19430
    },
    {
      "epoch": 0.82,
      "grad_norm": 10.4046297526499,
      "learning_rate": 9.104016843696688e-06,
      "loss": 1.1397,
      "step": 19435
    },
    {
      "epoch": 0.82,
      "grad_norm": 17.37654636027216,
      "learning_rate": 9.103313485580972e-06,
      "loss": 1.0904,
      "step": 19440
    },
    {
      "epoch": 0.82,
      "grad_norm": 33.344432938612684,
      "learning_rate": 9.102609878693444e-06,
      "loss": 1.0952,
      "step": 19445
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.74082419750613,
      "learning_rate": 9.101906023076765e-06,
      "loss": 1.0873,
      "step": 19450
    },
    {
      "epoch": 0.82,
      "grad_norm": 14.362731668300325,
      "learning_rate": 9.101201918773607e-06,
      "loss": 1.116,
      "step": 19455
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.310362478294449,
      "learning_rate": 9.100497565826656e-06,
      "loss": 1.1179,
      "step": 19460
    },
    {
      "epoch": 0.82,
      "grad_norm": 17.532089238639685,
      "learning_rate": 9.099792964278616e-06,
      "loss": 1.0442,
      "step": 19465
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.546423523225817,
      "learning_rate": 9.099088114172206e-06,
      "loss": 1.0908,
      "step": 19470
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.724064341744679,
      "learning_rate": 9.098383015550155e-06,
      "loss": 1.0426,
      "step": 19475
    },
    {
      "epoch": 0.82,
      "grad_norm": 23.444118248180988,
      "learning_rate": 9.097677668455218e-06,
      "loss": 1.088,
      "step": 19480
    },
    {
      "epoch": 0.82,
      "grad_norm": 39.30190879515092,
      "learning_rate": 9.096972072930153e-06,
      "loss": 1.1005,
      "step": 19485
    },
    {
      "epoch": 0.82,
      "grad_norm": 28.292805688068835,
      "learning_rate": 9.096266229017738e-06,
      "loss": 1.0969,
      "step": 19490
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.74188068243222,
      "learning_rate": 9.095560136760768e-06,
      "loss": 1.0869,
      "step": 19495
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.535740606916772,
      "learning_rate": 9.09485379620205e-06,
      "loss": 1.0821,
      "step": 19500
    },
    {
      "epoch": 0.83,
      "grad_norm": 23.276572860000982,
      "learning_rate": 9.09414720738441e-06,
      "loss": 1.1266,
      "step": 19505
    },
    {
      "epoch": 0.83,
      "grad_norm": 40.002289558763835,
      "learning_rate": 9.093440370350682e-06,
      "loss": 1.0953,
      "step": 19510
    },
    {
      "epoch": 0.83,
      "grad_norm": 28.71328740336301,
      "learning_rate": 9.092733285143723e-06,
      "loss": 1.1166,
      "step": 19515
    },
    {
      "epoch": 0.83,
      "grad_norm": 45.9264411176706,
      "learning_rate": 9.092025951806402e-06,
      "loss": 1.1115,
      "step": 19520
    },
    {
      "epoch": 0.83,
      "grad_norm": 21.601372558443252,
      "learning_rate": 9.091318370381599e-06,
      "loss": 1.0908,
      "step": 19525
    },
    {
      "epoch": 0.83,
      "grad_norm": 27.49187266331318,
      "learning_rate": 9.090610540912214e-06,
      "loss": 1.0823,
      "step": 19530
    },
    {
      "epoch": 0.83,
      "grad_norm": 35.236884920197994,
      "learning_rate": 9.089902463441163e-06,
      "loss": 1.1576,
      "step": 19535
    },
    {
      "epoch": 0.83,
      "grad_norm": 26.254648166086003,
      "learning_rate": 9.089194138011372e-06,
      "loss": 1.0969,
      "step": 19540
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.892745445502996,
      "learning_rate": 9.088485564665788e-06,
      "loss": 1.0708,
      "step": 19545
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.370896317391928,
      "learning_rate": 9.087776743447364e-06,
      "loss": 1.0686,
      "step": 19550
    },
    {
      "epoch": 0.83,
      "grad_norm": 21.883055982097662,
      "learning_rate": 9.087067674399079e-06,
      "loss": 1.09,
      "step": 19555
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.08360488336419,
      "learning_rate": 9.086358357563919e-06,
      "loss": 1.0829,
      "step": 19560
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.738685106889233,
      "learning_rate": 9.085648792984887e-06,
      "loss": 1.0873,
      "step": 19565
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.61050377511876,
      "learning_rate": 9.084938980705005e-06,
      "loss": 1.1301,
      "step": 19570
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.007318275262184,
      "learning_rate": 9.084228920767306e-06,
      "loss": 1.0537,
      "step": 19575
    },
    {
      "epoch": 0.83,
      "grad_norm": 15.244836727040267,
      "learning_rate": 9.083518613214836e-06,
      "loss": 1.0601,
      "step": 19580
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.324853940949712,
      "learning_rate": 9.082808058090662e-06,
      "loss": 1.1292,
      "step": 19585
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.4246808006599485,
      "learning_rate": 9.082097255437863e-06,
      "loss": 1.0518,
      "step": 19590
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.2451595886324975,
      "learning_rate": 9.08138620529953e-06,
      "loss": 1.0926,
      "step": 19595
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.879622339407153,
      "learning_rate": 9.080674907718773e-06,
      "loss": 1.1288,
      "step": 19600
    },
    {
      "epoch": 0.83,
      "grad_norm": 13.63277475712674,
      "learning_rate": 9.079963362738717e-06,
      "loss": 1.0906,
      "step": 19605
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.9457377690941655,
      "learning_rate": 9.0792515704025e-06,
      "loss": 1.0607,
      "step": 19610
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.61193743839909,
      "learning_rate": 9.078539530753277e-06,
      "loss": 1.0933,
      "step": 19615
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.934530538656906,
      "learning_rate": 9.077827243834216e-06,
      "loss": 1.0515,
      "step": 19620
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.350044952936317,
      "learning_rate": 9.0771147096885e-06,
      "loss": 1.1146,
      "step": 19625
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.394082976889703,
      "learning_rate": 9.076401928359328e-06,
      "loss": 1.0514,
      "step": 19630
    },
    {
      "epoch": 0.83,
      "grad_norm": 16.23560948344088,
      "learning_rate": 9.075688899889916e-06,
      "loss": 1.1166,
      "step": 19635
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.958315161605933,
      "learning_rate": 9.074975624323491e-06,
      "loss": 1.0914,
      "step": 19640
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.566861986679763,
      "learning_rate": 9.074262101703297e-06,
      "loss": 1.0859,
      "step": 19645
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.632693449285916,
      "learning_rate": 9.073548332072592e-06,
      "loss": 1.0791,
      "step": 19650
    },
    {
      "epoch": 0.83,
      "grad_norm": 20.65064323140121,
      "learning_rate": 9.072834315474652e-06,
      "loss": 1.1455,
      "step": 19655
    },
    {
      "epoch": 0.83,
      "grad_norm": 24.610704855283625,
      "learning_rate": 9.072120051952765e-06,
      "loss": 1.1107,
      "step": 19660
    },
    {
      "epoch": 0.83,
      "grad_norm": 26.838465839568748,
      "learning_rate": 9.071405541550232e-06,
      "loss": 1.0736,
      "step": 19665
    },
    {
      "epoch": 0.83,
      "grad_norm": 18.03428719682385,
      "learning_rate": 9.070690784310375e-06,
      "loss": 1.1318,
      "step": 19670
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.369698976471521,
      "learning_rate": 9.069975780276525e-06,
      "loss": 1.1233,
      "step": 19675
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.880803190191005,
      "learning_rate": 9.069260529492034e-06,
      "loss": 1.1,
      "step": 19680
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.2379545085251,
      "learning_rate": 9.068545032000262e-06,
      "loss": 1.0436,
      "step": 19685
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.4150443652031015,
      "learning_rate": 9.06782928784459e-06,
      "loss": 1.0594,
      "step": 19690
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.920069532461167,
      "learning_rate": 9.06711329706841e-06,
      "loss": 1.0735,
      "step": 19695
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.10497312710696,
      "learning_rate": 9.06639705971513e-06,
      "loss": 1.1091,
      "step": 19700
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.655612335696453,
      "learning_rate": 9.065680575828175e-06,
      "loss": 1.068,
      "step": 19705
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.769799576207244,
      "learning_rate": 9.064963845450982e-06,
      "loss": 1.0921,
      "step": 19710
    },
    {
      "epoch": 0.83,
      "grad_norm": 12.515708396778855,
      "learning_rate": 9.064246868627007e-06,
      "loss": 1.1151,
      "step": 19715
    },
    {
      "epoch": 0.83,
      "grad_norm": 11.04131509292738,
      "learning_rate": 9.063529645399714e-06,
      "loss": 1.0905,
      "step": 19720
    },
    {
      "epoch": 0.83,
      "grad_norm": 10.315296274757085,
      "learning_rate": 9.062812175812588e-06,
      "loss": 1.0856,
      "step": 19725
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.128517727264331,
      "learning_rate": 9.062094459909128e-06,
      "loss": 1.0619,
      "step": 19730
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.66805833480779,
      "learning_rate": 9.061376497732845e-06,
      "loss": 1.1245,
      "step": 19735
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.825044422790672,
      "learning_rate": 9.060658289327269e-06,
      "loss": 1.0863,
      "step": 19740
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.590008143801267,
      "learning_rate": 9.059939834735943e-06,
      "loss": 1.0848,
      "step": 19745
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.891054470172313,
      "learning_rate": 9.059221134002424e-06,
      "loss": 1.0208,
      "step": 19750
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.079366340057518,
      "learning_rate": 9.058502187170282e-06,
      "loss": 1.0624,
      "step": 19755
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.696786757151849,
      "learning_rate": 9.057782994283109e-06,
      "loss": 1.109,
      "step": 19760
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.791281576320251,
      "learning_rate": 9.057063555384505e-06,
      "loss": 1.091,
      "step": 19765
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.343303814871241,
      "learning_rate": 9.056343870518087e-06,
      "loss": 1.0976,
      "step": 19770
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.42922021394102,
      "learning_rate": 9.05562393972749e-06,
      "loss": 1.1127,
      "step": 19775
    },
    {
      "epoch": 0.84,
      "grad_norm": 20.434826362852455,
      "learning_rate": 9.05490376305636e-06,
      "loss": 1.086,
      "step": 19780
    },
    {
      "epoch": 0.84,
      "grad_norm": 15.140866189177508,
      "learning_rate": 9.054183340548357e-06,
      "loss": 1.0985,
      "step": 19785
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.213400280238245,
      "learning_rate": 9.053462672247162e-06,
      "loss": 1.1016,
      "step": 19790
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.6912766315098215,
      "learning_rate": 9.052741758196463e-06,
      "loss": 1.0957,
      "step": 19795
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.88233843925602,
      "learning_rate": 9.052020598439972e-06,
      "loss": 1.0864,
      "step": 19800
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.95720025396798,
      "learning_rate": 9.051299193021405e-06,
      "loss": 1.1033,
      "step": 19805
    },
    {
      "epoch": 0.84,
      "grad_norm": 17.654101836644557,
      "learning_rate": 9.050577541984504e-06,
      "loss": 1.0988,
      "step": 19810
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.800794635451995,
      "learning_rate": 9.049855645373015e-06,
      "loss": 1.104,
      "step": 19815
    },
    {
      "epoch": 0.84,
      "grad_norm": 31.641052757993517,
      "learning_rate": 9.049133503230708e-06,
      "loss": 1.0669,
      "step": 19820
    },
    {
      "epoch": 0.84,
      "grad_norm": 20.30315004564152,
      "learning_rate": 9.048411115601367e-06,
      "loss": 1.0696,
      "step": 19825
    },
    {
      "epoch": 0.84,
      "grad_norm": 44.01827957171962,
      "learning_rate": 9.047688482528783e-06,
      "loss": 1.1108,
      "step": 19830
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.505357928869747,
      "learning_rate": 9.046965604056768e-06,
      "loss": 1.0876,
      "step": 19835
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.48392857117296,
      "learning_rate": 9.046242480229151e-06,
      "loss": 1.0674,
      "step": 19840
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.178370075808223,
      "learning_rate": 9.045519111089769e-06,
      "loss": 1.0723,
      "step": 19845
    },
    {
      "epoch": 0.84,
      "grad_norm": 21.580231628102435,
      "learning_rate": 9.044795496682482e-06,
      "loss": 1.0773,
      "step": 19850
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.992579931934888,
      "learning_rate": 9.044071637051156e-06,
      "loss": 1.0985,
      "step": 19855
    },
    {
      "epoch": 0.84,
      "grad_norm": 21.68329726592509,
      "learning_rate": 9.043347532239682e-06,
      "loss": 1.034,
      "step": 19860
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.202868556941622,
      "learning_rate": 9.042623182291955e-06,
      "loss": 1.0997,
      "step": 19865
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.869264887140003,
      "learning_rate": 9.041898587251891e-06,
      "loss": 1.1441,
      "step": 19870
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.605558241622164,
      "learning_rate": 9.041173747163423e-06,
      "loss": 1.1082,
      "step": 19875
    },
    {
      "epoch": 0.84,
      "grad_norm": 15.410529170905154,
      "learning_rate": 9.040448662070496e-06,
      "loss": 1.0713,
      "step": 19880
    },
    {
      "epoch": 0.84,
      "grad_norm": 21.61015827706873,
      "learning_rate": 9.039723332017066e-06,
      "loss": 1.1334,
      "step": 19885
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.725539701215709,
      "learning_rate": 9.03899775704711e-06,
      "loss": 1.1025,
      "step": 19890
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.885791292808605,
      "learning_rate": 9.038271937204617e-06,
      "loss": 1.0914,
      "step": 19895
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.510368052497252,
      "learning_rate": 9.037545872533593e-06,
      "loss": 1.0794,
      "step": 19900
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.672695980168514,
      "learning_rate": 9.036819563078057e-06,
      "loss": 1.0642,
      "step": 19905
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.063951377464615,
      "learning_rate": 9.036093008882039e-06,
      "loss": 1.0824,
      "step": 19910
    },
    {
      "epoch": 0.84,
      "grad_norm": 22.049718130399935,
      "learning_rate": 9.035366209989592e-06,
      "loss": 1.092,
      "step": 19915
    },
    {
      "epoch": 0.84,
      "grad_norm": 27.384805349827268,
      "learning_rate": 9.034639166444776e-06,
      "loss": 1.1315,
      "step": 19920
    },
    {
      "epoch": 0.84,
      "grad_norm": 13.00327097821516,
      "learning_rate": 9.033911878291673e-06,
      "loss": 1.0554,
      "step": 19925
    },
    {
      "epoch": 0.84,
      "grad_norm": 13.525656956369206,
      "learning_rate": 9.033184345574375e-06,
      "loss": 1.1081,
      "step": 19930
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.343269884581191,
      "learning_rate": 9.032456568336992e-06,
      "loss": 1.0505,
      "step": 19935
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.742466878310966,
      "learning_rate": 9.031728546623644e-06,
      "loss": 1.106,
      "step": 19940
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.827341983796607,
      "learning_rate": 9.031000280478471e-06,
      "loss": 1.1043,
      "step": 19945
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.8023909340638,
      "learning_rate": 9.030271769945622e-06,
      "loss": 1.0994,
      "step": 19950
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.330851482313204,
      "learning_rate": 9.029543015069268e-06,
      "loss": 1.0535,
      "step": 19955
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.247238471473315,
      "learning_rate": 9.028814015893591e-06,
      "loss": 1.0763,
      "step": 19960
    },
    {
      "epoch": 0.84,
      "grad_norm": 31.42258428677921,
      "learning_rate": 9.02808477246279e-06,
      "loss": 1.1185,
      "step": 19965
    },
    {
      "epoch": 0.85,
      "grad_norm": 15.59514788468728,
      "learning_rate": 9.02735528482107e-06,
      "loss": 1.0878,
      "step": 19970
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.624107785535747,
      "learning_rate": 9.026625553012664e-06,
      "loss": 1.0662,
      "step": 19975
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.468863942961594,
      "learning_rate": 9.025895577081813e-06,
      "loss": 1.0798,
      "step": 19980
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.473507673820163,
      "learning_rate": 9.02516535707277e-06,
      "loss": 1.0545,
      "step": 19985
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.967272762152329,
      "learning_rate": 9.024434893029809e-06,
      "loss": 1.1045,
      "step": 19990
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.0224164179803745,
      "learning_rate": 9.023704184997216e-06,
      "loss": 1.082,
      "step": 19995
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.552637286883164,
      "learning_rate": 9.022973233019288e-06,
      "loss": 1.082,
      "step": 20000
    },
    {
      "epoch": 0.85,
      "grad_norm": 19.333768100992216,
      "learning_rate": 9.022242037140343e-06,
      "loss": 1.1083,
      "step": 20005
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.640705155336257,
      "learning_rate": 9.021510597404712e-06,
      "loss": 1.028,
      "step": 20010
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.2091170735482,
      "learning_rate": 9.020778913856738e-06,
      "loss": 1.1127,
      "step": 20015
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.803134728674725,
      "learning_rate": 9.020046986540784e-06,
      "loss": 1.0781,
      "step": 20020
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.9612602105252686,
      "learning_rate": 9.019314815501221e-06,
      "loss": 1.1101,
      "step": 20025
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.757119428537681,
      "learning_rate": 9.01858240078244e-06,
      "loss": 1.1053,
      "step": 20030
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.221655078730315,
      "learning_rate": 9.017849742428847e-06,
      "loss": 1.1389,
      "step": 20035
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.128617971391468,
      "learning_rate": 9.017116840484858e-06,
      "loss": 1.0491,
      "step": 20040
    },
    {
      "epoch": 0.85,
      "grad_norm": 12.523177373523286,
      "learning_rate": 9.016383694994908e-06,
      "loss": 1.0592,
      "step": 20045
    },
    {
      "epoch": 0.85,
      "grad_norm": 12.85040321334905,
      "learning_rate": 9.015650306003445e-06,
      "loss": 1.0505,
      "step": 20050
    },
    {
      "epoch": 0.85,
      "grad_norm": 15.760789253857968,
      "learning_rate": 9.014916673554932e-06,
      "loss": 1.0888,
      "step": 20055
    },
    {
      "epoch": 0.85,
      "grad_norm": 14.141240520302501,
      "learning_rate": 9.014182797693847e-06,
      "loss": 1.102,
      "step": 20060
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.243446148153748,
      "learning_rate": 9.013448678464685e-06,
      "loss": 1.0378,
      "step": 20065
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.655297097676386,
      "learning_rate": 9.01271431591195e-06,
      "loss": 1.0683,
      "step": 20070
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.262279623707647,
      "learning_rate": 9.011979710080168e-06,
      "loss": 1.0767,
      "step": 20075
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.256080630066792,
      "learning_rate": 9.01124486101387e-06,
      "loss": 1.0688,
      "step": 20080
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.997072070051807,
      "learning_rate": 9.010509768757615e-06,
      "loss": 1.0731,
      "step": 20085
    },
    {
      "epoch": 0.85,
      "grad_norm": 34.617390684592856,
      "learning_rate": 9.009774433355964e-06,
      "loss": 1.153,
      "step": 20090
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.104900195519913,
      "learning_rate": 9.009038854853503e-06,
      "loss": 1.1183,
      "step": 20095
    },
    {
      "epoch": 0.85,
      "grad_norm": 18.819050127253945,
      "learning_rate": 9.008303033294823e-06,
      "loss": 1.0961,
      "step": 20100
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.942577049123328,
      "learning_rate": 9.007566968724536e-06,
      "loss": 1.0912,
      "step": 20105
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.93438485176529,
      "learning_rate": 9.00683066118727e-06,
      "loss": 1.0767,
      "step": 20110
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.54712902837516,
      "learning_rate": 9.006094110727664e-06,
      "loss": 1.0883,
      "step": 20115
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.759340423207265,
      "learning_rate": 9.005357317390372e-06,
      "loss": 1.0946,
      "step": 20120
    },
    {
      "epoch": 0.85,
      "grad_norm": 17.119935450790077,
      "learning_rate": 9.004620281220063e-06,
      "loss": 1.0717,
      "step": 20125
    },
    {
      "epoch": 0.85,
      "grad_norm": 21.515732978868932,
      "learning_rate": 9.003883002261423e-06,
      "loss": 1.0819,
      "step": 20130
    },
    {
      "epoch": 0.85,
      "grad_norm": 12.557007819034817,
      "learning_rate": 9.00314548055915e-06,
      "loss": 1.1214,
      "step": 20135
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.591066386613737,
      "learning_rate": 9.002407716157959e-06,
      "loss": 1.0672,
      "step": 20140
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.8559643287969,
      "learning_rate": 9.001669709102577e-06,
      "loss": 1.0787,
      "step": 20145
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.420559727599148,
      "learning_rate": 9.000931459437749e-06,
      "loss": 1.0677,
      "step": 20150
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.7387590482845,
      "learning_rate": 9.00019296720823e-06,
      "loss": 1.1189,
      "step": 20155
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.497387769725192,
      "learning_rate": 8.999454232458795e-06,
      "loss": 1.1007,
      "step": 20160
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.911551653629037,
      "learning_rate": 8.998715255234232e-06,
      "loss": 1.079,
      "step": 20165
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.944551973676329,
      "learning_rate": 8.99797603557934e-06,
      "loss": 1.0517,
      "step": 20170
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.304741232955825,
      "learning_rate": 8.997236573538937e-06,
      "loss": 1.1277,
      "step": 20175
    },
    {
      "epoch": 0.85,
      "grad_norm": 13.778028106667294,
      "learning_rate": 8.996496869157856e-06,
      "loss": 1.0798,
      "step": 20180
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.231798086007403,
      "learning_rate": 8.995756922480942e-06,
      "loss": 1.0894,
      "step": 20185
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.043254808534116,
      "learning_rate": 8.995016733553056e-06,
      "loss": 1.0831,
      "step": 20190
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.879243494463987,
      "learning_rate": 8.994276302419072e-06,
      "loss": 1.0269,
      "step": 20195
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.297548316870222,
      "learning_rate": 8.993535629123881e-06,
      "loss": 1.0558,
      "step": 20200
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.946190612093227,
      "learning_rate": 8.992794713712389e-06,
      "loss": 1.1101,
      "step": 20205
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.822172476018839,
      "learning_rate": 8.992053556229516e-06,
      "loss": 1.0688,
      "step": 20210
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.79893351264903,
      "learning_rate": 8.991312156720193e-06,
      "loss": 1.0624,
      "step": 20215
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.109727100432633,
      "learning_rate": 8.99057051522937e-06,
      "loss": 1.1096,
      "step": 20220
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.79388252749503,
      "learning_rate": 8.989828631802014e-06,
      "loss": 1.0705,
      "step": 20225
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.354789730293937,
      "learning_rate": 8.989086506483098e-06,
      "loss": 1.0845,
      "step": 20230
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.389109037603895,
      "learning_rate": 8.988344139317618e-06,
      "loss": 1.0408,
      "step": 20235
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.848361395776826,
      "learning_rate": 8.987601530350582e-06,
      "loss": 1.0563,
      "step": 20240
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.706417136899323,
      "learning_rate": 8.986858679627011e-06,
      "loss": 1.0645,
      "step": 20245
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.886295755280162,
      "learning_rate": 8.98611558719194e-06,
      "loss": 1.092,
      "step": 20250
    },
    {
      "epoch": 0.86,
      "grad_norm": 14.00277406789332,
      "learning_rate": 8.985372253090423e-06,
      "loss": 1.0701,
      "step": 20255
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.568020195797528,
      "learning_rate": 8.984628677367527e-06,
      "loss": 1.0825,
      "step": 20260
    },
    {
      "epoch": 0.86,
      "grad_norm": 12.433120214978809,
      "learning_rate": 8.98388486006833e-06,
      "loss": 1.0918,
      "step": 20265
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.155095513664751,
      "learning_rate": 8.983140801237931e-06,
      "loss": 1.0906,
      "step": 20270
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.708769156445875,
      "learning_rate": 8.982396500921436e-06,
      "loss": 1.0612,
      "step": 20275
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.12344335717567,
      "learning_rate": 8.981651959163972e-06,
      "loss": 1.0705,
      "step": 20280
    },
    {
      "epoch": 0.86,
      "grad_norm": 16.48037028658968,
      "learning_rate": 8.980907176010678e-06,
      "loss": 1.0742,
      "step": 20285
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.42077654837551,
      "learning_rate": 8.980162151506708e-06,
      "loss": 1.0267,
      "step": 20290
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.632192408542217,
      "learning_rate": 8.979416885697231e-06,
      "loss": 1.0954,
      "step": 20295
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.966806795829951,
      "learning_rate": 8.97867137862743e-06,
      "loss": 1.0995,
      "step": 20300
    },
    {
      "epoch": 0.86,
      "grad_norm": 49.69778209689961,
      "learning_rate": 8.977925630342503e-06,
      "loss": 1.0752,
      "step": 20305
    },
    {
      "epoch": 0.86,
      "grad_norm": 13.307026872944558,
      "learning_rate": 8.977179640887661e-06,
      "loss": 1.1463,
      "step": 20310
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.21257520093946,
      "learning_rate": 8.976433410308134e-06,
      "loss": 1.0799,
      "step": 20315
    },
    {
      "epoch": 0.86,
      "grad_norm": 33.002611892588924,
      "learning_rate": 8.975686938649162e-06,
      "loss": 1.1548,
      "step": 20320
    },
    {
      "epoch": 0.86,
      "grad_norm": 20.35768011153258,
      "learning_rate": 8.974940225956002e-06,
      "loss": 1.1126,
      "step": 20325
    },
    {
      "epoch": 0.86,
      "grad_norm": 14.607463469824511,
      "learning_rate": 8.974193272273925e-06,
      "loss": 1.0883,
      "step": 20330
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.865027666199963,
      "learning_rate": 8.973446077648215e-06,
      "loss": 1.039,
      "step": 20335
    },
    {
      "epoch": 0.86,
      "grad_norm": 16.69134700460327,
      "learning_rate": 8.972698642124174e-06,
      "loss": 1.0965,
      "step": 20340
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.399351492280459,
      "learning_rate": 8.971950965747117e-06,
      "loss": 1.0769,
      "step": 20345
    },
    {
      "epoch": 0.86,
      "grad_norm": 11.902657752331681,
      "learning_rate": 8.971203048562372e-06,
      "loss": 1.0457,
      "step": 20350
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.524370329936866,
      "learning_rate": 8.970454890615283e-06,
      "loss": 1.067,
      "step": 20355
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.86432894769377,
      "learning_rate": 8.969706491951211e-06,
      "loss": 1.099,
      "step": 20360
    },
    {
      "epoch": 0.86,
      "grad_norm": 18.858133900544203,
      "learning_rate": 8.968957852615527e-06,
      "loss": 1.0831,
      "step": 20365
    },
    {
      "epoch": 0.86,
      "grad_norm": 14.070009841432515,
      "learning_rate": 8.968208972653619e-06,
      "loss": 1.0457,
      "step": 20370
    },
    {
      "epoch": 0.86,
      "grad_norm": 25.502989972825734,
      "learning_rate": 8.967459852110889e-06,
      "loss": 1.1033,
      "step": 20375
    },
    {
      "epoch": 0.86,
      "grad_norm": 17.446731739851604,
      "learning_rate": 8.966710491032756e-06,
      "loss": 1.0618,
      "step": 20380
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.701372549347564,
      "learning_rate": 8.965960889464649e-06,
      "loss": 1.0618,
      "step": 20385
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.508137373487315,
      "learning_rate": 8.965211047452016e-06,
      "loss": 1.054,
      "step": 20390
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.562783184307362,
      "learning_rate": 8.964460965040318e-06,
      "loss": 1.0725,
      "step": 20395
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.357401594470959,
      "learning_rate": 8.963710642275029e-06,
      "loss": 1.0351,
      "step": 20400
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.856682878737185,
      "learning_rate": 8.962960079201639e-06,
      "loss": 1.1016,
      "step": 20405
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.200709766732144,
      "learning_rate": 8.962209275865653e-06,
      "loss": 1.0679,
      "step": 20410
    },
    {
      "epoch": 0.86,
      "grad_norm": 12.876027765183197,
      "learning_rate": 8.961458232312588e-06,
      "loss": 1.0743,
      "step": 20415
    },
    {
      "epoch": 0.86,
      "grad_norm": 9.675125683171963,
      "learning_rate": 8.960706948587981e-06,
      "loss": 1.1276,
      "step": 20420
    },
    {
      "epoch": 0.86,
      "grad_norm": 16.350548429975284,
      "learning_rate": 8.959955424737378e-06,
      "loss": 1.0713,
      "step": 20425
    },
    {
      "epoch": 0.86,
      "grad_norm": 8.117774312324682,
      "learning_rate": 8.959203660806342e-06,
      "loss": 1.0598,
      "step": 20430
    },
    {
      "epoch": 0.86,
      "grad_norm": 12.957991636458253,
      "learning_rate": 8.95845165684045e-06,
      "loss": 1.0998,
      "step": 20435
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.229193068692858,
      "learning_rate": 8.957699412885295e-06,
      "loss": 1.0535,
      "step": 20440
    },
    {
      "epoch": 0.87,
      "grad_norm": 12.668239749634456,
      "learning_rate": 8.956946928986481e-06,
      "loss": 1.0856,
      "step": 20445
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.851890856640829,
      "learning_rate": 8.95619420518963e-06,
      "loss": 1.0966,
      "step": 20450
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.811605278014109,
      "learning_rate": 8.955441241540379e-06,
      "loss": 1.1098,
      "step": 20455
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.046129936556467,
      "learning_rate": 8.954688038084376e-06,
      "loss": 1.0611,
      "step": 20460
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.500104935071587,
      "learning_rate": 8.953934594867287e-06,
      "loss": 1.0544,
      "step": 20465
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.481932886297046,
      "learning_rate": 8.95318091193479e-06,
      "loss": 1.0846,
      "step": 20470
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.080761304040258,
      "learning_rate": 8.952426989332577e-06,
      "loss": 1.0182,
      "step": 20475
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.224281468221,
      "learning_rate": 8.95167282710636e-06,
      "loss": 1.0712,
      "step": 20480
    },
    {
      "epoch": 0.87,
      "grad_norm": 18.84219314552434,
      "learning_rate": 8.950918425301858e-06,
      "loss": 1.0775,
      "step": 20485
    },
    {
      "epoch": 0.87,
      "grad_norm": 31.204784337293685,
      "learning_rate": 8.950163783964812e-06,
      "loss": 1.0813,
      "step": 20490
    },
    {
      "epoch": 0.87,
      "grad_norm": 20.484143649162505,
      "learning_rate": 8.949408903140968e-06,
      "loss": 1.0541,
      "step": 20495
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.115122733564604,
      "learning_rate": 8.948653782876098e-06,
      "loss": 1.0568,
      "step": 20500
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.505647491943554,
      "learning_rate": 8.94789842321598e-06,
      "loss": 1.0812,
      "step": 20505
    },
    {
      "epoch": 0.87,
      "grad_norm": 10.40810364648984,
      "learning_rate": 8.94714282420641e-06,
      "loss": 1.085,
      "step": 20510
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.230703497318224,
      "learning_rate": 8.946386985893196e-06,
      "loss": 1.0775,
      "step": 20515
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.705929919561878,
      "learning_rate": 8.945630908322164e-06,
      "loss": 1.0563,
      "step": 20520
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.141083517598648,
      "learning_rate": 8.944874591539152e-06,
      "loss": 1.0818,
      "step": 20525
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.1663110330952415,
      "learning_rate": 8.944118035590014e-06,
      "loss": 1.0462,
      "step": 20530
    },
    {
      "epoch": 0.87,
      "grad_norm": 8.749706182628607,
      "learning_rate": 8.943361240520616e-06,
      "loss": 1.0998,
      "step": 20535
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.910245931412528,
      "learning_rate": 8.942604206376844e-06,
      "loss": 1.0616,
      "step": 20540
    },
    {
      "epoch": 0.87,
      "grad_norm": 20.380305710353692,
      "learning_rate": 8.94184693320459e-06,
      "loss": 1.0853,
      "step": 20545
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.088292089443374,
      "learning_rate": 8.94108942104977e-06,
      "loss": 1.0674,
      "step": 20550
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.908595489520451,
      "learning_rate": 8.940331669958304e-06,
      "loss": 1.0982,
      "step": 20555
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.231517270423463,
      "learning_rate": 8.939573679976137e-06,
      "loss": 1.0693,
      "step": 20560
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.75705596330069,
      "learning_rate": 8.938815451149224e-06,
      "loss": 1.0941,
      "step": 20565
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.741464463568612,
      "learning_rate": 8.93805698352353e-06,
      "loss": 1.0467,
      "step": 20570
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.68267101065953,
      "learning_rate": 8.937298277145041e-06,
      "loss": 1.1012,
      "step": 20575
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.17126534449252,
      "learning_rate": 8.936539332059755e-06,
      "loss": 1.1025,
      "step": 20580
    },
    {
      "epoch": 0.87,
      "grad_norm": 12.702948326580303,
      "learning_rate": 8.935780148313686e-06,
      "loss": 1.098,
      "step": 20585
    },
    {
      "epoch": 0.87,
      "grad_norm": 12.86970600561228,
      "learning_rate": 8.93502072595286e-06,
      "loss": 1.0963,
      "step": 20590
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.156514746335783,
      "learning_rate": 8.934261065023318e-06,
      "loss": 1.0958,
      "step": 20595
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.158055846196725,
      "learning_rate": 8.933501165571115e-06,
      "loss": 1.0649,
      "step": 20600
    },
    {
      "epoch": 0.87,
      "grad_norm": 12.187059209033082,
      "learning_rate": 8.932741027642324e-06,
      "loss": 1.0818,
      "step": 20605
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.930266801849595,
      "learning_rate": 8.93198065128303e-06,
      "loss": 1.0357,
      "step": 20610
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.394440513658248,
      "learning_rate": 8.931220036539329e-06,
      "loss": 1.0529,
      "step": 20615
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.629478752452526,
      "learning_rate": 8.930459183457338e-06,
      "loss": 1.1006,
      "step": 20620
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.066311021056976,
      "learning_rate": 8.929698092083184e-06,
      "loss": 1.0507,
      "step": 20625
    },
    {
      "epoch": 0.87,
      "grad_norm": 19.557018415209853,
      "learning_rate": 8.92893676246301e-06,
      "loss": 1.0749,
      "step": 20630
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.864813280868297,
      "learning_rate": 8.928175194642974e-06,
      "loss": 1.06,
      "step": 20635
    },
    {
      "epoch": 0.87,
      "grad_norm": 34.294989888131774,
      "learning_rate": 8.927413388669246e-06,
      "loss": 1.0518,
      "step": 20640
    },
    {
      "epoch": 0.87,
      "grad_norm": 33.18504417784154,
      "learning_rate": 8.926651344588014e-06,
      "loss": 1.12,
      "step": 20645
    },
    {
      "epoch": 0.87,
      "grad_norm": 52.22895972154031,
      "learning_rate": 8.925889062445475e-06,
      "loss": 1.0405,
      "step": 20650
    },
    {
      "epoch": 0.87,
      "grad_norm": 27.638755830578024,
      "learning_rate": 8.925126542287848e-06,
      "loss": 1.0766,
      "step": 20655
    },
    {
      "epoch": 0.87,
      "grad_norm": 16.14746930318794,
      "learning_rate": 8.924363784161358e-06,
      "loss": 1.0963,
      "step": 20660
    },
    {
      "epoch": 0.87,
      "grad_norm": 23.38002364267718,
      "learning_rate": 8.923600788112255e-06,
      "loss": 1.0448,
      "step": 20665
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.527568486030331,
      "learning_rate": 8.922837554186791e-06,
      "loss": 1.0695,
      "step": 20670
    },
    {
      "epoch": 0.88,
      "grad_norm": 23.44621025035625,
      "learning_rate": 8.922074082431242e-06,
      "loss": 1.0786,
      "step": 20675
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.3400610424946,
      "learning_rate": 8.921310372891896e-06,
      "loss": 1.0836,
      "step": 20680
    },
    {
      "epoch": 0.88,
      "grad_norm": 16.164554952646025,
      "learning_rate": 8.92054642561505e-06,
      "loss": 1.1359,
      "step": 20685
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.931753405279445,
      "learning_rate": 8.919782240647024e-06,
      "loss": 1.0583,
      "step": 20690
    },
    {
      "epoch": 0.88,
      "grad_norm": 17.801030716947963,
      "learning_rate": 8.919017818034146e-06,
      "loss": 1.0792,
      "step": 20695
    },
    {
      "epoch": 0.88,
      "grad_norm": 15.31334068495149,
      "learning_rate": 8.918253157822762e-06,
      "loss": 1.0739,
      "step": 20700
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.938565593692502,
      "learning_rate": 8.917488260059231e-06,
      "loss": 1.0941,
      "step": 20705
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.405551159369352,
      "learning_rate": 8.916723124789925e-06,
      "loss": 1.0762,
      "step": 20710
    },
    {
      "epoch": 0.88,
      "grad_norm": 19.204296672551017,
      "learning_rate": 8.915957752061233e-06,
      "loss": 1.1,
      "step": 20715
    },
    {
      "epoch": 0.88,
      "grad_norm": 14.21333323378113,
      "learning_rate": 8.91519214191956e-06,
      "loss": 1.0697,
      "step": 20720
    },
    {
      "epoch": 0.88,
      "grad_norm": 18.83871330536022,
      "learning_rate": 8.914426294411317e-06,
      "loss": 1.0631,
      "step": 20725
    },
    {
      "epoch": 0.88,
      "grad_norm": 25.934339745583106,
      "learning_rate": 8.913660209582939e-06,
      "loss": 1.0939,
      "step": 20730
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.257537411450811,
      "learning_rate": 8.912893887480871e-06,
      "loss": 1.041,
      "step": 20735
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.142897804106785,
      "learning_rate": 8.912127328151571e-06,
      "loss": 1.0361,
      "step": 20740
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.132320895741621,
      "learning_rate": 8.911360531641518e-06,
      "loss": 1.062,
      "step": 20745
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.179315535499878,
      "learning_rate": 8.910593497997194e-06,
      "loss": 1.0799,
      "step": 20750
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.187907941619817,
      "learning_rate": 8.909826227265106e-06,
      "loss": 1.0989,
      "step": 20755
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.177659043854928,
      "learning_rate": 8.90905871949177e-06,
      "loss": 1.06,
      "step": 20760
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.5550911277253405,
      "learning_rate": 8.908290974723721e-06,
      "loss": 1.0565,
      "step": 20765
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.946757083052128,
      "learning_rate": 8.907522993007501e-06,
      "loss": 1.0353,
      "step": 20770
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.52549635103769,
      "learning_rate": 8.90675477438967e-06,
      "loss": 1.0856,
      "step": 20775
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.454723011371246,
      "learning_rate": 8.905986318916806e-06,
      "loss": 1.0717,
      "step": 20780
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.053309484669723,
      "learning_rate": 8.905217626635498e-06,
      "loss": 1.0865,
      "step": 20785
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.761567338458246,
      "learning_rate": 8.904448697592348e-06,
      "loss": 1.0641,
      "step": 20790
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.520055402460125,
      "learning_rate": 8.903679531833975e-06,
      "loss": 1.0307,
      "step": 20795
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.033838696527171,
      "learning_rate": 8.90291012940701e-06,
      "loss": 1.0553,
      "step": 20800
    },
    {
      "epoch": 0.88,
      "grad_norm": 15.765954286507517,
      "learning_rate": 8.9021404903581e-06,
      "loss": 1.1015,
      "step": 20805
    },
    {
      "epoch": 0.88,
      "grad_norm": 19.209895388757598,
      "learning_rate": 8.901370614733906e-06,
      "loss": 1.0945,
      "step": 20810
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.406341842835518,
      "learning_rate": 8.900600502581103e-06,
      "loss": 1.0553,
      "step": 20815
    },
    {
      "epoch": 0.88,
      "grad_norm": 16.402156409278557,
      "learning_rate": 8.899830153946383e-06,
      "loss": 1.0928,
      "step": 20820
    },
    {
      "epoch": 0.88,
      "grad_norm": 26.428071666534358,
      "learning_rate": 8.899059568876447e-06,
      "loss": 1.0769,
      "step": 20825
    },
    {
      "epoch": 0.88,
      "grad_norm": 15.63819326871164,
      "learning_rate": 8.898288747418013e-06,
      "loss": 1.0898,
      "step": 20830
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.303785869266925,
      "learning_rate": 8.897517689617816e-06,
      "loss": 1.0838,
      "step": 20835
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.821092322354956,
      "learning_rate": 8.896746395522604e-06,
      "loss": 1.0604,
      "step": 20840
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.341660470921059,
      "learning_rate": 8.895974865179134e-06,
      "loss": 1.0286,
      "step": 20845
    },
    {
      "epoch": 0.88,
      "grad_norm": 10.596173663255243,
      "learning_rate": 8.895203098634185e-06,
      "loss": 1.0588,
      "step": 20850
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.366538384175701,
      "learning_rate": 8.894431095934546e-06,
      "loss": 1.0295,
      "step": 20855
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.623112085436764,
      "learning_rate": 8.89365885712702e-06,
      "loss": 1.0751,
      "step": 20860
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.643617595513142,
      "learning_rate": 8.892886382258428e-06,
      "loss": 1.0702,
      "step": 20865
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.66271425253117,
      "learning_rate": 8.8921136713756e-06,
      "loss": 1.0499,
      "step": 20870
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.159326185861467,
      "learning_rate": 8.891340724525386e-06,
      "loss": 1.0308,
      "step": 20875
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.833823525018116,
      "learning_rate": 8.890567541754644e-06,
      "loss": 1.025,
      "step": 20880
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.745727709978524,
      "learning_rate": 8.889794123110255e-06,
      "loss": 1.0511,
      "step": 20885
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.938843311877473,
      "learning_rate": 8.889020468639105e-06,
      "loss": 1.0931,
      "step": 20890
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.913676987686792,
      "learning_rate": 8.888246578388098e-06,
      "loss": 1.0751,
      "step": 20895
    },
    {
      "epoch": 0.88,
      "grad_norm": 22.48758205215652,
      "learning_rate": 8.887472452404155e-06,
      "loss": 1.0831,
      "step": 20900
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.087265746936609,
      "learning_rate": 8.886698090734209e-06,
      "loss": 1.0422,
      "step": 20905
    },
    {
      "epoch": 0.88,
      "grad_norm": 12.119597605665433,
      "learning_rate": 8.885923493425207e-06,
      "loss": 1.0428,
      "step": 20910
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.421824227753902,
      "learning_rate": 8.885148660524107e-06,
      "loss": 1.0522,
      "step": 20915
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.855683202356163,
      "learning_rate": 8.88437359207789e-06,
      "loss": 1.0706,
      "step": 20920
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.3552971430271,
      "learning_rate": 8.883598288133543e-06,
      "loss": 1.023,
      "step": 20925
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.112546080983584,
      "learning_rate": 8.882822748738073e-06,
      "loss": 1.0557,
      "step": 20930
    },
    {
      "epoch": 0.89,
      "grad_norm": 20.83501223995967,
      "learning_rate": 8.882046973938495e-06,
      "loss": 1.0992,
      "step": 20935
    },
    {
      "epoch": 0.89,
      "grad_norm": 15.898196183156685,
      "learning_rate": 8.881270963781845e-06,
      "loss": 1.067,
      "step": 20940
    },
    {
      "epoch": 0.89,
      "grad_norm": 20.368831475959936,
      "learning_rate": 8.880494718315169e-06,
      "loss": 1.0645,
      "step": 20945
    },
    {
      "epoch": 0.89,
      "grad_norm": 14.840503456194964,
      "learning_rate": 8.879718237585528e-06,
      "loss": 1.0425,
      "step": 20950
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.94553073434144,
      "learning_rate": 8.878941521639999e-06,
      "loss": 1.0555,
      "step": 20955
    },
    {
      "epoch": 0.89,
      "grad_norm": 21.304225090622737,
      "learning_rate": 8.87816457052567e-06,
      "loss": 1.0813,
      "step": 20960
    },
    {
      "epoch": 0.89,
      "grad_norm": 17.063780005158844,
      "learning_rate": 8.877387384289648e-06,
      "loss": 1.0206,
      "step": 20965
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.71493471733423,
      "learning_rate": 8.876609962979051e-06,
      "loss": 1.0705,
      "step": 20970
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.457989573601514,
      "learning_rate": 8.875832306641007e-06,
      "loss": 1.0568,
      "step": 20975
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.199547787306146,
      "learning_rate": 8.875054415322672e-06,
      "loss": 1.069,
      "step": 20980
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.397217092721782,
      "learning_rate": 8.874276289071198e-06,
      "loss": 1.0304,
      "step": 20985
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.190563130924475,
      "learning_rate": 8.873497927933768e-06,
      "loss": 1.1021,
      "step": 20990
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.791445325662123,
      "learning_rate": 8.872719331957566e-06,
      "loss": 1.0497,
      "step": 20995
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.97302100044505,
      "learning_rate": 8.871940501189798e-06,
      "loss": 1.0483,
      "step": 21000
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.535502584319712,
      "learning_rate": 8.871161435677684e-06,
      "loss": 1.0322,
      "step": 21005
    },
    {
      "epoch": 0.89,
      "grad_norm": 35.578972990516014,
      "learning_rate": 8.870382135468456e-06,
      "loss": 1.1031,
      "step": 21010
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.2327016161187885,
      "learning_rate": 8.869602600609358e-06,
      "loss": 1.0906,
      "step": 21015
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.810473207120032,
      "learning_rate": 8.868822831147651e-06,
      "loss": 1.0523,
      "step": 21020
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.844436916856838,
      "learning_rate": 8.868042827130616e-06,
      "loss": 1.0779,
      "step": 21025
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.218708977326758,
      "learning_rate": 8.867262588605536e-06,
      "loss": 1.0639,
      "step": 21030
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.533044933174893,
      "learning_rate": 8.866482115619717e-06,
      "loss": 1.0608,
      "step": 21035
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.454596487583646,
      "learning_rate": 8.865701408220475e-06,
      "loss": 1.0782,
      "step": 21040
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.99224609192941,
      "learning_rate": 8.864920466455145e-06,
      "loss": 1.0687,
      "step": 21045
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.927393602973079,
      "learning_rate": 8.86413929037107e-06,
      "loss": 1.1232,
      "step": 21050
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.648906703651733,
      "learning_rate": 8.863357880015614e-06,
      "loss": 1.0797,
      "step": 21055
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.729621479341857,
      "learning_rate": 8.862576235436148e-06,
      "loss": 1.0756,
      "step": 21060
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.686381069102199,
      "learning_rate": 8.861794356680062e-06,
      "loss": 1.0806,
      "step": 21065
    },
    {
      "epoch": 0.89,
      "grad_norm": 19.142315008884896,
      "learning_rate": 8.861012243794763e-06,
      "loss": 1.053,
      "step": 21070
    },
    {
      "epoch": 0.89,
      "grad_norm": 21.590932178107565,
      "learning_rate": 8.86022989682766e-06,
      "loss": 1.069,
      "step": 21075
    },
    {
      "epoch": 0.89,
      "grad_norm": 15.617893747057055,
      "learning_rate": 8.859447315826193e-06,
      "loss": 1.0569,
      "step": 21080
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.787898382927495,
      "learning_rate": 8.858664500837802e-06,
      "loss": 1.0509,
      "step": 21085
    },
    {
      "epoch": 0.89,
      "grad_norm": 11.134315365923271,
      "learning_rate": 8.857881451909946e-06,
      "loss": 1.0495,
      "step": 21090
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.63660454211466,
      "learning_rate": 8.857098169090106e-06,
      "loss": 1.0577,
      "step": 21095
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.141387720627309,
      "learning_rate": 8.85631465242576e-06,
      "loss": 1.0862,
      "step": 21100
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.027946337922725,
      "learning_rate": 8.85553090196442e-06,
      "loss": 0.993,
      "step": 21105
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.698045784393106,
      "learning_rate": 8.854746917753598e-06,
      "loss": 1.0491,
      "step": 21110
    },
    {
      "epoch": 0.89,
      "grad_norm": 14.001669152120114,
      "learning_rate": 8.853962699840823e-06,
      "loss": 1.0904,
      "step": 21115
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.743278108038464,
      "learning_rate": 8.853178248273643e-06,
      "loss": 1.033,
      "step": 21120
    },
    {
      "epoch": 0.89,
      "grad_norm": 4.834385892911956,
      "learning_rate": 8.852393563099615e-06,
      "loss": 1.0709,
      "step": 21125
    },
    {
      "epoch": 0.89,
      "grad_norm": 13.304727341112551,
      "learning_rate": 8.851608644366313e-06,
      "loss": 1.0335,
      "step": 21130
    },
    {
      "epoch": 0.89,
      "grad_norm": 55.55135213246576,
      "learning_rate": 8.850823492121326e-06,
      "loss": 1.0712,
      "step": 21135
    },
    {
      "epoch": 0.89,
      "grad_norm": 21.269407661946715,
      "learning_rate": 8.850038106412252e-06,
      "loss": 1.0979,
      "step": 21140
    },
    {
      "epoch": 0.89,
      "grad_norm": 12.780294573712636,
      "learning_rate": 8.849252487286707e-06,
      "loss": 1.0481,
      "step": 21145
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.307682314681879,
      "learning_rate": 8.848466634792324e-06,
      "loss": 1.0805,
      "step": 21150
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.452304683137468,
      "learning_rate": 8.847680548976744e-06,
      "loss": 1.0428,
      "step": 21155
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.633935619732695,
      "learning_rate": 8.846894229887625e-06,
      "loss": 1.033,
      "step": 21160
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.035233577059684,
      "learning_rate": 8.846107677572643e-06,
      "loss": 1.0812,
      "step": 21165
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.810927330439723,
      "learning_rate": 8.84532089207948e-06,
      "loss": 1.1059,
      "step": 21170
    },
    {
      "epoch": 0.9,
      "grad_norm": 12.457373964066747,
      "learning_rate": 8.844533873455839e-06,
      "loss": 1.0532,
      "step": 21175
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.29557635870949,
      "learning_rate": 8.843746621749431e-06,
      "loss": 1.0529,
      "step": 21180
    },
    {
      "epoch": 0.9,
      "grad_norm": 18.903483485020317,
      "learning_rate": 8.84295913700799e-06,
      "loss": 1.1045,
      "step": 21185
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.697674756133697,
      "learning_rate": 8.842171419279254e-06,
      "loss": 1.0873,
      "step": 21190
    },
    {
      "epoch": 0.9,
      "grad_norm": 11.637893625317709,
      "learning_rate": 8.841383468610986e-06,
      "loss": 1.0582,
      "step": 21195
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.852088128804523,
      "learning_rate": 8.840595285050952e-06,
      "loss": 1.0517,
      "step": 21200
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.642060777790125,
      "learning_rate": 8.839806868646939e-06,
      "loss": 1.0442,
      "step": 21205
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.771363577990281,
      "learning_rate": 8.839018219446745e-06,
      "loss": 1.0318,
      "step": 21210
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.191485223243958,
      "learning_rate": 8.838229337498184e-06,
      "loss": 1.0469,
      "step": 21215
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.381567428825308,
      "learning_rate": 8.837440222849085e-06,
      "loss": 1.1077,
      "step": 21220
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.872507242014159,
      "learning_rate": 8.83665087554729e-06,
      "loss": 1.0596,
      "step": 21225
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.729158805788173,
      "learning_rate": 8.835861295640651e-06,
      "loss": 1.0328,
      "step": 21230
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.881584941590122,
      "learning_rate": 8.835071483177043e-06,
      "loss": 1.0925,
      "step": 21235
    },
    {
      "epoch": 0.9,
      "grad_norm": 20.221204981328434,
      "learning_rate": 8.834281438204346e-06,
      "loss": 1.0432,
      "step": 21240
    },
    {
      "epoch": 0.9,
      "grad_norm": 25.58926523591417,
      "learning_rate": 8.83349116077046e-06,
      "loss": 1.1169,
      "step": 21245
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.861630965217138,
      "learning_rate": 8.832700650923298e-06,
      "loss": 1.022,
      "step": 21250
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.387333370723109,
      "learning_rate": 8.831909908710783e-06,
      "loss": 1.0343,
      "step": 21255
    },
    {
      "epoch": 0.9,
      "grad_norm": 12.849297724015658,
      "learning_rate": 8.83111893418086e-06,
      "loss": 1.0543,
      "step": 21260
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.427884669005525,
      "learning_rate": 8.83032772738148e-06,
      "loss": 1.0992,
      "step": 21265
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.618057487027922,
      "learning_rate": 8.829536288360611e-06,
      "loss": 1.0839,
      "step": 21270
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.942745729346651,
      "learning_rate": 8.828744617166238e-06,
      "loss": 1.0663,
      "step": 21275
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.342060660292795,
      "learning_rate": 8.827952713846357e-06,
      "loss": 1.0415,
      "step": 21280
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.8713240571949274,
      "learning_rate": 8.827160578448979e-06,
      "loss": 1.0678,
      "step": 21285
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.684412898819534,
      "learning_rate": 8.826368211022128e-06,
      "loss": 1.0481,
      "step": 21290
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.399008127084763,
      "learning_rate": 8.825575611613843e-06,
      "loss": 1.0225,
      "step": 21295
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.377956228972538,
      "learning_rate": 8.824782780272178e-06,
      "loss": 1.1008,
      "step": 21300
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.2756897674620165,
      "learning_rate": 8.823989717045198e-06,
      "loss": 1.0509,
      "step": 21305
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.507591575698728,
      "learning_rate": 8.823196421980987e-06,
      "loss": 1.0821,
      "step": 21310
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.85916782231805,
      "learning_rate": 8.822402895127638e-06,
      "loss": 1.0823,
      "step": 21315
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.744323366967096,
      "learning_rate": 8.82160913653326e-06,
      "loss": 1.0612,
      "step": 21320
    },
    {
      "epoch": 0.9,
      "grad_norm": 16.969147456939872,
      "learning_rate": 8.820815146245978e-06,
      "loss": 1.0899,
      "step": 21325
    },
    {
      "epoch": 0.9,
      "grad_norm": 15.716666278893374,
      "learning_rate": 8.820020924313928e-06,
      "loss": 1.0519,
      "step": 21330
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.511279862971171,
      "learning_rate": 8.819226470785262e-06,
      "loss": 1.0858,
      "step": 21335
    },
    {
      "epoch": 0.9,
      "grad_norm": 10.637362562293369,
      "learning_rate": 8.818431785708143e-06,
      "loss": 1.0587,
      "step": 21340
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.951445253670879,
      "learning_rate": 8.817636869130755e-06,
      "loss": 1.0846,
      "step": 21345
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.889855922987261,
      "learning_rate": 8.816841721101287e-06,
      "loss": 1.0456,
      "step": 21350
    },
    {
      "epoch": 0.9,
      "grad_norm": 8.067959870935121,
      "learning_rate": 8.816046341667948e-06,
      "loss": 1.0435,
      "step": 21355
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.687627150407593,
      "learning_rate": 8.815250730878962e-06,
      "loss": 1.0366,
      "step": 21360
    },
    {
      "epoch": 0.9,
      "grad_norm": 18.289593121532008,
      "learning_rate": 8.81445488878256e-06,
      "loss": 1.0521,
      "step": 21365
    },
    {
      "epoch": 0.9,
      "grad_norm": 17.720960834379003,
      "learning_rate": 8.813658815426995e-06,
      "loss": 1.0795,
      "step": 21370
    },
    {
      "epoch": 0.9,
      "grad_norm": 13.44152781084223,
      "learning_rate": 8.81286251086053e-06,
      "loss": 1.0637,
      "step": 21375
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.492694841534716,
      "learning_rate": 8.812065975131442e-06,
      "loss": 1.0366,
      "step": 21380
    },
    {
      "epoch": 0.91,
      "grad_norm": 23.36785605756628,
      "learning_rate": 8.811269208288022e-06,
      "loss": 1.0937,
      "step": 21385
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.642373998336643,
      "learning_rate": 8.810472210378578e-06,
      "loss": 1.0141,
      "step": 21390
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.094154983879217,
      "learning_rate": 8.809674981451427e-06,
      "loss": 1.0569,
      "step": 21395
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.460687961498408,
      "learning_rate": 8.808877521554901e-06,
      "loss": 1.0272,
      "step": 21400
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.8291843082098564,
      "learning_rate": 8.808079830737354e-06,
      "loss": 1.041,
      "step": 21405
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.623937533126616,
      "learning_rate": 8.807281909047142e-06,
      "loss": 0.9993,
      "step": 21410
    },
    {
      "epoch": 0.91,
      "grad_norm": 15.794917263344798,
      "learning_rate": 8.806483756532645e-06,
      "loss": 1.0967,
      "step": 21415
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.515565518273104,
      "learning_rate": 8.805685373242248e-06,
      "loss": 1.0765,
      "step": 21420
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.285868561488702,
      "learning_rate": 8.804886759224356e-06,
      "loss": 1.0596,
      "step": 21425
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.851472600076748,
      "learning_rate": 8.804087914527389e-06,
      "loss": 1.0179,
      "step": 21430
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.729506306413892,
      "learning_rate": 8.803288839199778e-06,
      "loss": 1.0859,
      "step": 21435
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.726209158536673,
      "learning_rate": 8.802489533289966e-06,
      "loss": 1.0451,
      "step": 21440
    },
    {
      "epoch": 0.91,
      "grad_norm": 29.801373754501615,
      "learning_rate": 8.801689996846413e-06,
      "loss": 1.0383,
      "step": 21445
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.899255609396299,
      "learning_rate": 8.800890229917597e-06,
      "loss": 1.0664,
      "step": 21450
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.096716507522391,
      "learning_rate": 8.800090232552e-06,
      "loss": 1.0756,
      "step": 21455
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.469882665770605,
      "learning_rate": 8.799290004798126e-06,
      "loss": 1.101,
      "step": 21460
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.940804734570921,
      "learning_rate": 8.79848954670449e-06,
      "loss": 1.0801,
      "step": 21465
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.168900541435045,
      "learning_rate": 8.797688858319623e-06,
      "loss": 1.0362,
      "step": 21470
    },
    {
      "epoch": 0.91,
      "grad_norm": 12.83710785361824,
      "learning_rate": 8.796887939692067e-06,
      "loss": 1.0336,
      "step": 21475
    },
    {
      "epoch": 0.91,
      "grad_norm": 27.06704901373965,
      "learning_rate": 8.796086790870377e-06,
      "loss": 0.9858,
      "step": 21480
    },
    {
      "epoch": 0.91,
      "grad_norm": 28.989457394553632,
      "learning_rate": 8.795285411903128e-06,
      "loss": 1.032,
      "step": 21485
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.901257782767786,
      "learning_rate": 8.794483802838905e-06,
      "loss": 1.076,
      "step": 21490
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.684125418272892,
      "learning_rate": 8.793681963726305e-06,
      "loss": 1.067,
      "step": 21495
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.352511766000096,
      "learning_rate": 8.792879894613942e-06,
      "loss": 1.1057,
      "step": 21500
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.871832773556916,
      "learning_rate": 8.792077595550443e-06,
      "loss": 1.0517,
      "step": 21505
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.278087552393924,
      "learning_rate": 8.791275066584448e-06,
      "loss": 1.0908,
      "step": 21510
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.343478919696157,
      "learning_rate": 8.790472307764615e-06,
      "loss": 1.0214,
      "step": 21515
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.111609858482743,
      "learning_rate": 8.789669319139612e-06,
      "loss": 1.0369,
      "step": 21520
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.406706354094906,
      "learning_rate": 8.78886610075812e-06,
      "loss": 1.01,
      "step": 21525
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.335915396451932,
      "learning_rate": 8.788062652668838e-06,
      "loss": 1.0848,
      "step": 21530
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.420963162954823,
      "learning_rate": 8.787258974920473e-06,
      "loss": 1.1064,
      "step": 21535
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.615029492454818,
      "learning_rate": 8.786455067561755e-06,
      "loss": 1.0473,
      "step": 21540
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.365865015705364,
      "learning_rate": 8.78565093064142e-06,
      "loss": 1.0412,
      "step": 21545
    },
    {
      "epoch": 0.91,
      "grad_norm": 17.57346314875629,
      "learning_rate": 8.784846564208216e-06,
      "loss": 1.0716,
      "step": 21550
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.0366668798692755,
      "learning_rate": 8.784041968310916e-06,
      "loss": 1.0769,
      "step": 21555
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.139804577220695,
      "learning_rate": 8.783237142998298e-06,
      "loss": 1.0266,
      "step": 21560
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.52747414563638,
      "learning_rate": 8.782432088319155e-06,
      "loss": 1.0808,
      "step": 21565
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.587613788154729,
      "learning_rate": 8.781626804322297e-06,
      "loss": 1.0394,
      "step": 21570
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.6715182651851785,
      "learning_rate": 8.780821291056543e-06,
      "loss": 1.041,
      "step": 21575
    },
    {
      "epoch": 0.91,
      "grad_norm": 4.674414577419238,
      "learning_rate": 8.780015548570732e-06,
      "loss": 1.0008,
      "step": 21580
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.574997036691096,
      "learning_rate": 8.779209576913714e-06,
      "loss": 1.0319,
      "step": 21585
    },
    {
      "epoch": 0.91,
      "grad_norm": 16.10578121438595,
      "learning_rate": 8.77840337613435e-06,
      "loss": 1.0339,
      "step": 21590
    },
    {
      "epoch": 0.91,
      "grad_norm": 16.73778871115213,
      "learning_rate": 8.777596946281518e-06,
      "loss": 1.0627,
      "step": 21595
    },
    {
      "epoch": 0.91,
      "grad_norm": 14.389316835856235,
      "learning_rate": 8.776790287404111e-06,
      "loss": 1.0496,
      "step": 21600
    },
    {
      "epoch": 0.91,
      "grad_norm": 12.491064468821286,
      "learning_rate": 8.775983399551036e-06,
      "loss": 1.0559,
      "step": 21605
    },
    {
      "epoch": 0.91,
      "grad_norm": 15.80902430708104,
      "learning_rate": 8.775176282771209e-06,
      "loss": 1.0155,
      "step": 21610
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.154023048005005,
      "learning_rate": 8.774368937113562e-06,
      "loss": 1.0482,
      "step": 21615
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.3376485078668905,
      "learning_rate": 8.773561362627046e-06,
      "loss": 1.0403,
      "step": 21620
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.685281741953801,
      "learning_rate": 8.77275355936062e-06,
      "loss": 1.0422,
      "step": 21625
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.784705236886882,
      "learning_rate": 8.771945527363258e-06,
      "loss": 1.0651,
      "step": 21630
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.468583235877373,
      "learning_rate": 8.77113726668395e-06,
      "loss": 1.0873,
      "step": 21635
    },
    {
      "epoch": 0.92,
      "grad_norm": 13.840071593082772,
      "learning_rate": 8.770328777371697e-06,
      "loss": 1.1091,
      "step": 21640
    },
    {
      "epoch": 0.92,
      "grad_norm": 23.321627193632175,
      "learning_rate": 8.769520059475517e-06,
      "loss": 1.0721,
      "step": 21645
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.336669417801813,
      "learning_rate": 8.768711113044439e-06,
      "loss": 1.0281,
      "step": 21650
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.04609472650862,
      "learning_rate": 8.767901938127508e-06,
      "loss": 1.0611,
      "step": 21655
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.885480966822543,
      "learning_rate": 8.76709253477378e-06,
      "loss": 1.0695,
      "step": 21660
    },
    {
      "epoch": 0.92,
      "grad_norm": 18.06587988887282,
      "learning_rate": 8.766282903032328e-06,
      "loss": 1.0506,
      "step": 21665
    },
    {
      "epoch": 0.92,
      "grad_norm": 16.495498752347274,
      "learning_rate": 8.765473042952237e-06,
      "loss": 1.0431,
      "step": 21670
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.47121866701029,
      "learning_rate": 8.764662954582608e-06,
      "loss": 1.0271,
      "step": 21675
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.207834602336261,
      "learning_rate": 8.763852637972552e-06,
      "loss": 1.0483,
      "step": 21680
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.929564142189032,
      "learning_rate": 8.763042093171199e-06,
      "loss": 1.0721,
      "step": 21685
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.444481860338322,
      "learning_rate": 8.762231320227687e-06,
      "loss": 1.0309,
      "step": 21690
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.929177762962361,
      "learning_rate": 8.761420319191175e-06,
      "loss": 1.0396,
      "step": 21695
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.111111542085034,
      "learning_rate": 8.760609090110826e-06,
      "loss": 1.0546,
      "step": 21700
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.70990168497997,
      "learning_rate": 8.759797633035825e-06,
      "loss": 1.0612,
      "step": 21705
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.59013134631506,
      "learning_rate": 8.758985948015367e-06,
      "loss": 1.0218,
      "step": 21710
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.552035113880665,
      "learning_rate": 8.758174035098668e-06,
      "loss": 1.0847,
      "step": 21715
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.140368859277036,
      "learning_rate": 8.757361894334943e-06,
      "loss": 1.0572,
      "step": 21720
    },
    {
      "epoch": 0.92,
      "grad_norm": 57.63361491524216,
      "learning_rate": 8.756549525773438e-06,
      "loss": 1.0483,
      "step": 21725
    },
    {
      "epoch": 0.92,
      "grad_norm": 47.2841645970153,
      "learning_rate": 8.755736929463397e-06,
      "loss": 1.0694,
      "step": 21730
    },
    {
      "epoch": 0.92,
      "grad_norm": 15.940659935050457,
      "learning_rate": 8.75492410545409e-06,
      "loss": 1.0751,
      "step": 21735
    },
    {
      "epoch": 0.92,
      "grad_norm": 11.61599805270187,
      "learning_rate": 8.754111053794795e-06,
      "loss": 1.0377,
      "step": 21740
    },
    {
      "epoch": 0.92,
      "grad_norm": 22.077601283259092,
      "learning_rate": 8.753297774534804e-06,
      "loss": 1.1211,
      "step": 21745
    },
    {
      "epoch": 0.92,
      "grad_norm": 33.27846151912245,
      "learning_rate": 8.752484267723427e-06,
      "loss": 1.012,
      "step": 21750
    },
    {
      "epoch": 0.92,
      "grad_norm": 32.92893552543508,
      "learning_rate": 8.75167053340998e-06,
      "loss": 1.0694,
      "step": 21755
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.979473432915224,
      "learning_rate": 8.7508565716438e-06,
      "loss": 1.0257,
      "step": 21760
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.061015184293739,
      "learning_rate": 8.750042382474235e-06,
      "loss": 1.0365,
      "step": 21765
    },
    {
      "epoch": 0.92,
      "grad_norm": 19.22056404639587,
      "learning_rate": 8.749227965950647e-06,
      "loss": 1.0847,
      "step": 21770
    },
    {
      "epoch": 0.92,
      "grad_norm": 44.50404057302526,
      "learning_rate": 8.74841332212241e-06,
      "loss": 1.071,
      "step": 21775
    },
    {
      "epoch": 0.92,
      "grad_norm": 21.410855373385008,
      "learning_rate": 8.747598451038914e-06,
      "loss": 1.0629,
      "step": 21780
    },
    {
      "epoch": 0.92,
      "grad_norm": 9.127279653696853,
      "learning_rate": 8.746783352749564e-06,
      "loss": 1.0882,
      "step": 21785
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.852493897897254,
      "learning_rate": 8.745968027303775e-06,
      "loss": 1.0656,
      "step": 21790
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.363963186886291,
      "learning_rate": 8.74515247475098e-06,
      "loss": 1.0656,
      "step": 21795
    },
    {
      "epoch": 0.92,
      "grad_norm": 22.089935622392925,
      "learning_rate": 8.744336695140621e-06,
      "loss": 1.0812,
      "step": 21800
    },
    {
      "epoch": 0.92,
      "grad_norm": 19.19163037925424,
      "learning_rate": 8.743520688522158e-06,
      "loss": 1.0956,
      "step": 21805
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.047441321112834,
      "learning_rate": 8.742704454945063e-06,
      "loss": 1.0352,
      "step": 21810
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.062421142821726,
      "learning_rate": 8.74188799445882e-06,
      "loss": 1.0091,
      "step": 21815
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.495731062251956,
      "learning_rate": 8.741071307112931e-06,
      "loss": 1.0263,
      "step": 21820
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.042633889615112,
      "learning_rate": 8.740254392956908e-06,
      "loss": 1.0816,
      "step": 21825
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.076738491993376,
      "learning_rate": 8.739437252040279e-06,
      "loss": 1.0315,
      "step": 21830
    },
    {
      "epoch": 0.92,
      "grad_norm": 10.015667998514347,
      "learning_rate": 8.738619884412584e-06,
      "loss": 1.0495,
      "step": 21835
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.3949257086583255,
      "learning_rate": 8.737802290123377e-06,
      "loss": 1.042,
      "step": 21840
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.764006044340583,
      "learning_rate": 8.736984469222228e-06,
      "loss": 1.0415,
      "step": 21845
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.09804184512544,
      "learning_rate": 8.736166421758719e-06,
      "loss": 1.0495,
      "step": 21850
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.974383652800656,
      "learning_rate": 8.735348147782444e-06,
      "loss": 1.053,
      "step": 21855
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.009311942839261,
      "learning_rate": 8.734529647343014e-06,
      "loss": 1.0612,
      "step": 21860
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.121323101727387,
      "learning_rate": 8.733710920490051e-06,
      "loss": 1.076,
      "step": 21865
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.79092212251893,
      "learning_rate": 8.732891967273194e-06,
      "loss": 1.0497,
      "step": 21870
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.935487296396415,
      "learning_rate": 8.732072787742092e-06,
      "loss": 1.0392,
      "step": 21875
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.793977192089978,
      "learning_rate": 8.731253381946411e-06,
      "loss": 1.049,
      "step": 21880
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.48089571565006,
      "learning_rate": 8.730433749935826e-06,
      "loss": 1.0616,
      "step": 21885
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.745497063889569,
      "learning_rate": 8.729613891760033e-06,
      "loss": 1.0816,
      "step": 21890
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.890019287663492,
      "learning_rate": 8.728793807468737e-06,
      "loss": 1.0768,
      "step": 21895
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.265183064840851,
      "learning_rate": 8.727973497111654e-06,
      "loss": 1.0326,
      "step": 21900
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.815145211579305,
      "learning_rate": 8.72715296073852e-06,
      "loss": 1.0755,
      "step": 21905
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.82723645844355,
      "learning_rate": 8.72633219839908e-06,
      "loss": 1.0264,
      "step": 21910
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.72244192439977,
      "learning_rate": 8.725511210143096e-06,
      "loss": 1.0188,
      "step": 21915
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.078545180241331,
      "learning_rate": 8.724689996020338e-06,
      "loss": 1.0749,
      "step": 21920
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.70268875774656,
      "learning_rate": 8.7238685560806e-06,
      "loss": 1.0633,
      "step": 21925
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.593572157882693,
      "learning_rate": 8.72304689037368e-06,
      "loss": 1.0389,
      "step": 21930
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.637762503754141,
      "learning_rate": 8.722224998949394e-06,
      "loss": 1.0724,
      "step": 21935
    },
    {
      "epoch": 0.93,
      "grad_norm": 17.670081107075355,
      "learning_rate": 8.72140288185757e-06,
      "loss": 1.0172,
      "step": 21940
    },
    {
      "epoch": 0.93,
      "grad_norm": 35.55222885560427,
      "learning_rate": 8.720580539148051e-06,
      "loss": 1.0569,
      "step": 21945
    },
    {
      "epoch": 0.93,
      "grad_norm": 19.40025731407459,
      "learning_rate": 8.719757970870692e-06,
      "loss": 1.0306,
      "step": 21950
    },
    {
      "epoch": 0.93,
      "grad_norm": 15.076429975283755,
      "learning_rate": 8.718935177075364e-06,
      "loss": 1.0261,
      "step": 21955
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.542991044235364,
      "learning_rate": 8.718112157811953e-06,
      "loss": 1.069,
      "step": 21960
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.838222235628757,
      "learning_rate": 8.717288913130353e-06,
      "loss": 1.0041,
      "step": 21965
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.170399537129438,
      "learning_rate": 8.716465443080476e-06,
      "loss": 1.075,
      "step": 21970
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.4674115154313,
      "learning_rate": 8.715641747712246e-06,
      "loss": 1.0142,
      "step": 21975
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.669374867755228,
      "learning_rate": 8.714817827075602e-06,
      "loss": 1.0923,
      "step": 21980
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.820976552613582,
      "learning_rate": 8.713993681220496e-06,
      "loss": 1.0408,
      "step": 21985
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.140315928607075,
      "learning_rate": 8.713169310196893e-06,
      "loss": 1.0784,
      "step": 21990
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.374115522248512,
      "learning_rate": 8.71234471405477e-06,
      "loss": 1.0612,
      "step": 21995
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.971696584577922,
      "learning_rate": 8.711519892844124e-06,
      "loss": 1.0208,
      "step": 22000
    },
    {
      "epoch": 0.93,
      "grad_norm": 13.450381433773453,
      "learning_rate": 8.710694846614958e-06,
      "loss": 1.028,
      "step": 22005
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.070905848438207,
      "learning_rate": 8.709869575417296e-06,
      "loss": 1.0564,
      "step": 22010
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.282452083221917,
      "learning_rate": 8.709044079301167e-06,
      "loss": 1.0273,
      "step": 22015
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.6394521627122,
      "learning_rate": 8.708218358316622e-06,
      "loss": 1.0433,
      "step": 22020
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.823966610599468,
      "learning_rate": 8.707392412513723e-06,
      "loss": 1.1086,
      "step": 22025
    },
    {
      "epoch": 0.93,
      "grad_norm": 17.6737234213204,
      "learning_rate": 8.706566241942538e-06,
      "loss": 1.0758,
      "step": 22030
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.96099275751703,
      "learning_rate": 8.705739846653163e-06,
      "loss": 1.0564,
      "step": 22035
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.396464325059147,
      "learning_rate": 8.704913226695694e-06,
      "loss": 1.0872,
      "step": 22040
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.155400502711098,
      "learning_rate": 8.704086382120253e-06,
      "loss": 1.0688,
      "step": 22045
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.276094930929235,
      "learning_rate": 8.703259312976964e-06,
      "loss": 0.9951,
      "step": 22050
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.959162644711834,
      "learning_rate": 8.702432019315969e-06,
      "loss": 1.0726,
      "step": 22055
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.8817955997993,
      "learning_rate": 8.701604501187429e-06,
      "loss": 1.0269,
      "step": 22060
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.704106881361415,
      "learning_rate": 8.700776758641511e-06,
      "loss": 1.0408,
      "step": 22065
    },
    {
      "epoch": 0.93,
      "grad_norm": 9.948920096532369,
      "learning_rate": 8.699948791728398e-06,
      "loss": 1.0354,
      "step": 22070
    },
    {
      "epoch": 0.93,
      "grad_norm": 20.575258315768465,
      "learning_rate": 8.69912060049829e-06,
      "loss": 1.083,
      "step": 22075
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.109688586745168,
      "learning_rate": 8.698292185001395e-06,
      "loss": 1.0532,
      "step": 22080
    },
    {
      "epoch": 0.93,
      "grad_norm": 14.700664626707857,
      "learning_rate": 8.697463545287939e-06,
      "loss": 1.0509,
      "step": 22085
    },
    {
      "epoch": 0.93,
      "grad_norm": 31.00422435019389,
      "learning_rate": 8.696634681408159e-06,
      "loss": 1.0516,
      "step": 22090
    },
    {
      "epoch": 0.94,
      "grad_norm": 30.538540018804785,
      "learning_rate": 8.695805593412307e-06,
      "loss": 1.035,
      "step": 22095
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.582995982042233,
      "learning_rate": 8.694976281350648e-06,
      "loss": 0.9977,
      "step": 22100
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.216779362198505,
      "learning_rate": 8.694146745273461e-06,
      "loss": 1.0347,
      "step": 22105
    },
    {
      "epoch": 0.94,
      "grad_norm": 43.73094129805623,
      "learning_rate": 8.693316985231039e-06,
      "loss": 1.0219,
      "step": 22110
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.16376083750654,
      "learning_rate": 8.692487001273688e-06,
      "loss": 1.0525,
      "step": 22115
    },
    {
      "epoch": 0.94,
      "grad_norm": 34.72791944055447,
      "learning_rate": 8.691656793451726e-06,
      "loss": 1.056,
      "step": 22120
    },
    {
      "epoch": 0.94,
      "grad_norm": 69.58577127840131,
      "learning_rate": 8.690826361815486e-06,
      "loss": 1.0973,
      "step": 22125
    },
    {
      "epoch": 0.94,
      "grad_norm": 65.60558608180274,
      "learning_rate": 8.689995706415316e-06,
      "loss": 1.084,
      "step": 22130
    },
    {
      "epoch": 0.94,
      "grad_norm": 31.721758687253875,
      "learning_rate": 8.689164827301575e-06,
      "loss": 1.037,
      "step": 22135
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.7644748360270235,
      "learning_rate": 8.688333724524637e-06,
      "loss": 1.0589,
      "step": 22140
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.808601941765356,
      "learning_rate": 8.687502398134892e-06,
      "loss": 1.0694,
      "step": 22145
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.527289182132591,
      "learning_rate": 8.686670848182736e-06,
      "loss": 1.0818,
      "step": 22150
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.663093006318125,
      "learning_rate": 8.685839074718588e-06,
      "loss": 1.0566,
      "step": 22155
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.395977401398287,
      "learning_rate": 8.685007077792871e-06,
      "loss": 1.0494,
      "step": 22160
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.56342243810745,
      "learning_rate": 8.684174857456031e-06,
      "loss": 1.0433,
      "step": 22165
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.210971882092373,
      "learning_rate": 8.683342413758522e-06,
      "loss": 1.0578,
      "step": 22170
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.57597147072617,
      "learning_rate": 8.68250974675081e-06,
      "loss": 1.0456,
      "step": 22175
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.802007712555317,
      "learning_rate": 8.681676856483383e-06,
      "loss": 1.072,
      "step": 22180
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.752925562526565,
      "learning_rate": 8.680843743006731e-06,
      "loss": 1.0251,
      "step": 22185
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.8426808897513665,
      "learning_rate": 8.680010406371364e-06,
      "loss": 1.0249,
      "step": 22190
    },
    {
      "epoch": 0.94,
      "grad_norm": 14.329783195261392,
      "learning_rate": 8.679176846627807e-06,
      "loss": 1.0299,
      "step": 22195
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.776929723652254,
      "learning_rate": 8.678343063826594e-06,
      "loss": 1.0303,
      "step": 22200
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.368130237537494,
      "learning_rate": 8.677509058018278e-06,
      "loss": 1.0773,
      "step": 22205
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.61928727770007,
      "learning_rate": 8.676674829253419e-06,
      "loss": 1.0222,
      "step": 22210
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.233023249819664,
      "learning_rate": 8.675840377582595e-06,
      "loss": 1.0411,
      "step": 22215
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.97172455881486,
      "learning_rate": 8.675005703056399e-06,
      "loss": 1.044,
      "step": 22220
    },
    {
      "epoch": 0.94,
      "grad_norm": 13.314626258237514,
      "learning_rate": 8.674170805725428e-06,
      "loss": 1.059,
      "step": 22225
    },
    {
      "epoch": 0.94,
      "grad_norm": 15.545766082051502,
      "learning_rate": 8.673335685640305e-06,
      "loss": 0.9978,
      "step": 22230
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.86523492801638,
      "learning_rate": 8.672500342851661e-06,
      "loss": 1.0564,
      "step": 22235
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.914569423381977,
      "learning_rate": 8.671664777410138e-06,
      "loss": 1.0691,
      "step": 22240
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.773933891915756,
      "learning_rate": 8.670828989366395e-06,
      "loss": 1.1061,
      "step": 22245
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.446393775155094,
      "learning_rate": 8.669992978771104e-06,
      "loss": 1.0216,
      "step": 22250
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.594977449224079,
      "learning_rate": 8.669156745674946e-06,
      "loss": 1.0548,
      "step": 22255
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.22305430863255,
      "learning_rate": 8.668320290128626e-06,
      "loss": 1.0312,
      "step": 22260
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.739578516044126,
      "learning_rate": 8.66748361218285e-06,
      "loss": 1.0732,
      "step": 22265
    },
    {
      "epoch": 0.94,
      "grad_norm": 12.354408314238693,
      "learning_rate": 8.666646711888348e-06,
      "loss": 1.0553,
      "step": 22270
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.903784659517549,
      "learning_rate": 8.665809589295855e-06,
      "loss": 1.1459,
      "step": 22275
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.243230755908847,
      "learning_rate": 8.664972244456122e-06,
      "loss": 1.072,
      "step": 22280
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.100263469256676,
      "learning_rate": 8.664134677419921e-06,
      "loss": 1.0762,
      "step": 22285
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.910965601986049,
      "learning_rate": 8.663296888238027e-06,
      "loss": 1.0732,
      "step": 22290
    },
    {
      "epoch": 0.94,
      "grad_norm": 10.029501501277,
      "learning_rate": 8.66245887696123e-06,
      "loss": 1.0584,
      "step": 22295
    },
    {
      "epoch": 0.94,
      "grad_norm": 28.751099622813843,
      "learning_rate": 8.661620643640343e-06,
      "loss": 1.0465,
      "step": 22300
    },
    {
      "epoch": 0.94,
      "grad_norm": 17.95335348479722,
      "learning_rate": 8.660782188326181e-06,
      "loss": 1.0192,
      "step": 22305
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.530815863475906,
      "learning_rate": 8.659943511069578e-06,
      "loss": 1.0489,
      "step": 22310
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.403582955689854,
      "learning_rate": 8.659104611921381e-06,
      "loss": 1.0601,
      "step": 22315
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.75370249460377,
      "learning_rate": 8.65826549093245e-06,
      "loss": 1.0346,
      "step": 22320
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.142803175559164,
      "learning_rate": 8.657426148153656e-06,
      "loss": 1.0341,
      "step": 22325
    },
    {
      "epoch": 0.95,
      "grad_norm": 16.139827316661293,
      "learning_rate": 8.656586583635888e-06,
      "loss": 1.0633,
      "step": 22330
    },
    {
      "epoch": 0.95,
      "grad_norm": 15.591077376580577,
      "learning_rate": 8.655746797430047e-06,
      "loss": 1.033,
      "step": 22335
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.795394436701578,
      "learning_rate": 8.654906789587048e-06,
      "loss": 1.0995,
      "step": 22340
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.812144871268645,
      "learning_rate": 8.654066560157814e-06,
      "loss": 1.0595,
      "step": 22345
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.381837875600517,
      "learning_rate": 8.653226109193289e-06,
      "loss": 1.0755,
      "step": 22350
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.724953554206786,
      "learning_rate": 8.652385436744426e-06,
      "loss": 1.055,
      "step": 22355
    },
    {
      "epoch": 0.95,
      "grad_norm": 19.75986423849521,
      "learning_rate": 8.651544542862192e-06,
      "loss": 1.0178,
      "step": 22360
    },
    {
      "epoch": 0.95,
      "grad_norm": 14.122213502198173,
      "learning_rate": 8.650703427597568e-06,
      "loss": 1.0619,
      "step": 22365
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.715685344258217,
      "learning_rate": 8.649862091001548e-06,
      "loss": 1.0256,
      "step": 22370
    },
    {
      "epoch": 0.95,
      "grad_norm": 25.253990303156407,
      "learning_rate": 8.649020533125142e-06,
      "loss": 1.0669,
      "step": 22375
    },
    {
      "epoch": 0.95,
      "grad_norm": 18.860335000019642,
      "learning_rate": 8.64817875401937e-06,
      "loss": 1.0242,
      "step": 22380
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.2157275145466135,
      "learning_rate": 8.647336753735264e-06,
      "loss": 1.0696,
      "step": 22385
    },
    {
      "epoch": 0.95,
      "grad_norm": 16.67555358222662,
      "learning_rate": 8.646494532323876e-06,
      "loss": 1.0699,
      "step": 22390
    },
    {
      "epoch": 0.95,
      "grad_norm": 17.85617248554722,
      "learning_rate": 8.645652089836265e-06,
      "loss": 1.075,
      "step": 22395
    },
    {
      "epoch": 0.95,
      "grad_norm": 28.737897039503693,
      "learning_rate": 8.644809426323505e-06,
      "loss": 1.049,
      "step": 22400
    },
    {
      "epoch": 0.95,
      "grad_norm": 13.60442217055186,
      "learning_rate": 8.643966541836688e-06,
      "loss": 1.0492,
      "step": 22405
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.916508236432474,
      "learning_rate": 8.643123436426912e-06,
      "loss": 1.0932,
      "step": 22410
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.646594956507517,
      "learning_rate": 8.642280110145294e-06,
      "loss": 1.0767,
      "step": 22415
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.448217132402646,
      "learning_rate": 8.641436563042961e-06,
      "loss": 1.0644,
      "step": 22420
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.597621449451742,
      "learning_rate": 8.640592795171056e-06,
      "loss": 1.0572,
      "step": 22425
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.789746424483436,
      "learning_rate": 8.639748806580732e-06,
      "loss": 1.0236,
      "step": 22430
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.924856195661939,
      "learning_rate": 8.638904597323161e-06,
      "loss": 1.0459,
      "step": 22435
    },
    {
      "epoch": 0.95,
      "grad_norm": 10.299525784868806,
      "learning_rate": 8.638060167449522e-06,
      "loss": 1.0626,
      "step": 22440
    },
    {
      "epoch": 0.95,
      "grad_norm": 14.895224853085512,
      "learning_rate": 8.63721551701101e-06,
      "loss": 1.028,
      "step": 22445
    },
    {
      "epoch": 0.95,
      "grad_norm": 11.478161503300399,
      "learning_rate": 8.636370646058838e-06,
      "loss": 1.0458,
      "step": 22450
    },
    {
      "epoch": 0.95,
      "grad_norm": 17.274506746035797,
      "learning_rate": 8.635525554644224e-06,
      "loss": 1.0211,
      "step": 22455
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.771203714954781,
      "learning_rate": 8.634680242818402e-06,
      "loss": 1.0848,
      "step": 22460
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.557339811779501,
      "learning_rate": 8.633834710632626e-06,
      "loss": 1.0283,
      "step": 22465
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.975748174040248,
      "learning_rate": 8.632988958138155e-06,
      "loss": 1.0806,
      "step": 22470
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.3011142207786,
      "learning_rate": 8.632142985386263e-06,
      "loss": 1.0379,
      "step": 22475
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.9229304586236555,
      "learning_rate": 8.631296792428242e-06,
      "loss": 1.0316,
      "step": 22480
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.602299842531993,
      "learning_rate": 8.630450379315392e-06,
      "loss": 1.0479,
      "step": 22485
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.920623212628911,
      "learning_rate": 8.629603746099032e-06,
      "loss": 1.0572,
      "step": 22490
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.6966586530370416,
      "learning_rate": 8.628756892830485e-06,
      "loss": 1.0222,
      "step": 22495
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.145376790688642,
      "learning_rate": 8.627909819561098e-06,
      "loss": 1.0162,
      "step": 22500
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.611645023317141,
      "learning_rate": 8.627062526342225e-06,
      "loss": 1.0677,
      "step": 22505
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.939565094224612,
      "learning_rate": 8.626215013225234e-06,
      "loss": 1.07,
      "step": 22510
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.724080822028831,
      "learning_rate": 8.625367280261508e-06,
      "loss": 1.0568,
      "step": 22515
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.8959575722004764,
      "learning_rate": 8.624519327502444e-06,
      "loss": 1.0604,
      "step": 22520
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.802583091897265,
      "learning_rate": 8.623671154999448e-06,
      "loss": 1.0278,
      "step": 22525
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.773076392246919,
      "learning_rate": 8.622822762803944e-06,
      "loss": 1.0675,
      "step": 22530
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.8317559020018885,
      "learning_rate": 8.621974150967367e-06,
      "loss": 1.0364,
      "step": 22535
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.339746924570607,
      "learning_rate": 8.621125319541168e-06,
      "loss": 1.023,
      "step": 22540
    },
    {
      "epoch": 0.95,
      "grad_norm": 16.008673226914222,
      "learning_rate": 8.620276268576806e-06,
      "loss": 1.0415,
      "step": 22545
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.598222235446073,
      "learning_rate": 8.619426998125758e-06,
      "loss": 1.0118,
      "step": 22550
    },
    {
      "epoch": 0.95,
      "grad_norm": 4.626896345472039,
      "learning_rate": 8.618577508239512e-06,
      "loss": 1.0377,
      "step": 22555
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.368129709555259,
      "learning_rate": 8.617727798969571e-06,
      "loss": 1.0723,
      "step": 22560
    },
    {
      "epoch": 0.95,
      "grad_norm": 13.792895549901294,
      "learning_rate": 8.61687787036745e-06,
      "loss": 1.051,
      "step": 22565
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.861500876880135,
      "learning_rate": 8.616027722484678e-06,
      "loss": 1.0185,
      "step": 22570
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.978222204899534,
      "learning_rate": 8.615177355372794e-06,
      "loss": 1.034,
      "step": 22575
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.727826171694468,
      "learning_rate": 8.614326769083359e-06,
      "loss": 1.0456,
      "step": 22580
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.943462237602688,
      "learning_rate": 8.613475963667937e-06,
      "loss": 1.096,
      "step": 22585
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.078749701164291,
      "learning_rate": 8.612624939178111e-06,
      "loss": 1.0733,
      "step": 22590
    },
    {
      "epoch": 0.96,
      "grad_norm": 23.593619568594583,
      "learning_rate": 8.611773695665477e-06,
      "loss": 1.0045,
      "step": 22595
    },
    {
      "epoch": 0.96,
      "grad_norm": 25.877109034643134,
      "learning_rate": 8.610922233181642e-06,
      "loss": 1.0181,
      "step": 22600
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.4353597828378994,
      "learning_rate": 8.61007055177823e-06,
      "loss": 1.0615,
      "step": 22605
    },
    {
      "epoch": 0.96,
      "grad_norm": 17.499574040833348,
      "learning_rate": 8.609218651506872e-06,
      "loss": 1.013,
      "step": 22610
    },
    {
      "epoch": 0.96,
      "grad_norm": 13.659242476773333,
      "learning_rate": 8.60836653241922e-06,
      "loss": 1.0199,
      "step": 22615
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.608400698627674,
      "learning_rate": 8.60751419456693e-06,
      "loss": 1.0506,
      "step": 22620
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.702057135255766,
      "learning_rate": 8.606661638001686e-06,
      "loss": 1.0716,
      "step": 22625
    },
    {
      "epoch": 0.96,
      "grad_norm": 18.393959498606716,
      "learning_rate": 8.605808862775168e-06,
      "loss": 1.0154,
      "step": 22630
    },
    {
      "epoch": 0.96,
      "grad_norm": 20.833851670917028,
      "learning_rate": 8.60495586893908e-06,
      "loss": 0.9968,
      "step": 22635
    },
    {
      "epoch": 0.96,
      "grad_norm": 16.3987131118621,
      "learning_rate": 8.604102656545138e-06,
      "loss": 1.0248,
      "step": 22640
    },
    {
      "epoch": 0.96,
      "grad_norm": 25.58544936694752,
      "learning_rate": 8.603249225645066e-06,
      "loss": 1.0616,
      "step": 22645
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.823068895473675,
      "learning_rate": 8.602395576290608e-06,
      "loss": 1.1135,
      "step": 22650
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.178873329500044,
      "learning_rate": 8.601541708533517e-06,
      "loss": 1.0362,
      "step": 22655
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.781944605070839,
      "learning_rate": 8.600687622425563e-06,
      "loss": 1.0649,
      "step": 22660
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.097052698035258,
      "learning_rate": 8.599833318018522e-06,
      "loss": 1.0343,
      "step": 22665
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.273011000979423,
      "learning_rate": 8.598978795364192e-06,
      "loss": 1.021,
      "step": 22670
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.48210656093624,
      "learning_rate": 8.598124054514379e-06,
      "loss": 1.0797,
      "step": 22675
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.864921184004051,
      "learning_rate": 8.597269095520902e-06,
      "loss": 1.0049,
      "step": 22680
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.527082078116695,
      "learning_rate": 8.596413918435595e-06,
      "loss": 1.0382,
      "step": 22685
    },
    {
      "epoch": 0.96,
      "grad_norm": 12.188234754386398,
      "learning_rate": 8.595558523310307e-06,
      "loss": 1.0364,
      "step": 22690
    },
    {
      "epoch": 0.96,
      "grad_norm": 18.7546662979122,
      "learning_rate": 8.594702910196899e-06,
      "loss": 1.0623,
      "step": 22695
    },
    {
      "epoch": 0.96,
      "grad_norm": 20.44543714615951,
      "learning_rate": 8.593847079147239e-06,
      "loss": 1.0609,
      "step": 22700
    },
    {
      "epoch": 0.96,
      "grad_norm": 18.597882662544343,
      "learning_rate": 8.592991030213217e-06,
      "loss": 1.0423,
      "step": 22705
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.733908470621946,
      "learning_rate": 8.592134763446733e-06,
      "loss": 1.0656,
      "step": 22710
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.51079960986976,
      "learning_rate": 8.5912782788997e-06,
      "loss": 1.1192,
      "step": 22715
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.5463955593727405,
      "learning_rate": 8.590421576624042e-06,
      "loss": 1.0244,
      "step": 22720
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.199460656469777,
      "learning_rate": 8.589564656671701e-06,
      "loss": 1.0145,
      "step": 22725
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.914944971914244,
      "learning_rate": 8.588707519094628e-06,
      "loss": 1.0361,
      "step": 22730
    },
    {
      "epoch": 0.96,
      "grad_norm": 15.26421108734658,
      "learning_rate": 8.587850163944788e-06,
      "loss": 1.0387,
      "step": 22735
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.116683932140762,
      "learning_rate": 8.586992591274162e-06,
      "loss": 1.0479,
      "step": 22740
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.364069259977557,
      "learning_rate": 8.586134801134741e-06,
      "loss": 1.024,
      "step": 22745
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.5665919065188785,
      "learning_rate": 8.58527679357853e-06,
      "loss": 1.0422,
      "step": 22750
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.114018625468076,
      "learning_rate": 8.584418568657547e-06,
      "loss": 1.0539,
      "step": 22755
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.03998921719011,
      "learning_rate": 8.583560126423827e-06,
      "loss": 1.0758,
      "step": 22760
    },
    {
      "epoch": 0.96,
      "grad_norm": 19.54082984279,
      "learning_rate": 8.58270146692941e-06,
      "loss": 1.057,
      "step": 22765
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.759790183732632,
      "learning_rate": 8.581842590226355e-06,
      "loss": 0.986,
      "step": 22770
    },
    {
      "epoch": 0.96,
      "grad_norm": 10.08746887017593,
      "learning_rate": 8.580983496366736e-06,
      "loss": 1.0062,
      "step": 22775
    },
    {
      "epoch": 0.96,
      "grad_norm": 14.319336390442473,
      "learning_rate": 8.580124185402636e-06,
      "loss": 1.0391,
      "step": 22780
    },
    {
      "epoch": 0.96,
      "grad_norm": 11.715863968001175,
      "learning_rate": 8.579264657386152e-06,
      "loss": 1.0558,
      "step": 22785
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.9620813893898434,
      "learning_rate": 8.578404912369396e-06,
      "loss": 1.0364,
      "step": 22790
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.853247406950727,
      "learning_rate": 8.57754495040449e-06,
      "loss": 1.0449,
      "step": 22795
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.9325868163133855,
      "learning_rate": 8.57668477154357e-06,
      "loss": 1.0235,
      "step": 22800
    },
    {
      "epoch": 0.97,
      "grad_norm": 14.217207569081221,
      "learning_rate": 8.57582437583879e-06,
      "loss": 1.0196,
      "step": 22805
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.6218494476714085,
      "learning_rate": 8.57496376334231e-06,
      "loss": 1.0021,
      "step": 22810
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.609844277642452,
      "learning_rate": 8.574102934106308e-06,
      "loss": 1.0401,
      "step": 22815
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.298393187484265,
      "learning_rate": 8.573241888182973e-06,
      "loss": 1.0818,
      "step": 22820
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.640461642732947,
      "learning_rate": 8.572380625624506e-06,
      "loss": 1.0781,
      "step": 22825
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.461278443647827,
      "learning_rate": 8.571519146483128e-06,
      "loss": 1.0423,
      "step": 22830
    },
    {
      "epoch": 0.97,
      "grad_norm": 14.584826313438121,
      "learning_rate": 8.57065745081106e-06,
      "loss": 1.0371,
      "step": 22835
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.432323586901132,
      "learning_rate": 8.56979553866055e-06,
      "loss": 1.0653,
      "step": 22840
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.274285011381062,
      "learning_rate": 8.568933410083853e-06,
      "loss": 1.0197,
      "step": 22845
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.258067392149234,
      "learning_rate": 8.568071065133236e-06,
      "loss": 1.0112,
      "step": 22850
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.439167452221515,
      "learning_rate": 8.56720850386098e-06,
      "loss": 1.0398,
      "step": 22855
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.939122885874223,
      "learning_rate": 8.566345726319378e-06,
      "loss": 1.0363,
      "step": 22860
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.57590426612838,
      "learning_rate": 8.565482732560741e-06,
      "loss": 0.9979,
      "step": 22865
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.291005880357732,
      "learning_rate": 8.56461952263739e-06,
      "loss": 1.0466,
      "step": 22870
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.933556562433127,
      "learning_rate": 8.563756096601655e-06,
      "loss": 1.0146,
      "step": 22875
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.7314152230084545,
      "learning_rate": 8.562892454505886e-06,
      "loss": 1.0047,
      "step": 22880
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.589110578264134,
      "learning_rate": 8.562028596402443e-06,
      "loss": 0.9974,
      "step": 22885
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.708398943177179,
      "learning_rate": 8.561164522343697e-06,
      "loss": 1.0338,
      "step": 22890
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.799722849333198,
      "learning_rate": 8.560300232382037e-06,
      "loss": 1.0386,
      "step": 22895
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.503503606800315,
      "learning_rate": 8.559435726569862e-06,
      "loss": 1.058,
      "step": 22900
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.459599596724459,
      "learning_rate": 8.558571004959582e-06,
      "loss": 0.9967,
      "step": 22905
    },
    {
      "epoch": 0.97,
      "grad_norm": 13.447624775927896,
      "learning_rate": 8.557706067603623e-06,
      "loss": 1.0293,
      "step": 22910
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.15624737203624,
      "learning_rate": 8.556840914554428e-06,
      "loss": 1.0353,
      "step": 22915
    },
    {
      "epoch": 0.97,
      "grad_norm": 16.9682109648794,
      "learning_rate": 8.555975545864444e-06,
      "loss": 0.9997,
      "step": 22920
    },
    {
      "epoch": 0.97,
      "grad_norm": 29.099404126642884,
      "learning_rate": 8.555109961586137e-06,
      "loss": 1.0416,
      "step": 22925
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.2843312621134935,
      "learning_rate": 8.554244161771982e-06,
      "loss": 1.0187,
      "step": 22930
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.78562255681709,
      "learning_rate": 8.553378146474477e-06,
      "loss": 1.0735,
      "step": 22935
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.797934466693091,
      "learning_rate": 8.55251191574612e-06,
      "loss": 1.078,
      "step": 22940
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.736168536629084,
      "learning_rate": 8.55164546963943e-06,
      "loss": 1.0345,
      "step": 22945
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.937905640880309,
      "learning_rate": 8.550778808206937e-06,
      "loss": 1.0122,
      "step": 22950
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.165484689090224,
      "learning_rate": 8.549911931501185e-06,
      "loss": 1.0478,
      "step": 22955
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.9071265823273125,
      "learning_rate": 8.549044839574727e-06,
      "loss": 1.0159,
      "step": 22960
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.3988932790998545,
      "learning_rate": 8.548177532480135e-06,
      "loss": 1.0035,
      "step": 22965
    },
    {
      "epoch": 0.97,
      "grad_norm": 14.548916511792239,
      "learning_rate": 8.547310010269993e-06,
      "loss": 1.0499,
      "step": 22970
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.064724224720091,
      "learning_rate": 8.546442272996891e-06,
      "loss": 1.0223,
      "step": 22975
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.0437475407919905,
      "learning_rate": 8.545574320713442e-06,
      "loss": 1.0337,
      "step": 22980
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.2573158804506805,
      "learning_rate": 8.544706153472266e-06,
      "loss": 1.0309,
      "step": 22985
    },
    {
      "epoch": 0.97,
      "grad_norm": 16.754930512289445,
      "learning_rate": 8.543837771325997e-06,
      "loss": 1.0839,
      "step": 22990
    },
    {
      "epoch": 0.97,
      "grad_norm": 16.255288409879782,
      "learning_rate": 8.542969174327283e-06,
      "loss": 1.0398,
      "step": 22995
    },
    {
      "epoch": 0.97,
      "grad_norm": 15.428333095801877,
      "learning_rate": 8.542100362528783e-06,
      "loss": 1.0421,
      "step": 23000
    },
    {
      "epoch": 0.97,
      "grad_norm": 21.36290421824013,
      "learning_rate": 8.541231335983171e-06,
      "loss": 0.9987,
      "step": 23005
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.899910202340994,
      "learning_rate": 8.540362094743135e-06,
      "loss": 1.0552,
      "step": 23010
    },
    {
      "epoch": 0.97,
      "grad_norm": 9.980222913180013,
      "learning_rate": 8.539492638861375e-06,
      "loss": 1.0167,
      "step": 23015
    },
    {
      "epoch": 0.97,
      "grad_norm": 18.49719666601898,
      "learning_rate": 8.5386229683906e-06,
      "loss": 1.0419,
      "step": 23020
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.104077917230324,
      "learning_rate": 8.537753083383539e-06,
      "loss": 1.0512,
      "step": 23025
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.7957468631237115,
      "learning_rate": 8.53688298389293e-06,
      "loss": 1.0305,
      "step": 23030
    },
    {
      "epoch": 0.97,
      "grad_norm": 16.937029322802367,
      "learning_rate": 8.536012669971523e-06,
      "loss": 1.0241,
      "step": 23035
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.726734339725752,
      "learning_rate": 8.535142141672083e-06,
      "loss": 1.0711,
      "step": 23040
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.590678251498707,
      "learning_rate": 8.534271399047389e-06,
      "loss": 1.0331,
      "step": 23045
    },
    {
      "epoch": 0.98,
      "grad_norm": 27.443177665631616,
      "learning_rate": 8.53340044215023e-06,
      "loss": 1.057,
      "step": 23050
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.709461693344375,
      "learning_rate": 8.532529271033412e-06,
      "loss": 1.0476,
      "step": 23055
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.52151672361176,
      "learning_rate": 8.531657885749747e-06,
      "loss": 1.05,
      "step": 23060
    },
    {
      "epoch": 0.98,
      "grad_norm": 28.16659603080454,
      "learning_rate": 8.530786286352068e-06,
      "loss": 1.0471,
      "step": 23065
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.819881132559455,
      "learning_rate": 8.529914472893217e-06,
      "loss": 0.9943,
      "step": 23070
    },
    {
      "epoch": 0.98,
      "grad_norm": 15.970063304840865,
      "learning_rate": 8.52904244542605e-06,
      "loss": 1.0482,
      "step": 23075
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.066594950420572,
      "learning_rate": 8.528170204003434e-06,
      "loss": 1.0257,
      "step": 23080
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.304265522139255,
      "learning_rate": 8.527297748678252e-06,
      "loss": 1.0259,
      "step": 23085
    },
    {
      "epoch": 0.98,
      "grad_norm": 16.77287028967692,
      "learning_rate": 8.526425079503395e-06,
      "loss": 1.0389,
      "step": 23090
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.90383393067753,
      "learning_rate": 8.525552196531775e-06,
      "loss": 1.0915,
      "step": 23095
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.969770022144005,
      "learning_rate": 8.524679099816308e-06,
      "loss": 1.0145,
      "step": 23100
    },
    {
      "epoch": 0.98,
      "grad_norm": 14.615720422693036,
      "learning_rate": 8.523805789409933e-06,
      "loss": 1.0382,
      "step": 23105
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.896574009718107,
      "learning_rate": 8.52293226536559e-06,
      "loss": 1.0231,
      "step": 23110
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.03820018486783,
      "learning_rate": 8.522058527736242e-06,
      "loss": 1.0995,
      "step": 23115
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.7266038124399055,
      "learning_rate": 8.52118457657486e-06,
      "loss": 1.0219,
      "step": 23120
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.03469004883265,
      "learning_rate": 8.520310411934427e-06,
      "loss": 1.0416,
      "step": 23125
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.051595793411523,
      "learning_rate": 8.519436033867947e-06,
      "loss": 0.9894,
      "step": 23130
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.454713193809196,
      "learning_rate": 8.518561442428424e-06,
      "loss": 1.003,
      "step": 23135
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.884689564512003,
      "learning_rate": 8.517686637668888e-06,
      "loss": 1.0639,
      "step": 23140
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.012427419432159,
      "learning_rate": 8.51681161964237e-06,
      "loss": 1.0001,
      "step": 23145
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.043031054799105,
      "learning_rate": 8.515936388401924e-06,
      "loss": 1.0307,
      "step": 23150
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.265952046332991,
      "learning_rate": 8.51506094400061e-06,
      "loss": 1.0292,
      "step": 23155
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.608910602343494,
      "learning_rate": 8.514185286491506e-06,
      "loss": 1.0632,
      "step": 23160
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.752875066980779,
      "learning_rate": 8.5133094159277e-06,
      "loss": 1.0472,
      "step": 23165
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.073088789417399,
      "learning_rate": 8.512433332362291e-06,
      "loss": 1.0378,
      "step": 23170
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.039947304869808,
      "learning_rate": 8.511557035848397e-06,
      "loss": 1.0537,
      "step": 23175
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.8651125297603715,
      "learning_rate": 8.510680526439142e-06,
      "loss": 1.0348,
      "step": 23180
    },
    {
      "epoch": 0.98,
      "grad_norm": 15.136058309539601,
      "learning_rate": 8.509803804187669e-06,
      "loss": 1.0417,
      "step": 23185
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.431586827139782,
      "learning_rate": 8.50892686914713e-06,
      "loss": 1.0127,
      "step": 23190
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.9659224370651085,
      "learning_rate": 8.50804972137069e-06,
      "loss": 1.0283,
      "step": 23195
    },
    {
      "epoch": 0.98,
      "grad_norm": 12.362531898597064,
      "learning_rate": 8.507172360911529e-06,
      "loss": 1.0001,
      "step": 23200
    },
    {
      "epoch": 0.98,
      "grad_norm": 20.03845547654598,
      "learning_rate": 8.506294787822837e-06,
      "loss": 1.0498,
      "step": 23205
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.951606531113273,
      "learning_rate": 8.505417002157823e-06,
      "loss": 0.9917,
      "step": 23210
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.601171206012808,
      "learning_rate": 8.504539003969701e-06,
      "loss": 1.0504,
      "step": 23215
    },
    {
      "epoch": 0.98,
      "grad_norm": 9.241876040040765,
      "learning_rate": 8.5036607933117e-06,
      "loss": 1.0188,
      "step": 23220
    },
    {
      "epoch": 0.98,
      "grad_norm": 17.39617153236078,
      "learning_rate": 8.502782370237068e-06,
      "loss": 1.0297,
      "step": 23225
    },
    {
      "epoch": 0.98,
      "grad_norm": 19.921580882180873,
      "learning_rate": 8.501903734799058e-06,
      "loss": 1.0572,
      "step": 23230
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.593051124968328,
      "learning_rate": 8.501024887050939e-06,
      "loss": 1.0269,
      "step": 23235
    },
    {
      "epoch": 0.98,
      "grad_norm": 18.783322251033454,
      "learning_rate": 8.500145827045994e-06,
      "loss": 1.0638,
      "step": 23240
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.308844649106488,
      "learning_rate": 8.499266554837518e-06,
      "loss": 0.9901,
      "step": 23245
    },
    {
      "epoch": 0.98,
      "grad_norm": 14.12850652294799,
      "learning_rate": 8.498387070478816e-06,
      "loss": 1.0395,
      "step": 23250
    },
    {
      "epoch": 0.98,
      "grad_norm": 17.50235764908691,
      "learning_rate": 8.497507374023212e-06,
      "loss": 1.0566,
      "step": 23255
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.817943554467996,
      "learning_rate": 8.49662746552404e-06,
      "loss": 1.0042,
      "step": 23260
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.51947684277591,
      "learning_rate": 8.495747345034644e-06,
      "loss": 1.0333,
      "step": 23265
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.151864044990613,
      "learning_rate": 8.49486701260838e-06,
      "loss": 1.0171,
      "step": 23270
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.793646968134784,
      "learning_rate": 8.493986468298626e-06,
      "loss": 0.9703,
      "step": 23275
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.268255583096469,
      "learning_rate": 8.493105712158767e-06,
      "loss": 0.9962,
      "step": 23280
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.290962586119373,
      "learning_rate": 8.492224744242194e-06,
      "loss": 1.0246,
      "step": 23285
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.521967868598463,
      "learning_rate": 8.491343564602324e-06,
      "loss": 1.0415,
      "step": 23290
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.948269786217876,
      "learning_rate": 8.490462173292576e-06,
      "loss": 1.0004,
      "step": 23295
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.609054069180209,
      "learning_rate": 8.489580570366389e-06,
      "loss": 1.046,
      "step": 23300
    },
    {
      "epoch": 0.99,
      "grad_norm": 27.12831081982543,
      "learning_rate": 8.48869875587721e-06,
      "loss": 1.0055,
      "step": 23305
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.078232057142618,
      "learning_rate": 8.487816729878501e-06,
      "loss": 1.043,
      "step": 23310
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.08809490334739,
      "learning_rate": 8.48693449242374e-06,
      "loss": 1.0341,
      "step": 23315
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.181579037380805,
      "learning_rate": 8.48605204356641e-06,
      "loss": 0.9927,
      "step": 23320
    },
    {
      "epoch": 0.99,
      "grad_norm": 19.513917064074878,
      "learning_rate": 8.485169383360014e-06,
      "loss": 0.9949,
      "step": 23325
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.671188730020175,
      "learning_rate": 8.484286511858062e-06,
      "loss": 1.0067,
      "step": 23330
    },
    {
      "epoch": 0.99,
      "grad_norm": 15.877637792063496,
      "learning_rate": 8.483403429114085e-06,
      "loss": 1.0481,
      "step": 23335
    },
    {
      "epoch": 0.99,
      "grad_norm": 28.355402625944034,
      "learning_rate": 8.482520135181617e-06,
      "loss": 1.0114,
      "step": 23340
    },
    {
      "epoch": 0.99,
      "grad_norm": 32.307020953693794,
      "learning_rate": 8.481636630114212e-06,
      "loss": 1.0086,
      "step": 23345
    },
    {
      "epoch": 0.99,
      "grad_norm": 22.947329093412538,
      "learning_rate": 8.480752913965434e-06,
      "loss": 1.0109,
      "step": 23350
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.741749530268085,
      "learning_rate": 8.479868986788858e-06,
      "loss": 1.0693,
      "step": 23355
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.808812079773028,
      "learning_rate": 8.478984848638077e-06,
      "loss": 0.9898,
      "step": 23360
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.817170577309794,
      "learning_rate": 8.478100499566693e-06,
      "loss": 1.0201,
      "step": 23365
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.042753637957894,
      "learning_rate": 8.47721593962832e-06,
      "loss": 1.0125,
      "step": 23370
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.082615930342339,
      "learning_rate": 8.476331168876588e-06,
      "loss": 1.0727,
      "step": 23375
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.407795793199803,
      "learning_rate": 8.475446187365136e-06,
      "loss": 1.0423,
      "step": 23380
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.912832440040569,
      "learning_rate": 8.474560995147618e-06,
      "loss": 1.0174,
      "step": 23385
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.54450292386654,
      "learning_rate": 8.473675592277702e-06,
      "loss": 1.0222,
      "step": 23390
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.27306608562583,
      "learning_rate": 8.47278997880907e-06,
      "loss": 1.0343,
      "step": 23395
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.273221841701524,
      "learning_rate": 8.471904154795407e-06,
      "loss": 1.0499,
      "step": 23400
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.892920524724065,
      "learning_rate": 8.471018120290422e-06,
      "loss": 1.0146,
      "step": 23405
    },
    {
      "epoch": 0.99,
      "grad_norm": 14.493540436041627,
      "learning_rate": 8.470131875347834e-06,
      "loss": 1.0415,
      "step": 23410
    },
    {
      "epoch": 0.99,
      "grad_norm": 17.336811305379513,
      "learning_rate": 8.469245420021373e-06,
      "loss": 1.0267,
      "step": 23415
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.635234002713541,
      "learning_rate": 8.46835875436478e-06,
      "loss": 1.023,
      "step": 23420
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.00175757248925,
      "learning_rate": 8.467471878431812e-06,
      "loss": 1.022,
      "step": 23425
    },
    {
      "epoch": 0.99,
      "grad_norm": 10.20532814275073,
      "learning_rate": 8.466584792276237e-06,
      "loss": 1.0721,
      "step": 23430
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.8541351706445,
      "learning_rate": 8.465697495951839e-06,
      "loss": 0.9971,
      "step": 23435
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.010403253158212,
      "learning_rate": 8.464809989512409e-06,
      "loss": 1.0142,
      "step": 23440
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.404646170758126,
      "learning_rate": 8.463922273011755e-06,
      "loss": 1.0646,
      "step": 23445
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.467273917314427,
      "learning_rate": 8.463034346503697e-06,
      "loss": 1.0091,
      "step": 23450
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.875605929914122,
      "learning_rate": 8.46214621004207e-06,
      "loss": 1.0182,
      "step": 23455
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.200797996398627,
      "learning_rate": 8.461257863680715e-06,
      "loss": 0.9846,
      "step": 23460
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.054734007179876,
      "learning_rate": 8.46036930747349e-06,
      "loss": 1.0155,
      "step": 23465
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.136718430267301,
      "learning_rate": 8.459480541474268e-06,
      "loss": 1.0221,
      "step": 23470
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.302955018043857,
      "learning_rate": 8.45859156573693e-06,
      "loss": 1.0227,
      "step": 23475
    },
    {
      "epoch": 0.99,
      "grad_norm": 11.30483994874588,
      "learning_rate": 8.457702380315373e-06,
      "loss": 1.0444,
      "step": 23480
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.413382685398307,
      "learning_rate": 8.456812985263506e-06,
      "loss": 1.0122,
      "step": 23485
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.459691452536524,
      "learning_rate": 8.455923380635249e-06,
      "loss": 1.0255,
      "step": 23490
    },
    {
      "epoch": 0.99,
      "grad_norm": 19.29235383257498,
      "learning_rate": 8.455033566484539e-06,
      "loss": 1.0588,
      "step": 23495
    },
    {
      "epoch": 0.99,
      "grad_norm": 38.757148731564264,
      "learning_rate": 8.45414354286532e-06,
      "loss": 1.0682,
      "step": 23500
    },
    {
      "epoch": 0.99,
      "grad_norm": 16.13836724915053,
      "learning_rate": 8.453253309831553e-06,
      "loss": 1.0186,
      "step": 23505
    },
    {
      "epoch": 0.99,
      "grad_norm": 12.270015011940659,
      "learning_rate": 8.452362867437209e-06,
      "loss": 1.0316,
      "step": 23510
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.937654674803527,
      "learning_rate": 8.451472215736273e-06,
      "loss": 1.0214,
      "step": 23515
    },
    {
      "epoch": 1.0,
      "grad_norm": 14.76738521883225,
      "learning_rate": 8.450581354782745e-06,
      "loss": 1.0293,
      "step": 23520
    },
    {
      "epoch": 1.0,
      "grad_norm": 13.023266108812958,
      "learning_rate": 8.449690284630633e-06,
      "loss": 1.036,
      "step": 23525
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.959896773654664,
      "learning_rate": 8.448799005333961e-06,
      "loss": 1.0692,
      "step": 23530
    },
    {
      "epoch": 1.0,
      "grad_norm": 6.580881282273475,
      "learning_rate": 8.447907516946764e-06,
      "loss": 1.0389,
      "step": 23535
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.030960711324296,
      "learning_rate": 8.44701581952309e-06,
      "loss": 0.9856,
      "step": 23540
    },
    {
      "epoch": 1.0,
      "grad_norm": 21.20139650692982,
      "learning_rate": 8.446123913116999e-06,
      "loss": 1.0042,
      "step": 23545
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.68008802117942,
      "learning_rate": 8.445231797782567e-06,
      "loss": 1.0087,
      "step": 23550
    },
    {
      "epoch": 1.0,
      "grad_norm": 15.610982203593359,
      "learning_rate": 8.44433947357388e-06,
      "loss": 1.0116,
      "step": 23555
    },
    {
      "epoch": 1.0,
      "grad_norm": 9.94457946694081,
      "learning_rate": 8.443446940545036e-06,
      "loss": 1.0129,
      "step": 23560
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.809800318645303,
      "learning_rate": 8.442554198750145e-06,
      "loss": 1.0405,
      "step": 23565
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.147377555924988,
      "learning_rate": 8.441661248243337e-06,
      "loss": 1.0282,
      "step": 23570
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.115743486277359,
      "learning_rate": 8.440768089078743e-06,
      "loss": 1.0095,
      "step": 23575
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.05918066235286,
      "learning_rate": 8.439874721310514e-06,
      "loss": 1.0088,
      "step": 23580
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.611997174602991,
      "learning_rate": 8.438981144992814e-06,
      "loss": 1.0275,
      "step": 23585
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.143948913638901,
      "learning_rate": 8.438087360179818e-06,
      "loss": 1.0135,
      "step": 23590
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.9681691944224085,
      "learning_rate": 8.437193366925713e-06,
      "loss": 1.0233,
      "step": 23595
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.939220956826862,
      "learning_rate": 8.436299165284698e-06,
      "loss": 1.0449,
      "step": 23600
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.410859961631028,
      "learning_rate": 8.435404755310985e-06,
      "loss": 1.0485,
      "step": 23605
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.554943205074981,
      "learning_rate": 8.434510137058804e-06,
      "loss": 1.0411,
      "step": 23610
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.287320141489989,
      "learning_rate": 8.433615310582388e-06,
      "loss": 1.0516,
      "step": 23615
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.950023929476283,
      "learning_rate": 8.43272027593599e-06,
      "loss": 1.0374,
      "step": 23620
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.932865995739235,
      "learning_rate": 8.431825033173874e-06,
      "loss": 1.0034,
      "step": 23625
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.063048243522644,
      "eval_runtime": 37.3271,
      "eval_samples_per_second": 31.318,
      "eval_steps_per_second": 3.938,
      "step": 23628
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.47134981684741,
      "learning_rate": 8.430929582350313e-06,
      "loss": 0.9972,
      "step": 23630
    },
    {
      "epoch": 1.0,
      "grad_norm": 11.484846539009633,
      "learning_rate": 8.4300339235196e-06,
      "loss": 0.8836,
      "step": 23635
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.735414921089642,
      "learning_rate": 8.429138056736033e-06,
      "loss": 0.9112,
      "step": 23640
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.9708377499865,
      "learning_rate": 8.428241982053925e-06,
      "loss": 0.8925,
      "step": 23645
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.24996479615225,
      "learning_rate": 8.427345699527605e-06,
      "loss": 0.9281,
      "step": 23650
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.352222500245511,
      "learning_rate": 8.42644920921141e-06,
      "loss": 0.7964,
      "step": 23655
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.509460803468615,
      "learning_rate": 8.425552511159694e-06,
      "loss": 0.8545,
      "step": 23660
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.492095378442041,
      "learning_rate": 8.424655605426819e-06,
      "loss": 0.8666,
      "step": 23665
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.613763619418842,
      "learning_rate": 8.42375849206716e-06,
      "loss": 0.928,
      "step": 23670
    },
    {
      "epoch": 1.0,
      "grad_norm": 18.052390896206354,
      "learning_rate": 8.422861171135111e-06,
      "loss": 0.8648,
      "step": 23675
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.554035526215605,
      "learning_rate": 8.421963642685072e-06,
      "loss": 0.9003,
      "step": 23680
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.701076948439349,
      "learning_rate": 8.421065906771454e-06,
      "loss": 0.8336,
      "step": 23685
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.894132029624963,
      "learning_rate": 8.42016796344869e-06,
      "loss": 0.8511,
      "step": 23690
    },
    {
      "epoch": 1.0,
      "grad_norm": 11.075267667723061,
      "learning_rate": 8.419269812771216e-06,
      "loss": 0.9159,
      "step": 23695
    },
    {
      "epoch": 1.0,
      "grad_norm": 19.865335345987205,
      "learning_rate": 8.418371454793487e-06,
      "loss": 0.8974,
      "step": 23700
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.09374081122974,
      "learning_rate": 8.417472889569965e-06,
      "loss": 0.9152,
      "step": 23705
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.109418472952596,
      "learning_rate": 8.416574117155125e-06,
      "loss": 0.8536,
      "step": 23710
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.715071149704299,
      "learning_rate": 8.415675137603463e-06,
      "loss": 0.8788,
      "step": 23715
    },
    {
      "epoch": 1.0,
      "grad_norm": 15.849180203379067,
      "learning_rate": 8.414775950969478e-06,
      "loss": 0.8406,
      "step": 23720
    },
    {
      "epoch": 1.0,
      "grad_norm": 12.692252844840338,
      "learning_rate": 8.413876557307686e-06,
      "loss": 0.8736,
      "step": 23725
    },
    {
      "epoch": 1.0,
      "grad_norm": 18.299949279276852,
      "learning_rate": 8.412976956672614e-06,
      "loss": 0.885,
      "step": 23730
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.010611493789524,
      "learning_rate": 8.412077149118801e-06,
      "loss": 0.8484,
      "step": 23735
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.596850204273611,
      "learning_rate": 8.411177134700802e-06,
      "loss": 0.8757,
      "step": 23740
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.284237471238228,
      "learning_rate": 8.41027691347318e-06,
      "loss": 0.8257,
      "step": 23745
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.74659215368054,
      "learning_rate": 8.409376485490517e-06,
      "loss": 0.872,
      "step": 23750
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.976388229287632,
      "learning_rate": 8.4084758508074e-06,
      "loss": 0.8681,
      "step": 23755
    },
    {
      "epoch": 1.01,
      "grad_norm": 10.09592169311273,
      "learning_rate": 8.40757500947843e-06,
      "loss": 0.8713,
      "step": 23760
    },
    {
      "epoch": 1.01,
      "grad_norm": 10.861659943765588,
      "learning_rate": 8.406673961558225e-06,
      "loss": 0.8717,
      "step": 23765
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.276349622542162,
      "learning_rate": 8.405772707101414e-06,
      "loss": 0.853,
      "step": 23770
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.178001037523651,
      "learning_rate": 8.404871246162635e-06,
      "loss": 0.9112,
      "step": 23775
    },
    {
      "epoch": 1.01,
      "grad_norm": 16.444284443936446,
      "learning_rate": 8.403969578796542e-06,
      "loss": 0.8815,
      "step": 23780
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.544319636872853,
      "learning_rate": 8.4030677050578e-06,
      "loss": 0.9059,
      "step": 23785
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.0101367787362845,
      "learning_rate": 8.402165625001087e-06,
      "loss": 0.8507,
      "step": 23790
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.235502305671002,
      "learning_rate": 8.401263338681092e-06,
      "loss": 0.8854,
      "step": 23795
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.992561316105224,
      "learning_rate": 8.40036084615252e-06,
      "loss": 0.8775,
      "step": 23800
    },
    {
      "epoch": 1.01,
      "grad_norm": 15.58149670533829,
      "learning_rate": 8.399458147470088e-06,
      "loss": 0.8842,
      "step": 23805
    },
    {
      "epoch": 1.01,
      "grad_norm": 10.087206777389202,
      "learning_rate": 8.39855524268852e-06,
      "loss": 0.8692,
      "step": 23810
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.162604667308198,
      "learning_rate": 8.397652131862558e-06,
      "loss": 0.8706,
      "step": 23815
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.219265727431301,
      "learning_rate": 8.396748815046957e-06,
      "loss": 0.8956,
      "step": 23820
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.046611143807226,
      "learning_rate": 8.39584529229648e-06,
      "loss": 0.8421,
      "step": 23825
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.220776564853208,
      "learning_rate": 8.394941563665907e-06,
      "loss": 0.8411,
      "step": 23830
    },
    {
      "epoch": 1.01,
      "grad_norm": 9.533136360348138,
      "learning_rate": 8.394037629210025e-06,
      "loss": 0.852,
      "step": 23835
    },
    {
      "epoch": 1.01,
      "grad_norm": 6.349075164678321,
      "learning_rate": 8.393133488983641e-06,
      "loss": 0.9081,
      "step": 23840
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.6848376405254655,
      "learning_rate": 8.392229143041568e-06,
      "loss": 0.8643,
      "step": 23845
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.2565669840665885,
      "learning_rate": 8.391324591438634e-06,
      "loss": 0.8966,
      "step": 23850
    },
    {
      "epoch": 1.01,
      "grad_norm": 15.204771299936896,
      "learning_rate": 8.390419834229679e-06,
      "loss": 0.8765,
      "step": 23855
    },
    {
      "epoch": 1.01,
      "grad_norm": 17.42822410463676,
      "learning_rate": 8.389514871469558e-06,
      "loss": 0.868,
      "step": 23860
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.026132565859108,
      "learning_rate": 8.388609703213132e-06,
      "loss": 0.8364,
      "step": 23865
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.946876204519523,
      "learning_rate": 8.387704329515283e-06,
      "loss": 0.8676,
      "step": 23870
    },
    {
      "epoch": 1.01,
      "grad_norm": 9.11263701971295,
      "learning_rate": 8.386798750430898e-06,
      "loss": 0.9468,
      "step": 23875
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.571974886967865,
      "learning_rate": 8.385892966014881e-06,
      "loss": 0.8369,
      "step": 23880
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.907853064488422,
      "learning_rate": 8.384986976322147e-06,
      "loss": 0.8241,
      "step": 23885
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.1588994651635,
      "learning_rate": 8.384080781407625e-06,
      "loss": 0.8467,
      "step": 23890
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.9195596915901065,
      "learning_rate": 8.38317438132625e-06,
      "loss": 0.865,
      "step": 23895
    },
    {
      "epoch": 1.01,
      "grad_norm": 4.6919349248440545,
      "learning_rate": 8.38226777613298e-06,
      "loss": 0.8336,
      "step": 23900
    },
    {
      "epoch": 1.01,
      "grad_norm": 10.480509141963909,
      "learning_rate": 8.381360965882777e-06,
      "loss": 0.8655,
      "step": 23905
    },
    {
      "epoch": 1.01,
      "grad_norm": 14.76175781260011,
      "learning_rate": 8.38045395063062e-06,
      "loss": 0.878,
      "step": 23910
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.420032128307376,
      "learning_rate": 8.379546730431495e-06,
      "loss": 0.8942,
      "step": 23915
    },
    {
      "epoch": 1.01,
      "grad_norm": 7.726736496268883,
      "learning_rate": 8.37863930534041e-06,
      "loss": 0.8963,
      "step": 23920
    },
    {
      "epoch": 1.01,
      "grad_norm": 24.13648194752957,
      "learning_rate": 8.377731675412372e-06,
      "loss": 0.8677,
      "step": 23925
    },
    {
      "epoch": 1.01,
      "grad_norm": 11.382567562865813,
      "learning_rate": 8.376823840702415e-06,
      "loss": 0.8477,
      "step": 23930
    },
    {
      "epoch": 1.01,
      "grad_norm": 31.36303889985935,
      "learning_rate": 8.375915801265575e-06,
      "loss": 0.933,
      "step": 23935
    },
    {
      "epoch": 1.01,
      "grad_norm": 17.0140828667438,
      "learning_rate": 8.375007557156905e-06,
      "loss": 0.8857,
      "step": 23940
    },
    {
      "epoch": 1.01,
      "grad_norm": 27.187262423191953,
      "learning_rate": 8.374099108431466e-06,
      "loss": 0.9232,
      "step": 23945
    },
    {
      "epoch": 1.01,
      "grad_norm": 20.594163618995726,
      "learning_rate": 8.373190455144337e-06,
      "loss": 0.8905,
      "step": 23950
    },
    {
      "epoch": 1.01,
      "grad_norm": 13.488371883894047,
      "learning_rate": 8.372281597350607e-06,
      "loss": 0.8435,
      "step": 23955
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.094525503585649,
      "learning_rate": 8.371372535105378e-06,
      "loss": 0.8701,
      "step": 23960
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.30858375756817,
      "learning_rate": 8.370463268463761e-06,
      "loss": 0.89,
      "step": 23965
    },
    {
      "epoch": 1.01,
      "grad_norm": 8.673000248740133,
      "learning_rate": 8.369553797480886e-06,
      "loss": 0.8482,
      "step": 23970
    },
    {
      "epoch": 1.01,
      "grad_norm": 5.159525108034029,
      "learning_rate": 8.368644122211889e-06,
      "loss": 0.8542,
      "step": 23975
    },
    {
      "epoch": 1.01,
      "grad_norm": 11.864629066153018,
      "learning_rate": 8.36773424271192e-06,
      "loss": 0.8511,
      "step": 23980
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.87637759382671,
      "learning_rate": 8.366824159036146e-06,
      "loss": 0.879,
      "step": 23985
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.836157320522113,
      "learning_rate": 8.365913871239739e-06,
      "loss": 0.8592,
      "step": 23990
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.937417032202544,
      "learning_rate": 8.36500337937789e-06,
      "loss": 0.8748,
      "step": 23995
    },
    {
      "epoch": 1.02,
      "grad_norm": 12.291599742739209,
      "learning_rate": 8.364092683505798e-06,
      "loss": 0.8749,
      "step": 24000
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.168247710412413,
      "learning_rate": 8.363181783678675e-06,
      "loss": 0.8529,
      "step": 24005
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.343512792677986,
      "learning_rate": 8.362270679951747e-06,
      "loss": 0.8965,
      "step": 24010
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.684304435280374,
      "learning_rate": 8.361359372380251e-06,
      "loss": 0.8765,
      "step": 24015
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.507017609525433,
      "learning_rate": 8.36044786101944e-06,
      "loss": 0.8369,
      "step": 24020
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.8398627627755655,
      "learning_rate": 8.359536145924572e-06,
      "loss": 0.8664,
      "step": 24025
    },
    {
      "epoch": 1.02,
      "grad_norm": 14.605598797139457,
      "learning_rate": 8.358624227150925e-06,
      "loss": 0.8827,
      "step": 24030
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.244928171356866,
      "learning_rate": 8.357712104753782e-06,
      "loss": 0.8636,
      "step": 24035
    },
    {
      "epoch": 1.02,
      "grad_norm": 11.498267151570188,
      "learning_rate": 8.356799778788445e-06,
      "loss": 0.8643,
      "step": 24040
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.779164263837633,
      "learning_rate": 8.355887249310225e-06,
      "loss": 0.9066,
      "step": 24045
    },
    {
      "epoch": 1.02,
      "grad_norm": 11.43695863623216,
      "learning_rate": 8.354974516374445e-06,
      "loss": 0.9021,
      "step": 24050
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.908179428442711,
      "learning_rate": 8.354061580036446e-06,
      "loss": 0.8977,
      "step": 24055
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.990950689941788,
      "learning_rate": 8.353148440351571e-06,
      "loss": 0.9168,
      "step": 24060
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.1383915589603655,
      "learning_rate": 8.352235097375184e-06,
      "loss": 0.8929,
      "step": 24065
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.947358046783045,
      "learning_rate": 8.351321551162655e-06,
      "loss": 0.8628,
      "step": 24070
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.968321599730195,
      "learning_rate": 8.350407801769373e-06,
      "loss": 0.9497,
      "step": 24075
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.109981609425329,
      "learning_rate": 8.349493849250736e-06,
      "loss": 0.842,
      "step": 24080
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.835631065221668,
      "learning_rate": 8.348579693662153e-06,
      "loss": 0.8613,
      "step": 24085
    },
    {
      "epoch": 1.02,
      "grad_norm": 9.931005737393345,
      "learning_rate": 8.347665335059047e-06,
      "loss": 0.9413,
      "step": 24090
    },
    {
      "epoch": 1.02,
      "grad_norm": 8.835393998349327,
      "learning_rate": 8.346750773496853e-06,
      "loss": 0.861,
      "step": 24095
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.334493493491368,
      "learning_rate": 8.345836009031017e-06,
      "loss": 0.8629,
      "step": 24100
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.7080029750118,
      "learning_rate": 8.344921041716997e-06,
      "loss": 0.8705,
      "step": 24105
    },
    {
      "epoch": 1.02,
      "grad_norm": 26.046007950988546,
      "learning_rate": 8.34400587161027e-06,
      "loss": 0.8762,
      "step": 24110
    },
    {
      "epoch": 1.02,
      "grad_norm": 48.86154741646651,
      "learning_rate": 8.343090498766316e-06,
      "loss": 0.8744,
      "step": 24115
    },
    {
      "epoch": 1.02,
      "grad_norm": 23.730019677936074,
      "learning_rate": 8.342174923240634e-06,
      "loss": 0.8784,
      "step": 24120
    },
    {
      "epoch": 1.02,
      "grad_norm": 27.889318036607673,
      "learning_rate": 8.34125914508873e-06,
      "loss": 0.9226,
      "step": 24125
    },
    {
      "epoch": 1.02,
      "grad_norm": 30.59932174572011,
      "learning_rate": 8.340343164366125e-06,
      "loss": 0.9453,
      "step": 24130
    },
    {
      "epoch": 1.02,
      "grad_norm": 24.94479690396093,
      "learning_rate": 8.339426981128354e-06,
      "loss": 0.8866,
      "step": 24135
    },
    {
      "epoch": 1.02,
      "grad_norm": 9.937662057467135,
      "learning_rate": 8.338510595430962e-06,
      "loss": 0.9114,
      "step": 24140
    },
    {
      "epoch": 1.02,
      "grad_norm": 13.338887814379154,
      "learning_rate": 8.337594007329506e-06,
      "loss": 0.8807,
      "step": 24145
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.492255273441464,
      "learning_rate": 8.336677216879557e-06,
      "loss": 0.9021,
      "step": 24150
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.94147429487286,
      "learning_rate": 8.335760224136696e-06,
      "loss": 0.8701,
      "step": 24155
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.783563256906834,
      "learning_rate": 8.33484302915652e-06,
      "loss": 0.8889,
      "step": 24160
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.469533691528355,
      "learning_rate": 8.333925631994633e-06,
      "loss": 0.8853,
      "step": 24165
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.800084500142283,
      "learning_rate": 8.333008032706656e-06,
      "loss": 0.8391,
      "step": 24170
    },
    {
      "epoch": 1.02,
      "grad_norm": 10.945085626406911,
      "learning_rate": 8.332090231348219e-06,
      "loss": 0.913,
      "step": 24175
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.300160162872269,
      "learning_rate": 8.331172227974966e-06,
      "loss": 0.8493,
      "step": 24180
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.756158405453012,
      "learning_rate": 8.330254022642554e-06,
      "loss": 0.8468,
      "step": 24185
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.748301353505016,
      "learning_rate": 8.32933561540665e-06,
      "loss": 0.8723,
      "step": 24190
    },
    {
      "epoch": 1.02,
      "grad_norm": 5.629260730013607,
      "learning_rate": 8.328417006322935e-06,
      "loss": 0.8901,
      "step": 24195
    },
    {
      "epoch": 1.02,
      "grad_norm": 7.4393011270578295,
      "learning_rate": 8.327498195447102e-06,
      "loss": 0.8814,
      "step": 24200
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.196855903772969,
      "learning_rate": 8.326579182834852e-06,
      "loss": 0.8143,
      "step": 24205
    },
    {
      "epoch": 1.02,
      "grad_norm": 10.643205037603853,
      "learning_rate": 8.325659968541906e-06,
      "loss": 0.8483,
      "step": 24210
    },
    {
      "epoch": 1.02,
      "grad_norm": 6.312005926330905,
      "learning_rate": 8.324740552623993e-06,
      "loss": 0.9047,
      "step": 24215
    },
    {
      "epoch": 1.03,
      "grad_norm": 11.760047657221973,
      "learning_rate": 8.323820935136852e-06,
      "loss": 0.9146,
      "step": 24220
    },
    {
      "epoch": 1.03,
      "grad_norm": 11.162311325856344,
      "learning_rate": 8.322901116136239e-06,
      "loss": 0.8813,
      "step": 24225
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.387668670128087,
      "learning_rate": 8.32198109567792e-06,
      "loss": 0.8187,
      "step": 24230
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.998859866395861,
      "learning_rate": 8.321060873817672e-06,
      "loss": 0.8478,
      "step": 24235
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.736757584286885,
      "learning_rate": 8.320140450611287e-06,
      "loss": 0.9175,
      "step": 24240
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.359264818184678,
      "learning_rate": 8.319219826114566e-06,
      "loss": 0.8786,
      "step": 24245
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.57252863320637,
      "learning_rate": 8.318299000383322e-06,
      "loss": 0.8691,
      "step": 24250
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.69145506179168,
      "learning_rate": 8.317377973473385e-06,
      "loss": 0.8591,
      "step": 24255
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.217263280146151,
      "learning_rate": 8.316456745440594e-06,
      "loss": 0.8945,
      "step": 24260
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.473210643455328,
      "learning_rate": 8.315535316340798e-06,
      "loss": 0.8751,
      "step": 24265
    },
    {
      "epoch": 1.03,
      "grad_norm": 12.182122488829963,
      "learning_rate": 8.314613686229863e-06,
      "loss": 0.8868,
      "step": 24270
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.966629692121335,
      "learning_rate": 8.313691855163664e-06,
      "loss": 0.8943,
      "step": 24275
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.868992189440033,
      "learning_rate": 8.312769823198089e-06,
      "loss": 0.8496,
      "step": 24280
    },
    {
      "epoch": 1.03,
      "grad_norm": 11.166057713549508,
      "learning_rate": 8.311847590389034e-06,
      "loss": 0.8979,
      "step": 24285
    },
    {
      "epoch": 1.03,
      "grad_norm": 9.742813899773127,
      "learning_rate": 8.31092515679242e-06,
      "loss": 0.8984,
      "step": 24290
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.312501326377522,
      "learning_rate": 8.310002522464163e-06,
      "loss": 0.8948,
      "step": 24295
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.400950836308031,
      "learning_rate": 8.309079687460204e-06,
      "loss": 0.9205,
      "step": 24300
    },
    {
      "epoch": 1.03,
      "grad_norm": 10.180131985970997,
      "learning_rate": 8.30815665183649e-06,
      "loss": 0.903,
      "step": 24305
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.731119138366711,
      "learning_rate": 8.307233415648984e-06,
      "loss": 0.864,
      "step": 24310
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.865497128228217,
      "learning_rate": 8.306309978953657e-06,
      "loss": 0.855,
      "step": 24315
    },
    {
      "epoch": 1.03,
      "grad_norm": 8.02747360707304,
      "learning_rate": 8.305386341806497e-06,
      "loss": 0.8963,
      "step": 24320
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.767420368749239,
      "learning_rate": 8.304462504263497e-06,
      "loss": 0.8671,
      "step": 24325
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.9610402111887355,
      "learning_rate": 8.303538466380669e-06,
      "loss": 0.8605,
      "step": 24330
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.627551178682409,
      "learning_rate": 8.302614228214036e-06,
      "loss": 0.9156,
      "step": 24335
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.96799829639129,
      "learning_rate": 8.30168978981963e-06,
      "loss": 0.8803,
      "step": 24340
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.73228482673005,
      "learning_rate": 8.300765151253499e-06,
      "loss": 0.843,
      "step": 24345
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.923786369193929,
      "learning_rate": 8.299840312571696e-06,
      "loss": 0.8605,
      "step": 24350
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.291284169978704,
      "learning_rate": 8.298915273830297e-06,
      "loss": 0.9109,
      "step": 24355
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.5994469548330725,
      "learning_rate": 8.297990035085384e-06,
      "loss": 0.8545,
      "step": 24360
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.832755989735144,
      "learning_rate": 8.297064596393047e-06,
      "loss": 0.8877,
      "step": 24365
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.573937446762367,
      "learning_rate": 8.296138957809396e-06,
      "loss": 0.9379,
      "step": 24370
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.121601295767272,
      "learning_rate": 8.295213119390548e-06,
      "loss": 0.9013,
      "step": 24375
    },
    {
      "epoch": 1.03,
      "grad_norm": 18.78337432516734,
      "learning_rate": 8.294287081192639e-06,
      "loss": 0.8541,
      "step": 24380
    },
    {
      "epoch": 1.03,
      "grad_norm": 15.399830764568895,
      "learning_rate": 8.293360843271802e-06,
      "loss": 0.8751,
      "step": 24385
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.652853585889718,
      "learning_rate": 8.292434405684202e-06,
      "loss": 0.8888,
      "step": 24390
    },
    {
      "epoch": 1.03,
      "grad_norm": 7.943937360585071,
      "learning_rate": 8.291507768486002e-06,
      "loss": 0.8678,
      "step": 24395
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.1991155539661325,
      "learning_rate": 8.290580931733381e-06,
      "loss": 0.9089,
      "step": 24400
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.580522247341891,
      "learning_rate": 8.28965389548253e-06,
      "loss": 0.9016,
      "step": 24405
    },
    {
      "epoch": 1.03,
      "grad_norm": 9.447246254266961,
      "learning_rate": 8.288726659789655e-06,
      "loss": 0.9149,
      "step": 24410
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.189605830176344,
      "learning_rate": 8.287799224710968e-06,
      "loss": 0.9519,
      "step": 24415
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.573528428371724,
      "learning_rate": 8.286871590302698e-06,
      "loss": 0.8535,
      "step": 24420
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.554531673121981,
      "learning_rate": 8.285943756621088e-06,
      "loss": 0.8538,
      "step": 24425
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.21153748010033,
      "learning_rate": 8.285015723722384e-06,
      "loss": 0.8856,
      "step": 24430
    },
    {
      "epoch": 1.03,
      "grad_norm": 5.415904023147494,
      "learning_rate": 8.284087491662856e-06,
      "loss": 0.859,
      "step": 24435
    },
    {
      "epoch": 1.03,
      "grad_norm": 6.072848005480355,
      "learning_rate": 8.283159060498776e-06,
      "loss": 0.8839,
      "step": 24440
    },
    {
      "epoch": 1.03,
      "grad_norm": 12.743067454312094,
      "learning_rate": 8.282230430286432e-06,
      "loss": 0.8586,
      "step": 24445
    },
    {
      "epoch": 1.03,
      "grad_norm": 13.953526049570026,
      "learning_rate": 8.281301601082126e-06,
      "loss": 0.878,
      "step": 24450
    },
    {
      "epoch": 1.03,
      "grad_norm": 23.909154613931833,
      "learning_rate": 8.280372572942169e-06,
      "loss": 0.8804,
      "step": 24455
    },
    {
      "epoch": 1.04,
      "grad_norm": 10.077103246066287,
      "learning_rate": 8.279443345922885e-06,
      "loss": 0.8442,
      "step": 24460
    },
    {
      "epoch": 1.04,
      "grad_norm": 33.90951048034864,
      "learning_rate": 8.278513920080612e-06,
      "loss": 0.8686,
      "step": 24465
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.879425851014187,
      "learning_rate": 8.277584295471697e-06,
      "loss": 0.9286,
      "step": 24470
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.75295514247547,
      "learning_rate": 8.2766544721525e-06,
      "loss": 0.9079,
      "step": 24475
    },
    {
      "epoch": 1.04,
      "grad_norm": 24.358493897548712,
      "learning_rate": 8.275724450179394e-06,
      "loss": 0.9027,
      "step": 24480
    },
    {
      "epoch": 1.04,
      "grad_norm": 13.294865160711101,
      "learning_rate": 8.274794229608764e-06,
      "loss": 0.8849,
      "step": 24485
    },
    {
      "epoch": 1.04,
      "grad_norm": 13.786356372237712,
      "learning_rate": 8.273863810497007e-06,
      "loss": 0.8372,
      "step": 24490
    },
    {
      "epoch": 1.04,
      "grad_norm": 17.221597869640195,
      "learning_rate": 8.27293319290053e-06,
      "loss": 0.8598,
      "step": 24495
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.197777378344038,
      "learning_rate": 8.272002376875753e-06,
      "loss": 0.8733,
      "step": 24500
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.144404771237298,
      "learning_rate": 8.271071362479112e-06,
      "loss": 0.8741,
      "step": 24505
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.387627758409195,
      "learning_rate": 8.270140149767048e-06,
      "loss": 0.8637,
      "step": 24510
    },
    {
      "epoch": 1.04,
      "grad_norm": 8.07399330336378,
      "learning_rate": 8.26920873879602e-06,
      "loss": 0.9039,
      "step": 24515
    },
    {
      "epoch": 1.04,
      "grad_norm": 17.435600119165812,
      "learning_rate": 8.268277129622494e-06,
      "loss": 0.8738,
      "step": 24520
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.579910558029948,
      "learning_rate": 8.267345322302956e-06,
      "loss": 0.9057,
      "step": 24525
    },
    {
      "epoch": 1.04,
      "grad_norm": 15.406908301573534,
      "learning_rate": 8.266413316893894e-06,
      "loss": 0.8414,
      "step": 24530
    },
    {
      "epoch": 1.04,
      "grad_norm": 11.09430482992784,
      "learning_rate": 8.265481113451812e-06,
      "loss": 0.8645,
      "step": 24535
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.578750850188795,
      "learning_rate": 8.26454871203323e-06,
      "loss": 0.839,
      "step": 24540
    },
    {
      "epoch": 1.04,
      "grad_norm": 18.169412030873623,
      "learning_rate": 8.263616112694676e-06,
      "loss": 0.8567,
      "step": 24545
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.213727805025681,
      "learning_rate": 8.262683315492691e-06,
      "loss": 0.888,
      "step": 24550
    },
    {
      "epoch": 1.04,
      "grad_norm": 21.114962103237136,
      "learning_rate": 8.261750320483827e-06,
      "loss": 0.8644,
      "step": 24555
    },
    {
      "epoch": 1.04,
      "grad_norm": 15.248806360276067,
      "learning_rate": 8.260817127724649e-06,
      "loss": 0.8877,
      "step": 24560
    },
    {
      "epoch": 1.04,
      "grad_norm": 16.55383841480191,
      "learning_rate": 8.259883737271732e-06,
      "loss": 0.8702,
      "step": 24565
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.683249351725927,
      "learning_rate": 8.258950149181669e-06,
      "loss": 0.8733,
      "step": 24570
    },
    {
      "epoch": 1.04,
      "grad_norm": 13.11788429551088,
      "learning_rate": 8.258016363511057e-06,
      "loss": 0.8659,
      "step": 24575
    },
    {
      "epoch": 1.04,
      "grad_norm": 14.36547218796155,
      "learning_rate": 8.257082380316508e-06,
      "loss": 0.8629,
      "step": 24580
    },
    {
      "epoch": 1.04,
      "grad_norm": 10.80398287640242,
      "learning_rate": 8.25614819965465e-06,
      "loss": 0.8544,
      "step": 24585
    },
    {
      "epoch": 1.04,
      "grad_norm": 22.711621755931816,
      "learning_rate": 8.255213821582117e-06,
      "loss": 0.8532,
      "step": 24590
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.824880181766071,
      "learning_rate": 8.25427924615556e-06,
      "loss": 0.9031,
      "step": 24595
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.654799644382605,
      "learning_rate": 8.253344473431637e-06,
      "loss": 0.867,
      "step": 24600
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.611546900082446,
      "learning_rate": 8.252409503467023e-06,
      "loss": 0.8845,
      "step": 24605
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.925623162805631,
      "learning_rate": 8.251474336318399e-06,
      "loss": 0.8977,
      "step": 24610
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.173953985280534,
      "learning_rate": 8.250538972042463e-06,
      "loss": 0.9142,
      "step": 24615
    },
    {
      "epoch": 1.04,
      "grad_norm": 5.157813884794019,
      "learning_rate": 8.249603410695925e-06,
      "loss": 0.8477,
      "step": 24620
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.965555155137428,
      "learning_rate": 8.248667652335507e-06,
      "loss": 0.8383,
      "step": 24625
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.527702275654706,
      "learning_rate": 8.247731697017936e-06,
      "loss": 0.8629,
      "step": 24630
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.443119807927224,
      "learning_rate": 8.246795544799957e-06,
      "loss": 0.8774,
      "step": 24635
    },
    {
      "epoch": 1.04,
      "grad_norm": 13.053775959688798,
      "learning_rate": 8.245859195738331e-06,
      "loss": 0.8763,
      "step": 24640
    },
    {
      "epoch": 1.04,
      "grad_norm": 10.423611695402935,
      "learning_rate": 8.24492264988982e-06,
      "loss": 0.8962,
      "step": 24645
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.565582406802536,
      "learning_rate": 8.24398590731121e-06,
      "loss": 0.9159,
      "step": 24650
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.873725323121731,
      "learning_rate": 8.243048968059287e-06,
      "loss": 0.9205,
      "step": 24655
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.5362223458521145,
      "learning_rate": 8.24211183219086e-06,
      "loss": 0.8893,
      "step": 24660
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.171531771944546,
      "learning_rate": 8.241174499762741e-06,
      "loss": 0.8547,
      "step": 24665
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.601137342930457,
      "learning_rate": 8.24023697083176e-06,
      "loss": 0.8464,
      "step": 24670
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.705145527319646,
      "learning_rate": 8.239299245454757e-06,
      "loss": 0.8584,
      "step": 24675
    },
    {
      "epoch": 1.04,
      "grad_norm": 6.238972983584677,
      "learning_rate": 8.238361323688581e-06,
      "loss": 0.8719,
      "step": 24680
    },
    {
      "epoch": 1.04,
      "grad_norm": 7.811439509500927,
      "learning_rate": 8.237423205590099e-06,
      "loss": 0.8768,
      "step": 24685
    },
    {
      "epoch": 1.04,
      "grad_norm": 9.287015520252488,
      "learning_rate": 8.23648489121618e-06,
      "loss": 0.8771,
      "step": 24690
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.292480737217653,
      "learning_rate": 8.235546380623718e-06,
      "loss": 0.8561,
      "step": 24695
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.994876796592222,
      "learning_rate": 8.23460767386961e-06,
      "loss": 0.8518,
      "step": 24700
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.109196999249068,
      "learning_rate": 8.233668771010766e-06,
      "loss": 0.8619,
      "step": 24705
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.142546213784474,
      "learning_rate": 8.232729672104109e-06,
      "loss": 0.9101,
      "step": 24710
    },
    {
      "epoch": 1.05,
      "grad_norm": 15.1050883400635,
      "learning_rate": 8.231790377206576e-06,
      "loss": 0.8911,
      "step": 24715
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.217023777828318,
      "learning_rate": 8.23085088637511e-06,
      "loss": 0.885,
      "step": 24720
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.437660790147719,
      "learning_rate": 8.229911199666671e-06,
      "loss": 0.8989,
      "step": 24725
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.611352558849838,
      "learning_rate": 8.228971317138232e-06,
      "loss": 0.8816,
      "step": 24730
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.536254236386037,
      "learning_rate": 8.228031238846774e-06,
      "loss": 0.8339,
      "step": 24735
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.415014858555917,
      "learning_rate": 8.227090964849288e-06,
      "loss": 0.8912,
      "step": 24740
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.810009369170103,
      "learning_rate": 8.226150495202782e-06,
      "loss": 0.8767,
      "step": 24745
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.885314497446922,
      "learning_rate": 8.225209829964278e-06,
      "loss": 0.8767,
      "step": 24750
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.8135293528946255,
      "learning_rate": 8.2242689691908e-06,
      "loss": 0.8604,
      "step": 24755
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.933352369879482,
      "learning_rate": 8.223327912939392e-06,
      "loss": 0.9145,
      "step": 24760
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.413063637484613,
      "learning_rate": 8.222386661267107e-06,
      "loss": 0.877,
      "step": 24765
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.793822763722874,
      "learning_rate": 8.221445214231011e-06,
      "loss": 0.8263,
      "step": 24770
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.125289140801641,
      "learning_rate": 8.22050357188818e-06,
      "loss": 0.8419,
      "step": 24775
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.8686965745449315,
      "learning_rate": 8.219561734295705e-06,
      "loss": 0.8715,
      "step": 24780
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.041420031010027,
      "learning_rate": 8.218619701510684e-06,
      "loss": 0.8789,
      "step": 24785
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.712998761765499,
      "learning_rate": 8.217677473590234e-06,
      "loss": 0.8794,
      "step": 24790
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.914252320453015,
      "learning_rate": 8.216735050591473e-06,
      "loss": 0.8808,
      "step": 24795
    },
    {
      "epoch": 1.05,
      "grad_norm": 4.09387911224469,
      "learning_rate": 8.215792432571544e-06,
      "loss": 0.8852,
      "step": 24800
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.891490035289086,
      "learning_rate": 8.214849619587591e-06,
      "loss": 0.8498,
      "step": 24805
    },
    {
      "epoch": 1.05,
      "grad_norm": 12.913462709974251,
      "learning_rate": 8.213906611696773e-06,
      "loss": 0.9566,
      "step": 24810
    },
    {
      "epoch": 1.05,
      "grad_norm": 11.33936747214587,
      "learning_rate": 8.212963408956267e-06,
      "loss": 0.9205,
      "step": 24815
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.92715170899013,
      "learning_rate": 8.212020011423254e-06,
      "loss": 0.8525,
      "step": 24820
    },
    {
      "epoch": 1.05,
      "grad_norm": 13.999563790375015,
      "learning_rate": 8.211076419154928e-06,
      "loss": 0.8383,
      "step": 24825
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.517823968007406,
      "learning_rate": 8.210132632208499e-06,
      "loss": 0.8912,
      "step": 24830
    },
    {
      "epoch": 1.05,
      "grad_norm": 10.537348980898546,
      "learning_rate": 8.20918865064118e-06,
      "loss": 0.8533,
      "step": 24835
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.28702612414382,
      "learning_rate": 8.20824447451021e-06,
      "loss": 0.8776,
      "step": 24840
    },
    {
      "epoch": 1.05,
      "grad_norm": 27.98022112669318,
      "learning_rate": 8.207300103872826e-06,
      "loss": 0.8754,
      "step": 24845
    },
    {
      "epoch": 1.05,
      "grad_norm": 10.189362496980022,
      "learning_rate": 8.206355538786284e-06,
      "loss": 0.9066,
      "step": 24850
    },
    {
      "epoch": 1.05,
      "grad_norm": 11.89154233633554,
      "learning_rate": 8.20541077930785e-06,
      "loss": 0.8611,
      "step": 24855
    },
    {
      "epoch": 1.05,
      "grad_norm": 18.76840398692148,
      "learning_rate": 8.204465825494804e-06,
      "loss": 0.8816,
      "step": 24860
    },
    {
      "epoch": 1.05,
      "grad_norm": 19.340724963940847,
      "learning_rate": 8.203520677404433e-06,
      "loss": 0.8457,
      "step": 24865
    },
    {
      "epoch": 1.05,
      "grad_norm": 13.177593523558741,
      "learning_rate": 8.20257533509404e-06,
      "loss": 0.8507,
      "step": 24870
    },
    {
      "epoch": 1.05,
      "grad_norm": 17.13913612038679,
      "learning_rate": 8.201629798620939e-06,
      "loss": 0.8685,
      "step": 24875
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.837962530031618,
      "learning_rate": 8.20068406804245e-06,
      "loss": 0.8457,
      "step": 24880
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.285817982113224,
      "learning_rate": 8.199738143415918e-06,
      "loss": 0.8517,
      "step": 24885
    },
    {
      "epoch": 1.05,
      "grad_norm": 6.5602116364480105,
      "learning_rate": 8.198792024798685e-06,
      "loss": 0.8723,
      "step": 24890
    },
    {
      "epoch": 1.05,
      "grad_norm": 10.394414026933362,
      "learning_rate": 8.197845712248116e-06,
      "loss": 0.8562,
      "step": 24895
    },
    {
      "epoch": 1.05,
      "grad_norm": 5.070725000625126,
      "learning_rate": 8.19689920582158e-06,
      "loss": 0.8566,
      "step": 24900
    },
    {
      "epoch": 1.05,
      "grad_norm": 7.396003301168955,
      "learning_rate": 8.195952505576461e-06,
      "loss": 0.8886,
      "step": 24905
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.784735276861356,
      "learning_rate": 8.195005611570157e-06,
      "loss": 0.8725,
      "step": 24910
    },
    {
      "epoch": 1.05,
      "grad_norm": 10.17933778434733,
      "learning_rate": 8.194058523860073e-06,
      "loss": 0.8548,
      "step": 24915
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.985055575213194,
      "learning_rate": 8.19311124250363e-06,
      "loss": 0.862,
      "step": 24920
    },
    {
      "epoch": 1.05,
      "grad_norm": 8.461189066968783,
      "learning_rate": 8.192163767558256e-06,
      "loss": 0.8753,
      "step": 24925
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.047149930240928,
      "learning_rate": 8.191216099081399e-06,
      "loss": 0.8741,
      "step": 24930
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.525203834218405,
      "learning_rate": 8.190268237130508e-06,
      "loss": 0.8811,
      "step": 24935
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.377899310343711,
      "learning_rate": 8.189320181763051e-06,
      "loss": 0.8645,
      "step": 24940
    },
    {
      "epoch": 1.06,
      "grad_norm": 16.156141805125113,
      "learning_rate": 8.188371933036505e-06,
      "loss": 0.8466,
      "step": 24945
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.755348336243058,
      "learning_rate": 8.187423491008362e-06,
      "loss": 0.8533,
      "step": 24950
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.298152600986602,
      "learning_rate": 8.18647485573612e-06,
      "loss": 0.8541,
      "step": 24955
    },
    {
      "epoch": 1.06,
      "grad_norm": 9.952340785786104,
      "learning_rate": 8.185526027277294e-06,
      "loss": 0.9046,
      "step": 24960
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.040647229546104,
      "learning_rate": 8.184577005689408e-06,
      "loss": 0.8904,
      "step": 24965
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.248428851700628,
      "learning_rate": 8.183627791030001e-06,
      "loss": 0.854,
      "step": 24970
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.566589666009911,
      "learning_rate": 8.182678383356617e-06,
      "loss": 0.8572,
      "step": 24975
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.080922202072892,
      "learning_rate": 8.181728782726817e-06,
      "loss": 0.945,
      "step": 24980
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.390552877339715,
      "learning_rate": 8.180778989198174e-06,
      "loss": 0.8783,
      "step": 24985
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.028702812717927,
      "learning_rate": 8.17982900282827e-06,
      "loss": 0.893,
      "step": 24990
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.214799107210826,
      "learning_rate": 8.178878823674701e-06,
      "loss": 0.8929,
      "step": 24995
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.447727340755326,
      "learning_rate": 8.17792845179507e-06,
      "loss": 0.8788,
      "step": 25000
    },
    {
      "epoch": 1.06,
      "grad_norm": 13.083600188989156,
      "learning_rate": 8.176977887246999e-06,
      "loss": 0.8993,
      "step": 25005
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.524152034080783,
      "learning_rate": 8.176027130088118e-06,
      "loss": 0.8704,
      "step": 25010
    },
    {
      "epoch": 1.06,
      "grad_norm": 14.462433914343498,
      "learning_rate": 8.175076180376066e-06,
      "loss": 0.9058,
      "step": 25015
    },
    {
      "epoch": 1.06,
      "grad_norm": 9.796374398794727,
      "learning_rate": 8.174125038168498e-06,
      "loss": 0.9308,
      "step": 25020
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.760889357393078,
      "learning_rate": 8.173173703523078e-06,
      "loss": 0.8737,
      "step": 25025
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.660280960448217,
      "learning_rate": 8.172222176497485e-06,
      "loss": 0.852,
      "step": 25030
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.746842373928137,
      "learning_rate": 8.171270457149405e-06,
      "loss": 0.8727,
      "step": 25035
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.147661343372264,
      "learning_rate": 8.170318545536538e-06,
      "loss": 0.8661,
      "step": 25040
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.330545017607854,
      "learning_rate": 8.169366441716595e-06,
      "loss": 0.8776,
      "step": 25045
    },
    {
      "epoch": 1.06,
      "grad_norm": 12.567453466442913,
      "learning_rate": 8.1684141457473e-06,
      "loss": 0.8896,
      "step": 25050
    },
    {
      "epoch": 1.06,
      "grad_norm": 4.853275954291904,
      "learning_rate": 8.16746165768639e-06,
      "loss": 0.8797,
      "step": 25055
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.67841009398702,
      "learning_rate": 8.16650897759161e-06,
      "loss": 0.8529,
      "step": 25060
    },
    {
      "epoch": 1.06,
      "grad_norm": 9.68369100127473,
      "learning_rate": 8.165556105520718e-06,
      "loss": 0.8797,
      "step": 25065
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.435460296155595,
      "learning_rate": 8.164603041531482e-06,
      "loss": 0.9023,
      "step": 25070
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.210080494681193,
      "learning_rate": 8.163649785681687e-06,
      "loss": 0.9002,
      "step": 25075
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.884237745329049,
      "learning_rate": 8.162696338029124e-06,
      "loss": 0.8902,
      "step": 25080
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.0294033902849815,
      "learning_rate": 8.161742698631598e-06,
      "loss": 0.9034,
      "step": 25085
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.97050328242212,
      "learning_rate": 8.160788867546923e-06,
      "loss": 0.8719,
      "step": 25090
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.110338959489951,
      "learning_rate": 8.159834844832932e-06,
      "loss": 0.8898,
      "step": 25095
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.663584537769211,
      "learning_rate": 8.158880630547463e-06,
      "loss": 0.8426,
      "step": 25100
    },
    {
      "epoch": 1.06,
      "grad_norm": 18.77526748751355,
      "learning_rate": 8.157926224748363e-06,
      "loss": 0.9022,
      "step": 25105
    },
    {
      "epoch": 1.06,
      "grad_norm": 12.125939002791226,
      "learning_rate": 8.156971627493499e-06,
      "loss": 0.8652,
      "step": 25110
    },
    {
      "epoch": 1.06,
      "grad_norm": 16.92518764101726,
      "learning_rate": 8.156016838840746e-06,
      "loss": 0.885,
      "step": 25115
    },
    {
      "epoch": 1.06,
      "grad_norm": 7.161537676888788,
      "learning_rate": 8.155061858847987e-06,
      "loss": 0.8601,
      "step": 25120
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.151492349625277,
      "learning_rate": 8.154106687573122e-06,
      "loss": 0.8006,
      "step": 25125
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.459551875399119,
      "learning_rate": 8.153151325074058e-06,
      "loss": 0.8585,
      "step": 25130
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.223074625967917,
      "learning_rate": 8.152195771408717e-06,
      "loss": 0.877,
      "step": 25135
    },
    {
      "epoch": 1.06,
      "grad_norm": 8.409045252801864,
      "learning_rate": 8.151240026635033e-06,
      "loss": 0.8777,
      "step": 25140
    },
    {
      "epoch": 1.06,
      "grad_norm": 17.709059014264987,
      "learning_rate": 8.150284090810948e-06,
      "loss": 0.8717,
      "step": 25145
    },
    {
      "epoch": 1.06,
      "grad_norm": 15.421726616259106,
      "learning_rate": 8.149327963994417e-06,
      "loss": 0.8602,
      "step": 25150
    },
    {
      "epoch": 1.06,
      "grad_norm": 5.232292331829905,
      "learning_rate": 8.14837164624341e-06,
      "loss": 0.8907,
      "step": 25155
    },
    {
      "epoch": 1.06,
      "grad_norm": 6.69605382126191,
      "learning_rate": 8.147415137615902e-06,
      "loss": 0.933,
      "step": 25160
    },
    {
      "epoch": 1.07,
      "grad_norm": 15.386834417103419,
      "learning_rate": 8.146458438169887e-06,
      "loss": 0.8893,
      "step": 25165
    },
    {
      "epoch": 1.07,
      "grad_norm": 13.622953122638997,
      "learning_rate": 8.145501547963364e-06,
      "loss": 0.8827,
      "step": 25170
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.482481357356965,
      "learning_rate": 8.144544467054347e-06,
      "loss": 0.8498,
      "step": 25175
    },
    {
      "epoch": 1.07,
      "grad_norm": 18.59822555373446,
      "learning_rate": 8.143587195500861e-06,
      "loss": 0.9383,
      "step": 25180
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.04318198282557,
      "learning_rate": 8.142629733360944e-06,
      "loss": 0.9337,
      "step": 25185
    },
    {
      "epoch": 1.07,
      "grad_norm": 19.16056344299214,
      "learning_rate": 8.141672080692645e-06,
      "loss": 0.8785,
      "step": 25190
    },
    {
      "epoch": 1.07,
      "grad_norm": 13.642227117383678,
      "learning_rate": 8.140714237554019e-06,
      "loss": 0.9065,
      "step": 25195
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.207305919383431,
      "learning_rate": 8.139756204003141e-06,
      "loss": 0.8765,
      "step": 25200
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.879306934148238,
      "learning_rate": 8.138797980098094e-06,
      "loss": 0.8544,
      "step": 25205
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.176265237688608,
      "learning_rate": 8.13783956589697e-06,
      "loss": 0.8793,
      "step": 25210
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.371276628003012,
      "learning_rate": 8.136880961457876e-06,
      "loss": 0.9012,
      "step": 25215
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.845515552368298,
      "learning_rate": 8.13592216683893e-06,
      "loss": 0.9124,
      "step": 25220
    },
    {
      "epoch": 1.07,
      "grad_norm": 42.02464244322,
      "learning_rate": 8.134963182098258e-06,
      "loss": 0.8717,
      "step": 25225
    },
    {
      "epoch": 1.07,
      "grad_norm": 27.624042002708634,
      "learning_rate": 8.134004007294005e-06,
      "loss": 0.8573,
      "step": 25230
    },
    {
      "epoch": 1.07,
      "grad_norm": 14.055011799851112,
      "learning_rate": 8.13304464248432e-06,
      "loss": 0.8961,
      "step": 25235
    },
    {
      "epoch": 1.07,
      "grad_norm": 27.219485466255527,
      "learning_rate": 8.132085087727369e-06,
      "loss": 0.8261,
      "step": 25240
    },
    {
      "epoch": 1.07,
      "grad_norm": 14.717331427640557,
      "learning_rate": 8.131125343081321e-06,
      "loss": 0.8214,
      "step": 25245
    },
    {
      "epoch": 1.07,
      "grad_norm": 16.060883445361377,
      "learning_rate": 8.130165408604369e-06,
      "loss": 0.8898,
      "step": 25250
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.371913244279149,
      "learning_rate": 8.129205284354707e-06,
      "loss": 0.8827,
      "step": 25255
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.520258042709106,
      "learning_rate": 8.128244970390546e-06,
      "loss": 0.9218,
      "step": 25260
    },
    {
      "epoch": 1.07,
      "grad_norm": 17.140680393577295,
      "learning_rate": 8.127284466770108e-06,
      "loss": 0.8814,
      "step": 25265
    },
    {
      "epoch": 1.07,
      "grad_norm": 12.450424514421915,
      "learning_rate": 8.126323773551624e-06,
      "loss": 0.8496,
      "step": 25270
    },
    {
      "epoch": 1.07,
      "grad_norm": 14.037960018758197,
      "learning_rate": 8.125362890793339e-06,
      "loss": 0.9003,
      "step": 25275
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.500772545688967,
      "learning_rate": 8.124401818553508e-06,
      "loss": 0.8611,
      "step": 25280
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.951178068148634,
      "learning_rate": 8.123440556890399e-06,
      "loss": 0.8481,
      "step": 25285
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.482921304706396,
      "learning_rate": 8.122479105862286e-06,
      "loss": 0.9034,
      "step": 25290
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.879728783353924,
      "learning_rate": 8.121517465527465e-06,
      "loss": 0.8653,
      "step": 25295
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.46199076765795,
      "learning_rate": 8.120555635944234e-06,
      "loss": 0.8787,
      "step": 25300
    },
    {
      "epoch": 1.07,
      "grad_norm": 14.245695291533346,
      "learning_rate": 8.119593617170906e-06,
      "loss": 0.872,
      "step": 25305
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.47106259526486,
      "learning_rate": 8.118631409265809e-06,
      "loss": 0.8575,
      "step": 25310
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.393189694720299,
      "learning_rate": 8.117669012287274e-06,
      "loss": 0.8337,
      "step": 25315
    },
    {
      "epoch": 1.07,
      "grad_norm": 11.706976207005242,
      "learning_rate": 8.11670642629365e-06,
      "loss": 0.8796,
      "step": 25320
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.298548483288503,
      "learning_rate": 8.115743651343295e-06,
      "loss": 0.8362,
      "step": 25325
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.62089790959027,
      "learning_rate": 8.11478068749458e-06,
      "loss": 0.8517,
      "step": 25330
    },
    {
      "epoch": 1.07,
      "grad_norm": 13.217788235010783,
      "learning_rate": 8.11381753480589e-06,
      "loss": 0.8825,
      "step": 25335
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.368228159237754,
      "learning_rate": 8.112854193335613e-06,
      "loss": 0.8542,
      "step": 25340
    },
    {
      "epoch": 1.07,
      "grad_norm": 11.153957631591267,
      "learning_rate": 8.111890663142154e-06,
      "loss": 0.8592,
      "step": 25345
    },
    {
      "epoch": 1.07,
      "grad_norm": 16.906510396923906,
      "learning_rate": 8.110926944283932e-06,
      "loss": 0.8709,
      "step": 25350
    },
    {
      "epoch": 1.07,
      "grad_norm": 21.694519542909006,
      "learning_rate": 8.109963036819373e-06,
      "loss": 0.8463,
      "step": 25355
    },
    {
      "epoch": 1.07,
      "grad_norm": 16.810842855949446,
      "learning_rate": 8.108998940806916e-06,
      "loss": 0.8817,
      "step": 25360
    },
    {
      "epoch": 1.07,
      "grad_norm": 10.398551787132956,
      "learning_rate": 8.108034656305011e-06,
      "loss": 0.8719,
      "step": 25365
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.639916160900846,
      "learning_rate": 8.107070183372118e-06,
      "loss": 0.8351,
      "step": 25370
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.263861379214783,
      "learning_rate": 8.106105522066715e-06,
      "loss": 0.8545,
      "step": 25375
    },
    {
      "epoch": 1.07,
      "grad_norm": 9.449413336189044,
      "learning_rate": 8.10514067244728e-06,
      "loss": 0.8616,
      "step": 25380
    },
    {
      "epoch": 1.07,
      "grad_norm": 5.941075232733145,
      "learning_rate": 8.104175634572316e-06,
      "loss": 0.8422,
      "step": 25385
    },
    {
      "epoch": 1.07,
      "grad_norm": 8.043681790486543,
      "learning_rate": 8.103210408500324e-06,
      "loss": 0.8653,
      "step": 25390
    },
    {
      "epoch": 1.07,
      "grad_norm": 6.449934044436362,
      "learning_rate": 8.102244994289827e-06,
      "loss": 0.8426,
      "step": 25395
    },
    {
      "epoch": 1.07,
      "grad_norm": 7.788802553175485,
      "learning_rate": 8.101279391999354e-06,
      "loss": 0.8309,
      "step": 25400
    },
    {
      "epoch": 1.08,
      "grad_norm": 10.688512780888843,
      "learning_rate": 8.100313601687448e-06,
      "loss": 0.8788,
      "step": 25405
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.769083051412717,
      "learning_rate": 8.099347623412659e-06,
      "loss": 0.8753,
      "step": 25410
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.876451117223648,
      "learning_rate": 8.098381457233553e-06,
      "loss": 0.8761,
      "step": 25415
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.8238583766418985,
      "learning_rate": 8.097415103208706e-06,
      "loss": 0.8621,
      "step": 25420
    },
    {
      "epoch": 1.08,
      "grad_norm": 10.976909261461111,
      "learning_rate": 8.096448561396706e-06,
      "loss": 0.8841,
      "step": 25425
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.512689370962415,
      "learning_rate": 8.095481831856148e-06,
      "loss": 0.8832,
      "step": 25430
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.39359475834383,
      "learning_rate": 8.094514914645646e-06,
      "loss": 0.8979,
      "step": 25435
    },
    {
      "epoch": 1.08,
      "grad_norm": 17.400975150587065,
      "learning_rate": 8.093547809823821e-06,
      "loss": 0.8933,
      "step": 25440
    },
    {
      "epoch": 1.08,
      "grad_norm": 17.616837411935776,
      "learning_rate": 8.092580517449305e-06,
      "loss": 0.7896,
      "step": 25445
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.758981529465788,
      "learning_rate": 8.091613037580739e-06,
      "loss": 0.887,
      "step": 25450
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.40101586473215,
      "learning_rate": 8.090645370276783e-06,
      "loss": 0.9036,
      "step": 25455
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.714757552984231,
      "learning_rate": 8.0896775155961e-06,
      "loss": 0.917,
      "step": 25460
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.728322393244592,
      "learning_rate": 8.088709473597373e-06,
      "loss": 0.8589,
      "step": 25465
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.721356900003896,
      "learning_rate": 8.087741244339287e-06,
      "loss": 0.8685,
      "step": 25470
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.501083457602186,
      "learning_rate": 8.086772827880547e-06,
      "loss": 0.8691,
      "step": 25475
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.5141982684354796,
      "learning_rate": 8.085804224279861e-06,
      "loss": 0.822,
      "step": 25480
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.28554874554032,
      "learning_rate": 8.084835433595954e-06,
      "loss": 0.8734,
      "step": 25485
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.309531759924036,
      "learning_rate": 8.083866455887562e-06,
      "loss": 0.8477,
      "step": 25490
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.522826605714632,
      "learning_rate": 8.08289729121343e-06,
      "loss": 0.8694,
      "step": 25495
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.906033594042015,
      "learning_rate": 8.081927939632317e-06,
      "loss": 0.8566,
      "step": 25500
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.350908645257273,
      "learning_rate": 8.080958401202991e-06,
      "loss": 0.8299,
      "step": 25505
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.3371257898694155,
      "learning_rate": 8.079988675984231e-06,
      "loss": 0.874,
      "step": 25510
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.735118510974356,
      "learning_rate": 8.079018764034832e-06,
      "loss": 0.8469,
      "step": 25515
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.1169073382059205,
      "learning_rate": 8.078048665413594e-06,
      "loss": 0.8729,
      "step": 25520
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.048799588499916,
      "learning_rate": 8.077078380179331e-06,
      "loss": 0.8951,
      "step": 25525
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.800440727369588,
      "learning_rate": 8.076107908390872e-06,
      "loss": 0.8499,
      "step": 25530
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.799500015142693,
      "learning_rate": 8.07513725010705e-06,
      "loss": 0.8327,
      "step": 25535
    },
    {
      "epoch": 1.08,
      "grad_norm": 7.476954778671344,
      "learning_rate": 8.074166405386714e-06,
      "loss": 0.9115,
      "step": 25540
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.886379614249767,
      "learning_rate": 8.073195374288725e-06,
      "loss": 0.8922,
      "step": 25545
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.47774889886093,
      "learning_rate": 8.072224156871953e-06,
      "loss": 0.8281,
      "step": 25550
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.910630046193432,
      "learning_rate": 8.07125275319528e-06,
      "loss": 0.866,
      "step": 25555
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.843111271943483,
      "learning_rate": 8.0702811633176e-06,
      "loss": 0.8739,
      "step": 25560
    },
    {
      "epoch": 1.08,
      "grad_norm": 10.183151655698447,
      "learning_rate": 8.069309387297814e-06,
      "loss": 0.8692,
      "step": 25565
    },
    {
      "epoch": 1.08,
      "grad_norm": 17.049354155448636,
      "learning_rate": 8.068337425194844e-06,
      "loss": 0.8894,
      "step": 25570
    },
    {
      "epoch": 1.08,
      "grad_norm": 13.408895707763493,
      "learning_rate": 8.067365277067612e-06,
      "loss": 0.8692,
      "step": 25575
    },
    {
      "epoch": 1.08,
      "grad_norm": 20.460245148921782,
      "learning_rate": 8.066392942975059e-06,
      "loss": 0.878,
      "step": 25580
    },
    {
      "epoch": 1.08,
      "grad_norm": 11.358839716949952,
      "learning_rate": 8.065420422976135e-06,
      "loss": 0.8634,
      "step": 25585
    },
    {
      "epoch": 1.08,
      "grad_norm": 4.455306316482305,
      "learning_rate": 8.0644477171298e-06,
      "loss": 0.8695,
      "step": 25590
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.568262201494377,
      "learning_rate": 8.063474825495026e-06,
      "loss": 0.8589,
      "step": 25595
    },
    {
      "epoch": 1.08,
      "grad_norm": 9.823863298858734,
      "learning_rate": 8.062501748130796e-06,
      "loss": 0.8362,
      "step": 25600
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.945576410242681,
      "learning_rate": 8.061528485096107e-06,
      "loss": 0.8614,
      "step": 25605
    },
    {
      "epoch": 1.08,
      "grad_norm": 6.855890023407875,
      "learning_rate": 8.060555036449965e-06,
      "loss": 0.8372,
      "step": 25610
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.994632923702126,
      "learning_rate": 8.059581402251384e-06,
      "loss": 0.8975,
      "step": 25615
    },
    {
      "epoch": 1.08,
      "grad_norm": 9.732606836065855,
      "learning_rate": 8.058607582559397e-06,
      "loss": 0.8667,
      "step": 25620
    },
    {
      "epoch": 1.08,
      "grad_norm": 5.514484607296094,
      "learning_rate": 8.05763357743304e-06,
      "loss": 0.8502,
      "step": 25625
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.33383914685765,
      "learning_rate": 8.056659386931367e-06,
      "loss": 0.8803,
      "step": 25630
    },
    {
      "epoch": 1.08,
      "grad_norm": 8.781920285047889,
      "learning_rate": 8.055685011113439e-06,
      "loss": 0.872,
      "step": 25635
    },
    {
      "epoch": 1.09,
      "grad_norm": 20.052349810923946,
      "learning_rate": 8.05471045003833e-06,
      "loss": 0.8793,
      "step": 25640
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.865399164145096,
      "learning_rate": 8.053735703765125e-06,
      "loss": 0.9175,
      "step": 25645
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.176753269916266,
      "learning_rate": 8.052760772352916e-06,
      "loss": 0.8524,
      "step": 25650
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.148275771150095,
      "learning_rate": 8.051785655860817e-06,
      "loss": 0.8461,
      "step": 25655
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.877476467046608,
      "learning_rate": 8.050810354347941e-06,
      "loss": 0.8777,
      "step": 25660
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.24096921947043,
      "learning_rate": 8.04983486787342e-06,
      "loss": 0.9171,
      "step": 25665
    },
    {
      "epoch": 1.09,
      "grad_norm": 13.776186080489287,
      "learning_rate": 8.048859196496395e-06,
      "loss": 0.8736,
      "step": 25670
    },
    {
      "epoch": 1.09,
      "grad_norm": 11.168948874751703,
      "learning_rate": 8.047883340276016e-06,
      "loss": 0.8428,
      "step": 25675
    },
    {
      "epoch": 1.09,
      "grad_norm": 16.58707484326654,
      "learning_rate": 8.046907299271451e-06,
      "loss": 0.8719,
      "step": 25680
    },
    {
      "epoch": 1.09,
      "grad_norm": 13.407123580900468,
      "learning_rate": 8.045931073541869e-06,
      "loss": 0.8842,
      "step": 25685
    },
    {
      "epoch": 1.09,
      "grad_norm": 11.688239916780878,
      "learning_rate": 8.04495466314646e-06,
      "loss": 0.8484,
      "step": 25690
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.0145413475936875,
      "learning_rate": 8.043978068144418e-06,
      "loss": 0.9074,
      "step": 25695
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.655220322296769,
      "learning_rate": 8.043001288594952e-06,
      "loss": 0.832,
      "step": 25700
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.655896965371052,
      "learning_rate": 8.04202432455728e-06,
      "loss": 0.8705,
      "step": 25705
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.646697002295684,
      "learning_rate": 8.041047176090635e-06,
      "loss": 0.8692,
      "step": 25710
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.943349605285469,
      "learning_rate": 8.040069843254258e-06,
      "loss": 0.8486,
      "step": 25715
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.3644949390245165,
      "learning_rate": 8.0390923261074e-06,
      "loss": 0.8522,
      "step": 25720
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.3719193998857095,
      "learning_rate": 8.038114624709328e-06,
      "loss": 0.8685,
      "step": 25725
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.86060402463312,
      "learning_rate": 8.037136739119315e-06,
      "loss": 0.8505,
      "step": 25730
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.0805585058419,
      "learning_rate": 8.036158669396646e-06,
      "loss": 0.9115,
      "step": 25735
    },
    {
      "epoch": 1.09,
      "grad_norm": 18.465607613503146,
      "learning_rate": 8.035180415600621e-06,
      "loss": 0.8402,
      "step": 25740
    },
    {
      "epoch": 1.09,
      "grad_norm": 16.352602004998136,
      "learning_rate": 8.034201977790548e-06,
      "loss": 0.8695,
      "step": 25745
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.370508561390915,
      "learning_rate": 8.033223356025746e-06,
      "loss": 0.8559,
      "step": 25750
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.251686379445301,
      "learning_rate": 8.032244550365549e-06,
      "loss": 0.8805,
      "step": 25755
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.399152481844639,
      "learning_rate": 8.031265560869293e-06,
      "loss": 0.8776,
      "step": 25760
    },
    {
      "epoch": 1.09,
      "grad_norm": 19.369761465997048,
      "learning_rate": 8.030286387596337e-06,
      "loss": 0.8584,
      "step": 25765
    },
    {
      "epoch": 1.09,
      "grad_norm": 23.655973894547344,
      "learning_rate": 8.029307030606043e-06,
      "loss": 0.8815,
      "step": 25770
    },
    {
      "epoch": 1.09,
      "grad_norm": 10.528715431816428,
      "learning_rate": 8.028327489957787e-06,
      "loss": 0.9026,
      "step": 25775
    },
    {
      "epoch": 1.09,
      "grad_norm": 13.824315673789933,
      "learning_rate": 8.027347765710956e-06,
      "loss": 0.8678,
      "step": 25780
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.271860445422015,
      "learning_rate": 8.026367857924945e-06,
      "loss": 0.8436,
      "step": 25785
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.070055992732557,
      "learning_rate": 8.025387766659169e-06,
      "loss": 0.8853,
      "step": 25790
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.787664628073545,
      "learning_rate": 8.024407491973041e-06,
      "loss": 0.8603,
      "step": 25795
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.48574233385568,
      "learning_rate": 8.023427033925998e-06,
      "loss": 0.8843,
      "step": 25800
    },
    {
      "epoch": 1.09,
      "grad_norm": 17.14355182985522,
      "learning_rate": 8.02244639257748e-06,
      "loss": 0.8622,
      "step": 25805
    },
    {
      "epoch": 1.09,
      "grad_norm": 22.698802917790847,
      "learning_rate": 8.021465567986937e-06,
      "loss": 0.8623,
      "step": 25810
    },
    {
      "epoch": 1.09,
      "grad_norm": 24.978878559406873,
      "learning_rate": 8.020484560213839e-06,
      "loss": 0.8768,
      "step": 25815
    },
    {
      "epoch": 1.09,
      "grad_norm": 7.859632622654838,
      "learning_rate": 8.019503369317658e-06,
      "loss": 0.8328,
      "step": 25820
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.156135491009852,
      "learning_rate": 8.018521995357881e-06,
      "loss": 0.8322,
      "step": 25825
    },
    {
      "epoch": 1.09,
      "grad_norm": 6.1718075339290275,
      "learning_rate": 8.01754043839401e-06,
      "loss": 0.8852,
      "step": 25830
    },
    {
      "epoch": 1.09,
      "grad_norm": 14.868105382476271,
      "learning_rate": 8.016558698485548e-06,
      "loss": 0.906,
      "step": 25835
    },
    {
      "epoch": 1.09,
      "grad_norm": 10.586508122436015,
      "learning_rate": 8.015576775692018e-06,
      "loss": 0.8814,
      "step": 25840
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.153466709562309,
      "learning_rate": 8.01459467007295e-06,
      "loss": 0.8317,
      "step": 25845
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.773015256721841,
      "learning_rate": 8.01361238168789e-06,
      "loss": 0.9061,
      "step": 25850
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.363952338448904,
      "learning_rate": 8.012629910596386e-06,
      "loss": 0.8457,
      "step": 25855
    },
    {
      "epoch": 1.09,
      "grad_norm": 5.3403625219143525,
      "learning_rate": 8.011647256858005e-06,
      "loss": 0.8697,
      "step": 25860
    },
    {
      "epoch": 1.09,
      "grad_norm": 11.559433307459656,
      "learning_rate": 8.010664420532322e-06,
      "loss": 0.8599,
      "step": 25865
    },
    {
      "epoch": 1.09,
      "grad_norm": 13.611868273622242,
      "learning_rate": 8.009681401678923e-06,
      "loss": 0.8354,
      "step": 25870
    },
    {
      "epoch": 1.1,
      "grad_norm": 13.184817812150095,
      "learning_rate": 8.008698200357405e-06,
      "loss": 0.888,
      "step": 25875
    },
    {
      "epoch": 1.1,
      "grad_norm": 13.778169734887975,
      "learning_rate": 8.00771481662738e-06,
      "loss": 0.8199,
      "step": 25880
    },
    {
      "epoch": 1.1,
      "grad_norm": 13.961395734334369,
      "learning_rate": 8.006731250548464e-06,
      "loss": 0.9039,
      "step": 25885
    },
    {
      "epoch": 1.1,
      "grad_norm": 20.475687786303837,
      "learning_rate": 8.005747502180288e-06,
      "loss": 0.8698,
      "step": 25890
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.470655738657399,
      "learning_rate": 8.004763571582497e-06,
      "loss": 0.842,
      "step": 25895
    },
    {
      "epoch": 1.1,
      "grad_norm": 9.670784407083827,
      "learning_rate": 8.003779458814739e-06,
      "loss": 0.8929,
      "step": 25900
    },
    {
      "epoch": 1.1,
      "grad_norm": 13.447836649571068,
      "learning_rate": 8.002795163936681e-06,
      "loss": 0.9021,
      "step": 25905
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.895677954365911,
      "learning_rate": 8.001810687007997e-06,
      "loss": 0.8484,
      "step": 25910
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.348058610796533,
      "learning_rate": 8.000826028088374e-06,
      "loss": 0.8678,
      "step": 25915
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.442044068857169,
      "learning_rate": 7.999841187237507e-06,
      "loss": 0.8438,
      "step": 25920
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.697303256082096,
      "learning_rate": 7.998856164515107e-06,
      "loss": 0.8572,
      "step": 25925
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.417868423191606,
      "learning_rate": 7.997870959980889e-06,
      "loss": 0.9233,
      "step": 25930
    },
    {
      "epoch": 1.1,
      "grad_norm": 9.660283134574927,
      "learning_rate": 7.996885573694585e-06,
      "loss": 0.8578,
      "step": 25935
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.413815718340507,
      "learning_rate": 7.99590000571594e-06,
      "loss": 0.8521,
      "step": 25940
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.66924631816616,
      "learning_rate": 7.9949142561047e-06,
      "loss": 0.8689,
      "step": 25945
    },
    {
      "epoch": 1.1,
      "grad_norm": 9.7151451929432,
      "learning_rate": 7.99392832492063e-06,
      "loss": 0.8809,
      "step": 25950
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.061240815707852,
      "learning_rate": 7.992942212223504e-06,
      "loss": 0.8851,
      "step": 25955
    },
    {
      "epoch": 1.1,
      "grad_norm": 15.479900462634221,
      "learning_rate": 7.991955918073109e-06,
      "loss": 0.8816,
      "step": 25960
    },
    {
      "epoch": 1.1,
      "grad_norm": 12.548875450606191,
      "learning_rate": 7.99096944252924e-06,
      "loss": 0.8834,
      "step": 25965
    },
    {
      "epoch": 1.1,
      "grad_norm": 22.6096576128708,
      "learning_rate": 7.989982785651702e-06,
      "loss": 0.8738,
      "step": 25970
    },
    {
      "epoch": 1.1,
      "grad_norm": 22.369854671946587,
      "learning_rate": 7.988995947500317e-06,
      "loss": 0.8493,
      "step": 25975
    },
    {
      "epoch": 1.1,
      "grad_norm": 27.00822721519099,
      "learning_rate": 7.988008928134911e-06,
      "loss": 0.8446,
      "step": 25980
    },
    {
      "epoch": 1.1,
      "grad_norm": 42.04359968538319,
      "learning_rate": 7.987021727615326e-06,
      "loss": 0.8649,
      "step": 25985
    },
    {
      "epoch": 1.1,
      "grad_norm": 39.66896717495057,
      "learning_rate": 7.986034346001413e-06,
      "loss": 0.8756,
      "step": 25990
    },
    {
      "epoch": 1.1,
      "grad_norm": 12.514514387072678,
      "learning_rate": 7.98504678335303e-06,
      "loss": 0.8687,
      "step": 25995
    },
    {
      "epoch": 1.1,
      "grad_norm": 21.961964168573633,
      "learning_rate": 7.984059039730057e-06,
      "loss": 0.9258,
      "step": 26000
    },
    {
      "epoch": 1.1,
      "grad_norm": 10.587983893889762,
      "learning_rate": 7.983071115192372e-06,
      "loss": 0.8884,
      "step": 26005
    },
    {
      "epoch": 1.1,
      "grad_norm": 10.909622389012378,
      "learning_rate": 7.982083009799873e-06,
      "loss": 0.862,
      "step": 26010
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.483467070218326,
      "learning_rate": 7.981094723612463e-06,
      "loss": 0.8729,
      "step": 26015
    },
    {
      "epoch": 1.1,
      "grad_norm": 10.129897051040773,
      "learning_rate": 7.980106256690063e-06,
      "loss": 0.8269,
      "step": 26020
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.308340333664583,
      "learning_rate": 7.9791176090926e-06,
      "loss": 0.8409,
      "step": 26025
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.864691547736599,
      "learning_rate": 7.97812878088001e-06,
      "loss": 0.8548,
      "step": 26030
    },
    {
      "epoch": 1.1,
      "grad_norm": 11.76402197889257,
      "learning_rate": 7.977139772112245e-06,
      "loss": 0.8513,
      "step": 26035
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.422863501669311,
      "learning_rate": 7.976150582849265e-06,
      "loss": 0.8474,
      "step": 26040
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.193621857563075,
      "learning_rate": 7.975161213151041e-06,
      "loss": 0.8691,
      "step": 26045
    },
    {
      "epoch": 1.1,
      "grad_norm": 10.366649439526833,
      "learning_rate": 7.974171663077557e-06,
      "loss": 0.8842,
      "step": 26050
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.381058963025625,
      "learning_rate": 7.973181932688805e-06,
      "loss": 0.8432,
      "step": 26055
    },
    {
      "epoch": 1.1,
      "grad_norm": 12.788441498061113,
      "learning_rate": 7.97219202204479e-06,
      "loss": 0.888,
      "step": 26060
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.723974431855365,
      "learning_rate": 7.97120193120553e-06,
      "loss": 0.8761,
      "step": 26065
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.515420774520167,
      "learning_rate": 7.970211660231046e-06,
      "loss": 0.8214,
      "step": 26070
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.359362255999275,
      "learning_rate": 7.969221209181379e-06,
      "loss": 0.8413,
      "step": 26075
    },
    {
      "epoch": 1.1,
      "grad_norm": 7.7258402674579205,
      "learning_rate": 7.968230578116578e-06,
      "loss": 0.936,
      "step": 26080
    },
    {
      "epoch": 1.1,
      "grad_norm": 6.88223033280284,
      "learning_rate": 7.967239767096699e-06,
      "loss": 0.8554,
      "step": 26085
    },
    {
      "epoch": 1.1,
      "grad_norm": 8.223486309806825,
      "learning_rate": 7.966248776181812e-06,
      "loss": 0.8501,
      "step": 26090
    },
    {
      "epoch": 1.1,
      "grad_norm": 9.35221063475463,
      "learning_rate": 7.965257605432e-06,
      "loss": 0.8774,
      "step": 26095
    },
    {
      "epoch": 1.1,
      "grad_norm": 5.706253774820753,
      "learning_rate": 7.964266254907353e-06,
      "loss": 0.8865,
      "step": 26100
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.810241986121962,
      "learning_rate": 7.963274724667977e-06,
      "loss": 0.8506,
      "step": 26105
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.524693264060233,
      "learning_rate": 7.962283014773981e-06,
      "loss": 0.8677,
      "step": 26110
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.647485460122754,
      "learning_rate": 7.961291125285492e-06,
      "loss": 0.8864,
      "step": 26115
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.342578105314222,
      "learning_rate": 7.960299056262645e-06,
      "loss": 0.8742,
      "step": 26120
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.883492960986341,
      "learning_rate": 7.959306807765586e-06,
      "loss": 0.8624,
      "step": 26125
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.2127920605974545,
      "learning_rate": 7.958314379854473e-06,
      "loss": 0.8546,
      "step": 26130
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.358431148667071,
      "learning_rate": 7.957321772589471e-06,
      "loss": 0.8813,
      "step": 26135
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.083872585523129,
      "learning_rate": 7.956328986030763e-06,
      "loss": 0.849,
      "step": 26140
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.352373832275167,
      "learning_rate": 7.955336020238536e-06,
      "loss": 0.8705,
      "step": 26145
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.708133325457954,
      "learning_rate": 7.954342875272991e-06,
      "loss": 0.8724,
      "step": 26150
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.127430736276292,
      "learning_rate": 7.953349551194339e-06,
      "loss": 0.8599,
      "step": 26155
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.438810760041203,
      "learning_rate": 7.952356048062804e-06,
      "loss": 0.8983,
      "step": 26160
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.715100680774751,
      "learning_rate": 7.95136236593862e-06,
      "loss": 0.8521,
      "step": 26165
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.458606231230873,
      "learning_rate": 7.950368504882027e-06,
      "loss": 0.871,
      "step": 26170
    },
    {
      "epoch": 1.11,
      "grad_norm": 11.749413807861309,
      "learning_rate": 7.949374464953282e-06,
      "loss": 0.8538,
      "step": 26175
    },
    {
      "epoch": 1.11,
      "grad_norm": 13.335430301614918,
      "learning_rate": 7.948380246212652e-06,
      "loss": 0.8431,
      "step": 26180
    },
    {
      "epoch": 1.11,
      "grad_norm": 21.461542093516865,
      "learning_rate": 7.947385848720412e-06,
      "loss": 0.828,
      "step": 26185
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.388743793404191,
      "learning_rate": 7.94639127253685e-06,
      "loss": 0.8589,
      "step": 26190
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.602818741191548,
      "learning_rate": 7.945396517722264e-06,
      "loss": 0.8574,
      "step": 26195
    },
    {
      "epoch": 1.11,
      "grad_norm": 14.37538850341335,
      "learning_rate": 7.944401584336962e-06,
      "loss": 0.8204,
      "step": 26200
    },
    {
      "epoch": 1.11,
      "grad_norm": 12.14909866165158,
      "learning_rate": 7.943406472441267e-06,
      "loss": 0.8342,
      "step": 26205
    },
    {
      "epoch": 1.11,
      "grad_norm": 17.678975504214247,
      "learning_rate": 7.942411182095506e-06,
      "loss": 0.8153,
      "step": 26210
    },
    {
      "epoch": 1.11,
      "grad_norm": 14.938447870937582,
      "learning_rate": 7.941415713360024e-06,
      "loss": 0.8592,
      "step": 26215
    },
    {
      "epoch": 1.11,
      "grad_norm": 16.556492004981912,
      "learning_rate": 7.940420066295169e-06,
      "loss": 0.8618,
      "step": 26220
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.675382693372199,
      "learning_rate": 7.939424240961308e-06,
      "loss": 0.8701,
      "step": 26225
    },
    {
      "epoch": 1.11,
      "grad_norm": 13.944574998291337,
      "learning_rate": 7.938428237418814e-06,
      "loss": 0.8584,
      "step": 26230
    },
    {
      "epoch": 1.11,
      "grad_norm": 12.331353040685906,
      "learning_rate": 7.937432055728072e-06,
      "loss": 0.8625,
      "step": 26235
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.699687380772396,
      "learning_rate": 7.936435695949474e-06,
      "loss": 0.8612,
      "step": 26240
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.0692182989667876,
      "learning_rate": 7.935439158143432e-06,
      "loss": 0.8632,
      "step": 26245
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.4494704988481075,
      "learning_rate": 7.934442442370361e-06,
      "loss": 0.8861,
      "step": 26250
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.7805451667261645,
      "learning_rate": 7.933445548690687e-06,
      "loss": 0.897,
      "step": 26255
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.79196521993131,
      "learning_rate": 7.932448477164849e-06,
      "loss": 0.8042,
      "step": 26260
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.499576239836002,
      "learning_rate": 7.931451227853299e-06,
      "loss": 0.856,
      "step": 26265
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.230448573932353,
      "learning_rate": 7.930453800816495e-06,
      "loss": 0.8556,
      "step": 26270
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.982385423422259,
      "learning_rate": 7.929456196114911e-06,
      "loss": 0.8132,
      "step": 26275
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.12457874234298,
      "learning_rate": 7.928458413809024e-06,
      "loss": 0.8544,
      "step": 26280
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.930002930423708,
      "learning_rate": 7.927460453959333e-06,
      "loss": 0.8689,
      "step": 26285
    },
    {
      "epoch": 1.11,
      "grad_norm": 7.788295212520432,
      "learning_rate": 7.926462316626335e-06,
      "loss": 0.8523,
      "step": 26290
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.37551487558383,
      "learning_rate": 7.925464001870549e-06,
      "loss": 0.8706,
      "step": 26295
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.554335521737757,
      "learning_rate": 7.924465509752494e-06,
      "loss": 0.8771,
      "step": 26300
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.320329767579542,
      "learning_rate": 7.923466840332715e-06,
      "loss": 0.848,
      "step": 26305
    },
    {
      "epoch": 1.11,
      "grad_norm": 6.353197471583026,
      "learning_rate": 7.922467993671749e-06,
      "loss": 0.8652,
      "step": 26310
    },
    {
      "epoch": 1.11,
      "grad_norm": 12.707890319678297,
      "learning_rate": 7.921468969830157e-06,
      "loss": 0.832,
      "step": 26315
    },
    {
      "epoch": 1.11,
      "grad_norm": 16.082846990192856,
      "learning_rate": 7.920469768868508e-06,
      "loss": 0.8951,
      "step": 26320
    },
    {
      "epoch": 1.11,
      "grad_norm": 12.155509125144548,
      "learning_rate": 7.919470390847379e-06,
      "loss": 0.8326,
      "step": 26325
    },
    {
      "epoch": 1.11,
      "grad_norm": 8.4691323602816,
      "learning_rate": 7.91847083582736e-06,
      "loss": 0.846,
      "step": 26330
    },
    {
      "epoch": 1.11,
      "grad_norm": 11.498080104062787,
      "learning_rate": 7.91747110386905e-06,
      "loss": 0.8174,
      "step": 26335
    },
    {
      "epoch": 1.11,
      "grad_norm": 16.724565374138287,
      "learning_rate": 7.916471195033062e-06,
      "loss": 0.8793,
      "step": 26340
    },
    {
      "epoch": 1.11,
      "grad_norm": 5.084562589331612,
      "learning_rate": 7.915471109380015e-06,
      "loss": 0.9458,
      "step": 26345
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.153207951653013,
      "learning_rate": 7.914470846970543e-06,
      "loss": 0.8639,
      "step": 26350
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.044821088268971,
      "learning_rate": 7.91347040786529e-06,
      "loss": 0.8634,
      "step": 26355
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.313021099573543,
      "learning_rate": 7.912469792124905e-06,
      "loss": 0.8509,
      "step": 26360
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.112470298646595,
      "learning_rate": 7.911468999810057e-06,
      "loss": 0.8506,
      "step": 26365
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.219758190596584,
      "learning_rate": 7.91046803098142e-06,
      "loss": 0.8513,
      "step": 26370
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.242959216042871,
      "learning_rate": 7.90946688569968e-06,
      "loss": 0.8752,
      "step": 26375
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.937742429038265,
      "learning_rate": 7.908465564025531e-06,
      "loss": 0.8602,
      "step": 26380
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.861255864363088,
      "learning_rate": 7.907464066019682e-06,
      "loss": 0.878,
      "step": 26385
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.479599626662508,
      "learning_rate": 7.906462391742852e-06,
      "loss": 0.868,
      "step": 26390
    },
    {
      "epoch": 1.12,
      "grad_norm": 10.55122443045248,
      "learning_rate": 7.905460541255767e-06,
      "loss": 0.8502,
      "step": 26395
    },
    {
      "epoch": 1.12,
      "grad_norm": 10.507384465887881,
      "learning_rate": 7.904458514619169e-06,
      "loss": 0.821,
      "step": 26400
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.619197406412334,
      "learning_rate": 7.903456311893807e-06,
      "loss": 0.8539,
      "step": 26405
    },
    {
      "epoch": 1.12,
      "grad_norm": 9.14679174209778,
      "learning_rate": 7.90245393314044e-06,
      "loss": 0.9354,
      "step": 26410
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.852325797491746,
      "learning_rate": 7.901451378419841e-06,
      "loss": 0.8481,
      "step": 26415
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.236739403660115,
      "learning_rate": 7.900448647792791e-06,
      "loss": 0.8399,
      "step": 26420
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.985129513759088,
      "learning_rate": 7.899445741320084e-06,
      "loss": 0.8593,
      "step": 26425
    },
    {
      "epoch": 1.12,
      "grad_norm": 9.215414650311526,
      "learning_rate": 7.898442659062522e-06,
      "loss": 0.8446,
      "step": 26430
    },
    {
      "epoch": 1.12,
      "grad_norm": 10.799833885818252,
      "learning_rate": 7.897439401080919e-06,
      "loss": 0.8644,
      "step": 26435
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.404066915088412,
      "learning_rate": 7.896435967436101e-06,
      "loss": 0.8331,
      "step": 26440
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.662322846064061,
      "learning_rate": 7.8954323581889e-06,
      "loss": 0.8761,
      "step": 26445
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.529573161338292,
      "learning_rate": 7.894428573400165e-06,
      "loss": 0.8992,
      "step": 26450
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.923357730500779,
      "learning_rate": 7.89342461313075e-06,
      "loss": 0.8588,
      "step": 26455
    },
    {
      "epoch": 1.12,
      "grad_norm": 9.421254811733112,
      "learning_rate": 7.892420477441527e-06,
      "loss": 0.9093,
      "step": 26460
    },
    {
      "epoch": 1.12,
      "grad_norm": 9.978084914221466,
      "learning_rate": 7.891416166393367e-06,
      "loss": 0.8671,
      "step": 26465
    },
    {
      "epoch": 1.12,
      "grad_norm": 10.665992272479397,
      "learning_rate": 7.890411680047163e-06,
      "loss": 0.8236,
      "step": 26470
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.566813411659888,
      "learning_rate": 7.889407018463812e-06,
      "loss": 0.837,
      "step": 26475
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.553667692697166,
      "learning_rate": 7.888402181704226e-06,
      "loss": 0.8303,
      "step": 26480
    },
    {
      "epoch": 1.12,
      "grad_norm": 11.124962721895892,
      "learning_rate": 7.887397169829323e-06,
      "loss": 0.8545,
      "step": 26485
    },
    {
      "epoch": 1.12,
      "grad_norm": 8.453632999725867,
      "learning_rate": 7.886391982900034e-06,
      "loss": 0.878,
      "step": 26490
    },
    {
      "epoch": 1.12,
      "grad_norm": 13.215457009175124,
      "learning_rate": 7.885386620977302e-06,
      "loss": 0.8492,
      "step": 26495
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.655740398770599,
      "learning_rate": 7.88438108412208e-06,
      "loss": 0.8311,
      "step": 26500
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.960290676498367,
      "learning_rate": 7.883375372395326e-06,
      "loss": 0.9035,
      "step": 26505
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.793186051565295,
      "learning_rate": 7.882369485858018e-06,
      "loss": 0.8517,
      "step": 26510
    },
    {
      "epoch": 1.12,
      "grad_norm": 9.792223978566913,
      "learning_rate": 7.881363424571137e-06,
      "loss": 0.8673,
      "step": 26515
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.799364079032052,
      "learning_rate": 7.880357188595683e-06,
      "loss": 0.8701,
      "step": 26520
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.329275825766881,
      "learning_rate": 7.879350777992653e-06,
      "loss": 0.8771,
      "step": 26525
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.4701577246268664,
      "learning_rate": 7.878344192823069e-06,
      "loss": 0.8886,
      "step": 26530
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.890569453564953,
      "learning_rate": 7.877337433147954e-06,
      "loss": 0.9337,
      "step": 26535
    },
    {
      "epoch": 1.12,
      "grad_norm": 4.897192483503493,
      "learning_rate": 7.876330499028344e-06,
      "loss": 0.8643,
      "step": 26540
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.1961950066339355,
      "learning_rate": 7.875323390525291e-06,
      "loss": 0.8619,
      "step": 26545
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.560107205409909,
      "learning_rate": 7.87431610769985e-06,
      "loss": 0.8342,
      "step": 26550
    },
    {
      "epoch": 1.12,
      "grad_norm": 10.971556707739259,
      "learning_rate": 7.87330865061309e-06,
      "loss": 0.8848,
      "step": 26555
    },
    {
      "epoch": 1.12,
      "grad_norm": 7.719900411231033,
      "learning_rate": 7.872301019326089e-06,
      "loss": 0.8724,
      "step": 26560
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.089131466554082,
      "learning_rate": 7.871293213899937e-06,
      "loss": 0.8987,
      "step": 26565
    },
    {
      "epoch": 1.12,
      "grad_norm": 6.990412022404094,
      "learning_rate": 7.870285234395737e-06,
      "loss": 0.862,
      "step": 26570
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.748927013077684,
      "learning_rate": 7.869277080874597e-06,
      "loss": 0.8281,
      "step": 26575
    },
    {
      "epoch": 1.12,
      "grad_norm": 5.5222190643836315,
      "learning_rate": 7.86826875339764e-06,
      "loss": 0.8592,
      "step": 26580
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.554976429327962,
      "learning_rate": 7.867260252025997e-06,
      "loss": 0.9364,
      "step": 26585
    },
    {
      "epoch": 1.13,
      "grad_norm": 10.755445220111715,
      "learning_rate": 7.866251576820811e-06,
      "loss": 0.8656,
      "step": 26590
    },
    {
      "epoch": 1.13,
      "grad_norm": 9.20211654830716,
      "learning_rate": 7.865242727843233e-06,
      "loss": 0.8728,
      "step": 26595
    },
    {
      "epoch": 1.13,
      "grad_norm": 21.90600740447074,
      "learning_rate": 7.86423370515443e-06,
      "loss": 0.8825,
      "step": 26600
    },
    {
      "epoch": 1.13,
      "grad_norm": 24.250115247433797,
      "learning_rate": 7.863224508815575e-06,
      "loss": 0.8953,
      "step": 26605
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.776777047109916,
      "learning_rate": 7.862215138887852e-06,
      "loss": 0.8734,
      "step": 26610
    },
    {
      "epoch": 1.13,
      "grad_norm": 11.222521053195054,
      "learning_rate": 7.861205595432455e-06,
      "loss": 0.8521,
      "step": 26615
    },
    {
      "epoch": 1.13,
      "grad_norm": 15.4757726145903,
      "learning_rate": 7.860195878510592e-06,
      "loss": 0.8691,
      "step": 26620
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.753732597558064,
      "learning_rate": 7.859185988183475e-06,
      "loss": 0.8422,
      "step": 26625
    },
    {
      "epoch": 1.13,
      "grad_norm": 10.745038201599531,
      "learning_rate": 7.858175924512335e-06,
      "loss": 0.8609,
      "step": 26630
    },
    {
      "epoch": 1.13,
      "grad_norm": 12.237458005685918,
      "learning_rate": 7.85716568755841e-06,
      "loss": 0.9102,
      "step": 26635
    },
    {
      "epoch": 1.13,
      "grad_norm": 19.52675624499569,
      "learning_rate": 7.856155277382942e-06,
      "loss": 0.827,
      "step": 26640
    },
    {
      "epoch": 1.13,
      "grad_norm": 11.243646000641661,
      "learning_rate": 7.855144694047195e-06,
      "loss": 0.8395,
      "step": 26645
    },
    {
      "epoch": 1.13,
      "grad_norm": 13.421330366379776,
      "learning_rate": 7.854133937612433e-06,
      "loss": 0.857,
      "step": 26650
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.013234633397708,
      "learning_rate": 7.85312300813994e-06,
      "loss": 0.8337,
      "step": 26655
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.72309012691772,
      "learning_rate": 7.852111905691001e-06,
      "loss": 0.8483,
      "step": 26660
    },
    {
      "epoch": 1.13,
      "grad_norm": 11.988992360580719,
      "learning_rate": 7.85110063032692e-06,
      "loss": 0.8367,
      "step": 26665
    },
    {
      "epoch": 1.13,
      "grad_norm": 15.060130387338289,
      "learning_rate": 7.850089182109003e-06,
      "loss": 0.832,
      "step": 26670
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.612872499527278,
      "learning_rate": 7.849077561098576e-06,
      "loss": 0.8793,
      "step": 26675
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.667532171093065,
      "learning_rate": 7.848065767356972e-06,
      "loss": 0.8645,
      "step": 26680
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.807332830117759,
      "learning_rate": 7.847053800945525e-06,
      "loss": 0.8319,
      "step": 26685
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.385690809313313,
      "learning_rate": 7.846041661925597e-06,
      "loss": 0.8669,
      "step": 26690
    },
    {
      "epoch": 1.13,
      "grad_norm": 9.233918128482422,
      "learning_rate": 7.845029350358542e-06,
      "loss": 0.8701,
      "step": 26695
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.5241799427556675,
      "learning_rate": 7.844016866305739e-06,
      "loss": 0.8717,
      "step": 26700
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.654524793418459,
      "learning_rate": 7.84300420982857e-06,
      "loss": 0.8381,
      "step": 26705
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.317974951972202,
      "learning_rate": 7.841991380988433e-06,
      "loss": 0.7977,
      "step": 26710
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.922015417926553,
      "learning_rate": 7.840978379846728e-06,
      "loss": 0.9152,
      "step": 26715
    },
    {
      "epoch": 1.13,
      "grad_norm": 8.420550301219485,
      "learning_rate": 7.83996520646487e-06,
      "loss": 0.839,
      "step": 26720
    },
    {
      "epoch": 1.13,
      "grad_norm": 11.441949788537928,
      "learning_rate": 7.83895186090429e-06,
      "loss": 0.8481,
      "step": 26725
    },
    {
      "epoch": 1.13,
      "grad_norm": 13.784853697703845,
      "learning_rate": 7.837938343226421e-06,
      "loss": 0.8501,
      "step": 26730
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.99350935362089,
      "learning_rate": 7.836924653492709e-06,
      "loss": 0.9007,
      "step": 26735
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.391258245343709,
      "learning_rate": 7.835910791764611e-06,
      "loss": 0.8617,
      "step": 26740
    },
    {
      "epoch": 1.13,
      "grad_norm": 9.713336044989521,
      "learning_rate": 7.834896758103596e-06,
      "loss": 0.8839,
      "step": 26745
    },
    {
      "epoch": 1.13,
      "grad_norm": 13.715219668104211,
      "learning_rate": 7.83388255257114e-06,
      "loss": 0.8872,
      "step": 26750
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.033699236550981,
      "learning_rate": 7.832868175228733e-06,
      "loss": 0.8557,
      "step": 26755
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.8878581659656675,
      "learning_rate": 7.831853626137873e-06,
      "loss": 0.8921,
      "step": 26760
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.794052430097345,
      "learning_rate": 7.83083890536007e-06,
      "loss": 0.8657,
      "step": 26765
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.161985488106933,
      "learning_rate": 7.829824012956843e-06,
      "loss": 0.8564,
      "step": 26770
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.517373623150537,
      "learning_rate": 7.828808948989721e-06,
      "loss": 0.846,
      "step": 26775
    },
    {
      "epoch": 1.13,
      "grad_norm": 4.965864831707195,
      "learning_rate": 7.827793713520244e-06,
      "loss": 0.8263,
      "step": 26780
    },
    {
      "epoch": 1.13,
      "grad_norm": 7.572072833578794,
      "learning_rate": 7.826778306609963e-06,
      "loss": 0.8784,
      "step": 26785
    },
    {
      "epoch": 1.13,
      "grad_norm": 19.58769577147176,
      "learning_rate": 7.825762728320441e-06,
      "loss": 0.8592,
      "step": 26790
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.856431530750798,
      "learning_rate": 7.824746978713249e-06,
      "loss": 0.8463,
      "step": 26795
    },
    {
      "epoch": 1.13,
      "grad_norm": 13.415610109036363,
      "learning_rate": 7.823731057849968e-06,
      "loss": 0.8759,
      "step": 26800
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.355237783571644,
      "learning_rate": 7.82271496579219e-06,
      "loss": 0.9095,
      "step": 26805
    },
    {
      "epoch": 1.13,
      "grad_norm": 5.5291020808105475,
      "learning_rate": 7.821698702601518e-06,
      "loss": 0.8485,
      "step": 26810
    },
    {
      "epoch": 1.13,
      "grad_norm": 6.432170156810988,
      "learning_rate": 7.820682268339565e-06,
      "loss": 0.8143,
      "step": 26815
    },
    {
      "epoch": 1.14,
      "grad_norm": 12.885790040533996,
      "learning_rate": 7.819665663067955e-06,
      "loss": 0.9016,
      "step": 26820
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.308668580777147,
      "learning_rate": 7.818648886848322e-06,
      "loss": 0.8184,
      "step": 26825
    },
    {
      "epoch": 1.14,
      "grad_norm": 17.97692222708809,
      "learning_rate": 7.81763193974231e-06,
      "loss": 0.8614,
      "step": 26830
    },
    {
      "epoch": 1.14,
      "grad_norm": 19.73903313184167,
      "learning_rate": 7.816614821811573e-06,
      "loss": 0.9027,
      "step": 26835
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.2638924445763635,
      "learning_rate": 7.815597533117774e-06,
      "loss": 0.8326,
      "step": 26840
    },
    {
      "epoch": 1.14,
      "grad_norm": 14.284736657157557,
      "learning_rate": 7.814580073722592e-06,
      "loss": 0.8259,
      "step": 26845
    },
    {
      "epoch": 1.14,
      "grad_norm": 14.520932253709928,
      "learning_rate": 7.81356244368771e-06,
      "loss": 0.839,
      "step": 26850
    },
    {
      "epoch": 1.14,
      "grad_norm": 15.195141194142588,
      "learning_rate": 7.812544643074826e-06,
      "loss": 0.8753,
      "step": 26855
    },
    {
      "epoch": 1.14,
      "grad_norm": 15.99062710551827,
      "learning_rate": 7.811526671945644e-06,
      "loss": 0.8192,
      "step": 26860
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.50763504403253,
      "learning_rate": 7.81050853036188e-06,
      "loss": 0.814,
      "step": 26865
    },
    {
      "epoch": 1.14,
      "grad_norm": 13.57963237693707,
      "learning_rate": 7.809490218385265e-06,
      "loss": 0.8535,
      "step": 26870
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.699340883645577,
      "learning_rate": 7.808471736077532e-06,
      "loss": 0.8563,
      "step": 26875
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.223232849495145,
      "learning_rate": 7.80745308350043e-06,
      "loss": 0.8689,
      "step": 26880
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.2102132997441775,
      "learning_rate": 7.806434260715718e-06,
      "loss": 0.8486,
      "step": 26885
    },
    {
      "epoch": 1.14,
      "grad_norm": 10.111656855257813,
      "learning_rate": 7.805415267785164e-06,
      "loss": 0.8409,
      "step": 26890
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.789176996540388,
      "learning_rate": 7.804396104770544e-06,
      "loss": 0.8714,
      "step": 26895
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.654885199060267,
      "learning_rate": 7.803376771733651e-06,
      "loss": 0.8712,
      "step": 26900
    },
    {
      "epoch": 1.14,
      "grad_norm": 15.298537759433874,
      "learning_rate": 7.80235726873628e-06,
      "loss": 0.8876,
      "step": 26905
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.197520155448941,
      "learning_rate": 7.801337595840243e-06,
      "loss": 0.8779,
      "step": 26910
    },
    {
      "epoch": 1.14,
      "grad_norm": 9.409360441838256,
      "learning_rate": 7.80031775310736e-06,
      "loss": 0.8654,
      "step": 26915
    },
    {
      "epoch": 1.14,
      "grad_norm": 13.718151268956833,
      "learning_rate": 7.799297740599461e-06,
      "loss": 0.899,
      "step": 26920
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.515417767567687,
      "learning_rate": 7.798277558378383e-06,
      "loss": 0.8193,
      "step": 26925
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.335845840609774,
      "learning_rate": 7.797257206505982e-06,
      "loss": 0.8364,
      "step": 26930
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.953376764899409,
      "learning_rate": 7.796236685044113e-06,
      "loss": 0.859,
      "step": 26935
    },
    {
      "epoch": 1.14,
      "grad_norm": 11.149289751225545,
      "learning_rate": 7.795215994054655e-06,
      "loss": 0.8733,
      "step": 26940
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.44217338293341,
      "learning_rate": 7.79419513359948e-06,
      "loss": 0.9356,
      "step": 26945
    },
    {
      "epoch": 1.14,
      "grad_norm": 9.944065439393688,
      "learning_rate": 7.793174103740488e-06,
      "loss": 0.8772,
      "step": 26950
    },
    {
      "epoch": 1.14,
      "grad_norm": 15.660101175955392,
      "learning_rate": 7.792152904539576e-06,
      "loss": 0.8493,
      "step": 26955
    },
    {
      "epoch": 1.14,
      "grad_norm": 18.14015428218497,
      "learning_rate": 7.791131536058659e-06,
      "loss": 0.8227,
      "step": 26960
    },
    {
      "epoch": 1.14,
      "grad_norm": 14.479782528678234,
      "learning_rate": 7.790109998359656e-06,
      "loss": 0.863,
      "step": 26965
    },
    {
      "epoch": 1.14,
      "grad_norm": 8.04743827991232,
      "learning_rate": 7.789088291504505e-06,
      "loss": 0.8603,
      "step": 26970
    },
    {
      "epoch": 1.14,
      "grad_norm": 10.527038238649569,
      "learning_rate": 7.788066415555144e-06,
      "loss": 0.8582,
      "step": 26975
    },
    {
      "epoch": 1.14,
      "grad_norm": 11.425065904239588,
      "learning_rate": 7.78704437057353e-06,
      "loss": 0.8496,
      "step": 26980
    },
    {
      "epoch": 1.14,
      "grad_norm": 13.217916132253075,
      "learning_rate": 7.786022156621626e-06,
      "loss": 0.834,
      "step": 26985
    },
    {
      "epoch": 1.14,
      "grad_norm": 12.816355601444743,
      "learning_rate": 7.784999773761405e-06,
      "loss": 0.911,
      "step": 26990
    },
    {
      "epoch": 1.14,
      "grad_norm": 4.8496703016782545,
      "learning_rate": 7.78397722205485e-06,
      "loss": 0.8728,
      "step": 26995
    },
    {
      "epoch": 1.14,
      "grad_norm": 9.911400235634874,
      "learning_rate": 7.782954501563959e-06,
      "loss": 0.8628,
      "step": 27000
    },
    {
      "epoch": 1.14,
      "grad_norm": 22.1676903894186,
      "learning_rate": 7.781931612350732e-06,
      "loss": 0.8378,
      "step": 27005
    },
    {
      "epoch": 1.14,
      "grad_norm": 19.198931511428537,
      "learning_rate": 7.780908554477187e-06,
      "loss": 0.8387,
      "step": 27010
    },
    {
      "epoch": 1.14,
      "grad_norm": 16.119500222539244,
      "learning_rate": 7.779885328005347e-06,
      "loss": 0.8696,
      "step": 27015
    },
    {
      "epoch": 1.14,
      "grad_norm": 20.206130804311602,
      "learning_rate": 7.778861932997248e-06,
      "loss": 0.8735,
      "step": 27020
    },
    {
      "epoch": 1.14,
      "grad_norm": 14.646468899449912,
      "learning_rate": 7.777838369514934e-06,
      "loss": 0.8737,
      "step": 27025
    },
    {
      "epoch": 1.14,
      "grad_norm": 15.936270658117936,
      "learning_rate": 7.776814637620465e-06,
      "loss": 0.8123,
      "step": 27030
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.600928576146234,
      "learning_rate": 7.775790737375901e-06,
      "loss": 0.808,
      "step": 27035
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.7706822854091,
      "learning_rate": 7.774766668843323e-06,
      "loss": 0.8824,
      "step": 27040
    },
    {
      "epoch": 1.14,
      "grad_norm": 5.15918773156841,
      "learning_rate": 7.773742432084815e-06,
      "loss": 0.8399,
      "step": 27045
    },
    {
      "epoch": 1.14,
      "grad_norm": 7.21056195321401,
      "learning_rate": 7.772718027162473e-06,
      "loss": 0.8816,
      "step": 27050
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.77728011241169,
      "learning_rate": 7.771693454138405e-06,
      "loss": 0.8817,
      "step": 27055
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.484314690605723,
      "learning_rate": 7.770668713074726e-06,
      "loss": 0.8462,
      "step": 27060
    },
    {
      "epoch": 1.15,
      "grad_norm": 12.142716095242328,
      "learning_rate": 7.769643804033564e-06,
      "loss": 0.8604,
      "step": 27065
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.475203472027149,
      "learning_rate": 7.768618727077057e-06,
      "loss": 0.8166,
      "step": 27070
    },
    {
      "epoch": 1.15,
      "grad_norm": 14.945967883466762,
      "learning_rate": 7.767593482267352e-06,
      "loss": 0.8973,
      "step": 27075
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.1150238727608865,
      "learning_rate": 7.766568069666607e-06,
      "loss": 0.8458,
      "step": 27080
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.0843257541669145,
      "learning_rate": 7.765542489336989e-06,
      "loss": 0.8492,
      "step": 27085
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.154301201245478,
      "learning_rate": 7.764516741340675e-06,
      "loss": 0.8529,
      "step": 27090
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.359557595392436,
      "learning_rate": 7.763490825739857e-06,
      "loss": 0.8921,
      "step": 27095
    },
    {
      "epoch": 1.15,
      "grad_norm": 10.648640141356333,
      "learning_rate": 7.762464742596726e-06,
      "loss": 0.8638,
      "step": 27100
    },
    {
      "epoch": 1.15,
      "grad_norm": 17.84050128196062,
      "learning_rate": 7.761438491973499e-06,
      "loss": 0.8589,
      "step": 27105
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.360991389523278,
      "learning_rate": 7.760412073932387e-06,
      "loss": 0.814,
      "step": 27110
    },
    {
      "epoch": 1.15,
      "grad_norm": 14.856043847272236,
      "learning_rate": 7.759385488535623e-06,
      "loss": 0.8704,
      "step": 27115
    },
    {
      "epoch": 1.15,
      "grad_norm": 13.981582047051566,
      "learning_rate": 7.758358735845446e-06,
      "loss": 0.8762,
      "step": 27120
    },
    {
      "epoch": 1.15,
      "grad_norm": 15.52080847728403,
      "learning_rate": 7.757331815924103e-06,
      "loss": 0.875,
      "step": 27125
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.520287958726929,
      "learning_rate": 7.756304728833853e-06,
      "loss": 0.8626,
      "step": 27130
    },
    {
      "epoch": 1.15,
      "grad_norm": 13.67124852286572,
      "learning_rate": 7.755277474636967e-06,
      "loss": 0.8896,
      "step": 27135
    },
    {
      "epoch": 1.15,
      "grad_norm": 15.133427616989637,
      "learning_rate": 7.754250053395725e-06,
      "loss": 0.8867,
      "step": 27140
    },
    {
      "epoch": 1.15,
      "grad_norm": 12.203771082525684,
      "learning_rate": 7.753222465172413e-06,
      "loss": 0.8522,
      "step": 27145
    },
    {
      "epoch": 1.15,
      "grad_norm": 19.52042664030729,
      "learning_rate": 7.752194710029334e-06,
      "loss": 0.8567,
      "step": 27150
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.556782586130002,
      "learning_rate": 7.751166788028796e-06,
      "loss": 0.8211,
      "step": 27155
    },
    {
      "epoch": 1.15,
      "grad_norm": 4.438704807861239,
      "learning_rate": 7.75013869923312e-06,
      "loss": 0.8498,
      "step": 27160
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.342816116882408,
      "learning_rate": 7.749110443704636e-06,
      "loss": 0.8522,
      "step": 27165
    },
    {
      "epoch": 1.15,
      "grad_norm": 10.063566831098308,
      "learning_rate": 7.748082021505682e-06,
      "loss": 0.9002,
      "step": 27170
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.079935585329831,
      "learning_rate": 7.747053432698613e-06,
      "loss": 0.8668,
      "step": 27175
    },
    {
      "epoch": 1.15,
      "grad_norm": 11.911554457975734,
      "learning_rate": 7.746024677345782e-06,
      "loss": 0.882,
      "step": 27180
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.282740796062666,
      "learning_rate": 7.744995755509566e-06,
      "loss": 0.8524,
      "step": 27185
    },
    {
      "epoch": 1.15,
      "grad_norm": 9.311826855993818,
      "learning_rate": 7.743966667252343e-06,
      "loss": 0.8402,
      "step": 27190
    },
    {
      "epoch": 1.15,
      "grad_norm": 15.175072831580708,
      "learning_rate": 7.742937412636503e-06,
      "loss": 0.8775,
      "step": 27195
    },
    {
      "epoch": 1.15,
      "grad_norm": 26.816963976243624,
      "learning_rate": 7.741907991724447e-06,
      "loss": 0.9046,
      "step": 27200
    },
    {
      "epoch": 1.15,
      "grad_norm": 8.966737467527985,
      "learning_rate": 7.740878404578584e-06,
      "loss": 0.8657,
      "step": 27205
    },
    {
      "epoch": 1.15,
      "grad_norm": 11.449970539582631,
      "learning_rate": 7.73984865126134e-06,
      "loss": 0.9281,
      "step": 27210
    },
    {
      "epoch": 1.15,
      "grad_norm": 16.050798275405686,
      "learning_rate": 7.738818731835142e-06,
      "loss": 0.8304,
      "step": 27215
    },
    {
      "epoch": 1.15,
      "grad_norm": 24.048969562311843,
      "learning_rate": 7.737788646362432e-06,
      "loss": 0.8703,
      "step": 27220
    },
    {
      "epoch": 1.15,
      "grad_norm": 23.897415197635976,
      "learning_rate": 7.73675839490566e-06,
      "loss": 0.8251,
      "step": 27225
    },
    {
      "epoch": 1.15,
      "grad_norm": 6.999074851963846,
      "learning_rate": 7.735727977527287e-06,
      "loss": 0.881,
      "step": 27230
    },
    {
      "epoch": 1.15,
      "grad_norm": 5.68607491237158,
      "learning_rate": 7.734697394289787e-06,
      "loss": 0.8468,
      "step": 27235
    },
    {
      "epoch": 1.15,
      "grad_norm": 18.016385641529947,
      "learning_rate": 7.733666645255639e-06,
      "loss": 0.8517,
      "step": 27240
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.6008166815681415,
      "learning_rate": 7.732635730487333e-06,
      "loss": 0.8528,
      "step": 27245
    },
    {
      "epoch": 1.15,
      "grad_norm": 15.798518565649362,
      "learning_rate": 7.731604650047373e-06,
      "loss": 0.868,
      "step": 27250
    },
    {
      "epoch": 1.15,
      "grad_norm": 21.522841309858762,
      "learning_rate": 7.73057340399827e-06,
      "loss": 0.9072,
      "step": 27255
    },
    {
      "epoch": 1.15,
      "grad_norm": 26.643299711535217,
      "learning_rate": 7.729541992402546e-06,
      "loss": 0.855,
      "step": 27260
    },
    {
      "epoch": 1.15,
      "grad_norm": 13.253308131733148,
      "learning_rate": 7.728510415322727e-06,
      "loss": 0.8417,
      "step": 27265
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.263470141494027,
      "learning_rate": 7.727478672821362e-06,
      "loss": 0.8868,
      "step": 27270
    },
    {
      "epoch": 1.15,
      "grad_norm": 7.5953735070437265,
      "learning_rate": 7.726446764960998e-06,
      "loss": 0.9227,
      "step": 27275
    },
    {
      "epoch": 1.15,
      "grad_norm": 14.24061089109959,
      "learning_rate": 7.725414691804196e-06,
      "loss": 0.8622,
      "step": 27280
    },
    {
      "epoch": 1.15,
      "grad_norm": 9.712110940298755,
      "learning_rate": 7.72438245341353e-06,
      "loss": 0.8283,
      "step": 27285
    },
    {
      "epoch": 1.15,
      "grad_norm": 17.708881360823202,
      "learning_rate": 7.723350049851583e-06,
      "loss": 0.8316,
      "step": 27290
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.997175146017793,
      "learning_rate": 7.722317481180944e-06,
      "loss": 0.8628,
      "step": 27295
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.348698573207775,
      "learning_rate": 7.721284747464212e-06,
      "loss": 0.8732,
      "step": 27300
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.541107517511868,
      "learning_rate": 7.720251848764004e-06,
      "loss": 0.8598,
      "step": 27305
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.860351573617423,
      "learning_rate": 7.71921878514294e-06,
      "loss": 0.865,
      "step": 27310
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.330076277294611,
      "learning_rate": 7.718185556663648e-06,
      "loss": 0.847,
      "step": 27315
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.703301277916182,
      "learning_rate": 7.717152163388774e-06,
      "loss": 0.8661,
      "step": 27320
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.629398535030925,
      "learning_rate": 7.716118605380966e-06,
      "loss": 0.8911,
      "step": 27325
    },
    {
      "epoch": 1.16,
      "grad_norm": 9.98793569161913,
      "learning_rate": 7.71508488270289e-06,
      "loss": 0.8734,
      "step": 27330
    },
    {
      "epoch": 1.16,
      "grad_norm": 23.638872432904368,
      "learning_rate": 7.714050995417215e-06,
      "loss": 0.8318,
      "step": 27335
    },
    {
      "epoch": 1.16,
      "grad_norm": 11.340252996496963,
      "learning_rate": 7.713016943586622e-06,
      "loss": 0.8486,
      "step": 27340
    },
    {
      "epoch": 1.16,
      "grad_norm": 17.09403077891058,
      "learning_rate": 7.711982727273804e-06,
      "loss": 0.888,
      "step": 27345
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.7832109546675,
      "learning_rate": 7.71094834654146e-06,
      "loss": 0.8309,
      "step": 27350
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.883792935301827,
      "learning_rate": 7.709913801452305e-06,
      "loss": 0.8601,
      "step": 27355
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.286690897081227,
      "learning_rate": 7.708879092069057e-06,
      "loss": 0.863,
      "step": 27360
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.36152425118263,
      "learning_rate": 7.707844218454449e-06,
      "loss": 0.823,
      "step": 27365
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.890568786615989,
      "learning_rate": 7.706809180671221e-06,
      "loss": 0.8539,
      "step": 27370
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.262723437729285,
      "learning_rate": 7.70577397878213e-06,
      "loss": 0.8614,
      "step": 27375
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.86794575809123,
      "learning_rate": 7.70473861284993e-06,
      "loss": 0.8262,
      "step": 27380
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.403378586839154,
      "learning_rate": 7.703703082937392e-06,
      "loss": 0.864,
      "step": 27385
    },
    {
      "epoch": 1.16,
      "grad_norm": 11.868381746572243,
      "learning_rate": 7.702667389107305e-06,
      "loss": 0.8406,
      "step": 27390
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.952566892750742,
      "learning_rate": 7.701631531422453e-06,
      "loss": 0.7994,
      "step": 27395
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.885562734938025,
      "learning_rate": 7.700595509945639e-06,
      "loss": 0.8447,
      "step": 27400
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.025146300008063,
      "learning_rate": 7.699559324739673e-06,
      "loss": 0.9225,
      "step": 27405
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.423858067637031,
      "learning_rate": 7.69852297586738e-06,
      "loss": 0.8856,
      "step": 27410
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.020228151338761,
      "learning_rate": 7.697486463391585e-06,
      "loss": 0.8315,
      "step": 27415
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.793320892112957,
      "learning_rate": 7.696449787375135e-06,
      "loss": 0.8654,
      "step": 27420
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.8249425336504075,
      "learning_rate": 7.695412947880873e-06,
      "loss": 0.841,
      "step": 27425
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.28639020911608,
      "learning_rate": 7.694375944971666e-06,
      "loss": 0.8291,
      "step": 27430
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.81409659379978,
      "learning_rate": 7.693338778710382e-06,
      "loss": 0.8128,
      "step": 27435
    },
    {
      "epoch": 1.16,
      "grad_norm": 11.739210748964835,
      "learning_rate": 7.692301449159898e-06,
      "loss": 0.8782,
      "step": 27440
    },
    {
      "epoch": 1.16,
      "grad_norm": 9.854060773466067,
      "learning_rate": 7.691263956383111e-06,
      "loss": 0.8451,
      "step": 27445
    },
    {
      "epoch": 1.16,
      "grad_norm": 8.489044808294599,
      "learning_rate": 7.690226300442918e-06,
      "loss": 0.8194,
      "step": 27450
    },
    {
      "epoch": 1.16,
      "grad_norm": 9.09447171241856,
      "learning_rate": 7.689188481402227e-06,
      "loss": 0.8351,
      "step": 27455
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.745927544295226,
      "learning_rate": 7.688150499323961e-06,
      "loss": 0.8382,
      "step": 27460
    },
    {
      "epoch": 1.16,
      "grad_norm": 22.595565634608043,
      "learning_rate": 7.687112354271046e-06,
      "loss": 0.8696,
      "step": 27465
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.978275778853865,
      "learning_rate": 7.686074046306427e-06,
      "loss": 0.8688,
      "step": 27470
    },
    {
      "epoch": 1.16,
      "grad_norm": 19.67266723351257,
      "learning_rate": 7.685035575493049e-06,
      "loss": 0.8501,
      "step": 27475
    },
    {
      "epoch": 1.16,
      "grad_norm": 24.370771439683413,
      "learning_rate": 7.683996941893872e-06,
      "loss": 0.8403,
      "step": 27480
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.409431963560548,
      "learning_rate": 7.682958145571868e-06,
      "loss": 0.879,
      "step": 27485
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.888276552284849,
      "learning_rate": 7.681919186590015e-06,
      "loss": 0.8346,
      "step": 27490
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.086681885252242,
      "learning_rate": 7.6808800650113e-06,
      "loss": 0.8495,
      "step": 27495
    },
    {
      "epoch": 1.16,
      "grad_norm": 6.168482707177028,
      "learning_rate": 7.679840780898726e-06,
      "loss": 0.8301,
      "step": 27500
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.741519529628608,
      "learning_rate": 7.678801334315298e-06,
      "loss": 0.8459,
      "step": 27505
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.733702314062286,
      "learning_rate": 7.677761725324036e-06,
      "loss": 0.8592,
      "step": 27510
    },
    {
      "epoch": 1.16,
      "grad_norm": 7.683418825144909,
      "learning_rate": 7.676721953987967e-06,
      "loss": 0.8546,
      "step": 27515
    },
    {
      "epoch": 1.16,
      "grad_norm": 5.751840703909939,
      "learning_rate": 7.67568202037013e-06,
      "loss": 0.8836,
      "step": 27520
    },
    {
      "epoch": 1.16,
      "grad_norm": 10.254284232162648,
      "learning_rate": 7.674641924533575e-06,
      "loss": 0.8579,
      "step": 27525
    },
    {
      "epoch": 1.17,
      "grad_norm": 11.646552164090183,
      "learning_rate": 7.673601666541358e-06,
      "loss": 0.8455,
      "step": 27530
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.966369117631704,
      "learning_rate": 7.672561246456548e-06,
      "loss": 0.8395,
      "step": 27535
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.238456420578824,
      "learning_rate": 7.67152066434222e-06,
      "loss": 0.8474,
      "step": 27540
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.71478357741574,
      "learning_rate": 7.670479920261466e-06,
      "loss": 0.8308,
      "step": 27545
    },
    {
      "epoch": 1.17,
      "grad_norm": 13.103983033831756,
      "learning_rate": 7.66943901427738e-06,
      "loss": 0.8912,
      "step": 27550
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.275955979338747,
      "learning_rate": 7.668397946453068e-06,
      "loss": 0.895,
      "step": 27555
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.210058126881039,
      "learning_rate": 7.667356716851649e-06,
      "loss": 0.8681,
      "step": 27560
    },
    {
      "epoch": 1.17,
      "grad_norm": 12.554655831956167,
      "learning_rate": 7.66631532553625e-06,
      "loss": 0.8254,
      "step": 27565
    },
    {
      "epoch": 1.17,
      "grad_norm": 19.210384089349912,
      "learning_rate": 7.665273772570007e-06,
      "loss": 0.8186,
      "step": 27570
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.436548726078364,
      "learning_rate": 7.664232058016066e-06,
      "loss": 0.8753,
      "step": 27575
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.9379106745617385,
      "learning_rate": 7.663190181937582e-06,
      "loss": 0.8053,
      "step": 27580
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.686629557898449,
      "learning_rate": 7.662148144397724e-06,
      "loss": 0.8161,
      "step": 27585
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.991520918826527,
      "learning_rate": 7.661105945459662e-06,
      "loss": 0.8321,
      "step": 27590
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.32080752634358,
      "learning_rate": 7.660063585186588e-06,
      "loss": 0.8518,
      "step": 27595
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.671226647791415,
      "learning_rate": 7.659021063641692e-06,
      "loss": 0.8171,
      "step": 27600
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.871194076066952,
      "learning_rate": 7.657978380888184e-06,
      "loss": 0.8302,
      "step": 27605
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.228548617239947,
      "learning_rate": 7.656935536989273e-06,
      "loss": 0.839,
      "step": 27610
    },
    {
      "epoch": 1.17,
      "grad_norm": 13.934839079395935,
      "learning_rate": 7.65589253200819e-06,
      "loss": 0.8628,
      "step": 27615
    },
    {
      "epoch": 1.17,
      "grad_norm": 13.5939206736836,
      "learning_rate": 7.654849366008165e-06,
      "loss": 0.8727,
      "step": 27620
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.127171905459924,
      "learning_rate": 7.653806039052441e-06,
      "loss": 0.8345,
      "step": 27625
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.304199276197281,
      "learning_rate": 7.652762551204276e-06,
      "loss": 0.8103,
      "step": 27630
    },
    {
      "epoch": 1.17,
      "grad_norm": 15.504266219417596,
      "learning_rate": 7.651718902526932e-06,
      "loss": 0.851,
      "step": 27635
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.676149477882025,
      "learning_rate": 7.650675093083679e-06,
      "loss": 0.8925,
      "step": 27640
    },
    {
      "epoch": 1.17,
      "grad_norm": 10.977951798524913,
      "learning_rate": 7.649631122937807e-06,
      "loss": 0.8828,
      "step": 27645
    },
    {
      "epoch": 1.17,
      "grad_norm": 15.22570829620542,
      "learning_rate": 7.6485869921526e-06,
      "loss": 0.8213,
      "step": 27650
    },
    {
      "epoch": 1.17,
      "grad_norm": 25.062734538707396,
      "learning_rate": 7.647542700791368e-06,
      "loss": 0.8285,
      "step": 27655
    },
    {
      "epoch": 1.17,
      "grad_norm": 10.055435828160332,
      "learning_rate": 7.646498248917421e-06,
      "loss": 0.8802,
      "step": 27660
    },
    {
      "epoch": 1.17,
      "grad_norm": 16.89700212996837,
      "learning_rate": 7.64545363659408e-06,
      "loss": 0.8672,
      "step": 27665
    },
    {
      "epoch": 1.17,
      "grad_norm": 10.833302054850313,
      "learning_rate": 7.644408863884677e-06,
      "loss": 0.8414,
      "step": 27670
    },
    {
      "epoch": 1.17,
      "grad_norm": 15.597371231663407,
      "learning_rate": 7.643363930852555e-06,
      "loss": 0.8796,
      "step": 27675
    },
    {
      "epoch": 1.17,
      "grad_norm": 21.20763458877111,
      "learning_rate": 7.642318837561065e-06,
      "loss": 0.8104,
      "step": 27680
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.648144568217421,
      "learning_rate": 7.641273584073565e-06,
      "loss": 0.836,
      "step": 27685
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.369535311799559,
      "learning_rate": 7.640228170453427e-06,
      "loss": 0.887,
      "step": 27690
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.482130076599322,
      "learning_rate": 7.639182596764034e-06,
      "loss": 0.8058,
      "step": 27695
    },
    {
      "epoch": 1.17,
      "grad_norm": 15.201762077919843,
      "learning_rate": 7.638136863068774e-06,
      "loss": 0.8682,
      "step": 27700
    },
    {
      "epoch": 1.17,
      "grad_norm": 10.707946209431555,
      "learning_rate": 7.637090969431047e-06,
      "loss": 0.8088,
      "step": 27705
    },
    {
      "epoch": 1.17,
      "grad_norm": 19.13798040483209,
      "learning_rate": 7.636044915914261e-06,
      "loss": 0.8363,
      "step": 27710
    },
    {
      "epoch": 1.17,
      "grad_norm": 7.965068418045872,
      "learning_rate": 7.634998702581837e-06,
      "loss": 0.8334,
      "step": 27715
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.580006337187495,
      "learning_rate": 7.633952329497204e-06,
      "loss": 0.8491,
      "step": 27720
    },
    {
      "epoch": 1.17,
      "grad_norm": 6.764165416025875,
      "learning_rate": 7.6329057967238e-06,
      "loss": 0.8473,
      "step": 27725
    },
    {
      "epoch": 1.17,
      "grad_norm": 5.324253795361866,
      "learning_rate": 7.631859104325072e-06,
      "loss": 0.8369,
      "step": 27730
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.962681037737329,
      "learning_rate": 7.63081225236448e-06,
      "loss": 0.8451,
      "step": 27735
    },
    {
      "epoch": 1.17,
      "grad_norm": 9.590880885944284,
      "learning_rate": 7.62976524090549e-06,
      "loss": 0.8431,
      "step": 27740
    },
    {
      "epoch": 1.17,
      "grad_norm": 18.235409726344564,
      "learning_rate": 7.628718070011581e-06,
      "loss": 0.8286,
      "step": 27745
    },
    {
      "epoch": 1.17,
      "grad_norm": 8.764192789691922,
      "learning_rate": 7.627670739746236e-06,
      "loss": 0.8494,
      "step": 27750
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.783324195088866,
      "learning_rate": 7.626623250172955e-06,
      "loss": 0.8124,
      "step": 27755
    },
    {
      "epoch": 1.17,
      "grad_norm": 4.90817889085928,
      "learning_rate": 7.625575601355243e-06,
      "loss": 0.8559,
      "step": 27760
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.713028266312869,
      "learning_rate": 7.624527793356618e-06,
      "loss": 0.8384,
      "step": 27765
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.724507524271952,
      "learning_rate": 7.623479826240602e-06,
      "loss": 0.8265,
      "step": 27770
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.008352369710291,
      "learning_rate": 7.622431700070732e-06,
      "loss": 0.8551,
      "step": 27775
    },
    {
      "epoch": 1.18,
      "grad_norm": 11.534868258920003,
      "learning_rate": 7.6213834149105526e-06,
      "loss": 0.8483,
      "step": 27780
    },
    {
      "epoch": 1.18,
      "grad_norm": 9.829646302102221,
      "learning_rate": 7.620334970823618e-06,
      "loss": 0.8429,
      "step": 27785
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.588730456509883,
      "learning_rate": 7.619286367873493e-06,
      "loss": 0.8264,
      "step": 27790
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.1152142599429276,
      "learning_rate": 7.618237606123752e-06,
      "loss": 0.864,
      "step": 27795
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.686253510485167,
      "learning_rate": 7.617188685637975e-06,
      "loss": 0.841,
      "step": 27800
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.73788659006235,
      "learning_rate": 7.616139606479759e-06,
      "loss": 0.8917,
      "step": 27805
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.348402895636764,
      "learning_rate": 7.615090368712703e-06,
      "loss": 0.8361,
      "step": 27810
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.812146510386785,
      "learning_rate": 7.614040972400422e-06,
      "loss": 0.851,
      "step": 27815
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.157319708126221,
      "learning_rate": 7.612991417606538e-06,
      "loss": 0.8291,
      "step": 27820
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.418173409894394,
      "learning_rate": 7.6119417043946805e-06,
      "loss": 0.8654,
      "step": 27825
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.446844649442984,
      "learning_rate": 7.610891832828491e-06,
      "loss": 0.8539,
      "step": 27830
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.893240079957829,
      "learning_rate": 7.609841802971619e-06,
      "loss": 0.8362,
      "step": 27835
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.81288220380818,
      "learning_rate": 7.608791614887727e-06,
      "loss": 0.8395,
      "step": 27840
    },
    {
      "epoch": 1.18,
      "grad_norm": 27.360290542192228,
      "learning_rate": 7.607741268640484e-06,
      "loss": 0.8763,
      "step": 27845
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.969049667578217,
      "learning_rate": 7.606690764293569e-06,
      "loss": 0.819,
      "step": 27850
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.307470076649825,
      "learning_rate": 7.6056401019106716e-06,
      "loss": 0.8368,
      "step": 27855
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.839924621751747,
      "learning_rate": 7.60458928155549e-06,
      "loss": 0.8401,
      "step": 27860
    },
    {
      "epoch": 1.18,
      "grad_norm": 13.908765461839273,
      "learning_rate": 7.603538303291732e-06,
      "loss": 0.906,
      "step": 27865
    },
    {
      "epoch": 1.18,
      "grad_norm": 21.778484662244413,
      "learning_rate": 7.602487167183114e-06,
      "loss": 0.8406,
      "step": 27870
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.548158152746032,
      "learning_rate": 7.601435873293368e-06,
      "loss": 0.8429,
      "step": 27875
    },
    {
      "epoch": 1.18,
      "grad_norm": 9.501855722902018,
      "learning_rate": 7.6003844216862265e-06,
      "loss": 0.8602,
      "step": 27880
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.879120679560759,
      "learning_rate": 7.599332812425436e-06,
      "loss": 0.8546,
      "step": 27885
    },
    {
      "epoch": 1.18,
      "grad_norm": 17.821519682218852,
      "learning_rate": 7.598281045574755e-06,
      "loss": 0.835,
      "step": 27890
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.105085015962766,
      "learning_rate": 7.597229121197947e-06,
      "loss": 0.8288,
      "step": 27895
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.434859715371597,
      "learning_rate": 7.59617703935879e-06,
      "loss": 0.8117,
      "step": 27900
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.32888579585104,
      "learning_rate": 7.595124800121063e-06,
      "loss": 0.8642,
      "step": 27905
    },
    {
      "epoch": 1.18,
      "grad_norm": 13.581333073163272,
      "learning_rate": 7.594072403548567e-06,
      "loss": 0.8552,
      "step": 27910
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.36309030212805,
      "learning_rate": 7.5930198497051e-06,
      "loss": 0.8287,
      "step": 27915
    },
    {
      "epoch": 1.18,
      "grad_norm": 12.660316002134852,
      "learning_rate": 7.591967138654477e-06,
      "loss": 0.8964,
      "step": 27920
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.582443166828062,
      "learning_rate": 7.590914270460523e-06,
      "loss": 0.8729,
      "step": 27925
    },
    {
      "epoch": 1.18,
      "grad_norm": 5.762290179426884,
      "learning_rate": 7.589861245187067e-06,
      "loss": 0.8597,
      "step": 27930
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.380482049638994,
      "learning_rate": 7.588808062897955e-06,
      "loss": 0.7956,
      "step": 27935
    },
    {
      "epoch": 1.18,
      "grad_norm": 13.000222406937715,
      "learning_rate": 7.587754723657032e-06,
      "loss": 0.8011,
      "step": 27940
    },
    {
      "epoch": 1.18,
      "grad_norm": 11.397991076702855,
      "learning_rate": 7.586701227528165e-06,
      "loss": 0.8411,
      "step": 27945
    },
    {
      "epoch": 1.18,
      "grad_norm": 8.004559924893229,
      "learning_rate": 7.585647574575221e-06,
      "loss": 0.8328,
      "step": 27950
    },
    {
      "epoch": 1.18,
      "grad_norm": 12.925492474488394,
      "learning_rate": 7.584593764862081e-06,
      "loss": 0.8311,
      "step": 27955
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.690818949852794,
      "learning_rate": 7.583539798452635e-06,
      "loss": 0.8353,
      "step": 27960
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.513292805208883,
      "learning_rate": 7.582485675410781e-06,
      "loss": 0.826,
      "step": 27965
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.929384630067674,
      "learning_rate": 7.581431395800427e-06,
      "loss": 0.8414,
      "step": 27970
    },
    {
      "epoch": 1.18,
      "grad_norm": 7.098686476165517,
      "learning_rate": 7.580376959685491e-06,
      "loss": 0.8656,
      "step": 27975
    },
    {
      "epoch": 1.18,
      "grad_norm": 10.870683407795962,
      "learning_rate": 7.5793223671299e-06,
      "loss": 0.8121,
      "step": 27980
    },
    {
      "epoch": 1.18,
      "grad_norm": 4.745546473969652,
      "learning_rate": 7.578267618197594e-06,
      "loss": 0.8368,
      "step": 27985
    },
    {
      "epoch": 1.18,
      "grad_norm": 6.95174176893512,
      "learning_rate": 7.577212712952515e-06,
      "loss": 0.8737,
      "step": 27990
    },
    {
      "epoch": 1.18,
      "grad_norm": 12.8345052279372,
      "learning_rate": 7.576157651458622e-06,
      "loss": 0.8408,
      "step": 27995
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.005977161319825,
      "learning_rate": 7.575102433779877e-06,
      "loss": 0.8544,
      "step": 28000
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.986276334704683,
      "learning_rate": 7.574047059980258e-06,
      "loss": 0.8041,
      "step": 28005
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.412370995795932,
      "learning_rate": 7.572991530123748e-06,
      "loss": 0.8415,
      "step": 28010
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.374804435038488,
      "learning_rate": 7.571935844274338e-06,
      "loss": 0.8653,
      "step": 28015
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.6048587232680145,
      "learning_rate": 7.570880002496037e-06,
      "loss": 0.8162,
      "step": 28020
    },
    {
      "epoch": 1.19,
      "grad_norm": 12.194847011302116,
      "learning_rate": 7.569824004852852e-06,
      "loss": 0.805,
      "step": 28025
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.717125467874262,
      "learning_rate": 7.568767851408809e-06,
      "loss": 0.8487,
      "step": 28030
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.627311285886776,
      "learning_rate": 7.567711542227937e-06,
      "loss": 0.8773,
      "step": 28035
    },
    {
      "epoch": 1.19,
      "grad_norm": 13.873388470512738,
      "learning_rate": 7.566655077374278e-06,
      "loss": 0.8703,
      "step": 28040
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.4912792796707475,
      "learning_rate": 7.5655984569118805e-06,
      "loss": 0.8721,
      "step": 28045
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.932026194886099,
      "learning_rate": 7.564541680904806e-06,
      "loss": 0.8558,
      "step": 28050
    },
    {
      "epoch": 1.19,
      "grad_norm": 14.238880536736147,
      "learning_rate": 7.563484749417125e-06,
      "loss": 0.8727,
      "step": 28055
    },
    {
      "epoch": 1.19,
      "grad_norm": 14.376906356562099,
      "learning_rate": 7.5624276625129144e-06,
      "loss": 0.9071,
      "step": 28060
    },
    {
      "epoch": 1.19,
      "grad_norm": 17.00879540931177,
      "learning_rate": 7.561370420256262e-06,
      "loss": 0.875,
      "step": 28065
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.822858557009525,
      "learning_rate": 7.560313022711266e-06,
      "loss": 0.8789,
      "step": 28070
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.181121490015597,
      "learning_rate": 7.559255469942033e-06,
      "loss": 0.8142,
      "step": 28075
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.9937130752354575,
      "learning_rate": 7.558197762012679e-06,
      "loss": 0.8431,
      "step": 28080
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.780546674453072,
      "learning_rate": 7.557139898987331e-06,
      "loss": 0.8091,
      "step": 28085
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.779829060036907,
      "learning_rate": 7.556081880930122e-06,
      "loss": 0.9289,
      "step": 28090
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.509477277862589,
      "learning_rate": 7.555023707905198e-06,
      "loss": 0.9028,
      "step": 28095
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.440547603784951,
      "learning_rate": 7.553965379976714e-06,
      "loss": 0.8512,
      "step": 28100
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.764955861768891,
      "learning_rate": 7.552906897208829e-06,
      "loss": 0.87,
      "step": 28105
    },
    {
      "epoch": 1.19,
      "grad_norm": 15.635585903830606,
      "learning_rate": 7.551848259665721e-06,
      "loss": 0.8769,
      "step": 28110
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.8037009192162525,
      "learning_rate": 7.55078946741157e-06,
      "loss": 0.8418,
      "step": 28115
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.21938486468252,
      "learning_rate": 7.5497305205105665e-06,
      "loss": 0.851,
      "step": 28120
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.209014816444418,
      "learning_rate": 7.548671419026912e-06,
      "loss": 0.8407,
      "step": 28125
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.658079300211783,
      "learning_rate": 7.547612163024815e-06,
      "loss": 0.823,
      "step": 28130
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.426687639723021,
      "learning_rate": 7.5465527525685e-06,
      "loss": 0.849,
      "step": 28135
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.708656188891774,
      "learning_rate": 7.54549318772219e-06,
      "loss": 0.8513,
      "step": 28140
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.678332807196017,
      "learning_rate": 7.544433468550128e-06,
      "loss": 0.8162,
      "step": 28145
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.877331279988588,
      "learning_rate": 7.54337359511656e-06,
      "loss": 0.854,
      "step": 28150
    },
    {
      "epoch": 1.19,
      "grad_norm": 8.860323470715516,
      "learning_rate": 7.542313567485742e-06,
      "loss": 0.8003,
      "step": 28155
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.62422576715441,
      "learning_rate": 7.5412533857219405e-06,
      "loss": 0.8222,
      "step": 28160
    },
    {
      "epoch": 1.19,
      "grad_norm": 7.795437231638475,
      "learning_rate": 7.540193049889432e-06,
      "loss": 0.8784,
      "step": 28165
    },
    {
      "epoch": 1.19,
      "grad_norm": 12.866829067887812,
      "learning_rate": 7.5391325600525015e-06,
      "loss": 0.8931,
      "step": 28170
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.778881804171628,
      "learning_rate": 7.5380719162754425e-06,
      "loss": 0.8579,
      "step": 28175
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.744665997778414,
      "learning_rate": 7.537011118622559e-06,
      "loss": 0.8019,
      "step": 28180
    },
    {
      "epoch": 1.19,
      "grad_norm": 12.678200182813343,
      "learning_rate": 7.535950167158166e-06,
      "loss": 0.7903,
      "step": 28185
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.696953752553162,
      "learning_rate": 7.534889061946582e-06,
      "loss": 0.8018,
      "step": 28190
    },
    {
      "epoch": 1.19,
      "grad_norm": 14.251070906183966,
      "learning_rate": 7.5338278030521424e-06,
      "loss": 0.8586,
      "step": 28195
    },
    {
      "epoch": 1.19,
      "grad_norm": 11.622139844463858,
      "learning_rate": 7.532766390539187e-06,
      "loss": 0.8603,
      "step": 28200
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.615156466994754,
      "learning_rate": 7.531704824472066e-06,
      "loss": 0.8286,
      "step": 28205
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.816880063748879,
      "learning_rate": 7.530643104915139e-06,
      "loss": 0.8171,
      "step": 28210
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.461003187741392,
      "learning_rate": 7.529581231932772e-06,
      "loss": 0.8013,
      "step": 28215
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.178885886254754,
      "learning_rate": 7.528519205589349e-06,
      "loss": 0.8242,
      "step": 28220
    },
    {
      "epoch": 1.19,
      "grad_norm": 9.23610397110475,
      "learning_rate": 7.527457025949253e-06,
      "loss": 0.8435,
      "step": 28225
    },
    {
      "epoch": 1.19,
      "grad_norm": 4.469312700507896,
      "learning_rate": 7.526394693076884e-06,
      "loss": 0.8174,
      "step": 28230
    },
    {
      "epoch": 1.19,
      "grad_norm": 6.563343252192742,
      "learning_rate": 7.525332207036645e-06,
      "loss": 0.7947,
      "step": 28235
    },
    {
      "epoch": 1.2,
      "grad_norm": 10.18442601981919,
      "learning_rate": 7.524269567892954e-06,
      "loss": 0.8229,
      "step": 28240
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.452600003756412,
      "learning_rate": 7.5232067757102345e-06,
      "loss": 0.8295,
      "step": 28245
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.785289802378125,
      "learning_rate": 7.52214383055292e-06,
      "loss": 0.8537,
      "step": 28250
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.922889225903498,
      "learning_rate": 7.521080732485455e-06,
      "loss": 0.8666,
      "step": 28255
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.814412743718632,
      "learning_rate": 7.5200174815722895e-06,
      "loss": 0.8583,
      "step": 28260
    },
    {
      "epoch": 1.2,
      "grad_norm": 19.656432152275066,
      "learning_rate": 7.518954077877889e-06,
      "loss": 0.8562,
      "step": 28265
    },
    {
      "epoch": 1.2,
      "grad_norm": 25.79774643494239,
      "learning_rate": 7.517890521466722e-06,
      "loss": 0.8531,
      "step": 28270
    },
    {
      "epoch": 1.2,
      "grad_norm": 27.494003605576125,
      "learning_rate": 7.516826812403268e-06,
      "loss": 0.8245,
      "step": 28275
    },
    {
      "epoch": 1.2,
      "grad_norm": 16.629294015985117,
      "learning_rate": 7.515762950752019e-06,
      "loss": 0.8235,
      "step": 28280
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.772158858080155,
      "learning_rate": 7.514698936577474e-06,
      "loss": 0.8455,
      "step": 28285
    },
    {
      "epoch": 1.2,
      "grad_norm": 13.422855221250417,
      "learning_rate": 7.513634769944138e-06,
      "loss": 0.8694,
      "step": 28290
    },
    {
      "epoch": 1.2,
      "grad_norm": 34.3007385896735,
      "learning_rate": 7.512570450916531e-06,
      "loss": 0.8451,
      "step": 28295
    },
    {
      "epoch": 1.2,
      "grad_norm": 33.74170991704953,
      "learning_rate": 7.511505979559177e-06,
      "loss": 0.8868,
      "step": 28300
    },
    {
      "epoch": 1.2,
      "grad_norm": 26.5443760946743,
      "learning_rate": 7.5104413559366155e-06,
      "loss": 0.8272,
      "step": 28305
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.403280458128843,
      "learning_rate": 7.509376580113388e-06,
      "loss": 0.8915,
      "step": 28310
    },
    {
      "epoch": 1.2,
      "grad_norm": 13.799545885245172,
      "learning_rate": 7.508311652154051e-06,
      "loss": 0.8551,
      "step": 28315
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.428312638585163,
      "learning_rate": 7.507246572123167e-06,
      "loss": 0.8214,
      "step": 28320
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.798636870842741,
      "learning_rate": 7.506181340085308e-06,
      "loss": 0.8292,
      "step": 28325
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.950266598737123,
      "learning_rate": 7.505115956105056e-06,
      "loss": 0.8732,
      "step": 28330
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.303589137257754,
      "learning_rate": 7.504050420247003e-06,
      "loss": 0.8236,
      "step": 28335
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.490367762856796,
      "learning_rate": 7.5029847325757496e-06,
      "loss": 0.9007,
      "step": 28340
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.21824151629349,
      "learning_rate": 7.501918893155904e-06,
      "loss": 0.889,
      "step": 28345
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.194442878935659,
      "learning_rate": 7.500852902052086e-06,
      "loss": 0.827,
      "step": 28350
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.948833689413944,
      "learning_rate": 7.499786759328923e-06,
      "loss": 0.8176,
      "step": 28355
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.9000578222542215,
      "learning_rate": 7.498720465051051e-06,
      "loss": 0.8505,
      "step": 28360
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.87231332950506,
      "learning_rate": 7.4976540192831195e-06,
      "loss": 0.837,
      "step": 28365
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.398897433212114,
      "learning_rate": 7.49658742208978e-06,
      "loss": 0.832,
      "step": 28370
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.365084479002949,
      "learning_rate": 7.495520673535701e-06,
      "loss": 0.8384,
      "step": 28375
    },
    {
      "epoch": 1.2,
      "grad_norm": 7.220681949658558,
      "learning_rate": 7.494453773685553e-06,
      "loss": 0.8448,
      "step": 28380
    },
    {
      "epoch": 1.2,
      "grad_norm": 20.32257642371168,
      "learning_rate": 7.49338672260402e-06,
      "loss": 0.8133,
      "step": 28385
    },
    {
      "epoch": 1.2,
      "grad_norm": 12.690042424466201,
      "learning_rate": 7.492319520355796e-06,
      "loss": 0.933,
      "step": 28390
    },
    {
      "epoch": 1.2,
      "grad_norm": 24.03521144511985,
      "learning_rate": 7.491252167005581e-06,
      "loss": 0.7949,
      "step": 28395
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.7537894778255545,
      "learning_rate": 7.490184662618083e-06,
      "loss": 0.8727,
      "step": 28400
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.92990339266652,
      "learning_rate": 7.489117007258027e-06,
      "loss": 0.8297,
      "step": 28405
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.485700305949638,
      "learning_rate": 7.488049200990135e-06,
      "loss": 0.8353,
      "step": 28410
    },
    {
      "epoch": 1.2,
      "grad_norm": 9.359578029863616,
      "learning_rate": 7.486981243879152e-06,
      "loss": 0.881,
      "step": 28415
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.314239787702651,
      "learning_rate": 7.4859131359898195e-06,
      "loss": 0.8805,
      "step": 28420
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.5035130003843244,
      "learning_rate": 7.4848448773868975e-06,
      "loss": 0.8678,
      "step": 28425
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.368264559870695,
      "learning_rate": 7.483776468135151e-06,
      "loss": 0.8477,
      "step": 28430
    },
    {
      "epoch": 1.2,
      "grad_norm": 4.624408879333209,
      "learning_rate": 7.4827079082993514e-06,
      "loss": 0.8439,
      "step": 28435
    },
    {
      "epoch": 1.2,
      "grad_norm": 5.858398529538015,
      "learning_rate": 7.4816391979442845e-06,
      "loss": 0.8804,
      "step": 28440
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.2035357681283445,
      "learning_rate": 7.480570337134743e-06,
      "loss": 0.8525,
      "step": 28445
    },
    {
      "epoch": 1.2,
      "grad_norm": 15.272458704317378,
      "learning_rate": 7.47950132593553e-06,
      "loss": 0.8753,
      "step": 28450
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.887833676265465,
      "learning_rate": 7.478432164411454e-06,
      "loss": 0.8793,
      "step": 28455
    },
    {
      "epoch": 1.2,
      "grad_norm": 6.419626101916099,
      "learning_rate": 7.4773628526273364e-06,
      "loss": 0.8525,
      "step": 28460
    },
    {
      "epoch": 1.2,
      "grad_norm": 12.867686269089686,
      "learning_rate": 7.476293390648007e-06,
      "loss": 0.8639,
      "step": 28465
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.022212328657547,
      "learning_rate": 7.475223778538303e-06,
      "loss": 0.8157,
      "step": 28470
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.386911682796224,
      "learning_rate": 7.4741540163630735e-06,
      "loss": 0.8525,
      "step": 28475
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.593286990162671,
      "learning_rate": 7.473084104187173e-06,
      "loss": 0.8669,
      "step": 28480
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.373393707788437,
      "learning_rate": 7.472014042075469e-06,
      "loss": 0.8452,
      "step": 28485
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.581158035130508,
      "learning_rate": 7.470943830092834e-06,
      "loss": 0.8699,
      "step": 28490
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.790392270440567,
      "learning_rate": 7.469873468304152e-06,
      "loss": 0.8333,
      "step": 28495
    },
    {
      "epoch": 1.21,
      "grad_norm": 11.77505867533784,
      "learning_rate": 7.468802956774318e-06,
      "loss": 0.8345,
      "step": 28500
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.070534224632215,
      "learning_rate": 7.467732295568232e-06,
      "loss": 0.8408,
      "step": 28505
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.8857497670450325,
      "learning_rate": 7.466661484750807e-06,
      "loss": 0.8339,
      "step": 28510
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.434247891587606,
      "learning_rate": 7.465590524386962e-06,
      "loss": 0.8501,
      "step": 28515
    },
    {
      "epoch": 1.21,
      "grad_norm": 14.102312206253615,
      "learning_rate": 7.464519414541625e-06,
      "loss": 0.8666,
      "step": 28520
    },
    {
      "epoch": 1.21,
      "grad_norm": 11.500519354926956,
      "learning_rate": 7.463448155279737e-06,
      "loss": 0.8265,
      "step": 28525
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.548423021792267,
      "learning_rate": 7.4623767466662414e-06,
      "loss": 0.8194,
      "step": 28530
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.908599948378591,
      "learning_rate": 7.4613051887661e-06,
      "loss": 0.8574,
      "step": 28535
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.916027115139375,
      "learning_rate": 7.460233481644273e-06,
      "loss": 0.9028,
      "step": 28540
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.48309857593801,
      "learning_rate": 7.459161625365736e-06,
      "loss": 0.8178,
      "step": 28545
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.107350983721261,
      "learning_rate": 7.458089619995474e-06,
      "loss": 0.8632,
      "step": 28550
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.246321740012907,
      "learning_rate": 7.45701746559848e-06,
      "loss": 0.8429,
      "step": 28555
    },
    {
      "epoch": 1.21,
      "grad_norm": 10.082515216465119,
      "learning_rate": 7.455945162239754e-06,
      "loss": 0.8402,
      "step": 28560
    },
    {
      "epoch": 1.21,
      "grad_norm": 9.098963077268278,
      "learning_rate": 7.454872709984307e-06,
      "loss": 0.8279,
      "step": 28565
    },
    {
      "epoch": 1.21,
      "grad_norm": 15.86357991671177,
      "learning_rate": 7.4538001088971575e-06,
      "loss": 0.8225,
      "step": 28570
    },
    {
      "epoch": 1.21,
      "grad_norm": 9.59044500600705,
      "learning_rate": 7.452727359043337e-06,
      "loss": 0.8236,
      "step": 28575
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.317914184332231,
      "learning_rate": 7.4516544604878805e-06,
      "loss": 0.8112,
      "step": 28580
    },
    {
      "epoch": 1.21,
      "grad_norm": 11.64596292735616,
      "learning_rate": 7.450581413295835e-06,
      "loss": 0.8334,
      "step": 28585
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.6326296240670075,
      "learning_rate": 7.449508217532258e-06,
      "loss": 0.8371,
      "step": 28590
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.219978055554021,
      "learning_rate": 7.448434873262212e-06,
      "loss": 0.8624,
      "step": 28595
    },
    {
      "epoch": 1.21,
      "grad_norm": 12.57640479439711,
      "learning_rate": 7.447361380550774e-06,
      "loss": 0.8369,
      "step": 28600
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.213650905184059,
      "learning_rate": 7.446287739463022e-06,
      "loss": 0.835,
      "step": 28605
    },
    {
      "epoch": 1.21,
      "grad_norm": 12.94241083232243,
      "learning_rate": 7.44521395006405e-06,
      "loss": 0.8947,
      "step": 28610
    },
    {
      "epoch": 1.21,
      "grad_norm": 29.09712069062493,
      "learning_rate": 7.4441400124189614e-06,
      "loss": 0.8364,
      "step": 28615
    },
    {
      "epoch": 1.21,
      "grad_norm": 15.546233264346412,
      "learning_rate": 7.443065926592861e-06,
      "loss": 0.8553,
      "step": 28620
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.170031708672434,
      "learning_rate": 7.44199169265087e-06,
      "loss": 0.8634,
      "step": 28625
    },
    {
      "epoch": 1.21,
      "grad_norm": 15.110375012190065,
      "learning_rate": 7.440917310658116e-06,
      "loss": 0.8617,
      "step": 28630
    },
    {
      "epoch": 1.21,
      "grad_norm": 11.951723142896075,
      "learning_rate": 7.439842780679738e-06,
      "loss": 0.8464,
      "step": 28635
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.8626077438645785,
      "learning_rate": 7.438768102780876e-06,
      "loss": 0.8519,
      "step": 28640
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.357008880848115,
      "learning_rate": 7.437693277026688e-06,
      "loss": 0.8293,
      "step": 28645
    },
    {
      "epoch": 1.21,
      "grad_norm": 7.413801563252123,
      "learning_rate": 7.436618303482338e-06,
      "loss": 0.8374,
      "step": 28650
    },
    {
      "epoch": 1.21,
      "grad_norm": 9.057557074653534,
      "learning_rate": 7.435543182212996e-06,
      "loss": 0.8692,
      "step": 28655
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.495667929921721,
      "learning_rate": 7.434467913283846e-06,
      "loss": 0.8047,
      "step": 28660
    },
    {
      "epoch": 1.21,
      "grad_norm": 9.84849054490829,
      "learning_rate": 7.433392496760077e-06,
      "loss": 0.8497,
      "step": 28665
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.190450300692767,
      "learning_rate": 7.432316932706889e-06,
      "loss": 0.8823,
      "step": 28670
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.876990637220433,
      "learning_rate": 7.43124122118949e-06,
      "loss": 0.8129,
      "step": 28675
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.473585781561264,
      "learning_rate": 7.430165362273098e-06,
      "loss": 0.8109,
      "step": 28680
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.461410768968516,
      "learning_rate": 7.429089356022937e-06,
      "loss": 0.8424,
      "step": 28685
    },
    {
      "epoch": 1.21,
      "grad_norm": 5.26289224485984,
      "learning_rate": 7.428013202504243e-06,
      "loss": 0.84,
      "step": 28690
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.42143812914792,
      "learning_rate": 7.426936901782262e-06,
      "loss": 0.8242,
      "step": 28695
    },
    {
      "epoch": 1.21,
      "grad_norm": 4.815883699026703,
      "learning_rate": 7.425860453922246e-06,
      "loss": 0.8552,
      "step": 28700
    },
    {
      "epoch": 1.21,
      "grad_norm": 6.121663687823316,
      "learning_rate": 7.424783858989454e-06,
      "loss": 0.8945,
      "step": 28705
    },
    {
      "epoch": 1.22,
      "grad_norm": 12.180276211451368,
      "learning_rate": 7.42370711704916e-06,
      "loss": 0.8593,
      "step": 28710
    },
    {
      "epoch": 1.22,
      "grad_norm": 40.08373653153939,
      "learning_rate": 7.422630228166644e-06,
      "loss": 0.8565,
      "step": 28715
    },
    {
      "epoch": 1.22,
      "grad_norm": 31.2602347547327,
      "learning_rate": 7.421553192407192e-06,
      "loss": 0.8305,
      "step": 28720
    },
    {
      "epoch": 1.22,
      "grad_norm": 25.98790113741822,
      "learning_rate": 7.420476009836103e-06,
      "loss": 0.8505,
      "step": 28725
    },
    {
      "epoch": 1.22,
      "grad_norm": 13.550034102350098,
      "learning_rate": 7.419398680518683e-06,
      "loss": 0.8677,
      "step": 28730
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.176228148569209,
      "learning_rate": 7.418321204520249e-06,
      "loss": 0.8445,
      "step": 28735
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.58581753196945,
      "learning_rate": 7.417243581906123e-06,
      "loss": 0.8359,
      "step": 28740
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.281340271445188,
      "learning_rate": 7.4161658127416395e-06,
      "loss": 0.8229,
      "step": 28745
    },
    {
      "epoch": 1.22,
      "grad_norm": 9.837573024405199,
      "learning_rate": 7.415087897092138e-06,
      "loss": 0.8523,
      "step": 28750
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.172319214021831,
      "learning_rate": 7.414009835022973e-06,
      "loss": 0.8501,
      "step": 28755
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.252227960383379,
      "learning_rate": 7.412931626599501e-06,
      "loss": 0.8412,
      "step": 28760
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.176933885634275,
      "learning_rate": 7.411853271887093e-06,
      "loss": 0.8259,
      "step": 28765
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.312258475329861,
      "learning_rate": 7.410774770951125e-06,
      "loss": 0.7868,
      "step": 28770
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.067049698051338,
      "learning_rate": 7.409696123856983e-06,
      "loss": 0.831,
      "step": 28775
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.746892122131091,
      "learning_rate": 7.408617330670064e-06,
      "loss": 0.8188,
      "step": 28780
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.265860410818273,
      "learning_rate": 7.4075383914557705e-06,
      "loss": 0.8627,
      "step": 28785
    },
    {
      "epoch": 1.22,
      "grad_norm": 15.880108598758094,
      "learning_rate": 7.406459306279516e-06,
      "loss": 0.8239,
      "step": 28790
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.760247481397366,
      "learning_rate": 7.405380075206725e-06,
      "loss": 0.8611,
      "step": 28795
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.381576655412649,
      "learning_rate": 7.404300698302822e-06,
      "loss": 0.8453,
      "step": 28800
    },
    {
      "epoch": 1.22,
      "grad_norm": 13.684408087651821,
      "learning_rate": 7.40322117563325e-06,
      "loss": 0.7993,
      "step": 28805
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.921208671215888,
      "learning_rate": 7.402141507263458e-06,
      "loss": 0.7988,
      "step": 28810
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.275857173628684,
      "learning_rate": 7.4010616932589016e-06,
      "loss": 0.8506,
      "step": 28815
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.823043376800442,
      "learning_rate": 7.399981733685049e-06,
      "loss": 0.8692,
      "step": 28820
    },
    {
      "epoch": 1.22,
      "grad_norm": 10.193322236689127,
      "learning_rate": 7.398901628607371e-06,
      "loss": 0.8079,
      "step": 28825
    },
    {
      "epoch": 1.22,
      "grad_norm": 9.55495463812664,
      "learning_rate": 7.397821378091354e-06,
      "loss": 0.8619,
      "step": 28830
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.588245702214667,
      "learning_rate": 7.3967409822024905e-06,
      "loss": 0.8163,
      "step": 28835
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.691221586612508,
      "learning_rate": 7.39566044100628e-06,
      "loss": 0.8735,
      "step": 28840
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.551002387084731,
      "learning_rate": 7.394579754568235e-06,
      "loss": 0.8342,
      "step": 28845
    },
    {
      "epoch": 1.22,
      "grad_norm": 12.572532785115394,
      "learning_rate": 7.393498922953872e-06,
      "loss": 0.8181,
      "step": 28850
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.284503205119472,
      "learning_rate": 7.39241794622872e-06,
      "loss": 0.8516,
      "step": 28855
    },
    {
      "epoch": 1.22,
      "grad_norm": 13.032932299880988,
      "learning_rate": 7.391336824458315e-06,
      "loss": 0.8543,
      "step": 28860
    },
    {
      "epoch": 1.22,
      "grad_norm": 11.10857730651208,
      "learning_rate": 7.390255557708202e-06,
      "loss": 0.8265,
      "step": 28865
    },
    {
      "epoch": 1.22,
      "grad_norm": 9.231292686944624,
      "learning_rate": 7.389174146043936e-06,
      "loss": 0.8984,
      "step": 28870
    },
    {
      "epoch": 1.22,
      "grad_norm": 16.419166169581416,
      "learning_rate": 7.388092589531078e-06,
      "loss": 0.8584,
      "step": 28875
    },
    {
      "epoch": 1.22,
      "grad_norm": 31.25228712239634,
      "learning_rate": 7.387010888235202e-06,
      "loss": 0.8506,
      "step": 28880
    },
    {
      "epoch": 1.22,
      "grad_norm": 8.685911314917098,
      "learning_rate": 7.385929042221885e-06,
      "loss": 0.8793,
      "step": 28885
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.473763885931934,
      "learning_rate": 7.3848470515567206e-06,
      "loss": 0.8349,
      "step": 28890
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.14487750495818,
      "learning_rate": 7.383764916305303e-06,
      "loss": 0.8298,
      "step": 28895
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.4558479577407635,
      "learning_rate": 7.382682636533239e-06,
      "loss": 0.8345,
      "step": 28900
    },
    {
      "epoch": 1.22,
      "grad_norm": 12.822365263007674,
      "learning_rate": 7.381600212306147e-06,
      "loss": 0.8189,
      "step": 28905
    },
    {
      "epoch": 1.22,
      "grad_norm": 7.805840131674199,
      "learning_rate": 7.380517643689649e-06,
      "loss": 0.8419,
      "step": 28910
    },
    {
      "epoch": 1.22,
      "grad_norm": 10.689359727818442,
      "learning_rate": 7.379434930749377e-06,
      "loss": 0.8597,
      "step": 28915
    },
    {
      "epoch": 1.22,
      "grad_norm": 6.82115528686245,
      "learning_rate": 7.378352073550976e-06,
      "loss": 0.8676,
      "step": 28920
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.896500908341524,
      "learning_rate": 7.377269072160093e-06,
      "loss": 0.8595,
      "step": 28925
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.379893600618618,
      "learning_rate": 7.3761859266423896e-06,
      "loss": 0.8571,
      "step": 28930
    },
    {
      "epoch": 1.22,
      "grad_norm": 4.491961691159467,
      "learning_rate": 7.375102637063532e-06,
      "loss": 0.772,
      "step": 28935
    },
    {
      "epoch": 1.22,
      "grad_norm": 5.980315049821421,
      "learning_rate": 7.3740192034891975e-06,
      "loss": 0.8055,
      "step": 28940
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.955471521419596,
      "learning_rate": 7.3729356259850715e-06,
      "loss": 0.8667,
      "step": 28945
    },
    {
      "epoch": 1.23,
      "grad_norm": 16.38808048316717,
      "learning_rate": 7.371851904616848e-06,
      "loss": 0.8374,
      "step": 28950
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.522585680230388,
      "learning_rate": 7.3707680394502315e-06,
      "loss": 0.8402,
      "step": 28955
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.478789699465472,
      "learning_rate": 7.369684030550932e-06,
      "loss": 0.841,
      "step": 28960
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.115442217696387,
      "learning_rate": 7.368599877984669e-06,
      "loss": 0.8592,
      "step": 28965
    },
    {
      "epoch": 1.23,
      "grad_norm": 10.456542204353687,
      "learning_rate": 7.3675155818171715e-06,
      "loss": 0.8576,
      "step": 28970
    },
    {
      "epoch": 1.23,
      "grad_norm": 15.112071167837685,
      "learning_rate": 7.36643114211418e-06,
      "loss": 0.8768,
      "step": 28975
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.593518248100902,
      "learning_rate": 7.3653465589414375e-06,
      "loss": 0.8486,
      "step": 28980
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.888671014717431,
      "learning_rate": 7.364261832364701e-06,
      "loss": 0.7903,
      "step": 28985
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.9975094943152,
      "learning_rate": 7.363176962449734e-06,
      "loss": 0.8474,
      "step": 28990
    },
    {
      "epoch": 1.23,
      "grad_norm": 21.043961229223278,
      "learning_rate": 7.362091949262306e-06,
      "loss": 0.8114,
      "step": 28995
    },
    {
      "epoch": 1.23,
      "grad_norm": 27.24738938766072,
      "learning_rate": 7.361006792868205e-06,
      "loss": 0.8018,
      "step": 29000
    },
    {
      "epoch": 1.23,
      "grad_norm": 13.92275874520696,
      "learning_rate": 7.3599214933332155e-06,
      "loss": 0.8362,
      "step": 29005
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.243077377047206,
      "learning_rate": 7.358836050723135e-06,
      "loss": 0.8351,
      "step": 29010
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.20317227239975,
      "learning_rate": 7.357750465103775e-06,
      "loss": 0.8294,
      "step": 29015
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.333451603714722,
      "learning_rate": 7.3566647365409485e-06,
      "loss": 0.8288,
      "step": 29020
    },
    {
      "epoch": 1.23,
      "grad_norm": 12.045718477436173,
      "learning_rate": 7.355578865100482e-06,
      "loss": 0.8099,
      "step": 29025
    },
    {
      "epoch": 1.23,
      "grad_norm": 16.050989470551833,
      "learning_rate": 7.3544928508482075e-06,
      "loss": 0.8772,
      "step": 29030
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.577805652411866,
      "learning_rate": 7.353406693849965e-06,
      "loss": 0.8318,
      "step": 29035
    },
    {
      "epoch": 1.23,
      "grad_norm": 11.071099280151154,
      "learning_rate": 7.352320394171609e-06,
      "loss": 0.8489,
      "step": 29040
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.536610185534432,
      "learning_rate": 7.351233951878996e-06,
      "loss": 0.8477,
      "step": 29045
    },
    {
      "epoch": 1.23,
      "grad_norm": 13.565799746508434,
      "learning_rate": 7.350147367037993e-06,
      "loss": 0.8337,
      "step": 29050
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.637309960920233,
      "learning_rate": 7.349060639714478e-06,
      "loss": 0.8779,
      "step": 29055
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.14280788308352,
      "learning_rate": 7.347973769974336e-06,
      "loss": 0.8341,
      "step": 29060
    },
    {
      "epoch": 1.23,
      "grad_norm": 7.4523962552507115,
      "learning_rate": 7.346886757883461e-06,
      "loss": 0.8216,
      "step": 29065
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.225313854637131,
      "learning_rate": 7.345799603507754e-06,
      "loss": 0.8413,
      "step": 29070
    },
    {
      "epoch": 1.23,
      "grad_norm": 16.336602963552096,
      "learning_rate": 7.3447123069131275e-06,
      "loss": 0.8908,
      "step": 29075
    },
    {
      "epoch": 1.23,
      "grad_norm": 16.202059921097902,
      "learning_rate": 7.3436248681654995e-06,
      "loss": 0.8454,
      "step": 29080
    },
    {
      "epoch": 1.23,
      "grad_norm": 9.217272290208314,
      "learning_rate": 7.342537287330801e-06,
      "loss": 0.8698,
      "step": 29085
    },
    {
      "epoch": 1.23,
      "grad_norm": 6.3760256901915024,
      "learning_rate": 7.341449564474965e-06,
      "loss": 0.8415,
      "step": 29090
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.126219696437688,
      "learning_rate": 7.34036169966394e-06,
      "loss": 0.8746,
      "step": 29095
    },
    {
      "epoch": 1.23,
      "grad_norm": 9.569498847629877,
      "learning_rate": 7.339273692963679e-06,
      "loss": 0.8014,
      "step": 29100
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.820757424697069,
      "learning_rate": 7.338185544440143e-06,
      "loss": 0.8414,
      "step": 29105
    },
    {
      "epoch": 1.23,
      "grad_norm": 22.178867219973096,
      "learning_rate": 7.337097254159307e-06,
      "loss": 0.8348,
      "step": 29110
    },
    {
      "epoch": 1.23,
      "grad_norm": 25.718057039665243,
      "learning_rate": 7.3360088221871474e-06,
      "loss": 0.8279,
      "step": 29115
    },
    {
      "epoch": 1.23,
      "grad_norm": 19.84135891604384,
      "learning_rate": 7.334920248589654e-06,
      "loss": 0.8229,
      "step": 29120
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.5264950980969125,
      "learning_rate": 7.3338315334328244e-06,
      "loss": 0.8556,
      "step": 29125
    },
    {
      "epoch": 1.23,
      "grad_norm": 4.69602826829856,
      "learning_rate": 7.332742676782663e-06,
      "loss": 0.8663,
      "step": 29130
    },
    {
      "epoch": 1.23,
      "grad_norm": 10.38347927925216,
      "learning_rate": 7.331653678705184e-06,
      "loss": 0.8954,
      "step": 29135
    },
    {
      "epoch": 1.23,
      "grad_norm": 11.686709273365198,
      "learning_rate": 7.33056453926641e-06,
      "loss": 0.8736,
      "step": 29140
    },
    {
      "epoch": 1.23,
      "grad_norm": 8.637430529097813,
      "learning_rate": 7.3294752585323735e-06,
      "loss": 0.8411,
      "step": 29145
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.370082315510036,
      "learning_rate": 7.328385836569114e-06,
      "loss": 0.8307,
      "step": 29150
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.046395513314983,
      "learning_rate": 7.3272962734426785e-06,
      "loss": 0.8119,
      "step": 29155
    },
    {
      "epoch": 1.23,
      "grad_norm": 5.266396992448873,
      "learning_rate": 7.326206569219127e-06,
      "loss": 0.8174,
      "step": 29160
    },
    {
      "epoch": 1.23,
      "grad_norm": 14.051216831171775,
      "learning_rate": 7.3251167239645225e-06,
      "loss": 0.8065,
      "step": 29165
    },
    {
      "epoch": 1.23,
      "grad_norm": 10.62295564404839,
      "learning_rate": 7.3240267377449405e-06,
      "loss": 0.809,
      "step": 29170
    },
    {
      "epoch": 1.23,
      "grad_norm": 9.845446142338483,
      "learning_rate": 7.322936610626463e-06,
      "loss": 0.8275,
      "step": 29175
    },
    {
      "epoch": 1.23,
      "grad_norm": 9.239379955376876,
      "learning_rate": 7.32184634267518e-06,
      "loss": 0.8658,
      "step": 29180
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.858690595472981,
      "learning_rate": 7.320755933957193e-06,
      "loss": 0.8575,
      "step": 29185
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.105704610039998,
      "learning_rate": 7.3196653845386115e-06,
      "loss": 0.8372,
      "step": 29190
    },
    {
      "epoch": 1.24,
      "grad_norm": 13.853643856358639,
      "learning_rate": 7.318574694485548e-06,
      "loss": 0.8411,
      "step": 29195
    },
    {
      "epoch": 1.24,
      "grad_norm": 16.33363374073994,
      "learning_rate": 7.317483863864132e-06,
      "loss": 0.8372,
      "step": 29200
    },
    {
      "epoch": 1.24,
      "grad_norm": 11.276578318383763,
      "learning_rate": 7.316392892740495e-06,
      "loss": 0.8044,
      "step": 29205
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.407926808464365,
      "learning_rate": 7.31530178118078e-06,
      "loss": 0.8196,
      "step": 29210
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.106938516624817,
      "learning_rate": 7.314210529251138e-06,
      "loss": 0.8281,
      "step": 29215
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.936377290854027,
      "learning_rate": 7.313119137017728e-06,
      "loss": 0.8558,
      "step": 29220
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.726034332992367,
      "learning_rate": 7.31202760454672e-06,
      "loss": 0.8203,
      "step": 29225
    },
    {
      "epoch": 1.24,
      "grad_norm": 11.866780966485535,
      "learning_rate": 7.310935931904286e-06,
      "loss": 0.8235,
      "step": 29230
    },
    {
      "epoch": 1.24,
      "grad_norm": 8.152282577698498,
      "learning_rate": 7.309844119156615e-06,
      "loss": 0.8163,
      "step": 29235
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.4549454312139485,
      "learning_rate": 7.308752166369899e-06,
      "loss": 0.8583,
      "step": 29240
    },
    {
      "epoch": 1.24,
      "grad_norm": 9.723888347405351,
      "learning_rate": 7.307660073610339e-06,
      "loss": 0.8456,
      "step": 29245
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.48381014277602,
      "learning_rate": 7.306567840944148e-06,
      "loss": 0.8528,
      "step": 29250
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.821784134757839,
      "learning_rate": 7.305475468437541e-06,
      "loss": 0.8326,
      "step": 29255
    },
    {
      "epoch": 1.24,
      "grad_norm": 10.177411450799287,
      "learning_rate": 7.304382956156748e-06,
      "loss": 0.9001,
      "step": 29260
    },
    {
      "epoch": 1.24,
      "grad_norm": 12.197740145653627,
      "learning_rate": 7.3032903041680035e-06,
      "loss": 0.8796,
      "step": 29265
    },
    {
      "epoch": 1.24,
      "grad_norm": 15.146197711207414,
      "learning_rate": 7.302197512537553e-06,
      "loss": 0.8141,
      "step": 29270
    },
    {
      "epoch": 1.24,
      "grad_norm": 14.588899352718979,
      "learning_rate": 7.30110458133165e-06,
      "loss": 0.7912,
      "step": 29275
    },
    {
      "epoch": 1.24,
      "grad_norm": 18.317943918552075,
      "learning_rate": 7.300011510616554e-06,
      "loss": 0.8218,
      "step": 29280
    },
    {
      "epoch": 1.24,
      "grad_norm": 12.767231643223573,
      "learning_rate": 7.298918300458534e-06,
      "loss": 0.8558,
      "step": 29285
    },
    {
      "epoch": 1.24,
      "grad_norm": 13.071614531742537,
      "learning_rate": 7.297824950923869e-06,
      "loss": 0.8416,
      "step": 29290
    },
    {
      "epoch": 1.24,
      "grad_norm": 9.742379957112458,
      "learning_rate": 7.296731462078845e-06,
      "loss": 0.8609,
      "step": 29295
    },
    {
      "epoch": 1.24,
      "grad_norm": 12.509863489478066,
      "learning_rate": 7.295637833989759e-06,
      "loss": 0.878,
      "step": 29300
    },
    {
      "epoch": 1.24,
      "grad_norm": 12.372861781184831,
      "learning_rate": 7.294544066722913e-06,
      "loss": 0.8243,
      "step": 29305
    },
    {
      "epoch": 1.24,
      "grad_norm": 11.801422988956855,
      "learning_rate": 7.29345016034462e-06,
      "loss": 0.7955,
      "step": 29310
    },
    {
      "epoch": 1.24,
      "grad_norm": 16.522064923505,
      "learning_rate": 7.292356114921197e-06,
      "loss": 0.8169,
      "step": 29315
    },
    {
      "epoch": 1.24,
      "grad_norm": 16.30935027650137,
      "learning_rate": 7.291261930518977e-06,
      "loss": 0.8094,
      "step": 29320
    },
    {
      "epoch": 1.24,
      "grad_norm": 16.834573616715605,
      "learning_rate": 7.290167607204293e-06,
      "loss": 0.8556,
      "step": 29325
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.393644376665236,
      "learning_rate": 7.289073145043496e-06,
      "loss": 0.8021,
      "step": 29330
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.807999546801356,
      "learning_rate": 7.287978544102936e-06,
      "loss": 0.8571,
      "step": 29335
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.9648350108960955,
      "learning_rate": 7.286883804448975e-06,
      "loss": 0.8259,
      "step": 29340
    },
    {
      "epoch": 1.24,
      "grad_norm": 12.53822869513032,
      "learning_rate": 7.2857889261479864e-06,
      "loss": 0.8266,
      "step": 29345
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.9651634565440315,
      "learning_rate": 7.2846939092663475e-06,
      "loss": 0.8503,
      "step": 29350
    },
    {
      "epoch": 1.24,
      "grad_norm": 7.512055928241283,
      "learning_rate": 7.283598753870447e-06,
      "loss": 0.805,
      "step": 29355
    },
    {
      "epoch": 1.24,
      "grad_norm": 10.656744584243025,
      "learning_rate": 7.282503460026681e-06,
      "loss": 0.8722,
      "step": 29360
    },
    {
      "epoch": 1.24,
      "grad_norm": 12.702422182665197,
      "learning_rate": 7.281408027801453e-06,
      "loss": 0.8299,
      "step": 29365
    },
    {
      "epoch": 1.24,
      "grad_norm": 9.88179543244977,
      "learning_rate": 7.280312457261177e-06,
      "loss": 0.8746,
      "step": 29370
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.625958265756664,
      "learning_rate": 7.2792167484722716e-06,
      "loss": 0.8691,
      "step": 29375
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.175460045103117,
      "learning_rate": 7.278120901501171e-06,
      "loss": 0.8734,
      "step": 29380
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.0374724798087245,
      "learning_rate": 7.2770249164143085e-06,
      "loss": 0.8467,
      "step": 29385
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.9447351737332585,
      "learning_rate": 7.275928793278134e-06,
      "loss": 0.7984,
      "step": 29390
    },
    {
      "epoch": 1.24,
      "grad_norm": 5.765239433909384,
      "learning_rate": 7.274832532159101e-06,
      "loss": 0.84,
      "step": 29395
    },
    {
      "epoch": 1.24,
      "grad_norm": 18.969070211009956,
      "learning_rate": 7.27373613312367e-06,
      "loss": 0.8546,
      "step": 29400
    },
    {
      "epoch": 1.24,
      "grad_norm": 6.55334589889723,
      "learning_rate": 7.272639596238317e-06,
      "loss": 0.8623,
      "step": 29405
    },
    {
      "epoch": 1.24,
      "grad_norm": 10.854588120999194,
      "learning_rate": 7.271542921569519e-06,
      "loss": 0.8339,
      "step": 29410
    },
    {
      "epoch": 1.24,
      "grad_norm": 4.628159325090979,
      "learning_rate": 7.270446109183764e-06,
      "loss": 0.8348,
      "step": 29415
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.527769835290532,
      "learning_rate": 7.269349159147551e-06,
      "loss": 0.8596,
      "step": 29420
    },
    {
      "epoch": 1.25,
      "grad_norm": 12.660792412007858,
      "learning_rate": 7.268252071527381e-06,
      "loss": 0.8488,
      "step": 29425
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.912923489216101,
      "learning_rate": 7.26715484638977e-06,
      "loss": 0.7763,
      "step": 29430
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.99618155816127,
      "learning_rate": 7.266057483801239e-06,
      "loss": 0.8575,
      "step": 29435
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.256170042305894,
      "learning_rate": 7.264959983828317e-06,
      "loss": 0.8493,
      "step": 29440
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.858757837784894,
      "learning_rate": 7.263862346537544e-06,
      "loss": 0.8471,
      "step": 29445
    },
    {
      "epoch": 1.25,
      "grad_norm": 11.152206596373935,
      "learning_rate": 7.262764571995464e-06,
      "loss": 0.8254,
      "step": 29450
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.01167043438302,
      "learning_rate": 7.261666660268635e-06,
      "loss": 0.8428,
      "step": 29455
    },
    {
      "epoch": 1.25,
      "grad_norm": 14.02501198411762,
      "learning_rate": 7.2605686114236174e-06,
      "loss": 0.8207,
      "step": 29460
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.148897798329034,
      "learning_rate": 7.2594704255269835e-06,
      "loss": 0.8053,
      "step": 29465
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.856003729115084,
      "learning_rate": 7.258372102645313e-06,
      "loss": 0.8252,
      "step": 29470
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.605999149460576,
      "learning_rate": 7.257273642845194e-06,
      "loss": 0.8476,
      "step": 29475
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.936180719158974,
      "learning_rate": 7.256175046193223e-06,
      "loss": 0.8538,
      "step": 29480
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.694873933728037,
      "learning_rate": 7.2550763127560055e-06,
      "loss": 0.7998,
      "step": 29485
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.005500778403437,
      "learning_rate": 7.253977442600153e-06,
      "loss": 0.8358,
      "step": 29490
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.8578578427582,
      "learning_rate": 7.252878435792288e-06,
      "loss": 0.863,
      "step": 29495
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.807718655827233,
      "learning_rate": 7.251779292399039e-06,
      "loss": 0.8712,
      "step": 29500
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.528911006867819,
      "learning_rate": 7.250680012487045e-06,
      "loss": 0.8376,
      "step": 29505
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.188934788366203,
      "learning_rate": 7.2495805961229506e-06,
      "loss": 0.8171,
      "step": 29510
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.703753503479719,
      "learning_rate": 7.248481043373412e-06,
      "loss": 0.8633,
      "step": 29515
    },
    {
      "epoch": 1.25,
      "grad_norm": 20.157590759494898,
      "learning_rate": 7.24738135430509e-06,
      "loss": 0.8348,
      "step": 29520
    },
    {
      "epoch": 1.25,
      "grad_norm": 23.32965649754932,
      "learning_rate": 7.2462815289846575e-06,
      "loss": 0.8434,
      "step": 29525
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.713036855559205,
      "learning_rate": 7.2451815674787916e-06,
      "loss": 0.838,
      "step": 29530
    },
    {
      "epoch": 1.25,
      "grad_norm": 9.484524017693479,
      "learning_rate": 7.2440814698541805e-06,
      "loss": 0.8264,
      "step": 29535
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.676486139520441,
      "learning_rate": 7.242981236177521e-06,
      "loss": 0.7958,
      "step": 29540
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.701090946462555,
      "learning_rate": 7.2418808665155135e-06,
      "loss": 0.8532,
      "step": 29545
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.740492645623074,
      "learning_rate": 7.240780360934876e-06,
      "loss": 0.8624,
      "step": 29550
    },
    {
      "epoch": 1.25,
      "grad_norm": 11.348530755680857,
      "learning_rate": 7.2396797195023246e-06,
      "loss": 0.7957,
      "step": 29555
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.451479577235648,
      "learning_rate": 7.238578942284588e-06,
      "loss": 0.8064,
      "step": 29560
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.10087789848718,
      "learning_rate": 7.2374780293484045e-06,
      "loss": 0.8371,
      "step": 29565
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.446994852041026,
      "learning_rate": 7.2363769807605185e-06,
      "loss": 0.8292,
      "step": 29570
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.818436699407271,
      "learning_rate": 7.235275796587685e-06,
      "loss": 0.8299,
      "step": 29575
    },
    {
      "epoch": 1.25,
      "grad_norm": 15.23901620938986,
      "learning_rate": 7.2341744768966634e-06,
      "loss": 0.8569,
      "step": 29580
    },
    {
      "epoch": 1.25,
      "grad_norm": 30.841845289449214,
      "learning_rate": 7.233073021754224e-06,
      "loss": 0.8234,
      "step": 29585
    },
    {
      "epoch": 1.25,
      "grad_norm": 30.438206820369167,
      "learning_rate": 7.231971431227146e-06,
      "loss": 0.818,
      "step": 29590
    },
    {
      "epoch": 1.25,
      "grad_norm": 14.502394987712721,
      "learning_rate": 7.230869705382213e-06,
      "loss": 0.779,
      "step": 29595
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.794983742770824,
      "learning_rate": 7.229767844286222e-06,
      "loss": 0.8196,
      "step": 29600
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.516537531055367,
      "learning_rate": 7.228665848005975e-06,
      "loss": 0.8043,
      "step": 29605
    },
    {
      "epoch": 1.25,
      "grad_norm": 7.593921595982875,
      "learning_rate": 7.2275637166082825e-06,
      "loss": 0.8608,
      "step": 29610
    },
    {
      "epoch": 1.25,
      "grad_norm": 8.829725951709456,
      "learning_rate": 7.226461450159963e-06,
      "loss": 0.804,
      "step": 29615
    },
    {
      "epoch": 1.25,
      "grad_norm": 12.601119702098064,
      "learning_rate": 7.225359048727843e-06,
      "loss": 0.8669,
      "step": 29620
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.731621951276361,
      "learning_rate": 7.22425651237876e-06,
      "loss": 0.8656,
      "step": 29625
    },
    {
      "epoch": 1.25,
      "grad_norm": 5.035318345850336,
      "learning_rate": 7.223153841179555e-06,
      "loss": 0.8691,
      "step": 29630
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.7706115583553625,
      "learning_rate": 7.222051035197083e-06,
      "loss": 0.8032,
      "step": 29635
    },
    {
      "epoch": 1.25,
      "grad_norm": 4.871245622649054,
      "learning_rate": 7.220948094498201e-06,
      "loss": 0.8351,
      "step": 29640
    },
    {
      "epoch": 1.25,
      "grad_norm": 6.408729100339189,
      "learning_rate": 7.219845019149778e-06,
      "loss": 0.8328,
      "step": 29645
    },
    {
      "epoch": 1.25,
      "grad_norm": 10.043280881192205,
      "learning_rate": 7.21874180921869e-06,
      "loss": 0.8324,
      "step": 29650
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.569204109694414,
      "learning_rate": 7.217638464771822e-06,
      "loss": 0.834,
      "step": 29655
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.013292002671288,
      "learning_rate": 7.216534985876067e-06,
      "loss": 0.8142,
      "step": 29660
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.191731935573789,
      "learning_rate": 7.215431372598324e-06,
      "loss": 0.8634,
      "step": 29665
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.476164786543851,
      "learning_rate": 7.214327625005502e-06,
      "loss": 0.8276,
      "step": 29670
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.7686015597918985,
      "learning_rate": 7.213223743164519e-06,
      "loss": 0.8088,
      "step": 29675
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.524827207714984,
      "learning_rate": 7.212119727142301e-06,
      "loss": 0.7945,
      "step": 29680
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.400357228242433,
      "learning_rate": 7.21101557700578e-06,
      "loss": 0.822,
      "step": 29685
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.287508002072973,
      "learning_rate": 7.209911292821895e-06,
      "loss": 0.899,
      "step": 29690
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.858578429962108,
      "learning_rate": 7.208806874657599e-06,
      "loss": 0.8326,
      "step": 29695
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.9568991801922495,
      "learning_rate": 7.207702322579849e-06,
      "loss": 0.7937,
      "step": 29700
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.060902116530706,
      "learning_rate": 7.20659763665561e-06,
      "loss": 0.8261,
      "step": 29705
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.320079869779115,
      "learning_rate": 7.205492816951856e-06,
      "loss": 0.8031,
      "step": 29710
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.490553942136585,
      "learning_rate": 7.204387863535568e-06,
      "loss": 0.8394,
      "step": 29715
    },
    {
      "epoch": 1.26,
      "grad_norm": 22.94529087735934,
      "learning_rate": 7.203282776473738e-06,
      "loss": 0.8256,
      "step": 29720
    },
    {
      "epoch": 1.26,
      "grad_norm": 18.620025872079975,
      "learning_rate": 7.202177555833364e-06,
      "loss": 0.8276,
      "step": 29725
    },
    {
      "epoch": 1.26,
      "grad_norm": 12.20123193795549,
      "learning_rate": 7.20107220168145e-06,
      "loss": 0.8131,
      "step": 29730
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.500358002285791,
      "learning_rate": 7.199966714085012e-06,
      "loss": 0.8314,
      "step": 29735
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.023400003934056,
      "learning_rate": 7.198861093111072e-06,
      "loss": 0.8283,
      "step": 29740
    },
    {
      "epoch": 1.26,
      "grad_norm": 11.659605185452397,
      "learning_rate": 7.197755338826663e-06,
      "loss": 0.892,
      "step": 29745
    },
    {
      "epoch": 1.26,
      "grad_norm": 17.435221718273503,
      "learning_rate": 7.19664945129882e-06,
      "loss": 0.8528,
      "step": 29750
    },
    {
      "epoch": 1.26,
      "grad_norm": 12.609138868985822,
      "learning_rate": 7.195543430594592e-06,
      "loss": 0.8005,
      "step": 29755
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.566826661565516,
      "learning_rate": 7.194437276781032e-06,
      "loss": 0.8386,
      "step": 29760
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.40258414128485,
      "learning_rate": 7.193330989925205e-06,
      "loss": 0.8362,
      "step": 29765
    },
    {
      "epoch": 1.26,
      "grad_norm": 17.94998461763534,
      "learning_rate": 7.192224570094179e-06,
      "loss": 0.8383,
      "step": 29770
    },
    {
      "epoch": 1.26,
      "grad_norm": 17.446892458292783,
      "learning_rate": 7.1911180173550365e-06,
      "loss": 0.8736,
      "step": 29775
    },
    {
      "epoch": 1.26,
      "grad_norm": 17.194438449584965,
      "learning_rate": 7.190011331774862e-06,
      "loss": 0.8522,
      "step": 29780
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.15836325094222,
      "learning_rate": 7.188904513420751e-06,
      "loss": 0.8158,
      "step": 29785
    },
    {
      "epoch": 1.26,
      "grad_norm": 12.246171630612176,
      "learning_rate": 7.187797562359807e-06,
      "loss": 0.8552,
      "step": 29790
    },
    {
      "epoch": 1.26,
      "grad_norm": 11.480493201526757,
      "learning_rate": 7.186690478659142e-06,
      "loss": 0.8919,
      "step": 29795
    },
    {
      "epoch": 1.26,
      "grad_norm": 15.29436640897875,
      "learning_rate": 7.185583262385874e-06,
      "loss": 0.8495,
      "step": 29800
    },
    {
      "epoch": 1.26,
      "grad_norm": 16.75781068810704,
      "learning_rate": 7.184475913607131e-06,
      "loss": 0.7748,
      "step": 29805
    },
    {
      "epoch": 1.26,
      "grad_norm": 13.243322280388988,
      "learning_rate": 7.183368432390048e-06,
      "loss": 0.8363,
      "step": 29810
    },
    {
      "epoch": 1.26,
      "grad_norm": 12.32277189580591,
      "learning_rate": 7.1822608188017664e-06,
      "loss": 0.8474,
      "step": 29815
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.933760192679248,
      "learning_rate": 7.1811530729094415e-06,
      "loss": 0.8241,
      "step": 29820
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.748001295573413,
      "learning_rate": 7.180045194780229e-06,
      "loss": 0.7955,
      "step": 29825
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.705233959104965,
      "learning_rate": 7.1789371844812996e-06,
      "loss": 0.8613,
      "step": 29830
    },
    {
      "epoch": 1.26,
      "grad_norm": 7.360026036500931,
      "learning_rate": 7.1778290420798255e-06,
      "loss": 0.8901,
      "step": 29835
    },
    {
      "epoch": 1.26,
      "grad_norm": 8.24755712957183,
      "learning_rate": 7.176720767642992e-06,
      "loss": 0.8383,
      "step": 29840
    },
    {
      "epoch": 1.26,
      "grad_norm": 9.233027563859196,
      "learning_rate": 7.17561236123799e-06,
      "loss": 0.8075,
      "step": 29845
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.41474004468858,
      "learning_rate": 7.174503822932019e-06,
      "loss": 0.8191,
      "step": 29850
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.554102288736465,
      "learning_rate": 7.173395152792288e-06,
      "loss": 0.8487,
      "step": 29855
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.718494092386051,
      "learning_rate": 7.172286350886011e-06,
      "loss": 0.7789,
      "step": 29860
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.840213149751106,
      "learning_rate": 7.171177417280408e-06,
      "loss": 0.7893,
      "step": 29865
    },
    {
      "epoch": 1.26,
      "grad_norm": 6.0942096767544784,
      "learning_rate": 7.1700683520427164e-06,
      "loss": 0.8417,
      "step": 29870
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.823508151141095,
      "learning_rate": 7.168959155240171e-06,
      "loss": 0.8457,
      "step": 29875
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.889126439719206,
      "learning_rate": 7.167849826940022e-06,
      "loss": 0.8591,
      "step": 29880
    },
    {
      "epoch": 1.26,
      "grad_norm": 5.363740375639456,
      "learning_rate": 7.166740367209524e-06,
      "loss": 0.8409,
      "step": 29885
    },
    {
      "epoch": 1.26,
      "grad_norm": 4.949181665082757,
      "learning_rate": 7.165630776115938e-06,
      "loss": 0.7992,
      "step": 29890
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.274609147746191,
      "learning_rate": 7.164521053726539e-06,
      "loss": 0.8235,
      "step": 29895
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.2457391828557,
      "learning_rate": 7.163411200108602e-06,
      "loss": 0.8344,
      "step": 29900
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.03284665877464,
      "learning_rate": 7.1623012153294195e-06,
      "loss": 0.9036,
      "step": 29905
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.15690940550589,
      "learning_rate": 7.1611910994562816e-06,
      "loss": 0.8094,
      "step": 29910
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.7023573036996815,
      "learning_rate": 7.160080852556494e-06,
      "loss": 0.8336,
      "step": 29915
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.735035717766237,
      "learning_rate": 7.158970474697366e-06,
      "loss": 0.8675,
      "step": 29920
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.687302877373789,
      "learning_rate": 7.157859965946219e-06,
      "loss": 0.8061,
      "step": 29925
    },
    {
      "epoch": 1.27,
      "grad_norm": 10.597599726749966,
      "learning_rate": 7.156749326370379e-06,
      "loss": 0.8478,
      "step": 29930
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.678172636868652,
      "learning_rate": 7.155638556037179e-06,
      "loss": 0.8029,
      "step": 29935
    },
    {
      "epoch": 1.27,
      "grad_norm": 9.042772133185384,
      "learning_rate": 7.154527655013963e-06,
      "loss": 0.8299,
      "step": 29940
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.4137584243863035,
      "learning_rate": 7.153416623368082e-06,
      "loss": 0.8047,
      "step": 29945
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.779318593410042,
      "learning_rate": 7.152305461166893e-06,
      "loss": 0.8468,
      "step": 29950
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.683268593499577,
      "learning_rate": 7.151194168477765e-06,
      "loss": 0.8366,
      "step": 29955
    },
    {
      "epoch": 1.27,
      "grad_norm": 9.202841736955973,
      "learning_rate": 7.150082745368071e-06,
      "loss": 0.856,
      "step": 29960
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.74698312608711,
      "learning_rate": 7.148971191905193e-06,
      "loss": 0.8268,
      "step": 29965
    },
    {
      "epoch": 1.27,
      "grad_norm": 13.620053454352865,
      "learning_rate": 7.147859508156523e-06,
      "loss": 0.8727,
      "step": 29970
    },
    {
      "epoch": 1.27,
      "grad_norm": 10.30299209246428,
      "learning_rate": 7.146747694189455e-06,
      "loss": 0.8286,
      "step": 29975
    },
    {
      "epoch": 1.27,
      "grad_norm": 17.84742173687575,
      "learning_rate": 7.1456357500714e-06,
      "loss": 0.845,
      "step": 29980
    },
    {
      "epoch": 1.27,
      "grad_norm": 9.462707383785776,
      "learning_rate": 7.1445236758697685e-06,
      "loss": 0.831,
      "step": 29985
    },
    {
      "epoch": 1.27,
      "grad_norm": 11.923176002101156,
      "learning_rate": 7.143411471651984e-06,
      "loss": 0.8552,
      "step": 29990
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.617685315647928,
      "learning_rate": 7.142299137485477e-06,
      "loss": 0.8683,
      "step": 29995
    },
    {
      "epoch": 1.27,
      "grad_norm": 13.48666076840623,
      "learning_rate": 7.141186673437681e-06,
      "loss": 0.8094,
      "step": 30000
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.790314995126945,
      "learning_rate": 7.140074079576046e-06,
      "loss": 0.8444,
      "step": 30005
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.3584278837068755,
      "learning_rate": 7.1389613559680226e-06,
      "loss": 0.8139,
      "step": 30010
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.579526282844967,
      "learning_rate": 7.1378485026810726e-06,
      "loss": 0.8037,
      "step": 30015
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.497370313315846,
      "learning_rate": 7.136735519782666e-06,
      "loss": 0.8511,
      "step": 30020
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.868112884843894,
      "learning_rate": 7.135622407340278e-06,
      "loss": 0.8709,
      "step": 30025
    },
    {
      "epoch": 1.27,
      "grad_norm": 16.47655938002021,
      "learning_rate": 7.134509165421395e-06,
      "loss": 0.8252,
      "step": 30030
    },
    {
      "epoch": 1.27,
      "grad_norm": 12.863726584047223,
      "learning_rate": 7.133395794093507e-06,
      "loss": 0.7943,
      "step": 30035
    },
    {
      "epoch": 1.27,
      "grad_norm": 10.63944516989122,
      "learning_rate": 7.132282293424117e-06,
      "loss": 0.8772,
      "step": 30040
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.043440768260508,
      "learning_rate": 7.1311686634807335e-06,
      "loss": 0.8003,
      "step": 30045
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.016223481684054,
      "learning_rate": 7.13005490433087e-06,
      "loss": 0.8138,
      "step": 30050
    },
    {
      "epoch": 1.27,
      "grad_norm": 5.533192022029013,
      "learning_rate": 7.128941016042053e-06,
      "loss": 0.8365,
      "step": 30055
    },
    {
      "epoch": 1.27,
      "grad_norm": 6.085027975222191,
      "learning_rate": 7.12782699868181e-06,
      "loss": 0.8327,
      "step": 30060
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.976621528814561,
      "learning_rate": 7.126712852317688e-06,
      "loss": 0.7661,
      "step": 30065
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.752234974111703,
      "learning_rate": 7.125598577017229e-06,
      "loss": 0.8189,
      "step": 30070
    },
    {
      "epoch": 1.27,
      "grad_norm": 4.963205185987787,
      "learning_rate": 7.1244841728479875e-06,
      "loss": 0.8638,
      "step": 30075
    },
    {
      "epoch": 1.27,
      "grad_norm": 10.507446730850619,
      "learning_rate": 7.123369639877531e-06,
      "loss": 0.8139,
      "step": 30080
    },
    {
      "epoch": 1.27,
      "grad_norm": 15.032151920413579,
      "learning_rate": 7.122254978173426e-06,
      "loss": 0.8474,
      "step": 30085
    },
    {
      "epoch": 1.27,
      "grad_norm": 17.498418053837263,
      "learning_rate": 7.1211401878032546e-06,
      "loss": 0.8349,
      "step": 30090
    },
    {
      "epoch": 1.27,
      "grad_norm": 9.064477368120707,
      "learning_rate": 7.1200252688346e-06,
      "loss": 0.8027,
      "step": 30095
    },
    {
      "epoch": 1.27,
      "grad_norm": 8.782146255556922,
      "learning_rate": 7.11891022133506e-06,
      "loss": 0.8418,
      "step": 30100
    },
    {
      "epoch": 1.27,
      "grad_norm": 12.877113042272308,
      "learning_rate": 7.117795045372233e-06,
      "loss": 0.7941,
      "step": 30105
    },
    {
      "epoch": 1.27,
      "grad_norm": 7.611728568047225,
      "learning_rate": 7.116679741013731e-06,
      "loss": 0.8681,
      "step": 30110
    },
    {
      "epoch": 1.27,
      "grad_norm": 11.47638333755626,
      "learning_rate": 7.115564308327172e-06,
      "loss": 0.8081,
      "step": 30115
    },
    {
      "epoch": 1.27,
      "grad_norm": 9.840736983749753,
      "learning_rate": 7.11444874738018e-06,
      "loss": 0.7786,
      "step": 30120
    },
    {
      "epoch": 1.27,
      "grad_norm": 12.796416217118242,
      "learning_rate": 7.113333058240388e-06,
      "loss": 0.8456,
      "step": 30125
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.588679646362951,
      "learning_rate": 7.11221724097544e-06,
      "loss": 0.8299,
      "step": 30130
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.919185949709744,
      "learning_rate": 7.11110129565298e-06,
      "loss": 0.8144,
      "step": 30135
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.290741294604033,
      "learning_rate": 7.109985222340667e-06,
      "loss": 0.8372,
      "step": 30140
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.43352744007455,
      "learning_rate": 7.108869021106166e-06,
      "loss": 0.8083,
      "step": 30145
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.037044408829985,
      "learning_rate": 7.107752692017148e-06,
      "loss": 0.8528,
      "step": 30150
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.502362942761619,
      "learning_rate": 7.106636235141293e-06,
      "loss": 0.8467,
      "step": 30155
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.623877190625405,
      "learning_rate": 7.105519650546289e-06,
      "loss": 0.8181,
      "step": 30160
    },
    {
      "epoch": 1.28,
      "grad_norm": 14.75090657932128,
      "learning_rate": 7.104402938299829e-06,
      "loss": 0.827,
      "step": 30165
    },
    {
      "epoch": 1.28,
      "grad_norm": 16.44633905001911,
      "learning_rate": 7.103286098469618e-06,
      "loss": 0.8152,
      "step": 30170
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.399227906527695,
      "learning_rate": 7.102169131123369e-06,
      "loss": 0.7824,
      "step": 30175
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.7203933405543825,
      "learning_rate": 7.101052036328794e-06,
      "loss": 0.8211,
      "step": 30180
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.906991550785131,
      "learning_rate": 7.099934814153625e-06,
      "loss": 0.7989,
      "step": 30185
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.15562582352719,
      "learning_rate": 7.098817464665593e-06,
      "loss": 0.831,
      "step": 30190
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.22342610960438,
      "learning_rate": 7.09769998793244e-06,
      "loss": 0.8611,
      "step": 30195
    },
    {
      "epoch": 1.28,
      "grad_norm": 11.977201015932861,
      "learning_rate": 7.096582384021916e-06,
      "loss": 0.8598,
      "step": 30200
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.802049910883823,
      "learning_rate": 7.095464653001778e-06,
      "loss": 0.8494,
      "step": 30205
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.779810188745332,
      "learning_rate": 7.094346794939791e-06,
      "loss": 0.8334,
      "step": 30210
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.81770732352753,
      "learning_rate": 7.0932288099037265e-06,
      "loss": 0.816,
      "step": 30215
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.353533725321616,
      "learning_rate": 7.092110697961364e-06,
      "loss": 0.8193,
      "step": 30220
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.981228234484451,
      "learning_rate": 7.090992459180494e-06,
      "loss": 0.7971,
      "step": 30225
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.2959011153503495,
      "learning_rate": 7.089874093628909e-06,
      "loss": 0.7858,
      "step": 30230
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.176177150324127,
      "learning_rate": 7.088755601374414e-06,
      "loss": 0.8076,
      "step": 30235
    },
    {
      "epoch": 1.28,
      "grad_norm": 6.945037934640414,
      "learning_rate": 7.08763698248482e-06,
      "loss": 0.859,
      "step": 30240
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.7905779675592735,
      "learning_rate": 7.086518237027945e-06,
      "loss": 0.8367,
      "step": 30245
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.280954925373214,
      "learning_rate": 7.085399365071616e-06,
      "loss": 0.8146,
      "step": 30250
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.115545630346348,
      "learning_rate": 7.084280366683666e-06,
      "loss": 0.8332,
      "step": 30255
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.4397933176183795,
      "learning_rate": 7.083161241931936e-06,
      "loss": 0.8288,
      "step": 30260
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.447509744142868,
      "learning_rate": 7.082041990884276e-06,
      "loss": 0.8501,
      "step": 30265
    },
    {
      "epoch": 1.28,
      "grad_norm": 8.012698567214477,
      "learning_rate": 7.080922613608544e-06,
      "loss": 0.8345,
      "step": 30270
    },
    {
      "epoch": 1.28,
      "grad_norm": 15.480053757174062,
      "learning_rate": 7.0798031101726025e-06,
      "loss": 0.8226,
      "step": 30275
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.077484796241971,
      "learning_rate": 7.078683480644325e-06,
      "loss": 0.8806,
      "step": 30280
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.430139291054002,
      "learning_rate": 7.077563725091592e-06,
      "loss": 0.8396,
      "step": 30285
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.585228979579798,
      "learning_rate": 7.076443843582288e-06,
      "loss": 0.8394,
      "step": 30290
    },
    {
      "epoch": 1.28,
      "grad_norm": 9.457439343561308,
      "learning_rate": 7.075323836184311e-06,
      "loss": 0.8422,
      "step": 30295
    },
    {
      "epoch": 1.28,
      "grad_norm": 12.64569517508503,
      "learning_rate": 7.074203702965563e-06,
      "loss": 0.815,
      "step": 30300
    },
    {
      "epoch": 1.28,
      "grad_norm": 12.97579791063769,
      "learning_rate": 7.0730834439939545e-06,
      "loss": 0.8217,
      "step": 30305
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.42422861188891,
      "learning_rate": 7.071963059337403e-06,
      "loss": 0.7951,
      "step": 30310
    },
    {
      "epoch": 1.28,
      "grad_norm": 5.41837395054203,
      "learning_rate": 7.070842549063834e-06,
      "loss": 0.826,
      "step": 30315
    },
    {
      "epoch": 1.28,
      "grad_norm": 16.685378384653003,
      "learning_rate": 7.069721913241182e-06,
      "loss": 0.8537,
      "step": 30320
    },
    {
      "epoch": 1.28,
      "grad_norm": 32.95023878959854,
      "learning_rate": 7.068601151937387e-06,
      "loss": 0.8177,
      "step": 30325
    },
    {
      "epoch": 1.28,
      "grad_norm": 17.483250125378625,
      "learning_rate": 7.067480265220397e-06,
      "loss": 0.8383,
      "step": 30330
    },
    {
      "epoch": 1.28,
      "grad_norm": 14.165176240829023,
      "learning_rate": 7.066359253158169e-06,
      "loss": 0.8406,
      "step": 30335
    },
    {
      "epoch": 1.28,
      "grad_norm": 13.065664946436552,
      "learning_rate": 7.065238115818666e-06,
      "loss": 0.8286,
      "step": 30340
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.915332267403049,
      "learning_rate": 7.064116853269859e-06,
      "loss": 0.8271,
      "step": 30345
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.84254043980535,
      "learning_rate": 7.062995465579727e-06,
      "loss": 0.8456,
      "step": 30350
    },
    {
      "epoch": 1.28,
      "grad_norm": 19.634857644328868,
      "learning_rate": 7.0618739528162565e-06,
      "loss": 0.8412,
      "step": 30355
    },
    {
      "epoch": 1.28,
      "grad_norm": 10.70987988091038,
      "learning_rate": 7.0607523150474425e-06,
      "loss": 0.8314,
      "step": 30360
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.718628550490718,
      "learning_rate": 7.059630552341285e-06,
      "loss": 0.8076,
      "step": 30365
    },
    {
      "epoch": 1.29,
      "grad_norm": 11.204307126376989,
      "learning_rate": 7.058508664765794e-06,
      "loss": 0.8174,
      "step": 30370
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.41185352794296,
      "learning_rate": 7.057386652388987e-06,
      "loss": 0.8262,
      "step": 30375
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.2596983684043295,
      "learning_rate": 7.056264515278886e-06,
      "loss": 0.8243,
      "step": 30380
    },
    {
      "epoch": 1.29,
      "grad_norm": 22.85214537775442,
      "learning_rate": 7.055142253503527e-06,
      "loss": 0.8543,
      "step": 30385
    },
    {
      "epoch": 1.29,
      "grad_norm": 24.729839385395405,
      "learning_rate": 7.0540198671309436e-06,
      "loss": 0.8428,
      "step": 30390
    },
    {
      "epoch": 1.29,
      "grad_norm": 24.50980318297952,
      "learning_rate": 7.052897356229188e-06,
      "loss": 0.8386,
      "step": 30395
    },
    {
      "epoch": 1.29,
      "grad_norm": 14.45767582806743,
      "learning_rate": 7.051774720866314e-06,
      "loss": 0.8429,
      "step": 30400
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.898463530518916,
      "learning_rate": 7.05065196111038e-06,
      "loss": 0.8154,
      "step": 30405
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.331203312707453,
      "learning_rate": 7.049529077029458e-06,
      "loss": 0.8403,
      "step": 30410
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.660713071629328,
      "learning_rate": 7.048406068691627e-06,
      "loss": 0.8236,
      "step": 30415
    },
    {
      "epoch": 1.29,
      "grad_norm": 11.429330143684927,
      "learning_rate": 7.0472829361649675e-06,
      "loss": 0.7983,
      "step": 30420
    },
    {
      "epoch": 1.29,
      "grad_norm": 25.04186786502489,
      "learning_rate": 7.046159679517575e-06,
      "loss": 0.7984,
      "step": 30425
    },
    {
      "epoch": 1.29,
      "grad_norm": 12.136675449494412,
      "learning_rate": 7.045036298817549e-06,
      "loss": 0.8029,
      "step": 30430
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.714908734615776,
      "learning_rate": 7.043912794132995e-06,
      "loss": 0.8133,
      "step": 30435
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.108933449934547,
      "learning_rate": 7.042789165532029e-06,
      "loss": 0.7943,
      "step": 30440
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.69024776413526,
      "learning_rate": 7.041665413082774e-06,
      "loss": 0.796,
      "step": 30445
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.036148133406255,
      "learning_rate": 7.040541536853358e-06,
      "loss": 0.8585,
      "step": 30450
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.148474456546869,
      "learning_rate": 7.03941753691192e-06,
      "loss": 0.838,
      "step": 30455
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.940179552476616,
      "learning_rate": 7.038293413326605e-06,
      "loss": 0.8395,
      "step": 30460
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.512693903460496,
      "learning_rate": 7.037169166165561e-06,
      "loss": 0.8223,
      "step": 30465
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.41055574080407,
      "learning_rate": 7.036044795496953e-06,
      "loss": 0.8046,
      "step": 30470
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.567798351990528,
      "learning_rate": 7.0349203013889465e-06,
      "loss": 0.8236,
      "step": 30475
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.53895454166985,
      "learning_rate": 7.033795683909717e-06,
      "loss": 0.8042,
      "step": 30480
    },
    {
      "epoch": 1.29,
      "grad_norm": 11.051791032307255,
      "learning_rate": 7.032670943127445e-06,
      "loss": 0.8443,
      "step": 30485
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.9302784118252045,
      "learning_rate": 7.0315460791103204e-06,
      "loss": 0.8411,
      "step": 30490
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.642594464526272,
      "learning_rate": 7.030421091926545e-06,
      "loss": 0.8634,
      "step": 30495
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.024134146354589,
      "learning_rate": 7.029295981644315e-06,
      "loss": 0.851,
      "step": 30500
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.4433286682859245,
      "learning_rate": 7.02817074833185e-06,
      "loss": 0.8113,
      "step": 30505
    },
    {
      "epoch": 1.29,
      "grad_norm": 4.610507527751198,
      "learning_rate": 7.027045392057365e-06,
      "loss": 0.8569,
      "step": 30510
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.9668189217640215,
      "learning_rate": 7.025919912889088e-06,
      "loss": 0.8427,
      "step": 30515
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.236809348129294,
      "learning_rate": 7.024794310895256e-06,
      "loss": 0.8128,
      "step": 30520
    },
    {
      "epoch": 1.29,
      "grad_norm": 7.030674911060837,
      "learning_rate": 7.023668586144108e-06,
      "loss": 0.819,
      "step": 30525
    },
    {
      "epoch": 1.29,
      "grad_norm": 5.317666513898798,
      "learning_rate": 7.022542738703896e-06,
      "loss": 0.8243,
      "step": 30530
    },
    {
      "epoch": 1.29,
      "grad_norm": 24.203473246081643,
      "learning_rate": 7.0214167686428735e-06,
      "loss": 0.8614,
      "step": 30535
    },
    {
      "epoch": 1.29,
      "grad_norm": 24.28796578743221,
      "learning_rate": 7.020290676029309e-06,
      "loss": 0.8386,
      "step": 30540
    },
    {
      "epoch": 1.29,
      "grad_norm": 27.337473993683528,
      "learning_rate": 7.01916446093147e-06,
      "loss": 0.8346,
      "step": 30545
    },
    {
      "epoch": 1.29,
      "grad_norm": 12.401974137982343,
      "learning_rate": 7.018038123417636e-06,
      "loss": 0.8349,
      "step": 30550
    },
    {
      "epoch": 1.29,
      "grad_norm": 13.33358126598494,
      "learning_rate": 7.0169116635560985e-06,
      "loss": 0.868,
      "step": 30555
    },
    {
      "epoch": 1.29,
      "grad_norm": 12.388096772122134,
      "learning_rate": 7.0157850814151455e-06,
      "loss": 0.7887,
      "step": 30560
    },
    {
      "epoch": 1.29,
      "grad_norm": 14.810305112478773,
      "learning_rate": 7.01465837706308e-06,
      "loss": 0.787,
      "step": 30565
    },
    {
      "epoch": 1.29,
      "grad_norm": 24.886070183737026,
      "learning_rate": 7.013531550568214e-06,
      "loss": 0.809,
      "step": 30570
    },
    {
      "epoch": 1.29,
      "grad_norm": 16.114102943682518,
      "learning_rate": 7.01240460199886e-06,
      "loss": 0.8226,
      "step": 30575
    },
    {
      "epoch": 1.29,
      "grad_norm": 8.980425962840513,
      "learning_rate": 7.011277531423343e-06,
      "loss": 0.8512,
      "step": 30580
    },
    {
      "epoch": 1.29,
      "grad_norm": 11.86037313496318,
      "learning_rate": 7.010150338909992e-06,
      "loss": 0.7819,
      "step": 30585
    },
    {
      "epoch": 1.29,
      "grad_norm": 19.42306686722444,
      "learning_rate": 7.009023024527148e-06,
      "loss": 0.8218,
      "step": 30590
    },
    {
      "epoch": 1.29,
      "grad_norm": 6.555130073169696,
      "learning_rate": 7.007895588343157e-06,
      "loss": 0.8374,
      "step": 30595
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.336590353901856,
      "learning_rate": 7.00676803042637e-06,
      "loss": 0.8041,
      "step": 30600
    },
    {
      "epoch": 1.3,
      "grad_norm": 9.825797956082603,
      "learning_rate": 7.005640350845149e-06,
      "loss": 0.8263,
      "step": 30605
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.837724925885693,
      "learning_rate": 7.004512549667859e-06,
      "loss": 0.8452,
      "step": 30610
    },
    {
      "epoch": 1.3,
      "grad_norm": 9.104028869198386,
      "learning_rate": 7.0033846269628804e-06,
      "loss": 0.7596,
      "step": 30615
    },
    {
      "epoch": 1.3,
      "grad_norm": 10.70957553902516,
      "learning_rate": 7.002256582798592e-06,
      "loss": 0.8344,
      "step": 30620
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.675970869828424,
      "learning_rate": 7.001128417243384e-06,
      "loss": 0.8259,
      "step": 30625
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.142709495931403,
      "learning_rate": 7.000000130365656e-06,
      "loss": 0.8531,
      "step": 30630
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.312334922014445,
      "learning_rate": 6.99887172223381e-06,
      "loss": 0.8093,
      "step": 30635
    },
    {
      "epoch": 1.3,
      "grad_norm": 12.282191181222677,
      "learning_rate": 6.99774319291626e-06,
      "loss": 0.8398,
      "step": 30640
    },
    {
      "epoch": 1.3,
      "grad_norm": 16.754288082981116,
      "learning_rate": 6.996614542481426e-06,
      "loss": 0.8118,
      "step": 30645
    },
    {
      "epoch": 1.3,
      "grad_norm": 14.447771665892928,
      "learning_rate": 6.995485770997731e-06,
      "loss": 0.762,
      "step": 30650
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.355712766535142,
      "learning_rate": 6.994356878533612e-06,
      "loss": 0.8203,
      "step": 30655
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.837225438225289,
      "learning_rate": 6.9932278651575125e-06,
      "loss": 0.8345,
      "step": 30660
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.001692547152853,
      "learning_rate": 6.992098730937877e-06,
      "loss": 0.8173,
      "step": 30665
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.487802695080293,
      "learning_rate": 6.9909694759431635e-06,
      "loss": 0.8039,
      "step": 30670
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.525908000183807,
      "learning_rate": 6.989840100241836e-06,
      "loss": 0.8165,
      "step": 30675
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.635586668198554,
      "learning_rate": 6.988710603902363e-06,
      "loss": 0.8122,
      "step": 30680
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.8631152551421435,
      "learning_rate": 6.987580986993225e-06,
      "loss": 0.7868,
      "step": 30685
    },
    {
      "epoch": 1.3,
      "grad_norm": 16.55916805902345,
      "learning_rate": 6.9864512495829065e-06,
      "loss": 0.8027,
      "step": 30690
    },
    {
      "epoch": 1.3,
      "grad_norm": 10.958229928836165,
      "learning_rate": 6.9853213917398986e-06,
      "loss": 0.8197,
      "step": 30695
    },
    {
      "epoch": 1.3,
      "grad_norm": 13.783998417869906,
      "learning_rate": 6.984191413532704e-06,
      "loss": 0.8273,
      "step": 30700
    },
    {
      "epoch": 1.3,
      "grad_norm": 12.669461103046313,
      "learning_rate": 6.983061315029829e-06,
      "loss": 0.8137,
      "step": 30705
    },
    {
      "epoch": 1.3,
      "grad_norm": 6.760957022273189,
      "learning_rate": 6.981931096299786e-06,
      "loss": 0.8425,
      "step": 30710
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.963954319178731,
      "learning_rate": 6.980800757411102e-06,
      "loss": 0.7891,
      "step": 30715
    },
    {
      "epoch": 1.3,
      "grad_norm": 9.355479991868174,
      "learning_rate": 6.979670298432298e-06,
      "loss": 0.7645,
      "step": 30720
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.362935104138875,
      "learning_rate": 6.978539719431918e-06,
      "loss": 0.8449,
      "step": 30725
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.522756340793397,
      "learning_rate": 6.977409020478503e-06,
      "loss": 0.7937,
      "step": 30730
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.816311483771239,
      "learning_rate": 6.976278201640603e-06,
      "loss": 0.7903,
      "step": 30735
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.903836933197944,
      "learning_rate": 6.975147262986779e-06,
      "loss": 0.8355,
      "step": 30740
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.539756801411209,
      "learning_rate": 6.9740162045855915e-06,
      "loss": 0.8207,
      "step": 30745
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.026375055223772,
      "learning_rate": 6.972885026505617e-06,
      "loss": 0.7699,
      "step": 30750
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.996235742086593,
      "learning_rate": 6.971753728815435e-06,
      "loss": 0.8156,
      "step": 30755
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.754319185533612,
      "learning_rate": 6.970622311583633e-06,
      "loss": 0.8144,
      "step": 30760
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.998246538356853,
      "learning_rate": 6.969490774878804e-06,
      "loss": 0.7746,
      "step": 30765
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.404548546581281,
      "learning_rate": 6.96835911876955e-06,
      "loss": 0.8437,
      "step": 30770
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.122872351642947,
      "learning_rate": 6.967227343324482e-06,
      "loss": 0.8031,
      "step": 30775
    },
    {
      "epoch": 1.3,
      "grad_norm": 9.33870983485933,
      "learning_rate": 6.966095448612213e-06,
      "loss": 0.8507,
      "step": 30780
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.361076621921054,
      "learning_rate": 6.964963434701371e-06,
      "loss": 0.8085,
      "step": 30785
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.70317649130165,
      "learning_rate": 6.963831301660582e-06,
      "loss": 0.8134,
      "step": 30790
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.211607917963832,
      "learning_rate": 6.962699049558486e-06,
      "loss": 0.8037,
      "step": 30795
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.104388599288947,
      "learning_rate": 6.9615666784637285e-06,
      "loss": 0.8672,
      "step": 30800
    },
    {
      "epoch": 1.3,
      "grad_norm": 10.18235691078436,
      "learning_rate": 6.96043418844496e-06,
      "loss": 0.789,
      "step": 30805
    },
    {
      "epoch": 1.3,
      "grad_norm": 7.123682910712552,
      "learning_rate": 6.959301579570843e-06,
      "loss": 0.7993,
      "step": 30810
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.14090401788208,
      "learning_rate": 6.958168851910042e-06,
      "loss": 0.8333,
      "step": 30815
    },
    {
      "epoch": 1.3,
      "grad_norm": 5.895621629562826,
      "learning_rate": 6.9570360055312325e-06,
      "loss": 0.8915,
      "step": 30820
    },
    {
      "epoch": 1.3,
      "grad_norm": 8.662599902475666,
      "learning_rate": 6.955903040503093e-06,
      "loss": 0.8059,
      "step": 30825
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.45118165268154,
      "learning_rate": 6.954769956894314e-06,
      "loss": 0.8395,
      "step": 30830
    },
    {
      "epoch": 1.3,
      "grad_norm": 10.223819807189278,
      "learning_rate": 6.953636754773591e-06,
      "loss": 0.8309,
      "step": 30835
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.2764424779869055,
      "learning_rate": 6.952503434209624e-06,
      "loss": 0.8022,
      "step": 30840
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.830396664182078,
      "learning_rate": 6.951369995271127e-06,
      "loss": 0.8246,
      "step": 30845
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.56976739805219,
      "learning_rate": 6.9502364380268164e-06,
      "loss": 0.8331,
      "step": 30850
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.742287881718884,
      "learning_rate": 6.949102762545412e-06,
      "loss": 0.8292,
      "step": 30855
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.679770503302572,
      "learning_rate": 6.947968968895651e-06,
      "loss": 0.793,
      "step": 30860
    },
    {
      "epoch": 1.31,
      "grad_norm": 16.797062556412232,
      "learning_rate": 6.946835057146267e-06,
      "loss": 0.7925,
      "step": 30865
    },
    {
      "epoch": 1.31,
      "grad_norm": 12.40722891215225,
      "learning_rate": 6.9457010273660096e-06,
      "loss": 0.8365,
      "step": 30870
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.979602383123886,
      "learning_rate": 6.944566879623631e-06,
      "loss": 0.8228,
      "step": 30875
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.58948389218297,
      "learning_rate": 6.9434326139878885e-06,
      "loss": 0.8421,
      "step": 30880
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.845698518834456,
      "learning_rate": 6.942298230527552e-06,
      "loss": 0.8289,
      "step": 30885
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.303449986674132,
      "learning_rate": 6.941163729311395e-06,
      "loss": 0.8862,
      "step": 30890
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.038966819927321,
      "learning_rate": 6.9400291104082e-06,
      "loss": 0.8405,
      "step": 30895
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.761079627248277,
      "learning_rate": 6.938894373886753e-06,
      "loss": 0.8239,
      "step": 30900
    },
    {
      "epoch": 1.31,
      "grad_norm": 13.479778902798289,
      "learning_rate": 6.937759519815852e-06,
      "loss": 0.8446,
      "step": 30905
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.708365921329065,
      "learning_rate": 6.9366245482643e-06,
      "loss": 0.7883,
      "step": 30910
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.102379765514295,
      "learning_rate": 6.935489459300905e-06,
      "loss": 0.851,
      "step": 30915
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.862474080922375,
      "learning_rate": 6.934354252994485e-06,
      "loss": 0.8156,
      "step": 30920
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.808549772730339,
      "learning_rate": 6.933218929413865e-06,
      "loss": 0.8472,
      "step": 30925
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.891913876879858,
      "learning_rate": 6.932083488627875e-06,
      "loss": 0.792,
      "step": 30930
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.916705479239607,
      "learning_rate": 6.9309479307053565e-06,
      "loss": 0.7972,
      "step": 30935
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.463334166346825,
      "learning_rate": 6.92981225571515e-06,
      "loss": 0.8528,
      "step": 30940
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.324382622070353,
      "learning_rate": 6.928676463726113e-06,
      "loss": 0.7971,
      "step": 30945
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.315107254965493,
      "learning_rate": 6.927540554807102e-06,
      "loss": 0.8278,
      "step": 30950
    },
    {
      "epoch": 1.31,
      "grad_norm": 16.965968164940747,
      "learning_rate": 6.926404529026984e-06,
      "loss": 0.8127,
      "step": 30955
    },
    {
      "epoch": 1.31,
      "grad_norm": 21.841737829591953,
      "learning_rate": 6.9252683864546355e-06,
      "loss": 0.8154,
      "step": 30960
    },
    {
      "epoch": 1.31,
      "grad_norm": 33.836462859227524,
      "learning_rate": 6.924132127158936e-06,
      "loss": 0.8485,
      "step": 30965
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.765209986787647,
      "learning_rate": 6.922995751208772e-06,
      "loss": 0.7866,
      "step": 30970
    },
    {
      "epoch": 1.31,
      "grad_norm": 16.884119925220077,
      "learning_rate": 6.9218592586730405e-06,
      "loss": 0.8084,
      "step": 30975
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.365945345825806,
      "learning_rate": 6.920722649620644e-06,
      "loss": 0.816,
      "step": 30980
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.566912526098902,
      "learning_rate": 6.919585924120491e-06,
      "loss": 0.8345,
      "step": 30985
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.659700538285211,
      "learning_rate": 6.918449082241496e-06,
      "loss": 0.8196,
      "step": 30990
    },
    {
      "epoch": 1.31,
      "grad_norm": 15.584478451457812,
      "learning_rate": 6.917312124052586e-06,
      "loss": 0.8369,
      "step": 30995
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.950440858112083,
      "learning_rate": 6.916175049622687e-06,
      "loss": 0.7735,
      "step": 31000
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.609722652457403,
      "learning_rate": 6.915037859020741e-06,
      "loss": 0.8072,
      "step": 31005
    },
    {
      "epoch": 1.31,
      "grad_norm": 19.020926752614063,
      "learning_rate": 6.913900552315691e-06,
      "loss": 0.8706,
      "step": 31010
    },
    {
      "epoch": 1.31,
      "grad_norm": 11.174738472136832,
      "learning_rate": 6.912763129576486e-06,
      "loss": 0.8151,
      "step": 31015
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.180748671412141,
      "learning_rate": 6.911625590872089e-06,
      "loss": 0.8115,
      "step": 31020
    },
    {
      "epoch": 1.31,
      "grad_norm": 8.211292639061794,
      "learning_rate": 6.910487936271462e-06,
      "loss": 0.794,
      "step": 31025
    },
    {
      "epoch": 1.31,
      "grad_norm": 7.424041501824461,
      "learning_rate": 6.909350165843579e-06,
      "loss": 0.8302,
      "step": 31030
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.7380674197718315,
      "learning_rate": 6.90821227965742e-06,
      "loss": 0.8158,
      "step": 31035
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.434439417845401,
      "learning_rate": 6.907074277781971e-06,
      "loss": 0.8438,
      "step": 31040
    },
    {
      "epoch": 1.31,
      "grad_norm": 6.873490769424722,
      "learning_rate": 6.905936160286226e-06,
      "loss": 0.8051,
      "step": 31045
    },
    {
      "epoch": 1.31,
      "grad_norm": 19.66564185370898,
      "learning_rate": 6.904797927239185e-06,
      "loss": 0.8186,
      "step": 31050
    },
    {
      "epoch": 1.31,
      "grad_norm": 22.23823661026982,
      "learning_rate": 6.903659578709858e-06,
      "loss": 0.8318,
      "step": 31055
    },
    {
      "epoch": 1.31,
      "grad_norm": 19.702697828348985,
      "learning_rate": 6.902521114767257e-06,
      "loss": 0.834,
      "step": 31060
    },
    {
      "epoch": 1.31,
      "grad_norm": 21.443861857640417,
      "learning_rate": 6.901382535480406e-06,
      "loss": 0.858,
      "step": 31065
    },
    {
      "epoch": 1.31,
      "grad_norm": 16.64254718485684,
      "learning_rate": 6.900243840918332e-06,
      "loss": 0.7975,
      "step": 31070
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.359165475770751,
      "learning_rate": 6.89910503115007e-06,
      "loss": 0.8341,
      "step": 31075
    },
    {
      "epoch": 1.32,
      "grad_norm": 17.11748782735752,
      "learning_rate": 6.897966106244667e-06,
      "loss": 0.8206,
      "step": 31080
    },
    {
      "epoch": 1.32,
      "grad_norm": 24.011234453425086,
      "learning_rate": 6.896827066271167e-06,
      "loss": 0.8003,
      "step": 31085
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.080668993815845,
      "learning_rate": 6.89568791129863e-06,
      "loss": 0.8101,
      "step": 31090
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.34183228047882,
      "learning_rate": 6.89454864139612e-06,
      "loss": 0.7856,
      "step": 31095
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.579679750279931,
      "learning_rate": 6.893409256632704e-06,
      "loss": 0.8492,
      "step": 31100
    },
    {
      "epoch": 1.32,
      "grad_norm": 12.514889603919148,
      "learning_rate": 6.892269757077463e-06,
      "loss": 0.8367,
      "step": 31105
    },
    {
      "epoch": 1.32,
      "grad_norm": 13.40912054443203,
      "learning_rate": 6.89113014279948e-06,
      "loss": 0.8517,
      "step": 31110
    },
    {
      "epoch": 1.32,
      "grad_norm": 13.700540264779526,
      "learning_rate": 6.889990413867849e-06,
      "loss": 0.8318,
      "step": 31115
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.46024652359266,
      "learning_rate": 6.888850570351665e-06,
      "loss": 0.833,
      "step": 31120
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.324634788545719,
      "learning_rate": 6.8877106123200346e-06,
      "loss": 0.8094,
      "step": 31125
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.465528378515589,
      "learning_rate": 6.88657053984207e-06,
      "loss": 0.8523,
      "step": 31130
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.445692398449989,
      "learning_rate": 6.88543035298689e-06,
      "loss": 0.7675,
      "step": 31135
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.3969669811682675,
      "learning_rate": 6.884290051823623e-06,
      "loss": 0.7991,
      "step": 31140
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.45256981627089,
      "learning_rate": 6.883149636421401e-06,
      "loss": 0.8223,
      "step": 31145
    },
    {
      "epoch": 1.32,
      "grad_norm": 10.9331332734134,
      "learning_rate": 6.882009106849363e-06,
      "loss": 0.8136,
      "step": 31150
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.228833380550154,
      "learning_rate": 6.8808684631766565e-06,
      "loss": 0.824,
      "step": 31155
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.413588158623257,
      "learning_rate": 6.879727705472434e-06,
      "loss": 0.8265,
      "step": 31160
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.7725263977321015,
      "learning_rate": 6.878586833805859e-06,
      "loss": 0.8347,
      "step": 31165
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.932397816123524,
      "learning_rate": 6.877445848246097e-06,
      "loss": 0.8069,
      "step": 31170
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.010484090671348,
      "learning_rate": 6.876304748862323e-06,
      "loss": 0.8056,
      "step": 31175
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.490760312674396,
      "learning_rate": 6.875163535723721e-06,
      "loss": 0.8202,
      "step": 31180
    },
    {
      "epoch": 1.32,
      "grad_norm": 9.268119507485787,
      "learning_rate": 6.874022208899476e-06,
      "loss": 0.8172,
      "step": 31185
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.638722675772335,
      "learning_rate": 6.872880768458784e-06,
      "loss": 0.8377,
      "step": 31190
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.493596328948122,
      "learning_rate": 6.8717392144708495e-06,
      "loss": 0.8301,
      "step": 31195
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.465295497228103,
      "learning_rate": 6.870597547004877e-06,
      "loss": 0.8664,
      "step": 31200
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.191942900066486,
      "learning_rate": 6.869455766130087e-06,
      "loss": 0.7713,
      "step": 31205
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.217687216698285,
      "learning_rate": 6.8683138719157e-06,
      "loss": 0.7921,
      "step": 31210
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.683055698999031,
      "learning_rate": 6.867171864430947e-06,
      "loss": 0.8269,
      "step": 31215
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.951119368831036,
      "learning_rate": 6.866029743745064e-06,
      "loss": 0.8515,
      "step": 31220
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.333029550032008,
      "learning_rate": 6.864887509927294e-06,
      "loss": 0.7966,
      "step": 31225
    },
    {
      "epoch": 1.32,
      "grad_norm": 9.17231477234489,
      "learning_rate": 6.863745163046888e-06,
      "loss": 0.8462,
      "step": 31230
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.422157637656466,
      "learning_rate": 6.862602703173102e-06,
      "loss": 0.8648,
      "step": 31235
    },
    {
      "epoch": 1.32,
      "grad_norm": 10.05975947037493,
      "learning_rate": 6.8614601303751995e-06,
      "loss": 0.8076,
      "step": 31240
    },
    {
      "epoch": 1.32,
      "grad_norm": 13.68493938951246,
      "learning_rate": 6.8603174447224526e-06,
      "loss": 0.8154,
      "step": 31245
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.209584375849382,
      "learning_rate": 6.85917464628414e-06,
      "loss": 0.8378,
      "step": 31250
    },
    {
      "epoch": 1.32,
      "grad_norm": 10.00331145625349,
      "learning_rate": 6.858031735129544e-06,
      "loss": 0.807,
      "step": 31255
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.899734727129706,
      "learning_rate": 6.856888711327958e-06,
      "loss": 0.7955,
      "step": 31260
    },
    {
      "epoch": 1.32,
      "grad_norm": 5.229021843999381,
      "learning_rate": 6.8557455749486775e-06,
      "loss": 0.811,
      "step": 31265
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.365187739738876,
      "learning_rate": 6.854602326061011e-06,
      "loss": 0.7873,
      "step": 31270
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.667410717831659,
      "learning_rate": 6.853458964734268e-06,
      "loss": 0.7877,
      "step": 31275
    },
    {
      "epoch": 1.32,
      "grad_norm": 7.03533562104333,
      "learning_rate": 6.852315491037767e-06,
      "loss": 0.7636,
      "step": 31280
    },
    {
      "epoch": 1.32,
      "grad_norm": 4.863416256128626,
      "learning_rate": 6.851171905040832e-06,
      "loss": 0.8136,
      "step": 31285
    },
    {
      "epoch": 1.32,
      "grad_norm": 6.154900215702729,
      "learning_rate": 6.8500282068128e-06,
      "loss": 0.8148,
      "step": 31290
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.602310918845237,
      "learning_rate": 6.848884396423005e-06,
      "loss": 0.8085,
      "step": 31295
    },
    {
      "epoch": 1.32,
      "grad_norm": 14.449414666645763,
      "learning_rate": 6.847740473940797e-06,
      "loss": 0.8129,
      "step": 31300
    },
    {
      "epoch": 1.32,
      "grad_norm": 8.251807570340368,
      "learning_rate": 6.846596439435527e-06,
      "loss": 0.8183,
      "step": 31305
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.378056369690087,
      "learning_rate": 6.845452292976553e-06,
      "loss": 0.8724,
      "step": 31310
    },
    {
      "epoch": 1.33,
      "grad_norm": 17.095156356830913,
      "learning_rate": 6.8443080346332435e-06,
      "loss": 0.8339,
      "step": 31315
    },
    {
      "epoch": 1.33,
      "grad_norm": 12.430751516694825,
      "learning_rate": 6.84316366447497e-06,
      "loss": 0.776,
      "step": 31320
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.031436140997562,
      "learning_rate": 6.842019182571112e-06,
      "loss": 0.8029,
      "step": 31325
    },
    {
      "epoch": 1.33,
      "grad_norm": 19.97993466899796,
      "learning_rate": 6.840874588991056e-06,
      "loss": 0.8162,
      "step": 31330
    },
    {
      "epoch": 1.33,
      "grad_norm": 23.295443252191475,
      "learning_rate": 6.8397298838041985e-06,
      "loss": 0.8238,
      "step": 31335
    },
    {
      "epoch": 1.33,
      "grad_norm": 13.639961102082475,
      "learning_rate": 6.838585067079937e-06,
      "loss": 0.8396,
      "step": 31340
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.90193327258198,
      "learning_rate": 6.837440138887677e-06,
      "loss": 0.7631,
      "step": 31345
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.927493434624856,
      "learning_rate": 6.8362950992968344e-06,
      "loss": 0.8159,
      "step": 31350
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.506265271732505,
      "learning_rate": 6.835149948376829e-06,
      "loss": 0.7872,
      "step": 31355
    },
    {
      "epoch": 1.33,
      "grad_norm": 10.966219732554292,
      "learning_rate": 6.834004686197088e-06,
      "loss": 0.8225,
      "step": 31360
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.003876153384063,
      "learning_rate": 6.832859312827045e-06,
      "loss": 0.8427,
      "step": 31365
    },
    {
      "epoch": 1.33,
      "grad_norm": 15.756294919745246,
      "learning_rate": 6.831713828336141e-06,
      "loss": 0.8567,
      "step": 31370
    },
    {
      "epoch": 1.33,
      "grad_norm": 12.702167109653397,
      "learning_rate": 6.830568232793823e-06,
      "loss": 0.832,
      "step": 31375
    },
    {
      "epoch": 1.33,
      "grad_norm": 25.31887942303852,
      "learning_rate": 6.8294225262695445e-06,
      "loss": 0.8627,
      "step": 31380
    },
    {
      "epoch": 1.33,
      "grad_norm": 17.394052905117952,
      "learning_rate": 6.828276708832768e-06,
      "loss": 0.8083,
      "step": 31385
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.172379441602869,
      "learning_rate": 6.82713078055296e-06,
      "loss": 0.7886,
      "step": 31390
    },
    {
      "epoch": 1.33,
      "grad_norm": 14.484732433309771,
      "learning_rate": 6.8259847414995936e-06,
      "loss": 0.8096,
      "step": 31395
    },
    {
      "epoch": 1.33,
      "grad_norm": 18.410166146811918,
      "learning_rate": 6.824838591742152e-06,
      "loss": 0.8486,
      "step": 31400
    },
    {
      "epoch": 1.33,
      "grad_norm": 16.721298881434233,
      "learning_rate": 6.82369233135012e-06,
      "loss": 0.8369,
      "step": 31405
    },
    {
      "epoch": 1.33,
      "grad_norm": 18.900204059770356,
      "learning_rate": 6.822545960392996e-06,
      "loss": 0.827,
      "step": 31410
    },
    {
      "epoch": 1.33,
      "grad_norm": 25.611122503884104,
      "learning_rate": 6.821399478940278e-06,
      "loss": 0.8098,
      "step": 31415
    },
    {
      "epoch": 1.33,
      "grad_norm": 9.506566126164305,
      "learning_rate": 6.820252887061475e-06,
      "loss": 0.8295,
      "step": 31420
    },
    {
      "epoch": 1.33,
      "grad_norm": 6.010432172607462,
      "learning_rate": 6.819106184826101e-06,
      "loss": 0.8348,
      "step": 31425
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.793095339700078,
      "learning_rate": 6.817959372303678e-06,
      "loss": 0.7756,
      "step": 31430
    },
    {
      "epoch": 1.33,
      "grad_norm": 13.78281975395877,
      "learning_rate": 6.816812449563732e-06,
      "loss": 0.797,
      "step": 31435
    },
    {
      "epoch": 1.33,
      "grad_norm": 11.689210457184577,
      "learning_rate": 6.815665416675798e-06,
      "loss": 0.8204,
      "step": 31440
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.664428585515743,
      "learning_rate": 6.814518273709421e-06,
      "loss": 0.788,
      "step": 31445
    },
    {
      "epoch": 1.33,
      "grad_norm": 19.312871792742744,
      "learning_rate": 6.813371020734144e-06,
      "loss": 0.8318,
      "step": 31450
    },
    {
      "epoch": 1.33,
      "grad_norm": 19.445655064882253,
      "learning_rate": 6.812223657819523e-06,
      "loss": 0.7813,
      "step": 31455
    },
    {
      "epoch": 1.33,
      "grad_norm": 10.373334805431014,
      "learning_rate": 6.811076185035122e-06,
      "loss": 0.8369,
      "step": 31460
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.879837187896734,
      "learning_rate": 6.809928602450504e-06,
      "loss": 0.8434,
      "step": 31465
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.196774391129288,
      "learning_rate": 6.808780910135249e-06,
      "loss": 0.8484,
      "step": 31470
    },
    {
      "epoch": 1.33,
      "grad_norm": 7.397063609569595,
      "learning_rate": 6.807633108158932e-06,
      "loss": 0.7571,
      "step": 31475
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.959546346616855,
      "learning_rate": 6.806485196591146e-06,
      "loss": 0.8147,
      "step": 31480
    },
    {
      "epoch": 1.33,
      "grad_norm": 12.641369300705877,
      "learning_rate": 6.805337175501484e-06,
      "loss": 0.8616,
      "step": 31485
    },
    {
      "epoch": 1.33,
      "grad_norm": 4.557826201345145,
      "learning_rate": 6.804189044959546e-06,
      "loss": 0.764,
      "step": 31490
    },
    {
      "epoch": 1.33,
      "grad_norm": 14.186386222750375,
      "learning_rate": 6.8030408050349406e-06,
      "loss": 0.8312,
      "step": 31495
    },
    {
      "epoch": 1.33,
      "grad_norm": 15.337535082676023,
      "learning_rate": 6.801892455797283e-06,
      "loss": 0.7825,
      "step": 31500
    },
    {
      "epoch": 1.33,
      "grad_norm": 8.296646091300543,
      "learning_rate": 6.8007439973161925e-06,
      "loss": 0.774,
      "step": 31505
    },
    {
      "epoch": 1.33,
      "grad_norm": 9.166268796621337,
      "learning_rate": 6.7995954296612984e-06,
      "loss": 0.8601,
      "step": 31510
    },
    {
      "epoch": 1.33,
      "grad_norm": 16.27819911188128,
      "learning_rate": 6.798446752902233e-06,
      "loss": 0.8805,
      "step": 31515
    },
    {
      "epoch": 1.33,
      "grad_norm": 47.625725196408226,
      "learning_rate": 6.7972979671086405e-06,
      "loss": 0.8416,
      "step": 31520
    },
    {
      "epoch": 1.33,
      "grad_norm": 19.767712052270472,
      "learning_rate": 6.796149072350167e-06,
      "loss": 0.8429,
      "step": 31525
    },
    {
      "epoch": 1.33,
      "grad_norm": 20.79116171190734,
      "learning_rate": 6.795000068696463e-06,
      "loss": 0.8206,
      "step": 31530
    },
    {
      "epoch": 1.33,
      "grad_norm": 10.819522049356816,
      "learning_rate": 6.793850956217194e-06,
      "loss": 0.7943,
      "step": 31535
    },
    {
      "epoch": 1.33,
      "grad_norm": 5.518776462759827,
      "learning_rate": 6.7927017349820245e-06,
      "loss": 0.8211,
      "step": 31540
    },
    {
      "epoch": 1.34,
      "grad_norm": 16.422082918751077,
      "learning_rate": 6.791552405060631e-06,
      "loss": 0.7964,
      "step": 31545
    },
    {
      "epoch": 1.34,
      "grad_norm": 11.036342679420907,
      "learning_rate": 6.7904029665226915e-06,
      "loss": 0.8343,
      "step": 31550
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.569793418008702,
      "learning_rate": 6.789253419437893e-06,
      "loss": 0.784,
      "step": 31555
    },
    {
      "epoch": 1.34,
      "grad_norm": 11.588332425899969,
      "learning_rate": 6.788103763875932e-06,
      "loss": 0.778,
      "step": 31560
    },
    {
      "epoch": 1.34,
      "grad_norm": 19.174368261859875,
      "learning_rate": 6.786953999906505e-06,
      "loss": 0.8076,
      "step": 31565
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.4990375598683245,
      "learning_rate": 6.785804127599321e-06,
      "loss": 0.83,
      "step": 31570
    },
    {
      "epoch": 1.34,
      "grad_norm": 24.200845150802245,
      "learning_rate": 6.784654147024093e-06,
      "loss": 0.8255,
      "step": 31575
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.367909310517959,
      "learning_rate": 6.783504058250542e-06,
      "loss": 0.8669,
      "step": 31580
    },
    {
      "epoch": 1.34,
      "grad_norm": 15.12209636471602,
      "learning_rate": 6.782353861348393e-06,
      "loss": 0.7961,
      "step": 31585
    },
    {
      "epoch": 1.34,
      "grad_norm": 10.973117701752242,
      "learning_rate": 6.781203556387379e-06,
      "loss": 0.8136,
      "step": 31590
    },
    {
      "epoch": 1.34,
      "grad_norm": 20.1898445358115,
      "learning_rate": 6.78005314343724e-06,
      "loss": 0.8092,
      "step": 31595
    },
    {
      "epoch": 1.34,
      "grad_norm": 16.864538798387898,
      "learning_rate": 6.778902622567722e-06,
      "loss": 0.8317,
      "step": 31600
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.244042217321878,
      "learning_rate": 6.777751993848579e-06,
      "loss": 0.7993,
      "step": 31605
    },
    {
      "epoch": 1.34,
      "grad_norm": 12.433485844518673,
      "learning_rate": 6.776601257349569e-06,
      "loss": 0.819,
      "step": 31610
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.31247172500217,
      "learning_rate": 6.775450413140456e-06,
      "loss": 0.777,
      "step": 31615
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.339174850572915,
      "learning_rate": 6.774299461291016e-06,
      "loss": 0.8185,
      "step": 31620
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.299699819827479,
      "learning_rate": 6.773148401871026e-06,
      "loss": 0.8345,
      "step": 31625
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.900612523785766,
      "learning_rate": 6.771997234950272e-06,
      "loss": 0.8437,
      "step": 31630
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.039582811154645,
      "learning_rate": 6.770845960598544e-06,
      "loss": 0.8118,
      "step": 31635
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.560754764098867,
      "learning_rate": 6.769694578885643e-06,
      "loss": 0.7962,
      "step": 31640
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.896549377616818,
      "learning_rate": 6.768543089881371e-06,
      "loss": 0.8106,
      "step": 31645
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.601025156551856,
      "learning_rate": 6.767391493655541e-06,
      "loss": 0.8035,
      "step": 31650
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.892486634227984,
      "learning_rate": 6.766239790277971e-06,
      "loss": 0.8494,
      "step": 31655
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.433502059918096,
      "learning_rate": 6.765087979818486e-06,
      "loss": 0.797,
      "step": 31660
    },
    {
      "epoch": 1.34,
      "grad_norm": 11.0254929184585,
      "learning_rate": 6.7639360623469145e-06,
      "loss": 0.8413,
      "step": 31665
    },
    {
      "epoch": 1.34,
      "grad_norm": 11.887678097193918,
      "learning_rate": 6.762784037933097e-06,
      "loss": 0.8211,
      "step": 31670
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.780897689494343,
      "learning_rate": 6.761631906646874e-06,
      "loss": 0.7977,
      "step": 31675
    },
    {
      "epoch": 1.34,
      "grad_norm": 10.65914544739127,
      "learning_rate": 6.760479668558098e-06,
      "loss": 0.8169,
      "step": 31680
    },
    {
      "epoch": 1.34,
      "grad_norm": 18.286237221174733,
      "learning_rate": 6.759327323736626e-06,
      "loss": 0.7648,
      "step": 31685
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.581387026705098,
      "learning_rate": 6.758174872252321e-06,
      "loss": 0.7528,
      "step": 31690
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.822560259599014,
      "learning_rate": 6.757022314175052e-06,
      "loss": 0.777,
      "step": 31695
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.502515827187835,
      "learning_rate": 6.755869649574694e-06,
      "loss": 0.8465,
      "step": 31700
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.654670820696322,
      "learning_rate": 6.754716878521134e-06,
      "loss": 0.7904,
      "step": 31705
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.100792696165554,
      "learning_rate": 6.753564001084256e-06,
      "loss": 0.8189,
      "step": 31710
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.804483887970345,
      "learning_rate": 6.752411017333959e-06,
      "loss": 0.8051,
      "step": 31715
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.485755129905997,
      "learning_rate": 6.751257927340144e-06,
      "loss": 0.801,
      "step": 31720
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.089077139718888,
      "learning_rate": 6.75010473117272e-06,
      "loss": 0.791,
      "step": 31725
    },
    {
      "epoch": 1.34,
      "grad_norm": 6.114034195886247,
      "learning_rate": 6.7489514289016e-06,
      "loss": 0.8125,
      "step": 31730
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.276872543649539,
      "learning_rate": 6.747798020596707e-06,
      "loss": 0.823,
      "step": 31735
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.574592824064354,
      "learning_rate": 6.746644506327969e-06,
      "loss": 0.7833,
      "step": 31740
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.10504667784637,
      "learning_rate": 6.745490886165318e-06,
      "loss": 0.8155,
      "step": 31745
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.894681598238104,
      "learning_rate": 6.744337160178697e-06,
      "loss": 0.8316,
      "step": 31750
    },
    {
      "epoch": 1.34,
      "grad_norm": 4.020143291407074,
      "learning_rate": 6.743183328438052e-06,
      "loss": 0.8154,
      "step": 31755
    },
    {
      "epoch": 1.34,
      "grad_norm": 14.171505264071087,
      "learning_rate": 6.742029391013336e-06,
      "loss": 0.8082,
      "step": 31760
    },
    {
      "epoch": 1.34,
      "grad_norm": 9.594653998683382,
      "learning_rate": 6.7408753479745095e-06,
      "loss": 0.8284,
      "step": 31765
    },
    {
      "epoch": 1.34,
      "grad_norm": 5.287271512358595,
      "learning_rate": 6.7397211993915405e-06,
      "loss": 0.8329,
      "step": 31770
    },
    {
      "epoch": 1.34,
      "grad_norm": 7.8004329655422655,
      "learning_rate": 6.738566945334396e-06,
      "loss": 0.8088,
      "step": 31775
    },
    {
      "epoch": 1.34,
      "grad_norm": 10.190836102541125,
      "learning_rate": 6.737412585873062e-06,
      "loss": 0.8302,
      "step": 31780
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.129920044401575,
      "learning_rate": 6.736258121077518e-06,
      "loss": 0.8238,
      "step": 31785
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.708113228581283,
      "learning_rate": 6.7351035510177606e-06,
      "loss": 0.8417,
      "step": 31790
    },
    {
      "epoch": 1.35,
      "grad_norm": 14.25551287517732,
      "learning_rate": 6.733948875763785e-06,
      "loss": 0.8109,
      "step": 31795
    },
    {
      "epoch": 1.35,
      "grad_norm": 13.275068895355647,
      "learning_rate": 6.732794095385598e-06,
      "loss": 0.7882,
      "step": 31800
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.35647081621935,
      "learning_rate": 6.731639209953207e-06,
      "loss": 0.814,
      "step": 31805
    },
    {
      "epoch": 1.35,
      "grad_norm": 13.709381174935903,
      "learning_rate": 6.7304842195366324e-06,
      "loss": 0.8244,
      "step": 31810
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.114871848618827,
      "learning_rate": 6.729329124205897e-06,
      "loss": 0.8389,
      "step": 31815
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.939740472556877,
      "learning_rate": 6.7281739240310294e-06,
      "loss": 0.8285,
      "step": 31820
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.033439218103713,
      "learning_rate": 6.727018619082068e-06,
      "loss": 0.8475,
      "step": 31825
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.808520457385149,
      "learning_rate": 6.725863209429055e-06,
      "loss": 0.8773,
      "step": 31830
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.3878602119783565,
      "learning_rate": 6.72470769514204e-06,
      "loss": 0.7937,
      "step": 31835
    },
    {
      "epoch": 1.35,
      "grad_norm": 11.274461854370793,
      "learning_rate": 6.7235520762910775e-06,
      "loss": 0.8429,
      "step": 31840
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.755765416077909,
      "learning_rate": 6.722396352946228e-06,
      "loss": 0.8248,
      "step": 31845
    },
    {
      "epoch": 1.35,
      "grad_norm": 9.860363697027857,
      "learning_rate": 6.721240525177562e-06,
      "loss": 0.8068,
      "step": 31850
    },
    {
      "epoch": 1.35,
      "grad_norm": 18.114145051573207,
      "learning_rate": 6.720084593055154e-06,
      "loss": 0.8431,
      "step": 31855
    },
    {
      "epoch": 1.35,
      "grad_norm": 17.16302310924887,
      "learning_rate": 6.718928556649082e-06,
      "loss": 0.8022,
      "step": 31860
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.767305281369644,
      "learning_rate": 6.717772416029436e-06,
      "loss": 0.7999,
      "step": 31865
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.741653233161158,
      "learning_rate": 6.716616171266308e-06,
      "loss": 0.8258,
      "step": 31870
    },
    {
      "epoch": 1.35,
      "grad_norm": 10.462555329349241,
      "learning_rate": 6.715459822429799e-06,
      "loss": 0.8056,
      "step": 31875
    },
    {
      "epoch": 1.35,
      "grad_norm": 9.701989509259906,
      "learning_rate": 6.714303369590014e-06,
      "loss": 0.8043,
      "step": 31880
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.567365458568052,
      "learning_rate": 6.713146812817064e-06,
      "loss": 0.7893,
      "step": 31885
    },
    {
      "epoch": 1.35,
      "grad_norm": 19.794658080432235,
      "learning_rate": 6.71199015218107e-06,
      "loss": 0.8277,
      "step": 31890
    },
    {
      "epoch": 1.35,
      "grad_norm": 9.685438673188287,
      "learning_rate": 6.710833387752156e-06,
      "loss": 0.779,
      "step": 31895
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.298095262023212,
      "learning_rate": 6.709676519600452e-06,
      "loss": 0.8256,
      "step": 31900
    },
    {
      "epoch": 1.35,
      "grad_norm": 12.48602344525622,
      "learning_rate": 6.7085195477960976e-06,
      "loss": 0.8383,
      "step": 31905
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.767572478989148,
      "learning_rate": 6.707362472409234e-06,
      "loss": 0.7929,
      "step": 31910
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.309513982109161,
      "learning_rate": 6.706205293510015e-06,
      "loss": 0.8418,
      "step": 31915
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.306730080758769,
      "learning_rate": 6.705048011168593e-06,
      "loss": 0.8337,
      "step": 31920
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.682741573736404,
      "learning_rate": 6.703890625455134e-06,
      "loss": 0.8474,
      "step": 31925
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.602086176561714,
      "learning_rate": 6.702733136439805e-06,
      "loss": 0.818,
      "step": 31930
    },
    {
      "epoch": 1.35,
      "grad_norm": 9.483415185946203,
      "learning_rate": 6.701575544192781e-06,
      "loss": 0.8338,
      "step": 31935
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.170168058882046,
      "learning_rate": 6.7004178487842456e-06,
      "loss": 0.7898,
      "step": 31940
    },
    {
      "epoch": 1.35,
      "grad_norm": 12.91649822469014,
      "learning_rate": 6.699260050284382e-06,
      "loss": 0.8455,
      "step": 31945
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.6715116268511405,
      "learning_rate": 6.698102148763389e-06,
      "loss": 0.8018,
      "step": 31950
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.2767965603169396,
      "learning_rate": 6.696944144291464e-06,
      "loss": 0.8078,
      "step": 31955
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.6326577606380654,
      "learning_rate": 6.695786036938814e-06,
      "loss": 0.7884,
      "step": 31960
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.289676632047995,
      "learning_rate": 6.694627826775651e-06,
      "loss": 0.7961,
      "step": 31965
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.1555033170089555,
      "learning_rate": 6.693469513872196e-06,
      "loss": 0.8206,
      "step": 31970
    },
    {
      "epoch": 1.35,
      "grad_norm": 7.183753112534834,
      "learning_rate": 6.692311098298672e-06,
      "loss": 0.7579,
      "step": 31975
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.018166063272219,
      "learning_rate": 6.691152580125311e-06,
      "loss": 0.8111,
      "step": 31980
    },
    {
      "epoch": 1.35,
      "grad_norm": 5.173746213642771,
      "learning_rate": 6.689993959422349e-06,
      "loss": 0.7949,
      "step": 31985
    },
    {
      "epoch": 1.35,
      "grad_norm": 8.853147095611986,
      "learning_rate": 6.688835236260033e-06,
      "loss": 0.7895,
      "step": 31990
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.250482274041793,
      "learning_rate": 6.687676410708611e-06,
      "loss": 0.7942,
      "step": 31995
    },
    {
      "epoch": 1.35,
      "grad_norm": 13.971344879746761,
      "learning_rate": 6.68651748283834e-06,
      "loss": 0.8175,
      "step": 32000
    },
    {
      "epoch": 1.35,
      "grad_norm": 6.977647652881739,
      "learning_rate": 6.68535845271948e-06,
      "loss": 0.8118,
      "step": 32005
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.235039549092665,
      "learning_rate": 6.684199320422303e-06,
      "loss": 0.8022,
      "step": 32010
    },
    {
      "epoch": 1.35,
      "grad_norm": 4.290900853102789,
      "learning_rate": 6.6830400860170806e-06,
      "loss": 0.7892,
      "step": 32015
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.680708863022627,
      "learning_rate": 6.681880749574098e-06,
      "loss": 0.795,
      "step": 32020
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.290484996625954,
      "learning_rate": 6.6807213111636385e-06,
      "loss": 0.7846,
      "step": 32025
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.453024347844048,
      "learning_rate": 6.679561770855996e-06,
      "loss": 0.7906,
      "step": 32030
    },
    {
      "epoch": 1.36,
      "grad_norm": 16.617071394365883,
      "learning_rate": 6.678402128721472e-06,
      "loss": 0.8228,
      "step": 32035
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.750826967654463,
      "learning_rate": 6.677242384830369e-06,
      "loss": 0.83,
      "step": 32040
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.627319244546042,
      "learning_rate": 6.676082539253002e-06,
      "loss": 0.7833,
      "step": 32045
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.238031146101363,
      "learning_rate": 6.674922592059686e-06,
      "loss": 0.8148,
      "step": 32050
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.823354439990842,
      "learning_rate": 6.673762543320749e-06,
      "loss": 0.8188,
      "step": 32055
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.879124291301238,
      "learning_rate": 6.672602393106518e-06,
      "loss": 0.7962,
      "step": 32060
    },
    {
      "epoch": 1.36,
      "grad_norm": 10.836026514565734,
      "learning_rate": 6.67144214148733e-06,
      "loss": 0.8163,
      "step": 32065
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.396220305766949,
      "learning_rate": 6.67028178853353e-06,
      "loss": 0.8057,
      "step": 32070
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.4991739663200665,
      "learning_rate": 6.669121334315465e-06,
      "loss": 0.7804,
      "step": 32075
    },
    {
      "epoch": 1.36,
      "grad_norm": 10.449682550297233,
      "learning_rate": 6.667960778903489e-06,
      "loss": 0.7993,
      "step": 32080
    },
    {
      "epoch": 1.36,
      "grad_norm": 12.056767609423206,
      "learning_rate": 6.666800122367964e-06,
      "loss": 0.7833,
      "step": 32085
    },
    {
      "epoch": 1.36,
      "grad_norm": 11.56840000003019,
      "learning_rate": 6.665639364779259e-06,
      "loss": 0.8524,
      "step": 32090
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.607816670368353,
      "learning_rate": 6.664478506207743e-06,
      "loss": 0.7912,
      "step": 32095
    },
    {
      "epoch": 1.36,
      "grad_norm": 17.68101572851812,
      "learning_rate": 6.663317546723801e-06,
      "loss": 0.8102,
      "step": 32100
    },
    {
      "epoch": 1.36,
      "grad_norm": 11.92862765264717,
      "learning_rate": 6.6621564863978135e-06,
      "loss": 0.7908,
      "step": 32105
    },
    {
      "epoch": 1.36,
      "grad_norm": 21.238242795003643,
      "learning_rate": 6.660995325300175e-06,
      "loss": 0.7893,
      "step": 32110
    },
    {
      "epoch": 1.36,
      "grad_norm": 10.322622829820148,
      "learning_rate": 6.659834063501282e-06,
      "loss": 0.8,
      "step": 32115
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.277311317931259,
      "learning_rate": 6.6586727010715415e-06,
      "loss": 0.8447,
      "step": 32120
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.693527108013454,
      "learning_rate": 6.657511238081358e-06,
      "loss": 0.793,
      "step": 32125
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.5672305348038345,
      "learning_rate": 6.656349674601152e-06,
      "loss": 0.7956,
      "step": 32130
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.402174255570077,
      "learning_rate": 6.655188010701343e-06,
      "loss": 0.8127,
      "step": 32135
    },
    {
      "epoch": 1.36,
      "grad_norm": 12.081318049278226,
      "learning_rate": 6.654026246452361e-06,
      "loss": 0.7536,
      "step": 32140
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.81285903572821,
      "learning_rate": 6.6528643819246395e-06,
      "loss": 0.7832,
      "step": 32145
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.08482005069644,
      "learning_rate": 6.651702417188619e-06,
      "loss": 0.8138,
      "step": 32150
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.846238044570578,
      "learning_rate": 6.650540352314747e-06,
      "loss": 0.7794,
      "step": 32155
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.144626082558325,
      "learning_rate": 6.6493781873734755e-06,
      "loss": 0.8291,
      "step": 32160
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.047689191069621,
      "learning_rate": 6.6482159224352615e-06,
      "loss": 0.7648,
      "step": 32165
    },
    {
      "epoch": 1.36,
      "grad_norm": 8.111607108198218,
      "learning_rate": 6.647053557570573e-06,
      "loss": 0.8132,
      "step": 32170
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.472533579474425,
      "learning_rate": 6.645891092849877e-06,
      "loss": 0.7966,
      "step": 32175
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.631407652063096,
      "learning_rate": 6.644728528343653e-06,
      "loss": 0.7741,
      "step": 32180
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.765140595486738,
      "learning_rate": 6.643565864122384e-06,
      "loss": 0.7683,
      "step": 32185
    },
    {
      "epoch": 1.36,
      "grad_norm": 6.887997941692297,
      "learning_rate": 6.642403100256556e-06,
      "loss": 0.7975,
      "step": 32190
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.2645799983892445,
      "learning_rate": 6.6412402368166675e-06,
      "loss": 0.8247,
      "step": 32195
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.461615499203763,
      "learning_rate": 6.640077273873217e-06,
      "loss": 0.8026,
      "step": 32200
    },
    {
      "epoch": 1.36,
      "grad_norm": 7.594891055826621,
      "learning_rate": 6.638914211496712e-06,
      "loss": 0.8077,
      "step": 32205
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.787958265610801,
      "learning_rate": 6.637751049757666e-06,
      "loss": 0.8466,
      "step": 32210
    },
    {
      "epoch": 1.36,
      "grad_norm": 12.930173160126325,
      "learning_rate": 6.6365877887265996e-06,
      "loss": 0.8249,
      "step": 32215
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.560529549866392,
      "learning_rate": 6.635424428474035e-06,
      "loss": 0.8025,
      "step": 32220
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.00697433640159,
      "learning_rate": 6.634260969070505e-06,
      "loss": 0.7966,
      "step": 32225
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.437825235323011,
      "learning_rate": 6.633097410586546e-06,
      "loss": 0.7504,
      "step": 32230
    },
    {
      "epoch": 1.36,
      "grad_norm": 9.830814334802387,
      "learning_rate": 6.631933753092701e-06,
      "loss": 0.7846,
      "step": 32235
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.353841662231677,
      "learning_rate": 6.63076999665952e-06,
      "loss": 0.7476,
      "step": 32240
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.300028442633021,
      "learning_rate": 6.629606141357558e-06,
      "loss": 0.8064,
      "step": 32245
    },
    {
      "epoch": 1.36,
      "grad_norm": 4.673621634256113,
      "learning_rate": 6.628442187257376e-06,
      "loss": 0.8188,
      "step": 32250
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.183244474109752,
      "learning_rate": 6.62727813442954e-06,
      "loss": 0.7883,
      "step": 32255
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.792167352539829,
      "learning_rate": 6.626113982944624e-06,
      "loss": 0.8107,
      "step": 32260
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.570759933467777,
      "learning_rate": 6.624949732873208e-06,
      "loss": 0.7867,
      "step": 32265
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.631323941883451,
      "learning_rate": 6.6237853842858765e-06,
      "loss": 0.8215,
      "step": 32270
    },
    {
      "epoch": 1.37,
      "grad_norm": 9.801800503868858,
      "learning_rate": 6.622620937253219e-06,
      "loss": 0.8718,
      "step": 32275
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.473835377380429,
      "learning_rate": 6.621456391845835e-06,
      "loss": 0.8327,
      "step": 32280
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.480106037183473,
      "learning_rate": 6.620291748134325e-06,
      "loss": 0.7819,
      "step": 32285
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.17451115173463,
      "learning_rate": 6.6191270061893e-06,
      "loss": 0.7802,
      "step": 32290
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.728453902835553,
      "learning_rate": 6.617962166081372e-06,
      "loss": 0.8007,
      "step": 32295
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.4990375782794505,
      "learning_rate": 6.616797227881165e-06,
      "loss": 0.8069,
      "step": 32300
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.714997479769019,
      "learning_rate": 6.615632191659306e-06,
      "loss": 0.8232,
      "step": 32305
    },
    {
      "epoch": 1.37,
      "grad_norm": 18.637504225427705,
      "learning_rate": 6.614467057486423e-06,
      "loss": 0.7999,
      "step": 32310
    },
    {
      "epoch": 1.37,
      "grad_norm": 18.40429884734917,
      "learning_rate": 6.613301825433161e-06,
      "loss": 0.8194,
      "step": 32315
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.099649593287371,
      "learning_rate": 6.612136495570159e-06,
      "loss": 0.7648,
      "step": 32320
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.2940724434682975,
      "learning_rate": 6.610971067968071e-06,
      "loss": 0.7721,
      "step": 32325
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.1509816617884905,
      "learning_rate": 6.609805542697553e-06,
      "loss": 0.755,
      "step": 32330
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.161805000881345,
      "learning_rate": 6.608639919829265e-06,
      "loss": 0.8172,
      "step": 32335
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.367467846373024,
      "learning_rate": 6.607474199433879e-06,
      "loss": 0.7779,
      "step": 32340
    },
    {
      "epoch": 1.37,
      "grad_norm": 12.563513436686147,
      "learning_rate": 6.606308381582067e-06,
      "loss": 0.8556,
      "step": 32345
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.2996953397330815,
      "learning_rate": 6.605142466344509e-06,
      "loss": 0.8194,
      "step": 32350
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.9488242763629735,
      "learning_rate": 6.60397645379189e-06,
      "loss": 0.8183,
      "step": 32355
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.073979705043436,
      "learning_rate": 6.602810343994905e-06,
      "loss": 0.808,
      "step": 32360
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.573140832652359,
      "learning_rate": 6.6016441370242505e-06,
      "loss": 0.7939,
      "step": 32365
    },
    {
      "epoch": 1.37,
      "grad_norm": 13.969276038631799,
      "learning_rate": 6.600477832950628e-06,
      "loss": 0.8084,
      "step": 32370
    },
    {
      "epoch": 1.37,
      "grad_norm": 14.963157480677271,
      "learning_rate": 6.59931143184475e-06,
      "loss": 0.7922,
      "step": 32375
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.198505535026127,
      "learning_rate": 6.59814493377733e-06,
      "loss": 0.8133,
      "step": 32380
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.643582439644814,
      "learning_rate": 6.596978338819091e-06,
      "loss": 0.7805,
      "step": 32385
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.385407765723516,
      "learning_rate": 6.595811647040759e-06,
      "loss": 0.7762,
      "step": 32390
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.5015323830659995,
      "learning_rate": 6.5946448585130676e-06,
      "loss": 0.7523,
      "step": 32395
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.159939330906181,
      "learning_rate": 6.593477973306756e-06,
      "loss": 0.8023,
      "step": 32400
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.11369429154442,
      "learning_rate": 6.5923109914925684e-06,
      "loss": 0.804,
      "step": 32405
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.344438228089828,
      "learning_rate": 6.5911439131412544e-06,
      "loss": 0.8225,
      "step": 32410
    },
    {
      "epoch": 1.37,
      "grad_norm": 7.6978849016135245,
      "learning_rate": 6.589976738323574e-06,
      "loss": 0.7852,
      "step": 32415
    },
    {
      "epoch": 1.37,
      "grad_norm": 17.514977884739444,
      "learning_rate": 6.5888094671102854e-06,
      "loss": 0.8106,
      "step": 32420
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.908595176775958,
      "learning_rate": 6.58764209957216e-06,
      "loss": 0.7702,
      "step": 32425
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.83791187035084,
      "learning_rate": 6.58647463577997e-06,
      "loss": 0.8238,
      "step": 32430
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.778864084982646,
      "learning_rate": 6.585307075804496e-06,
      "loss": 0.786,
      "step": 32435
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.907398784986031,
      "learning_rate": 6.584139419716526e-06,
      "loss": 0.7727,
      "step": 32440
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.66029977546783,
      "learning_rate": 6.582971667586846e-06,
      "loss": 0.8112,
      "step": 32445
    },
    {
      "epoch": 1.37,
      "grad_norm": 9.886202823778657,
      "learning_rate": 6.581803819486258e-06,
      "loss": 0.7989,
      "step": 32450
    },
    {
      "epoch": 1.37,
      "grad_norm": 6.3331709971619885,
      "learning_rate": 6.5806358754855624e-06,
      "loss": 0.8022,
      "step": 32455
    },
    {
      "epoch": 1.37,
      "grad_norm": 9.500697958400549,
      "learning_rate": 6.579467835655571e-06,
      "loss": 0.7605,
      "step": 32460
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.54198043563717,
      "learning_rate": 6.578299700067098e-06,
      "loss": 0.8157,
      "step": 32465
    },
    {
      "epoch": 1.37,
      "grad_norm": 8.31600163649908,
      "learning_rate": 6.5771314687909605e-06,
      "loss": 0.828,
      "step": 32470
    },
    {
      "epoch": 1.37,
      "grad_norm": 10.572914577542127,
      "learning_rate": 6.57596314189799e-06,
      "loss": 0.8027,
      "step": 32475
    },
    {
      "epoch": 1.37,
      "grad_norm": 15.188386766580155,
      "learning_rate": 6.574794719459014e-06,
      "loss": 0.7974,
      "step": 32480
    },
    {
      "epoch": 1.37,
      "grad_norm": 5.662823127094822,
      "learning_rate": 6.573626201544876e-06,
      "loss": 0.8467,
      "step": 32485
    },
    {
      "epoch": 1.38,
      "grad_norm": 14.081916121729945,
      "learning_rate": 6.572457588226415e-06,
      "loss": 0.7898,
      "step": 32490
    },
    {
      "epoch": 1.38,
      "grad_norm": 18.785464812145943,
      "learning_rate": 6.571288879574482e-06,
      "loss": 0.8673,
      "step": 32495
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.755488643781528,
      "learning_rate": 6.570120075659934e-06,
      "loss": 0.7964,
      "step": 32500
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.166809908757483,
      "learning_rate": 6.568951176553628e-06,
      "loss": 0.824,
      "step": 32505
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.307587071819875,
      "learning_rate": 6.567782182326436e-06,
      "loss": 0.7766,
      "step": 32510
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.83883578174952,
      "learning_rate": 6.566613093049228e-06,
      "loss": 0.7994,
      "step": 32515
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.146078237845996,
      "learning_rate": 6.565443908792883e-06,
      "loss": 0.7952,
      "step": 32520
    },
    {
      "epoch": 1.38,
      "grad_norm": 11.816944415709806,
      "learning_rate": 6.564274629628284e-06,
      "loss": 0.8041,
      "step": 32525
    },
    {
      "epoch": 1.38,
      "grad_norm": 15.282933831006238,
      "learning_rate": 6.563105255626323e-06,
      "loss": 0.7669,
      "step": 32530
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.666109573171221,
      "learning_rate": 6.561935786857893e-06,
      "loss": 0.8105,
      "step": 32535
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.6800349433169925,
      "learning_rate": 6.560766223393898e-06,
      "loss": 0.7784,
      "step": 32540
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.496286880645884,
      "learning_rate": 6.559596565305245e-06,
      "loss": 0.8262,
      "step": 32545
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.879429061928658,
      "learning_rate": 6.558426812662846e-06,
      "loss": 0.7936,
      "step": 32550
    },
    {
      "epoch": 1.38,
      "grad_norm": 11.119821893339374,
      "learning_rate": 6.55725696553762e-06,
      "loss": 0.8015,
      "step": 32555
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.9960900871121,
      "learning_rate": 6.556087024000491e-06,
      "loss": 0.7697,
      "step": 32560
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.361444628954888,
      "learning_rate": 6.554916988122388e-06,
      "loss": 0.7796,
      "step": 32565
    },
    {
      "epoch": 1.38,
      "grad_norm": 14.211284584648379,
      "learning_rate": 6.553746857974249e-06,
      "loss": 0.8043,
      "step": 32570
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.360755360936613,
      "learning_rate": 6.552576633627016e-06,
      "loss": 0.825,
      "step": 32575
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.894446346638047,
      "learning_rate": 6.551406315151633e-06,
      "loss": 0.8297,
      "step": 32580
    },
    {
      "epoch": 1.38,
      "grad_norm": 10.420766007658528,
      "learning_rate": 6.550235902619055e-06,
      "loss": 0.7941,
      "step": 32585
    },
    {
      "epoch": 1.38,
      "grad_norm": 16.414845560769777,
      "learning_rate": 6.549065396100241e-06,
      "loss": 0.8052,
      "step": 32590
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.392968661482823,
      "learning_rate": 6.547894795666156e-06,
      "loss": 0.7983,
      "step": 32595
    },
    {
      "epoch": 1.38,
      "grad_norm": 11.40753938346193,
      "learning_rate": 6.546724101387769e-06,
      "loss": 0.8255,
      "step": 32600
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.237436926703271,
      "learning_rate": 6.545553313336055e-06,
      "loss": 0.7818,
      "step": 32605
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.162929333120063,
      "learning_rate": 6.544382431581997e-06,
      "loss": 0.8344,
      "step": 32610
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.813629755858621,
      "learning_rate": 6.54321145619658e-06,
      "loss": 0.826,
      "step": 32615
    },
    {
      "epoch": 1.38,
      "grad_norm": 15.184581253096587,
      "learning_rate": 6.542040387250799e-06,
      "loss": 0.8154,
      "step": 32620
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.568646136670125,
      "learning_rate": 6.54086922481565e-06,
      "loss": 0.8408,
      "step": 32625
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.703307532363448,
      "learning_rate": 6.539697968962141e-06,
      "loss": 0.7795,
      "step": 32630
    },
    {
      "epoch": 1.38,
      "grad_norm": 10.042921483392426,
      "learning_rate": 6.538526619761279e-06,
      "loss": 0.7783,
      "step": 32635
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.396146155657716,
      "learning_rate": 6.5373551772840786e-06,
      "loss": 0.7911,
      "step": 32640
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.770082468006368,
      "learning_rate": 6.536183641601563e-06,
      "loss": 0.7921,
      "step": 32645
    },
    {
      "epoch": 1.38,
      "grad_norm": 10.10295234548853,
      "learning_rate": 6.535012012784758e-06,
      "loss": 0.8245,
      "step": 32650
    },
    {
      "epoch": 1.38,
      "grad_norm": 20.26148071786032,
      "learning_rate": 6.533840290904696e-06,
      "loss": 0.8252,
      "step": 32655
    },
    {
      "epoch": 1.38,
      "grad_norm": 22.434907191319688,
      "learning_rate": 6.532668476032416e-06,
      "loss": 0.8516,
      "step": 32660
    },
    {
      "epoch": 1.38,
      "grad_norm": 25.311980031398974,
      "learning_rate": 6.531496568238959e-06,
      "loss": 0.8098,
      "step": 32665
    },
    {
      "epoch": 1.38,
      "grad_norm": 9.580428320557882,
      "learning_rate": 6.5303245675953784e-06,
      "loss": 0.8003,
      "step": 32670
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.371229164114469,
      "learning_rate": 6.529152474172727e-06,
      "loss": 0.8219,
      "step": 32675
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.770562744930191,
      "learning_rate": 6.527980288042065e-06,
      "loss": 0.8192,
      "step": 32680
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.029725948565768,
      "learning_rate": 6.526808009274459e-06,
      "loss": 0.8362,
      "step": 32685
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.7190400686814185,
      "learning_rate": 6.52563563794098e-06,
      "loss": 0.7996,
      "step": 32690
    },
    {
      "epoch": 1.38,
      "grad_norm": 8.036272784128906,
      "learning_rate": 6.5244631741127075e-06,
      "loss": 0.797,
      "step": 32695
    },
    {
      "epoch": 1.38,
      "grad_norm": 7.8832423450680675,
      "learning_rate": 6.52329061786072e-06,
      "loss": 0.8033,
      "step": 32700
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.469805645486168,
      "learning_rate": 6.522117969256113e-06,
      "loss": 0.7837,
      "step": 32705
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.756403888206813,
      "learning_rate": 6.520945228369976e-06,
      "loss": 0.8233,
      "step": 32710
    },
    {
      "epoch": 1.38,
      "grad_norm": 5.406053134866068,
      "learning_rate": 6.519772395273409e-06,
      "loss": 0.7705,
      "step": 32715
    },
    {
      "epoch": 1.38,
      "grad_norm": 6.10376853462327,
      "learning_rate": 6.518599470037519e-06,
      "loss": 0.7986,
      "step": 32720
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.431267666400869,
      "learning_rate": 6.517426452733416e-06,
      "loss": 0.7989,
      "step": 32725
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.482098569642227,
      "learning_rate": 6.516253343432218e-06,
      "loss": 0.8006,
      "step": 32730
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.557287201702734,
      "learning_rate": 6.5150801422050445e-06,
      "loss": 0.7984,
      "step": 32735
    },
    {
      "epoch": 1.39,
      "grad_norm": 14.528786229154706,
      "learning_rate": 6.513906849123025e-06,
      "loss": 0.8252,
      "step": 32740
    },
    {
      "epoch": 1.39,
      "grad_norm": 18.49454056620756,
      "learning_rate": 6.512733464257293e-06,
      "loss": 0.7925,
      "step": 32745
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.602707025626699,
      "learning_rate": 6.511559987678986e-06,
      "loss": 0.8401,
      "step": 32750
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.19021634647184,
      "learning_rate": 6.5103864194592514e-06,
      "loss": 0.7782,
      "step": 32755
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.666238686300844,
      "learning_rate": 6.509212759669237e-06,
      "loss": 0.8023,
      "step": 32760
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.3126717746543575,
      "learning_rate": 6.508039008380097e-06,
      "loss": 0.8175,
      "step": 32765
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.775910729010544,
      "learning_rate": 6.506865165662996e-06,
      "loss": 0.8035,
      "step": 32770
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.38812005156396,
      "learning_rate": 6.5056912315890965e-06,
      "loss": 0.7921,
      "step": 32775
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.748458645809002,
      "learning_rate": 6.504517206229574e-06,
      "loss": 0.787,
      "step": 32780
    },
    {
      "epoch": 1.39,
      "grad_norm": 22.702776966715145,
      "learning_rate": 6.503343089655603e-06,
      "loss": 0.7975,
      "step": 32785
    },
    {
      "epoch": 1.39,
      "grad_norm": 15.95715238560528,
      "learning_rate": 6.502168881938371e-06,
      "loss": 0.7758,
      "step": 32790
    },
    {
      "epoch": 1.39,
      "grad_norm": 16.985775748563615,
      "learning_rate": 6.500994583149064e-06,
      "loss": 0.8131,
      "step": 32795
    },
    {
      "epoch": 1.39,
      "grad_norm": 9.03461569930784,
      "learning_rate": 6.499820193358876e-06,
      "loss": 0.8242,
      "step": 32800
    },
    {
      "epoch": 1.39,
      "grad_norm": 17.662827897229604,
      "learning_rate": 6.498645712639009e-06,
      "loss": 0.8076,
      "step": 32805
    },
    {
      "epoch": 1.39,
      "grad_norm": 9.560416102776916,
      "learning_rate": 6.497471141060665e-06,
      "loss": 0.7785,
      "step": 32810
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.372175709025138,
      "learning_rate": 6.4962964786950575e-06,
      "loss": 0.7569,
      "step": 32815
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.437434248339504,
      "learning_rate": 6.495121725613402e-06,
      "loss": 0.8259,
      "step": 32820
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.539643541515003,
      "learning_rate": 6.49394688188692e-06,
      "loss": 0.8098,
      "step": 32825
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.9717784472017845,
      "learning_rate": 6.49277194758684e-06,
      "loss": 0.7639,
      "step": 32830
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.383438675293918,
      "learning_rate": 6.491596922784393e-06,
      "loss": 0.8322,
      "step": 32835
    },
    {
      "epoch": 1.39,
      "grad_norm": 10.059129819988367,
      "learning_rate": 6.4904218075508185e-06,
      "loss": 0.8008,
      "step": 32840
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.218741220432124,
      "learning_rate": 6.489246601957361e-06,
      "loss": 0.7665,
      "step": 32845
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.475843003876764,
      "learning_rate": 6.488071306075268e-06,
      "loss": 0.8084,
      "step": 32850
    },
    {
      "epoch": 1.39,
      "grad_norm": 8.571811936135655,
      "learning_rate": 6.486895919975795e-06,
      "loss": 0.8084,
      "step": 32855
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.851511390449732,
      "learning_rate": 6.485720443730201e-06,
      "loss": 0.7944,
      "step": 32860
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.9321290529365855,
      "learning_rate": 6.484544877409754e-06,
      "loss": 0.7924,
      "step": 32865
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.884257405141088,
      "learning_rate": 6.483369221085724e-06,
      "loss": 0.7733,
      "step": 32870
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.797512815579661,
      "learning_rate": 6.482193474829387e-06,
      "loss": 0.7908,
      "step": 32875
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.438727093393671,
      "learning_rate": 6.481017638712026e-06,
      "loss": 0.7749,
      "step": 32880
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.067616559359568,
      "learning_rate": 6.479841712804928e-06,
      "loss": 0.7907,
      "step": 32885
    },
    {
      "epoch": 1.39,
      "grad_norm": 5.043006135494658,
      "learning_rate": 6.478665697179386e-06,
      "loss": 0.8117,
      "step": 32890
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.1475784821434365,
      "learning_rate": 6.477489591906697e-06,
      "loss": 0.8447,
      "step": 32895
    },
    {
      "epoch": 1.39,
      "grad_norm": 14.548836230878772,
      "learning_rate": 6.476313397058168e-06,
      "loss": 0.7743,
      "step": 32900
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.937513110656233,
      "learning_rate": 6.475137112705106e-06,
      "loss": 0.7877,
      "step": 32905
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.10053323463538,
      "learning_rate": 6.473960738918826e-06,
      "loss": 0.7963,
      "step": 32910
    },
    {
      "epoch": 1.39,
      "grad_norm": 7.692046715571448,
      "learning_rate": 6.4727842757706474e-06,
      "loss": 0.8175,
      "step": 32915
    },
    {
      "epoch": 1.39,
      "grad_norm": 12.912291714255062,
      "learning_rate": 6.4716077233318965e-06,
      "loss": 0.8111,
      "step": 32920
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.605816354637375,
      "learning_rate": 6.470431081673905e-06,
      "loss": 0.7596,
      "step": 32925
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.432619590299122,
      "learning_rate": 6.469254350868007e-06,
      "loss": 0.8114,
      "step": 32930
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.793072470239717,
      "learning_rate": 6.468077530985547e-06,
      "loss": 0.838,
      "step": 32935
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.819586946104696,
      "learning_rate": 6.46690062209787e-06,
      "loss": 0.8069,
      "step": 32940
    },
    {
      "epoch": 1.39,
      "grad_norm": 4.539770561655789,
      "learning_rate": 6.465723624276329e-06,
      "loss": 0.7688,
      "step": 32945
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.46751269024532,
      "learning_rate": 6.464546537592282e-06,
      "loss": 0.7979,
      "step": 32950
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.475284238357774,
      "learning_rate": 6.463369362117093e-06,
      "loss": 0.8248,
      "step": 32955
    },
    {
      "epoch": 1.39,
      "grad_norm": 6.422390208930225,
      "learning_rate": 6.4621920979221285e-06,
      "loss": 0.7806,
      "step": 32960
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.490123617153963,
      "learning_rate": 6.4610147450787665e-06,
      "loss": 0.8019,
      "step": 32965
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.867896980274606,
      "learning_rate": 6.459837303658382e-06,
      "loss": 0.807,
      "step": 32970
    },
    {
      "epoch": 1.4,
      "grad_norm": 9.283771355576128,
      "learning_rate": 6.4586597737323635e-06,
      "loss": 0.7448,
      "step": 32975
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.306157580264152,
      "learning_rate": 6.4574821553720975e-06,
      "loss": 0.7908,
      "step": 32980
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.530920095575955,
      "learning_rate": 6.456304448648983e-06,
      "loss": 0.8084,
      "step": 32985
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.0158768451884805,
      "learning_rate": 6.4551266536344195e-06,
      "loss": 0.8082,
      "step": 32990
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.724381609113554,
      "learning_rate": 6.453948770399813e-06,
      "loss": 0.7738,
      "step": 32995
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.41866028276665,
      "learning_rate": 6.452770799016575e-06,
      "loss": 0.8257,
      "step": 33000
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.9074283272337205,
      "learning_rate": 6.451592739556123e-06,
      "loss": 0.7976,
      "step": 33005
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.9899901727741485,
      "learning_rate": 6.45041459208988e-06,
      "loss": 0.7905,
      "step": 33010
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.99282904133313,
      "learning_rate": 6.449236356689272e-06,
      "loss": 0.8027,
      "step": 33015
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.262841221526618,
      "learning_rate": 6.448058033425731e-06,
      "loss": 0.8057,
      "step": 33020
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.737270791244499,
      "learning_rate": 6.446879622370699e-06,
      "loss": 0.7611,
      "step": 33025
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.991359646274038,
      "learning_rate": 6.445701123595616e-06,
      "loss": 0.8194,
      "step": 33030
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.732435506791664,
      "learning_rate": 6.444522537171932e-06,
      "loss": 0.8227,
      "step": 33035
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.376097302555189,
      "learning_rate": 6.443343863171103e-06,
      "loss": 0.8081,
      "step": 33040
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.271384082786036,
      "learning_rate": 6.442165101664586e-06,
      "loss": 0.8102,
      "step": 33045
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.456062673521002,
      "learning_rate": 6.440986252723848e-06,
      "loss": 0.8188,
      "step": 33050
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.099204094506896,
      "learning_rate": 6.439807316420357e-06,
      "loss": 0.7717,
      "step": 33055
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.318946498438357,
      "learning_rate": 6.43862829282559e-06,
      "loss": 0.8177,
      "step": 33060
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.05900790819711,
      "learning_rate": 6.437449182011028e-06,
      "loss": 0.8013,
      "step": 33065
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.837506999870931,
      "learning_rate": 6.436269984048154e-06,
      "loss": 0.7632,
      "step": 33070
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.76772902700047,
      "learning_rate": 6.4350906990084625e-06,
      "loss": 0.8133,
      "step": 33075
    },
    {
      "epoch": 1.4,
      "grad_norm": 18.121361402924915,
      "learning_rate": 6.433911326963449e-06,
      "loss": 0.8207,
      "step": 33080
    },
    {
      "epoch": 1.4,
      "grad_norm": 12.163078268785592,
      "learning_rate": 6.432731867984618e-06,
      "loss": 0.7886,
      "step": 33085
    },
    {
      "epoch": 1.4,
      "grad_norm": 28.314550616970372,
      "learning_rate": 6.4315523221434705e-06,
      "loss": 0.821,
      "step": 33090
    },
    {
      "epoch": 1.4,
      "grad_norm": 10.949218419785698,
      "learning_rate": 6.430372689511525e-06,
      "loss": 0.7796,
      "step": 33095
    },
    {
      "epoch": 1.4,
      "grad_norm": 6.330186748409118,
      "learning_rate": 6.429192970160296e-06,
      "loss": 0.7771,
      "step": 33100
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.443302786957219,
      "learning_rate": 6.4280131641613064e-06,
      "loss": 0.8082,
      "step": 33105
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.890992968591108,
      "learning_rate": 6.426833271586086e-06,
      "loss": 0.8284,
      "step": 33110
    },
    {
      "epoch": 1.4,
      "grad_norm": 7.816618518452142,
      "learning_rate": 6.425653292506165e-06,
      "loss": 0.7816,
      "step": 33115
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.630217699055115,
      "learning_rate": 6.424473226993086e-06,
      "loss": 0.8186,
      "step": 33120
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.589063235720089,
      "learning_rate": 6.423293075118392e-06,
      "loss": 0.7795,
      "step": 33125
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.61802093160942,
      "learning_rate": 6.422112836953629e-06,
      "loss": 0.7753,
      "step": 33130
    },
    {
      "epoch": 1.4,
      "grad_norm": 14.050159465578894,
      "learning_rate": 6.4209325125703556e-06,
      "loss": 0.7584,
      "step": 33135
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.707300672888438,
      "learning_rate": 6.419752102040129e-06,
      "loss": 0.8287,
      "step": 33140
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.147396651089398,
      "learning_rate": 6.4185716054345136e-06,
      "loss": 0.8272,
      "step": 33145
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.016212261495118,
      "learning_rate": 6.417391022825082e-06,
      "loss": 0.7761,
      "step": 33150
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.292363560199518,
      "learning_rate": 6.416210354283407e-06,
      "loss": 0.7786,
      "step": 33155
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.878868743910045,
      "learning_rate": 6.415029599881068e-06,
      "loss": 0.7825,
      "step": 33160
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.149938082552027,
      "learning_rate": 6.413848759689654e-06,
      "loss": 0.7942,
      "step": 33165
    },
    {
      "epoch": 1.4,
      "grad_norm": 8.207944339852757,
      "learning_rate": 6.412667833780756e-06,
      "loss": 0.8172,
      "step": 33170
    },
    {
      "epoch": 1.4,
      "grad_norm": 10.37069909339702,
      "learning_rate": 6.4114868222259666e-06,
      "loss": 0.785,
      "step": 33175
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.2154684632893105,
      "learning_rate": 6.410305725096889e-06,
      "loss": 0.8063,
      "step": 33180
    },
    {
      "epoch": 1.4,
      "grad_norm": 11.45192542070303,
      "learning_rate": 6.409124542465129e-06,
      "loss": 0.7802,
      "step": 33185
    },
    {
      "epoch": 1.4,
      "grad_norm": 11.921757922717042,
      "learning_rate": 6.407943274402299e-06,
      "loss": 0.8465,
      "step": 33190
    },
    {
      "epoch": 1.4,
      "grad_norm": 5.37072460277755,
      "learning_rate": 6.406761920980016e-06,
      "loss": 0.8254,
      "step": 33195
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.617662375873842,
      "learning_rate": 6.4055804822699005e-06,
      "loss": 0.8255,
      "step": 33200
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.2688111173906105,
      "learning_rate": 6.404398958343582e-06,
      "loss": 0.7359,
      "step": 33205
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.436218622614176,
      "learning_rate": 6.403217349272691e-06,
      "loss": 0.7612,
      "step": 33210
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.851132645174651,
      "learning_rate": 6.402035655128864e-06,
      "loss": 0.803,
      "step": 33215
    },
    {
      "epoch": 1.41,
      "grad_norm": 10.082411408152227,
      "learning_rate": 6.400853875983747e-06,
      "loss": 0.8313,
      "step": 33220
    },
    {
      "epoch": 1.41,
      "grad_norm": 12.6743400529898,
      "learning_rate": 6.399672011908986e-06,
      "loss": 0.7687,
      "step": 33225
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.719349151513585,
      "learning_rate": 6.398490062976231e-06,
      "loss": 0.7814,
      "step": 33230
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.076231193956435,
      "learning_rate": 6.397308029257146e-06,
      "loss": 0.8014,
      "step": 33235
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.434478584553953,
      "learning_rate": 6.396125910823389e-06,
      "loss": 0.7796,
      "step": 33240
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.483799636494808,
      "learning_rate": 6.394943707746631e-06,
      "loss": 0.7927,
      "step": 33245
    },
    {
      "epoch": 1.41,
      "grad_norm": 15.5409722751984,
      "learning_rate": 6.393761420098546e-06,
      "loss": 0.7982,
      "step": 33250
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.8007651998966,
      "learning_rate": 6.392579047950811e-06,
      "loss": 0.7798,
      "step": 33255
    },
    {
      "epoch": 1.41,
      "grad_norm": 13.248744314510748,
      "learning_rate": 6.391396591375111e-06,
      "loss": 0.7997,
      "step": 33260
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.230725644655068,
      "learning_rate": 6.390214050443132e-06,
      "loss": 0.8061,
      "step": 33265
    },
    {
      "epoch": 1.41,
      "grad_norm": 19.291779957251343,
      "learning_rate": 6.389031425226572e-06,
      "loss": 0.8385,
      "step": 33270
    },
    {
      "epoch": 1.41,
      "grad_norm": 15.954965177255726,
      "learning_rate": 6.387848715797127e-06,
      "loss": 0.7882,
      "step": 33275
    },
    {
      "epoch": 1.41,
      "grad_norm": 14.123110945750108,
      "learning_rate": 6.3866659222265026e-06,
      "loss": 0.8327,
      "step": 33280
    },
    {
      "epoch": 1.41,
      "grad_norm": 17.895379547355656,
      "learning_rate": 6.385483044586409e-06,
      "loss": 0.7791,
      "step": 33285
    },
    {
      "epoch": 1.41,
      "grad_norm": 11.318118901413069,
      "learning_rate": 6.3843000829485576e-06,
      "loss": 0.7541,
      "step": 33290
    },
    {
      "epoch": 1.41,
      "grad_norm": 12.013842823949265,
      "learning_rate": 6.383117037384672e-06,
      "loss": 0.8191,
      "step": 33295
    },
    {
      "epoch": 1.41,
      "grad_norm": 15.153129798921043,
      "learning_rate": 6.381933907966473e-06,
      "loss": 0.8208,
      "step": 33300
    },
    {
      "epoch": 1.41,
      "grad_norm": 19.808212049826725,
      "learning_rate": 6.380750694765692e-06,
      "loss": 0.7517,
      "step": 33305
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.728000212955964,
      "learning_rate": 6.379567397854065e-06,
      "loss": 0.7933,
      "step": 33310
    },
    {
      "epoch": 1.41,
      "grad_norm": 9.240451849535159,
      "learning_rate": 6.378384017303327e-06,
      "loss": 0.81,
      "step": 33315
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.939889136556447,
      "learning_rate": 6.3772005531852295e-06,
      "loss": 0.8212,
      "step": 33320
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.358252774736982,
      "learning_rate": 6.376017005571518e-06,
      "loss": 0.7655,
      "step": 33325
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.220183882948222,
      "learning_rate": 6.374833374533949e-06,
      "loss": 0.7903,
      "step": 33330
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.928255244762764,
      "learning_rate": 6.373649660144281e-06,
      "loss": 0.7766,
      "step": 33335
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.0141816038703,
      "learning_rate": 6.372465862474282e-06,
      "loss": 0.8497,
      "step": 33340
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.483851405393595,
      "learning_rate": 6.371281981595719e-06,
      "loss": 0.7836,
      "step": 33345
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.344566871669768,
      "learning_rate": 6.370098017580369e-06,
      "loss": 0.7768,
      "step": 33350
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.714596951241924,
      "learning_rate": 6.368913970500013e-06,
      "loss": 0.8153,
      "step": 33355
    },
    {
      "epoch": 1.41,
      "grad_norm": 9.943582547085338,
      "learning_rate": 6.3677298404264335e-06,
      "loss": 0.806,
      "step": 33360
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.7785065165581875,
      "learning_rate": 6.366545627431424e-06,
      "loss": 0.7838,
      "step": 33365
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.010861643261537,
      "learning_rate": 6.365361331586778e-06,
      "loss": 0.7871,
      "step": 33370
    },
    {
      "epoch": 1.41,
      "grad_norm": 11.887829757289023,
      "learning_rate": 6.3641769529642964e-06,
      "loss": 0.8005,
      "step": 33375
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.764175795180371,
      "learning_rate": 6.3629924916357845e-06,
      "loss": 0.8547,
      "step": 33380
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.758475788918587,
      "learning_rate": 6.361807947673051e-06,
      "loss": 0.7809,
      "step": 33385
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.382764633663472,
      "learning_rate": 6.360623321147914e-06,
      "loss": 0.7832,
      "step": 33390
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.191875748999542,
      "learning_rate": 6.359438612132194e-06,
      "loss": 0.8141,
      "step": 33395
    },
    {
      "epoch": 1.41,
      "grad_norm": 5.410148922650489,
      "learning_rate": 6.358253820697714e-06,
      "loss": 0.7731,
      "step": 33400
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.330549440811483,
      "learning_rate": 6.357068946916307e-06,
      "loss": 0.8004,
      "step": 33405
    },
    {
      "epoch": 1.41,
      "grad_norm": 7.97921071145875,
      "learning_rate": 6.3558839908598055e-06,
      "loss": 0.8369,
      "step": 33410
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.6257965337114575,
      "learning_rate": 6.354698952600053e-06,
      "loss": 0.7489,
      "step": 33415
    },
    {
      "epoch": 1.41,
      "grad_norm": 8.38487796644744,
      "learning_rate": 6.353513832208893e-06,
      "loss": 0.8001,
      "step": 33420
    },
    {
      "epoch": 1.41,
      "grad_norm": 6.684241706869719,
      "learning_rate": 6.352328629758176e-06,
      "loss": 0.7593,
      "step": 33425
    },
    {
      "epoch": 1.41,
      "grad_norm": 4.1881531758514905,
      "learning_rate": 6.3511433453197575e-06,
      "loss": 0.7582,
      "step": 33430
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.04387711529924,
      "learning_rate": 6.349957978965497e-06,
      "loss": 0.8078,
      "step": 33435
    },
    {
      "epoch": 1.42,
      "grad_norm": 15.097050926566288,
      "learning_rate": 6.348772530767263e-06,
      "loss": 0.8075,
      "step": 33440
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.17846601168293,
      "learning_rate": 6.347587000796922e-06,
      "loss": 0.7599,
      "step": 33445
    },
    {
      "epoch": 1.42,
      "grad_norm": 12.224113310524332,
      "learning_rate": 6.346401389126351e-06,
      "loss": 0.7499,
      "step": 33450
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.185016709481542,
      "learning_rate": 6.345215695827429e-06,
      "loss": 0.8238,
      "step": 33455
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.8090163457243165,
      "learning_rate": 6.344029920972042e-06,
      "loss": 0.7777,
      "step": 33460
    },
    {
      "epoch": 1.42,
      "grad_norm": 12.755642457408706,
      "learning_rate": 6.342844064632081e-06,
      "loss": 0.8175,
      "step": 33465
    },
    {
      "epoch": 1.42,
      "grad_norm": 18.142957836402566,
      "learning_rate": 6.341658126879438e-06,
      "loss": 0.8104,
      "step": 33470
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.376409497988296,
      "learning_rate": 6.3404721077860155e-06,
      "loss": 0.7643,
      "step": 33475
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.703728206972405,
      "learning_rate": 6.339286007423718e-06,
      "loss": 0.7743,
      "step": 33480
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.314511191441458,
      "learning_rate": 6.338099825864455e-06,
      "loss": 0.7653,
      "step": 33485
    },
    {
      "epoch": 1.42,
      "grad_norm": 13.324673040779766,
      "learning_rate": 6.3369135631801415e-06,
      "loss": 0.8298,
      "step": 33490
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.39915612087775,
      "learning_rate": 6.335727219442696e-06,
      "loss": 0.772,
      "step": 33495
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.679108621575984,
      "learning_rate": 6.334540794724043e-06,
      "loss": 0.7848,
      "step": 33500
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.323469051210022,
      "learning_rate": 6.333354289096113e-06,
      "loss": 0.8046,
      "step": 33505
    },
    {
      "epoch": 1.42,
      "grad_norm": 10.796342203657705,
      "learning_rate": 6.3321677026308406e-06,
      "loss": 0.7659,
      "step": 33510
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.9143980655472905,
      "learning_rate": 6.3309810354001646e-06,
      "loss": 0.7437,
      "step": 33515
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.096871757892698,
      "learning_rate": 6.3297942874760275e-06,
      "loss": 0.7913,
      "step": 33520
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.783201381586201,
      "learning_rate": 6.328607458930381e-06,
      "loss": 0.762,
      "step": 33525
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.304642127169017,
      "learning_rate": 6.327420549835178e-06,
      "loss": 0.8302,
      "step": 33530
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.123473305508212,
      "learning_rate": 6.3262335602623755e-06,
      "loss": 0.8062,
      "step": 33535
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.328024947408747,
      "learning_rate": 6.32504649028394e-06,
      "loss": 0.7547,
      "step": 33540
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.497910507077813,
      "learning_rate": 6.323859339971839e-06,
      "loss": 0.7605,
      "step": 33545
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.957325703067418,
      "learning_rate": 6.3226721093980445e-06,
      "loss": 0.7616,
      "step": 33550
    },
    {
      "epoch": 1.42,
      "grad_norm": 13.985064366163243,
      "learning_rate": 6.321484798634538e-06,
      "loss": 0.8261,
      "step": 33555
    },
    {
      "epoch": 1.42,
      "grad_norm": 29.002979507589,
      "learning_rate": 6.320297407753299e-06,
      "loss": 0.8156,
      "step": 33560
    },
    {
      "epoch": 1.42,
      "grad_norm": 11.703443151285368,
      "learning_rate": 6.3191099368263186e-06,
      "loss": 0.7547,
      "step": 33565
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.021728980228457,
      "learning_rate": 6.317922385925588e-06,
      "loss": 0.7965,
      "step": 33570
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.664802376961735,
      "learning_rate": 6.316734755123106e-06,
      "loss": 0.7797,
      "step": 33575
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.963776082460479,
      "learning_rate": 6.3155470444908724e-06,
      "loss": 0.8411,
      "step": 33580
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.558474013703338,
      "learning_rate": 6.314359254100898e-06,
      "loss": 0.8267,
      "step": 33585
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.263537466413258,
      "learning_rate": 6.313171384025193e-06,
      "loss": 0.8876,
      "step": 33590
    },
    {
      "epoch": 1.42,
      "grad_norm": 10.799881519017585,
      "learning_rate": 6.311983434335775e-06,
      "loss": 0.8009,
      "step": 33595
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.724402504621985,
      "learning_rate": 6.310795405104666e-06,
      "loss": 0.7801,
      "step": 33600
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.567847531157446,
      "learning_rate": 6.309607296403892e-06,
      "loss": 0.7782,
      "step": 33605
    },
    {
      "epoch": 1.42,
      "grad_norm": 9.222203584142193,
      "learning_rate": 6.308419108305486e-06,
      "loss": 0.7709,
      "step": 33610
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.08037064499177,
      "learning_rate": 6.307230840881484e-06,
      "loss": 0.7934,
      "step": 33615
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.185159184413078,
      "learning_rate": 6.306042494203925e-06,
      "loss": 0.7643,
      "step": 33620
    },
    {
      "epoch": 1.42,
      "grad_norm": 5.7554089682114205,
      "learning_rate": 6.304854068344859e-06,
      "loss": 0.7933,
      "step": 33625
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.676777303939975,
      "learning_rate": 6.3036655633763325e-06,
      "loss": 0.8071,
      "step": 33630
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.938662027458149,
      "learning_rate": 6.302476979370404e-06,
      "loss": 0.7602,
      "step": 33635
    },
    {
      "epoch": 1.42,
      "grad_norm": 8.891418483254615,
      "learning_rate": 6.301288316399133e-06,
      "loss": 0.8057,
      "step": 33640
    },
    {
      "epoch": 1.42,
      "grad_norm": 11.726390402404421,
      "learning_rate": 6.300099574534582e-06,
      "loss": 0.7785,
      "step": 33645
    },
    {
      "epoch": 1.42,
      "grad_norm": 15.967132115859908,
      "learning_rate": 6.298910753848826e-06,
      "loss": 0.7672,
      "step": 33650
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.646060518597896,
      "learning_rate": 6.297721854413937e-06,
      "loss": 0.8515,
      "step": 33655
    },
    {
      "epoch": 1.42,
      "grad_norm": 7.883630140493178,
      "learning_rate": 6.296532876301995e-06,
      "loss": 0.7694,
      "step": 33660
    },
    {
      "epoch": 1.42,
      "grad_norm": 6.781839380970123,
      "learning_rate": 6.2953438195850816e-06,
      "loss": 0.8043,
      "step": 33665
    },
    {
      "epoch": 1.42,
      "grad_norm": 4.690140167382787,
      "learning_rate": 6.29415468433529e-06,
      "loss": 0.7567,
      "step": 33670
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.975010848973904,
      "learning_rate": 6.29296547062471e-06,
      "loss": 0.8031,
      "step": 33675
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.248268129993717,
      "learning_rate": 6.291776178525443e-06,
      "loss": 0.7715,
      "step": 33680
    },
    {
      "epoch": 1.43,
      "grad_norm": 13.015578254711917,
      "learning_rate": 6.2905868081095925e-06,
      "loss": 0.7465,
      "step": 33685
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.446898035359444,
      "learning_rate": 6.289397359449263e-06,
      "loss": 0.8119,
      "step": 33690
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.267318710835655,
      "learning_rate": 6.28820783261657e-06,
      "loss": 0.819,
      "step": 33695
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.5200918795535205,
      "learning_rate": 6.287018227683633e-06,
      "loss": 0.8024,
      "step": 33700
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.646725312417338,
      "learning_rate": 6.28582854472257e-06,
      "loss": 0.7962,
      "step": 33705
    },
    {
      "epoch": 1.43,
      "grad_norm": 14.331725110947344,
      "learning_rate": 6.28463878380551e-06,
      "loss": 0.7607,
      "step": 33710
    },
    {
      "epoch": 1.43,
      "grad_norm": 19.970792824740546,
      "learning_rate": 6.283448945004584e-06,
      "loss": 0.8046,
      "step": 33715
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.083635721352184,
      "learning_rate": 6.282259028391931e-06,
      "loss": 0.7686,
      "step": 33720
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.515417092340349,
      "learning_rate": 6.281069034039689e-06,
      "loss": 0.7954,
      "step": 33725
    },
    {
      "epoch": 1.43,
      "grad_norm": 22.857086869745768,
      "learning_rate": 6.279878962020005e-06,
      "loss": 0.7398,
      "step": 33730
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.969475322008545,
      "learning_rate": 6.27868881240503e-06,
      "loss": 0.7709,
      "step": 33735
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.196962414187352,
      "learning_rate": 6.2774985852669194e-06,
      "loss": 0.813,
      "step": 33740
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.95479997659839,
      "learning_rate": 6.27630828067783e-06,
      "loss": 0.7687,
      "step": 33745
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.166100612565272,
      "learning_rate": 6.27511789870993e-06,
      "loss": 0.8269,
      "step": 33750
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.991588145496196,
      "learning_rate": 6.27392743943539e-06,
      "loss": 0.796,
      "step": 33755
    },
    {
      "epoch": 1.43,
      "grad_norm": 17.550808065110896,
      "learning_rate": 6.27273690292638e-06,
      "loss": 0.7678,
      "step": 33760
    },
    {
      "epoch": 1.43,
      "grad_norm": 16.91998056655981,
      "learning_rate": 6.271546289255079e-06,
      "loss": 0.7924,
      "step": 33765
    },
    {
      "epoch": 1.43,
      "grad_norm": 23.228724521364768,
      "learning_rate": 6.270355598493675e-06,
      "loss": 0.8144,
      "step": 33770
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.352412549267018,
      "learning_rate": 6.269164830714352e-06,
      "loss": 0.7723,
      "step": 33775
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.100729005443817,
      "learning_rate": 6.267973985989303e-06,
      "loss": 0.7942,
      "step": 33780
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.343829188990839,
      "learning_rate": 6.266783064390726e-06,
      "loss": 0.794,
      "step": 33785
    },
    {
      "epoch": 1.43,
      "grad_norm": 19.850079813418155,
      "learning_rate": 6.2655920659908245e-06,
      "loss": 0.7717,
      "step": 33790
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.114012409801975,
      "learning_rate": 6.264400990861803e-06,
      "loss": 0.7782,
      "step": 33795
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.62194088276353,
      "learning_rate": 6.263209839075874e-06,
      "loss": 0.7711,
      "step": 33800
    },
    {
      "epoch": 1.43,
      "grad_norm": 16.459426670595082,
      "learning_rate": 6.262018610705254e-06,
      "loss": 0.758,
      "step": 33805
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.9227259749298025,
      "learning_rate": 6.260827305822163e-06,
      "loss": 0.7933,
      "step": 33810
    },
    {
      "epoch": 1.43,
      "grad_norm": 12.60348967551295,
      "learning_rate": 6.259635924498826e-06,
      "loss": 0.7749,
      "step": 33815
    },
    {
      "epoch": 1.43,
      "grad_norm": 6.571657138223135,
      "learning_rate": 6.2584444668074725e-06,
      "loss": 0.7442,
      "step": 33820
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.474382021308996,
      "learning_rate": 6.257252932820338e-06,
      "loss": 0.8083,
      "step": 33825
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.279164341693866,
      "learning_rate": 6.256061322609662e-06,
      "loss": 0.7926,
      "step": 33830
    },
    {
      "epoch": 1.43,
      "grad_norm": 11.978414178981435,
      "learning_rate": 6.254869636247689e-06,
      "loss": 0.7767,
      "step": 33835
    },
    {
      "epoch": 1.43,
      "grad_norm": 9.87756131730474,
      "learning_rate": 6.253677873806664e-06,
      "loss": 0.7771,
      "step": 33840
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.837892847089357,
      "learning_rate": 6.252486035358845e-06,
      "loss": 0.7666,
      "step": 33845
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.0688123275456425,
      "learning_rate": 6.251294120976485e-06,
      "loss": 0.8196,
      "step": 33850
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.980715273564473,
      "learning_rate": 6.250102130731848e-06,
      "loss": 0.8142,
      "step": 33855
    },
    {
      "epoch": 1.43,
      "grad_norm": 7.53340927478789,
      "learning_rate": 6.248910064697203e-06,
      "loss": 0.7988,
      "step": 33860
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.519232237945045,
      "learning_rate": 6.247717922944817e-06,
      "loss": 0.8561,
      "step": 33865
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.077288648296106,
      "learning_rate": 6.2465257055469705e-06,
      "loss": 0.7627,
      "step": 33870
    },
    {
      "epoch": 1.43,
      "grad_norm": 11.071584510719482,
      "learning_rate": 6.2453334125759416e-06,
      "loss": 0.8222,
      "step": 33875
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.3692577351745765,
      "learning_rate": 6.244141044104016e-06,
      "loss": 0.7532,
      "step": 33880
    },
    {
      "epoch": 1.43,
      "grad_norm": 8.042192469217852,
      "learning_rate": 6.2429486002034845e-06,
      "loss": 0.7934,
      "step": 33885
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.109208215289776,
      "learning_rate": 6.241756080946639e-06,
      "loss": 0.7858,
      "step": 33890
    },
    {
      "epoch": 1.43,
      "grad_norm": 5.403447878678749,
      "learning_rate": 6.240563486405783e-06,
      "loss": 0.7837,
      "step": 33895
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.588711799024247,
      "learning_rate": 6.239370816653214e-06,
      "loss": 0.8065,
      "step": 33900
    },
    {
      "epoch": 1.43,
      "grad_norm": 4.656594978787163,
      "learning_rate": 6.238178071761245e-06,
      "loss": 0.8167,
      "step": 33905
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.073669763022908,
      "learning_rate": 6.236985251802185e-06,
      "loss": 0.7854,
      "step": 33910
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.06583975803752,
      "learning_rate": 6.235792356848354e-06,
      "loss": 0.8506,
      "step": 33915
    },
    {
      "epoch": 1.44,
      "grad_norm": 10.439466141932854,
      "learning_rate": 6.2345993869720725e-06,
      "loss": 0.8071,
      "step": 33920
    },
    {
      "epoch": 1.44,
      "grad_norm": 9.110829511322205,
      "learning_rate": 6.233406342245666e-06,
      "loss": 0.7767,
      "step": 33925
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.670148774262995,
      "learning_rate": 6.2322132227414665e-06,
      "loss": 0.7891,
      "step": 33930
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.960863602233144,
      "learning_rate": 6.231020028531809e-06,
      "loss": 0.7794,
      "step": 33935
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.557495383739372,
      "learning_rate": 6.229826759689034e-06,
      "loss": 0.7617,
      "step": 33940
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.688243051270011,
      "learning_rate": 6.228633416285484e-06,
      "loss": 0.7753,
      "step": 33945
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.958934412467835,
      "learning_rate": 6.22743999839351e-06,
      "loss": 0.7936,
      "step": 33950
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.0968665325614175,
      "learning_rate": 6.226246506085464e-06,
      "loss": 0.8259,
      "step": 33955
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.001261741953069,
      "learning_rate": 6.225052939433703e-06,
      "loss": 0.7559,
      "step": 33960
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.27468128935905,
      "learning_rate": 6.2238592985105925e-06,
      "loss": 0.7481,
      "step": 33965
    },
    {
      "epoch": 1.44,
      "grad_norm": 10.000343085440802,
      "learning_rate": 6.2226655833884975e-06,
      "loss": 0.8172,
      "step": 33970
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.937616089023601,
      "learning_rate": 6.221471794139788e-06,
      "loss": 0.7364,
      "step": 33975
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.700235336896541,
      "learning_rate": 6.2202779308368454e-06,
      "loss": 0.7968,
      "step": 33980
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.88406638499069,
      "learning_rate": 6.219083993552043e-06,
      "loss": 0.8028,
      "step": 33985
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.659719834745839,
      "learning_rate": 6.2178899823577694e-06,
      "loss": 0.7831,
      "step": 33990
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.62628577536684,
      "learning_rate": 6.216695897326414e-06,
      "loss": 0.7824,
      "step": 33995
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.4910818638030925,
      "learning_rate": 6.215501738530371e-06,
      "loss": 0.8096,
      "step": 34000
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.167064205932446,
      "learning_rate": 6.214307506042038e-06,
      "loss": 0.7939,
      "step": 34005
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.487534658645625,
      "learning_rate": 6.213113199933816e-06,
      "loss": 0.7648,
      "step": 34010
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.433151402327117,
      "learning_rate": 6.2119188202781155e-06,
      "loss": 0.7668,
      "step": 34015
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.65246327594702,
      "learning_rate": 6.210724367147346e-06,
      "loss": 0.7781,
      "step": 34020
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.667865949709403,
      "learning_rate": 6.209529840613925e-06,
      "loss": 0.7894,
      "step": 34025
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.114879666857776,
      "learning_rate": 6.208335240750272e-06,
      "loss": 0.7945,
      "step": 34030
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.546163771367936,
      "learning_rate": 6.207140567628813e-06,
      "loss": 0.8149,
      "step": 34035
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.099522284386283,
      "learning_rate": 6.205945821321977e-06,
      "loss": 0.7798,
      "step": 34040
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.572394397884539,
      "learning_rate": 6.204751001902198e-06,
      "loss": 0.7947,
      "step": 34045
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.667708867504958,
      "learning_rate": 6.203556109441915e-06,
      "loss": 0.7913,
      "step": 34050
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.816099801353484,
      "learning_rate": 6.20236114401357e-06,
      "loss": 0.7463,
      "step": 34055
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.53170190313959,
      "learning_rate": 6.201166105689611e-06,
      "loss": 0.7813,
      "step": 34060
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.3808703746127735,
      "learning_rate": 6.199970994542489e-06,
      "loss": 0.7974,
      "step": 34065
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.787181093943561,
      "learning_rate": 6.19877581064466e-06,
      "loss": 0.7887,
      "step": 34070
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.3078981966530225,
      "learning_rate": 6.197580554068584e-06,
      "loss": 0.7488,
      "step": 34075
    },
    {
      "epoch": 1.44,
      "grad_norm": 12.39698128917764,
      "learning_rate": 6.196385224886726e-06,
      "loss": 0.7854,
      "step": 34080
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.721623734254808,
      "learning_rate": 6.195189823171558e-06,
      "loss": 0.7943,
      "step": 34085
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.604864372023594,
      "learning_rate": 6.1939943489955514e-06,
      "loss": 0.797,
      "step": 34090
    },
    {
      "epoch": 1.44,
      "grad_norm": 10.437604271693152,
      "learning_rate": 6.192798802431183e-06,
      "loss": 0.749,
      "step": 34095
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.005849860727069,
      "learning_rate": 6.191603183550937e-06,
      "loss": 0.811,
      "step": 34100
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.826263499121277,
      "learning_rate": 6.190407492427301e-06,
      "loss": 0.7877,
      "step": 34105
    },
    {
      "epoch": 1.44,
      "grad_norm": 11.518243211036426,
      "learning_rate": 6.189211729132764e-06,
      "loss": 0.7353,
      "step": 34110
    },
    {
      "epoch": 1.44,
      "grad_norm": 6.081839292047142,
      "learning_rate": 6.188015893739824e-06,
      "loss": 0.7289,
      "step": 34115
    },
    {
      "epoch": 1.44,
      "grad_norm": 8.62162030836602,
      "learning_rate": 6.186819986320977e-06,
      "loss": 0.7679,
      "step": 34120
    },
    {
      "epoch": 1.44,
      "grad_norm": 5.49739037988598,
      "learning_rate": 6.185624006948734e-06,
      "loss": 0.8126,
      "step": 34125
    },
    {
      "epoch": 1.44,
      "grad_norm": 7.365828539025667,
      "learning_rate": 6.184427955695597e-06,
      "loss": 0.7926,
      "step": 34130
    },
    {
      "epoch": 1.44,
      "grad_norm": 20.122613763498403,
      "learning_rate": 6.1832318326340836e-06,
      "loss": 0.7522,
      "step": 34135
    },
    {
      "epoch": 1.44,
      "grad_norm": 25.722705199570957,
      "learning_rate": 6.18203563783671e-06,
      "loss": 0.8103,
      "step": 34140
    },
    {
      "epoch": 1.45,
      "grad_norm": 14.894930124946361,
      "learning_rate": 6.180839371375995e-06,
      "loss": 0.8014,
      "step": 34145
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.68502490070025,
      "learning_rate": 6.179643033324469e-06,
      "loss": 0.7883,
      "step": 34150
    },
    {
      "epoch": 1.45,
      "grad_norm": 7.730206205907829,
      "learning_rate": 6.178446623754659e-06,
      "loss": 0.77,
      "step": 34155
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.860438992600811,
      "learning_rate": 6.177250142739104e-06,
      "loss": 0.7538,
      "step": 34160
    },
    {
      "epoch": 1.45,
      "grad_norm": 12.896813817108388,
      "learning_rate": 6.176053590350339e-06,
      "loss": 0.7473,
      "step": 34165
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.828523534967848,
      "learning_rate": 6.174856966660908e-06,
      "loss": 0.7885,
      "step": 34170
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.649846692872117,
      "learning_rate": 6.1736602717433615e-06,
      "loss": 0.7943,
      "step": 34175
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.597196832230006,
      "learning_rate": 6.1724635056702485e-06,
      "loss": 0.7731,
      "step": 34180
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.882036265353549,
      "learning_rate": 6.171266668514129e-06,
      "loss": 0.803,
      "step": 34185
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.740605598760074,
      "learning_rate": 6.170069760347559e-06,
      "loss": 0.7841,
      "step": 34190
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.567177106672672,
      "learning_rate": 6.168872781243107e-06,
      "loss": 0.7899,
      "step": 34195
    },
    {
      "epoch": 1.45,
      "grad_norm": 9.353479658910347,
      "learning_rate": 6.167675731273341e-06,
      "loss": 0.8117,
      "step": 34200
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.853490805474365,
      "learning_rate": 6.166478610510836e-06,
      "loss": 0.8074,
      "step": 34205
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.255780180375012,
      "learning_rate": 6.165281419028167e-06,
      "loss": 0.7763,
      "step": 34210
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.3549192222816,
      "learning_rate": 6.164084156897919e-06,
      "loss": 0.7495,
      "step": 34215
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.2778366458818855,
      "learning_rate": 6.162886824192678e-06,
      "loss": 0.7804,
      "step": 34220
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.294693646326014,
      "learning_rate": 6.161689420985034e-06,
      "loss": 0.7912,
      "step": 34225
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.602763280942825,
      "learning_rate": 6.160491947347582e-06,
      "loss": 0.772,
      "step": 34230
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.748799228167618,
      "learning_rate": 6.159294403352922e-06,
      "loss": 0.7842,
      "step": 34235
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.620880059970412,
      "learning_rate": 6.158096789073655e-06,
      "loss": 0.7685,
      "step": 34240
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.277112971517686,
      "learning_rate": 6.156899104582395e-06,
      "loss": 0.7652,
      "step": 34245
    },
    {
      "epoch": 1.45,
      "grad_norm": 10.026545398758971,
      "learning_rate": 6.155701349951747e-06,
      "loss": 0.7775,
      "step": 34250
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.7567221270075555,
      "learning_rate": 6.15450352525433e-06,
      "loss": 0.802,
      "step": 34255
    },
    {
      "epoch": 1.45,
      "grad_norm": 10.374829777110586,
      "learning_rate": 6.153305630562767e-06,
      "loss": 0.7814,
      "step": 34260
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.443677687440992,
      "learning_rate": 6.152107665949679e-06,
      "loss": 0.8103,
      "step": 34265
    },
    {
      "epoch": 1.45,
      "grad_norm": 11.498029050040106,
      "learning_rate": 6.1509096314876984e-06,
      "loss": 0.7773,
      "step": 34270
    },
    {
      "epoch": 1.45,
      "grad_norm": 24.0894683841244,
      "learning_rate": 6.149711527249457e-06,
      "loss": 0.7564,
      "step": 34275
    },
    {
      "epoch": 1.45,
      "grad_norm": 19.423177352698108,
      "learning_rate": 6.14851335330759e-06,
      "loss": 0.8138,
      "step": 34280
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.303172273041681,
      "learning_rate": 6.1473151097347435e-06,
      "loss": 0.8146,
      "step": 34285
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.256656282540311,
      "learning_rate": 6.146116796603561e-06,
      "loss": 0.8087,
      "step": 34290
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.758563319486756,
      "learning_rate": 6.1449184139866934e-06,
      "loss": 0.8088,
      "step": 34295
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.69419904807984,
      "learning_rate": 6.143719961956794e-06,
      "loss": 0.8175,
      "step": 34300
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.421471046486495,
      "learning_rate": 6.1425214405865245e-06,
      "loss": 0.7725,
      "step": 34305
    },
    {
      "epoch": 1.45,
      "grad_norm": 10.216531894528826,
      "learning_rate": 6.141322849948543e-06,
      "loss": 0.8278,
      "step": 34310
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.687174838359153,
      "learning_rate": 6.14012419011552e-06,
      "loss": 0.8016,
      "step": 34315
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.1806717544155845,
      "learning_rate": 6.1389254611601245e-06,
      "loss": 0.7771,
      "step": 34320
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.41011278936483,
      "learning_rate": 6.137726663155034e-06,
      "loss": 0.7781,
      "step": 34325
    },
    {
      "epoch": 1.45,
      "grad_norm": 12.714702846575664,
      "learning_rate": 6.136527796172928e-06,
      "loss": 0.7715,
      "step": 34330
    },
    {
      "epoch": 1.45,
      "grad_norm": 8.205782145134632,
      "learning_rate": 6.135328860286488e-06,
      "loss": 0.7896,
      "step": 34335
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.128060304488903,
      "learning_rate": 6.134129855568404e-06,
      "loss": 0.7967,
      "step": 34340
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.341593979154896,
      "learning_rate": 6.132930782091367e-06,
      "loss": 0.8202,
      "step": 34345
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.28490483128136,
      "learning_rate": 6.131731639928075e-06,
      "loss": 0.7953,
      "step": 34350
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.789051641452311,
      "learning_rate": 6.1305324291512265e-06,
      "loss": 0.7899,
      "step": 34355
    },
    {
      "epoch": 1.45,
      "grad_norm": 5.966130253862901,
      "learning_rate": 6.129333149833528e-06,
      "loss": 0.8268,
      "step": 34360
    },
    {
      "epoch": 1.45,
      "grad_norm": 6.465795886937925,
      "learning_rate": 6.128133802047686e-06,
      "loss": 0.7709,
      "step": 34365
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.5942301506624545,
      "learning_rate": 6.126934385866414e-06,
      "loss": 0.7303,
      "step": 34370
    },
    {
      "epoch": 1.45,
      "grad_norm": 4.3871066357512785,
      "learning_rate": 6.125734901362431e-06,
      "loss": 0.7267,
      "step": 34375
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.178169114929558,
      "learning_rate": 6.124535348608458e-06,
      "loss": 0.7922,
      "step": 34380
    },
    {
      "epoch": 1.46,
      "grad_norm": 12.911457921035186,
      "learning_rate": 6.123335727677217e-06,
      "loss": 0.794,
      "step": 34385
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.252914434087787,
      "learning_rate": 6.122136038641441e-06,
      "loss": 0.7794,
      "step": 34390
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.479340026538647,
      "learning_rate": 6.120936281573861e-06,
      "loss": 0.778,
      "step": 34395
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.626486971773706,
      "learning_rate": 6.119736456547217e-06,
      "loss": 0.7493,
      "step": 34400
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.593541292224104,
      "learning_rate": 6.1185365636342495e-06,
      "loss": 0.7461,
      "step": 34405
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.406773413162359,
      "learning_rate": 6.117336602907706e-06,
      "loss": 0.8254,
      "step": 34410
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.286827112078306,
      "learning_rate": 6.116136574440335e-06,
      "loss": 0.7656,
      "step": 34415
    },
    {
      "epoch": 1.46,
      "grad_norm": 12.977814096397365,
      "learning_rate": 6.1149364783048904e-06,
      "loss": 0.7774,
      "step": 34420
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.959256348545565,
      "learning_rate": 6.1137363145741315e-06,
      "loss": 0.7844,
      "step": 34425
    },
    {
      "epoch": 1.46,
      "grad_norm": 10.737034691521146,
      "learning_rate": 6.112536083320821e-06,
      "loss": 0.8205,
      "step": 34430
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.1674633446038065,
      "learning_rate": 6.111335784617725e-06,
      "loss": 0.7892,
      "step": 34435
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.030513230670901,
      "learning_rate": 6.110135418537612e-06,
      "loss": 0.8109,
      "step": 34440
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.567682056714353,
      "learning_rate": 6.108934985153261e-06,
      "loss": 0.7966,
      "step": 34445
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.307998245626137,
      "learning_rate": 6.107734484537448e-06,
      "loss": 0.7596,
      "step": 34450
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.76410913127957,
      "learning_rate": 6.106533916762955e-06,
      "loss": 0.7688,
      "step": 34455
    },
    {
      "epoch": 1.46,
      "grad_norm": 9.128187306238752,
      "learning_rate": 6.105333281902571e-06,
      "loss": 0.8032,
      "step": 34460
    },
    {
      "epoch": 1.46,
      "grad_norm": 11.524959566805476,
      "learning_rate": 6.104132580029086e-06,
      "loss": 0.7664,
      "step": 34465
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.718037816576605,
      "learning_rate": 6.1029318112152945e-06,
      "loss": 0.7488,
      "step": 34470
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.500707717709202,
      "learning_rate": 6.101730975533996e-06,
      "loss": 0.7445,
      "step": 34475
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.588835994024326,
      "learning_rate": 6.100530073057994e-06,
      "loss": 0.7783,
      "step": 34480
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.805799803838212,
      "learning_rate": 6.099329103860095e-06,
      "loss": 0.7925,
      "step": 34485
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.7701995916443085,
      "learning_rate": 6.098128068013112e-06,
      "loss": 0.8119,
      "step": 34490
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.0087936675207745,
      "learning_rate": 6.096926965589858e-06,
      "loss": 0.7929,
      "step": 34495
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.292341715882617,
      "learning_rate": 6.095725796663151e-06,
      "loss": 0.7722,
      "step": 34500
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.227944077813412,
      "learning_rate": 6.094524561305819e-06,
      "loss": 0.7675,
      "step": 34505
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.234618327444643,
      "learning_rate": 6.093323259590686e-06,
      "loss": 0.7882,
      "step": 34510
    },
    {
      "epoch": 1.46,
      "grad_norm": 10.553847309789164,
      "learning_rate": 6.092121891590586e-06,
      "loss": 0.7601,
      "step": 34515
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.055198094877963,
      "learning_rate": 6.0909204573783505e-06,
      "loss": 0.7575,
      "step": 34520
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.124209601561907,
      "learning_rate": 6.089718957026821e-06,
      "loss": 0.8143,
      "step": 34525
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.504296651792603,
      "learning_rate": 6.088517390608842e-06,
      "loss": 0.7681,
      "step": 34530
    },
    {
      "epoch": 1.46,
      "grad_norm": 9.588529299483298,
      "learning_rate": 6.0873157581972595e-06,
      "loss": 0.7848,
      "step": 34535
    },
    {
      "epoch": 1.46,
      "grad_norm": 7.346537961021802,
      "learning_rate": 6.086114059864926e-06,
      "loss": 0.7879,
      "step": 34540
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.823574832875575,
      "learning_rate": 6.084912295684696e-06,
      "loss": 0.8131,
      "step": 34545
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.620419284662399,
      "learning_rate": 6.08371046572943e-06,
      "loss": 0.7867,
      "step": 34550
    },
    {
      "epoch": 1.46,
      "grad_norm": 12.222807749306725,
      "learning_rate": 6.0825085700719895e-06,
      "loss": 0.7994,
      "step": 34555
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.639912162591293,
      "learning_rate": 6.0813066087852435e-06,
      "loss": 0.7547,
      "step": 34560
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.382533502959495,
      "learning_rate": 6.080104581942063e-06,
      "loss": 0.7785,
      "step": 34565
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.375276319096297,
      "learning_rate": 6.078902489615322e-06,
      "loss": 0.8186,
      "step": 34570
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.0707377002053535,
      "learning_rate": 6.077700331877903e-06,
      "loss": 0.8248,
      "step": 34575
    },
    {
      "epoch": 1.46,
      "grad_norm": 6.103196402738586,
      "learning_rate": 6.076498108802686e-06,
      "loss": 0.7597,
      "step": 34580
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.996083284476239,
      "learning_rate": 6.07529582046256e-06,
      "loss": 0.8063,
      "step": 34585
    },
    {
      "epoch": 1.46,
      "grad_norm": 10.66828348415592,
      "learning_rate": 6.074093466930416e-06,
      "loss": 0.7621,
      "step": 34590
    },
    {
      "epoch": 1.46,
      "grad_norm": 8.404291384536124,
      "learning_rate": 6.072891048279148e-06,
      "loss": 0.8191,
      "step": 34595
    },
    {
      "epoch": 1.46,
      "grad_norm": 24.259725752370393,
      "learning_rate": 6.071688564581657e-06,
      "loss": 0.7726,
      "step": 34600
    },
    {
      "epoch": 1.46,
      "grad_norm": 11.219175306689486,
      "learning_rate": 6.070486015910846e-06,
      "loss": 0.7823,
      "step": 34605
    },
    {
      "epoch": 1.46,
      "grad_norm": 10.525062508578785,
      "learning_rate": 6.069283402339619e-06,
      "loss": 0.7665,
      "step": 34610
    },
    {
      "epoch": 1.46,
      "grad_norm": 5.459301537740056,
      "learning_rate": 6.06808072394089e-06,
      "loss": 0.7652,
      "step": 34615
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.58255467813776,
      "learning_rate": 6.0668779807875726e-06,
      "loss": 0.7787,
      "step": 34620
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.191498946876796,
      "learning_rate": 6.065675172952587e-06,
      "loss": 0.7447,
      "step": 34625
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.617402950500728,
      "learning_rate": 6.064472300508853e-06,
      "loss": 0.7729,
      "step": 34630
    },
    {
      "epoch": 1.47,
      "grad_norm": 6.084350507353659,
      "learning_rate": 6.063269363529301e-06,
      "loss": 0.7681,
      "step": 34635
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.866721723374138,
      "learning_rate": 6.062066362086858e-06,
      "loss": 0.7451,
      "step": 34640
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.512566141460544,
      "learning_rate": 6.0608632962544586e-06,
      "loss": 0.7813,
      "step": 34645
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.4016321978687305,
      "learning_rate": 6.059660166105045e-06,
      "loss": 0.7689,
      "step": 34650
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.282949318772552,
      "learning_rate": 6.058456971711555e-06,
      "loss": 0.7594,
      "step": 34655
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.358737300231551,
      "learning_rate": 6.057253713146938e-06,
      "loss": 0.759,
      "step": 34660
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.115906535287775,
      "learning_rate": 6.056050390484141e-06,
      "loss": 0.7735,
      "step": 34665
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.665351400546173,
      "learning_rate": 6.054847003796121e-06,
      "loss": 0.8246,
      "step": 34670
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.024610394489488,
      "learning_rate": 6.053643553155834e-06,
      "loss": 0.8077,
      "step": 34675
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.884158504475991,
      "learning_rate": 6.052440038636242e-06,
      "loss": 0.8103,
      "step": 34680
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.782583386386305,
      "learning_rate": 6.051236460310309e-06,
      "loss": 0.8074,
      "step": 34685
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.209505905604866,
      "learning_rate": 6.050032818251008e-06,
      "loss": 0.7417,
      "step": 34690
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.357238451931976,
      "learning_rate": 6.048829112531308e-06,
      "loss": 0.7502,
      "step": 34695
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.644898695136965,
      "learning_rate": 6.047625343224191e-06,
      "loss": 0.8242,
      "step": 34700
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.857206336949919,
      "learning_rate": 6.0464215104026334e-06,
      "loss": 0.7721,
      "step": 34705
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.519163476186186,
      "learning_rate": 6.045217614139623e-06,
      "loss": 0.8065,
      "step": 34710
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.549848579505487,
      "learning_rate": 6.0440136545081485e-06,
      "loss": 0.793,
      "step": 34715
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.497533940545804,
      "learning_rate": 6.042809631581199e-06,
      "loss": 0.729,
      "step": 34720
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.700488283101436,
      "learning_rate": 6.041605545431774e-06,
      "loss": 0.761,
      "step": 34725
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.86190420513941,
      "learning_rate": 6.0404013961328726e-06,
      "loss": 0.7528,
      "step": 34730
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.055714871569262,
      "learning_rate": 6.0391971837575e-06,
      "loss": 0.7376,
      "step": 34735
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.41196321307138,
      "learning_rate": 6.037992908378664e-06,
      "loss": 0.7494,
      "step": 34740
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.0385730436908345,
      "learning_rate": 6.036788570069373e-06,
      "loss": 0.7191,
      "step": 34745
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.003093368622524,
      "learning_rate": 6.035584168902646e-06,
      "loss": 0.7571,
      "step": 34750
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.873117387778862,
      "learning_rate": 6.0343797049515e-06,
      "loss": 0.7493,
      "step": 34755
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.810647592350096,
      "learning_rate": 6.033175178288962e-06,
      "loss": 0.7659,
      "step": 34760
    },
    {
      "epoch": 1.47,
      "grad_norm": 10.284068096030106,
      "learning_rate": 6.031970588988057e-06,
      "loss": 0.7491,
      "step": 34765
    },
    {
      "epoch": 1.47,
      "grad_norm": 10.175686183762476,
      "learning_rate": 6.030765937121812e-06,
      "loss": 0.7783,
      "step": 34770
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.596337317241252,
      "learning_rate": 6.029561222763265e-06,
      "loss": 0.7874,
      "step": 34775
    },
    {
      "epoch": 1.47,
      "grad_norm": 13.7759997480062,
      "learning_rate": 6.028356445985453e-06,
      "loss": 0.777,
      "step": 34780
    },
    {
      "epoch": 1.47,
      "grad_norm": 8.139453311981725,
      "learning_rate": 6.027151606861422e-06,
      "loss": 0.7567,
      "step": 34785
    },
    {
      "epoch": 1.47,
      "grad_norm": 13.773812043866542,
      "learning_rate": 6.025946705464213e-06,
      "loss": 0.7813,
      "step": 34790
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.867209758105302,
      "learning_rate": 6.024741741866877e-06,
      "loss": 0.8086,
      "step": 34795
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.239636871008306,
      "learning_rate": 6.023536716142468e-06,
      "loss": 0.7967,
      "step": 34800
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.534730403953438,
      "learning_rate": 6.022331628364042e-06,
      "loss": 0.7739,
      "step": 34805
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.464949817251281,
      "learning_rate": 6.021126478604661e-06,
      "loss": 0.7884,
      "step": 34810
    },
    {
      "epoch": 1.47,
      "grad_norm": 5.426086292689652,
      "learning_rate": 6.01992126693739e-06,
      "loss": 0.7801,
      "step": 34815
    },
    {
      "epoch": 1.47,
      "grad_norm": 16.32247221820926,
      "learning_rate": 6.018715993435296e-06,
      "loss": 0.7359,
      "step": 34820
    },
    {
      "epoch": 1.47,
      "grad_norm": 13.79328730450873,
      "learning_rate": 6.017510658171453e-06,
      "loss": 0.7521,
      "step": 34825
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.808002527600671,
      "learning_rate": 6.016305261218936e-06,
      "loss": 0.7575,
      "step": 34830
    },
    {
      "epoch": 1.47,
      "grad_norm": 9.487535670880174,
      "learning_rate": 6.015099802650823e-06,
      "loss": 0.803,
      "step": 34835
    },
    {
      "epoch": 1.47,
      "grad_norm": 4.777230569040924,
      "learning_rate": 6.0138942825402e-06,
      "loss": 0.7478,
      "step": 34840
    },
    {
      "epoch": 1.47,
      "grad_norm": 7.67012680625771,
      "learning_rate": 6.012688700960154e-06,
      "loss": 0.7774,
      "step": 34845
    },
    {
      "epoch": 1.47,
      "grad_norm": 19.342245885094446,
      "learning_rate": 6.011483057983774e-06,
      "loss": 0.7549,
      "step": 34850
    },
    {
      "epoch": 1.48,
      "grad_norm": 26.812684259109933,
      "learning_rate": 6.010277353684156e-06,
      "loss": 0.7555,
      "step": 34855
    },
    {
      "epoch": 1.48,
      "grad_norm": 18.024609643387702,
      "learning_rate": 6.009071588134398e-06,
      "loss": 0.7879,
      "step": 34860
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.560421071687058,
      "learning_rate": 6.007865761407602e-06,
      "loss": 0.7976,
      "step": 34865
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.450221479476452,
      "learning_rate": 6.006659873576874e-06,
      "loss": 0.754,
      "step": 34870
    },
    {
      "epoch": 1.48,
      "grad_norm": 14.359205042335839,
      "learning_rate": 6.005453924715322e-06,
      "loss": 0.7661,
      "step": 34875
    },
    {
      "epoch": 1.48,
      "grad_norm": 10.690273356588177,
      "learning_rate": 6.004247914896062e-06,
      "loss": 0.7897,
      "step": 34880
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.1483865199357055,
      "learning_rate": 6.0030418441922076e-06,
      "loss": 0.7807,
      "step": 34885
    },
    {
      "epoch": 1.48,
      "grad_norm": 12.946443086295215,
      "learning_rate": 6.00183571267688e-06,
      "loss": 0.8033,
      "step": 34890
    },
    {
      "epoch": 1.48,
      "grad_norm": 18.642388402880393,
      "learning_rate": 6.000629520423206e-06,
      "loss": 0.7579,
      "step": 34895
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.727261823869453,
      "learning_rate": 5.999423267504309e-06,
      "loss": 0.7718,
      "step": 34900
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.157817298581853,
      "learning_rate": 5.998216953993327e-06,
      "loss": 0.7833,
      "step": 34905
    },
    {
      "epoch": 1.48,
      "grad_norm": 14.140947733949853,
      "learning_rate": 5.99701057996339e-06,
      "loss": 0.7482,
      "step": 34910
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.381456934094643,
      "learning_rate": 5.9958041454876355e-06,
      "loss": 0.8263,
      "step": 34915
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.905388882591197,
      "learning_rate": 5.9945976506392125e-06,
      "loss": 0.7652,
      "step": 34920
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.568367978560908,
      "learning_rate": 5.993391095491261e-06,
      "loss": 0.7709,
      "step": 34925
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.683460944199713,
      "learning_rate": 5.992184480116936e-06,
      "loss": 0.7457,
      "step": 34930
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.471657182779599,
      "learning_rate": 5.990977804589387e-06,
      "loss": 0.7851,
      "step": 34935
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.38249526274511,
      "learning_rate": 5.9897710689817725e-06,
      "loss": 0.8057,
      "step": 34940
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.210248195827969,
      "learning_rate": 5.988564273367254e-06,
      "loss": 0.7694,
      "step": 34945
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.663181419039169,
      "learning_rate": 5.987357417818997e-06,
      "loss": 0.7782,
      "step": 34950
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.47329177949332,
      "learning_rate": 5.986150502410166e-06,
      "loss": 0.7803,
      "step": 34955
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.061709888828411,
      "learning_rate": 5.984943527213937e-06,
      "loss": 0.8249,
      "step": 34960
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.429913146980737,
      "learning_rate": 5.98373649230348e-06,
      "loss": 0.7828,
      "step": 34965
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.04745980119588,
      "learning_rate": 5.9825293977519795e-06,
      "loss": 0.778,
      "step": 34970
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.11873690850023,
      "learning_rate": 5.981322243632614e-06,
      "loss": 0.7668,
      "step": 34975
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.6130197712584105,
      "learning_rate": 5.980115030018574e-06,
      "loss": 0.8051,
      "step": 34980
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.259833092708131,
      "learning_rate": 5.9789077569830455e-06,
      "loss": 0.7928,
      "step": 34985
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.543789940994861,
      "learning_rate": 5.977700424599222e-06,
      "loss": 0.7593,
      "step": 34990
    },
    {
      "epoch": 1.48,
      "grad_norm": 7.8286899833381955,
      "learning_rate": 5.976493032940303e-06,
      "loss": 0.7652,
      "step": 34995
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.972048611444596,
      "learning_rate": 5.975285582079486e-06,
      "loss": 0.7681,
      "step": 35000
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.319066407229352,
      "learning_rate": 5.97407807208998e-06,
      "loss": 0.7794,
      "step": 35005
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.7751126999389655,
      "learning_rate": 5.972870503044989e-06,
      "loss": 0.7534,
      "step": 35010
    },
    {
      "epoch": 1.48,
      "grad_norm": 10.447201816780197,
      "learning_rate": 5.971662875017723e-06,
      "loss": 0.766,
      "step": 35015
    },
    {
      "epoch": 1.48,
      "grad_norm": 6.4453953433613655,
      "learning_rate": 5.9704551880814e-06,
      "loss": 0.851,
      "step": 35020
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.963713827637875,
      "learning_rate": 5.969247442309238e-06,
      "loss": 0.7208,
      "step": 35025
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.378033749537752,
      "learning_rate": 5.968039637774459e-06,
      "loss": 0.7576,
      "step": 35030
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.343690921483835,
      "learning_rate": 5.9668317745502864e-06,
      "loss": 0.7738,
      "step": 35035
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.8343421442963574,
      "learning_rate": 5.965623852709953e-06,
      "loss": 0.8001,
      "step": 35040
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.718532206921225,
      "learning_rate": 5.96441587232669e-06,
      "loss": 0.7726,
      "step": 35045
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.1624921362561516,
      "learning_rate": 5.963207833473732e-06,
      "loss": 0.7792,
      "step": 35050
    },
    {
      "epoch": 1.48,
      "grad_norm": 11.661130437160107,
      "learning_rate": 5.961999736224321e-06,
      "loss": 0.7392,
      "step": 35055
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.733973181887682,
      "learning_rate": 5.960791580651699e-06,
      "loss": 0.793,
      "step": 35060
    },
    {
      "epoch": 1.48,
      "grad_norm": 5.627034840101632,
      "learning_rate": 5.959583366829114e-06,
      "loss": 0.7908,
      "step": 35065
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.341425503063352,
      "learning_rate": 5.958375094829817e-06,
      "loss": 0.7531,
      "step": 35070
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.120420558187537,
      "learning_rate": 5.957166764727059e-06,
      "loss": 0.7586,
      "step": 35075
    },
    {
      "epoch": 1.48,
      "grad_norm": 9.331085668020656,
      "learning_rate": 5.955958376594101e-06,
      "loss": 0.7536,
      "step": 35080
    },
    {
      "epoch": 1.48,
      "grad_norm": 8.018280502380685,
      "learning_rate": 5.954749930504201e-06,
      "loss": 0.7955,
      "step": 35085
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.438022521781885,
      "learning_rate": 5.953541426530627e-06,
      "loss": 0.7894,
      "step": 35090
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.0111579975110265,
      "learning_rate": 5.952332864746644e-06,
      "loss": 0.7839,
      "step": 35095
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.118333088525564,
      "learning_rate": 5.951124245225524e-06,
      "loss": 0.8123,
      "step": 35100
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.117332967827528,
      "learning_rate": 5.949915568040544e-06,
      "loss": 0.7557,
      "step": 35105
    },
    {
      "epoch": 1.49,
      "grad_norm": 13.600719817831894,
      "learning_rate": 5.948706833264981e-06,
      "loss": 0.7899,
      "step": 35110
    },
    {
      "epoch": 1.49,
      "grad_norm": 29.470359768645483,
      "learning_rate": 5.947498040972118e-06,
      "loss": 0.7932,
      "step": 35115
    },
    {
      "epoch": 1.49,
      "grad_norm": 31.794206009136563,
      "learning_rate": 5.946289191235236e-06,
      "loss": 0.7913,
      "step": 35120
    },
    {
      "epoch": 1.49,
      "grad_norm": 13.288278017831857,
      "learning_rate": 5.945080284127631e-06,
      "loss": 0.7705,
      "step": 35125
    },
    {
      "epoch": 1.49,
      "grad_norm": 19.42499878419053,
      "learning_rate": 5.943871319722592e-06,
      "loss": 0.8036,
      "step": 35130
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.994820360128343,
      "learning_rate": 5.9426622980934134e-06,
      "loss": 0.8007,
      "step": 35135
    },
    {
      "epoch": 1.49,
      "grad_norm": 10.017635858779661,
      "learning_rate": 5.9414532193133976e-06,
      "loss": 0.7784,
      "step": 35140
    },
    {
      "epoch": 1.49,
      "grad_norm": 13.203849381434013,
      "learning_rate": 5.940244083455846e-06,
      "loss": 0.7797,
      "step": 35145
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.848857003423155,
      "learning_rate": 5.939034890594065e-06,
      "loss": 0.7996,
      "step": 35150
    },
    {
      "epoch": 1.49,
      "grad_norm": 12.85879472578569,
      "learning_rate": 5.937825640801366e-06,
      "loss": 0.7635,
      "step": 35155
    },
    {
      "epoch": 1.49,
      "grad_norm": 20.381594119497006,
      "learning_rate": 5.936616334151059e-06,
      "loss": 0.7964,
      "step": 35160
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.358394741703964,
      "learning_rate": 5.935406970716464e-06,
      "loss": 0.7383,
      "step": 35165
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.735599362541418,
      "learning_rate": 5.934197550570898e-06,
      "loss": 0.7872,
      "step": 35170
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.846131333348263,
      "learning_rate": 5.932988073787687e-06,
      "loss": 0.745,
      "step": 35175
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.924324501919118,
      "learning_rate": 5.931778540440158e-06,
      "loss": 0.7518,
      "step": 35180
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.845810999800216,
      "learning_rate": 5.9305689506016406e-06,
      "loss": 0.7678,
      "step": 35185
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.601173023700071,
      "learning_rate": 5.92935930434547e-06,
      "loss": 0.7635,
      "step": 35190
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.8765654223981505,
      "learning_rate": 5.928149601744981e-06,
      "loss": 0.7924,
      "step": 35195
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.78094392533559,
      "learning_rate": 5.926939842873515e-06,
      "loss": 0.7401,
      "step": 35200
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.062577935214561,
      "learning_rate": 5.925730027804416e-06,
      "loss": 0.791,
      "step": 35205
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.121683566628302,
      "learning_rate": 5.924520156611035e-06,
      "loss": 0.7438,
      "step": 35210
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.52259579818607,
      "learning_rate": 5.923310229366718e-06,
      "loss": 0.7954,
      "step": 35215
    },
    {
      "epoch": 1.49,
      "grad_norm": 10.471378872273766,
      "learning_rate": 5.922100246144822e-06,
      "loss": 0.7706,
      "step": 35220
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.22188342090364,
      "learning_rate": 5.920890207018704e-06,
      "loss": 0.8041,
      "step": 35225
    },
    {
      "epoch": 1.49,
      "grad_norm": 4.522946450339393,
      "learning_rate": 5.919680112061725e-06,
      "loss": 0.7637,
      "step": 35230
    },
    {
      "epoch": 1.49,
      "grad_norm": 14.550441160637263,
      "learning_rate": 5.91846996134725e-06,
      "loss": 0.8375,
      "step": 35235
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.663077709231732,
      "learning_rate": 5.9172597549486485e-06,
      "loss": 0.7743,
      "step": 35240
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.256855744868679,
      "learning_rate": 5.916049492939288e-06,
      "loss": 0.7798,
      "step": 35245
    },
    {
      "epoch": 1.49,
      "grad_norm": 22.98557804025475,
      "learning_rate": 5.914839175392546e-06,
      "loss": 0.7192,
      "step": 35250
    },
    {
      "epoch": 1.49,
      "grad_norm": 12.068742861569723,
      "learning_rate": 5.913628802381799e-06,
      "loss": 0.8207,
      "step": 35255
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.117032859502086,
      "learning_rate": 5.912418373980429e-06,
      "loss": 0.7635,
      "step": 35260
    },
    {
      "epoch": 1.49,
      "grad_norm": 12.794649520892104,
      "learning_rate": 5.911207890261821e-06,
      "loss": 0.7625,
      "step": 35265
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.070369730109931,
      "learning_rate": 5.909997351299361e-06,
      "loss": 0.7956,
      "step": 35270
    },
    {
      "epoch": 1.49,
      "grad_norm": 9.570821767512408,
      "learning_rate": 5.908786757166445e-06,
      "loss": 0.8233,
      "step": 35275
    },
    {
      "epoch": 1.49,
      "grad_norm": 14.982739472421656,
      "learning_rate": 5.9075761079364625e-06,
      "loss": 0.7684,
      "step": 35280
    },
    {
      "epoch": 1.49,
      "grad_norm": 8.881102930158718,
      "learning_rate": 5.906365403682815e-06,
      "loss": 0.7595,
      "step": 35285
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.056373961808256,
      "learning_rate": 5.905154644478903e-06,
      "loss": 0.7726,
      "step": 35290
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.423977334871507,
      "learning_rate": 5.90394383039813e-06,
      "loss": 0.8075,
      "step": 35295
    },
    {
      "epoch": 1.49,
      "grad_norm": 5.092197116195112,
      "learning_rate": 5.902732961513906e-06,
      "loss": 0.7682,
      "step": 35300
    },
    {
      "epoch": 1.49,
      "grad_norm": 10.553450175873829,
      "learning_rate": 5.901522037899641e-06,
      "loss": 0.75,
      "step": 35305
    },
    {
      "epoch": 1.49,
      "grad_norm": 12.527406798476562,
      "learning_rate": 5.900311059628751e-06,
      "loss": 0.7762,
      "step": 35310
    },
    {
      "epoch": 1.49,
      "grad_norm": 7.113472707145592,
      "learning_rate": 5.899100026774654e-06,
      "loss": 0.7522,
      "step": 35315
    },
    {
      "epoch": 1.49,
      "grad_norm": 6.619710956856185,
      "learning_rate": 5.897888939410768e-06,
      "loss": 0.7874,
      "step": 35320
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.0232321403466615,
      "learning_rate": 5.896677797610524e-06,
      "loss": 0.791,
      "step": 35325
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.310499922417376,
      "learning_rate": 5.8954666014473435e-06,
      "loss": 0.7619,
      "step": 35330
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.281310666375591,
      "learning_rate": 5.894255350994663e-06,
      "loss": 0.8334,
      "step": 35335
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.293957115605188,
      "learning_rate": 5.893044046325915e-06,
      "loss": 0.805,
      "step": 35340
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.335017674432076,
      "learning_rate": 5.891832687514534e-06,
      "loss": 0.7716,
      "step": 35345
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.061030441380728,
      "learning_rate": 5.890621274633967e-06,
      "loss": 0.7489,
      "step": 35350
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.607128326135727,
      "learning_rate": 5.889409807757654e-06,
      "loss": 0.8118,
      "step": 35355
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.122108985717882,
      "learning_rate": 5.888198286959045e-06,
      "loss": 0.7844,
      "step": 35360
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.0880148218002486,
      "learning_rate": 5.8869867123115885e-06,
      "loss": 0.7751,
      "step": 35365
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.644669976822856,
      "learning_rate": 5.885775083888742e-06,
      "loss": 0.7638,
      "step": 35370
    },
    {
      "epoch": 1.5,
      "grad_norm": 12.381372015891664,
      "learning_rate": 5.884563401763961e-06,
      "loss": 0.7805,
      "step": 35375
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.768665337631596,
      "learning_rate": 5.883351666010706e-06,
      "loss": 0.7989,
      "step": 35380
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.557053215105897,
      "learning_rate": 5.882139876702442e-06,
      "loss": 0.7579,
      "step": 35385
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.783536033212442,
      "learning_rate": 5.880928033912634e-06,
      "loss": 0.8017,
      "step": 35390
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.6395420550262765,
      "learning_rate": 5.879716137714757e-06,
      "loss": 0.7291,
      "step": 35395
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.080613746063198,
      "learning_rate": 5.87850418818228e-06,
      "loss": 0.7793,
      "step": 35400
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.146536356446158,
      "learning_rate": 5.877292185388682e-06,
      "loss": 0.7679,
      "step": 35405
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.345736457139933,
      "learning_rate": 5.876080129407445e-06,
      "loss": 0.7139,
      "step": 35410
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.9984387349237687,
      "learning_rate": 5.874868020312047e-06,
      "loss": 0.7595,
      "step": 35415
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.89020555453679,
      "learning_rate": 5.8736558581759795e-06,
      "loss": 0.7791,
      "step": 35420
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.153144115118243,
      "learning_rate": 5.872443643072731e-06,
      "loss": 0.7639,
      "step": 35425
    },
    {
      "epoch": 1.5,
      "grad_norm": 10.81834548691463,
      "learning_rate": 5.871231375075793e-06,
      "loss": 0.8271,
      "step": 35430
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.401384550286797,
      "learning_rate": 5.8700190542586665e-06,
      "loss": 0.7826,
      "step": 35435
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.30012910191852,
      "learning_rate": 5.868806680694844e-06,
      "loss": 0.7556,
      "step": 35440
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.303222478019857,
      "learning_rate": 5.867594254457833e-06,
      "loss": 0.8094,
      "step": 35445
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.143857861405899,
      "learning_rate": 5.866381775621137e-06,
      "loss": 0.7679,
      "step": 35450
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.995035463516286,
      "learning_rate": 5.865169244258266e-06,
      "loss": 0.7654,
      "step": 35455
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.585811920156551,
      "learning_rate": 5.863956660442733e-06,
      "loss": 0.7812,
      "step": 35460
    },
    {
      "epoch": 1.5,
      "grad_norm": 7.083215085815355,
      "learning_rate": 5.862744024248053e-06,
      "loss": 0.7431,
      "step": 35465
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.60419105595136,
      "learning_rate": 5.861531335747743e-06,
      "loss": 0.766,
      "step": 35470
    },
    {
      "epoch": 1.5,
      "grad_norm": 9.882406985262726,
      "learning_rate": 5.8603185950153265e-06,
      "loss": 0.7674,
      "step": 35475
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.657436157098296,
      "learning_rate": 5.8591058021243275e-06,
      "loss": 0.8046,
      "step": 35480
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.455621908503985,
      "learning_rate": 5.857892957148275e-06,
      "loss": 0.736,
      "step": 35485
    },
    {
      "epoch": 1.5,
      "grad_norm": 12.332551964529308,
      "learning_rate": 5.856680060160698e-06,
      "loss": 0.7621,
      "step": 35490
    },
    {
      "epoch": 1.5,
      "grad_norm": 19.025869716264832,
      "learning_rate": 5.855467111235133e-06,
      "loss": 0.7743,
      "step": 35495
    },
    {
      "epoch": 1.5,
      "grad_norm": 16.007599014024574,
      "learning_rate": 5.8542541104451164e-06,
      "loss": 0.7752,
      "step": 35500
    },
    {
      "epoch": 1.5,
      "grad_norm": 14.225709600333296,
      "learning_rate": 5.85304105786419e-06,
      "loss": 0.7487,
      "step": 35505
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.745927281763014,
      "learning_rate": 5.851827953565897e-06,
      "loss": 0.7976,
      "step": 35510
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.112620909188447,
      "learning_rate": 5.8506147976237844e-06,
      "loss": 0.8209,
      "step": 35515
    },
    {
      "epoch": 1.5,
      "grad_norm": 6.796835108917794,
      "learning_rate": 5.849401590111403e-06,
      "loss": 0.7245,
      "step": 35520
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.365738500065053,
      "learning_rate": 5.848188331102303e-06,
      "loss": 0.7414,
      "step": 35525
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.127402861085467,
      "learning_rate": 5.846975020670045e-06,
      "loss": 0.7891,
      "step": 35530
    },
    {
      "epoch": 1.5,
      "grad_norm": 9.235317738855576,
      "learning_rate": 5.845761658888186e-06,
      "loss": 0.7956,
      "step": 35535
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.917507279621178,
      "learning_rate": 5.844548245830288e-06,
      "loss": 0.7963,
      "step": 35540
    },
    {
      "epoch": 1.5,
      "grad_norm": 13.690374060051681,
      "learning_rate": 5.843334781569918e-06,
      "loss": 0.7975,
      "step": 35545
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.480398716800464,
      "learning_rate": 5.8421212661806435e-06,
      "loss": 0.7493,
      "step": 35550
    },
    {
      "epoch": 1.5,
      "grad_norm": 8.442052432991687,
      "learning_rate": 5.840907699736039e-06,
      "loss": 0.7961,
      "step": 35555
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.520735819421966,
      "learning_rate": 5.8396940823096775e-06,
      "loss": 0.7847,
      "step": 35560
    },
    {
      "epoch": 1.51,
      "grad_norm": 27.124322271049827,
      "learning_rate": 5.838480413975135e-06,
      "loss": 0.7678,
      "step": 35565
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.384877079445805,
      "learning_rate": 5.837266694805996e-06,
      "loss": 0.762,
      "step": 35570
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.25118319333567,
      "learning_rate": 5.836052924875844e-06,
      "loss": 0.8021,
      "step": 35575
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.48196828136772,
      "learning_rate": 5.834839104258265e-06,
      "loss": 0.7533,
      "step": 35580
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.08614152008201,
      "learning_rate": 5.833625233026852e-06,
      "loss": 0.7676,
      "step": 35585
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.2639469462546495,
      "learning_rate": 5.8324113112551936e-06,
      "loss": 0.7847,
      "step": 35590
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.0812249878379125,
      "learning_rate": 5.831197339016893e-06,
      "loss": 0.7479,
      "step": 35595
    },
    {
      "epoch": 1.51,
      "grad_norm": 10.816922415291735,
      "learning_rate": 5.829983316385543e-06,
      "loss": 0.7472,
      "step": 35600
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.21452289376387,
      "learning_rate": 5.828769243434752e-06,
      "loss": 0.7534,
      "step": 35605
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.2607972272546375,
      "learning_rate": 5.827555120238121e-06,
      "loss": 0.7879,
      "step": 35610
    },
    {
      "epoch": 1.51,
      "grad_norm": 9.931020446241725,
      "learning_rate": 5.826340946869261e-06,
      "loss": 0.7352,
      "step": 35615
    },
    {
      "epoch": 1.51,
      "grad_norm": 10.8032613008011,
      "learning_rate": 5.825126723401784e-06,
      "loss": 0.7646,
      "step": 35620
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.782401302081939,
      "learning_rate": 5.823912449909303e-06,
      "loss": 0.7733,
      "step": 35625
    },
    {
      "epoch": 1.51,
      "grad_norm": 11.593323104936328,
      "learning_rate": 5.822698126465439e-06,
      "loss": 0.7898,
      "step": 35630
    },
    {
      "epoch": 1.51,
      "grad_norm": 14.355576140796922,
      "learning_rate": 5.821483753143809e-06,
      "loss": 0.8175,
      "step": 35635
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.388751730702089,
      "learning_rate": 5.8202693300180404e-06,
      "loss": 0.7494,
      "step": 35640
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.34556255886083,
      "learning_rate": 5.819054857161758e-06,
      "loss": 0.8051,
      "step": 35645
    },
    {
      "epoch": 1.51,
      "grad_norm": 15.569846832882002,
      "learning_rate": 5.817840334648591e-06,
      "loss": 0.7739,
      "step": 35650
    },
    {
      "epoch": 1.51,
      "grad_norm": 12.183574664236257,
      "learning_rate": 5.816625762552175e-06,
      "loss": 0.7563,
      "step": 35655
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.893357106700686,
      "learning_rate": 5.815411140946143e-06,
      "loss": 0.7582,
      "step": 35660
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.104639409493043,
      "learning_rate": 5.814196469904137e-06,
      "loss": 0.8255,
      "step": 35665
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.433111627799949,
      "learning_rate": 5.8129817494997976e-06,
      "loss": 0.8104,
      "step": 35670
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.42666332375167,
      "learning_rate": 5.811766979806768e-06,
      "loss": 0.7436,
      "step": 35675
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.9008503954022595,
      "learning_rate": 5.8105521608987e-06,
      "loss": 0.7785,
      "step": 35680
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.641816750575053,
      "learning_rate": 5.80933729284924e-06,
      "loss": 0.752,
      "step": 35685
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.415346017358438,
      "learning_rate": 5.808122375732044e-06,
      "loss": 0.7697,
      "step": 35690
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.149767847667662,
      "learning_rate": 5.806907409620768e-06,
      "loss": 0.7944,
      "step": 35695
    },
    {
      "epoch": 1.51,
      "grad_norm": 10.34104419587703,
      "learning_rate": 5.8056923945890745e-06,
      "loss": 0.7603,
      "step": 35700
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.582477482951482,
      "learning_rate": 5.804477330710623e-06,
      "loss": 0.7386,
      "step": 35705
    },
    {
      "epoch": 1.51,
      "grad_norm": 15.042520545906932,
      "learning_rate": 5.8032622180590805e-06,
      "loss": 0.8398,
      "step": 35710
    },
    {
      "epoch": 1.51,
      "grad_norm": 22.80175122182459,
      "learning_rate": 5.802047056708116e-06,
      "loss": 0.7194,
      "step": 35715
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.541306011039479,
      "learning_rate": 5.800831846731402e-06,
      "loss": 0.8017,
      "step": 35720
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.8057041667914,
      "learning_rate": 5.799616588202611e-06,
      "loss": 0.7824,
      "step": 35725
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.887167836404786,
      "learning_rate": 5.7984012811954225e-06,
      "loss": 0.7633,
      "step": 35730
    },
    {
      "epoch": 1.51,
      "grad_norm": 8.210730717948925,
      "learning_rate": 5.797185925783515e-06,
      "loss": 0.8119,
      "step": 35735
    },
    {
      "epoch": 1.51,
      "grad_norm": 17.731726587136,
      "learning_rate": 5.795970522040573e-06,
      "loss": 0.7673,
      "step": 35740
    },
    {
      "epoch": 1.51,
      "grad_norm": 16.73682256121197,
      "learning_rate": 5.794755070040283e-06,
      "loss": 0.7242,
      "step": 35745
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.269560649304046,
      "learning_rate": 5.793539569856335e-06,
      "loss": 0.7915,
      "step": 35750
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.779977761564655,
      "learning_rate": 5.79232402156242e-06,
      "loss": 0.7668,
      "step": 35755
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.117695598947614,
      "learning_rate": 5.7911084252322334e-06,
      "loss": 0.784,
      "step": 35760
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.387822661169172,
      "learning_rate": 5.789892780939475e-06,
      "loss": 0.784,
      "step": 35765
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.720664706631668,
      "learning_rate": 5.7886770887578404e-06,
      "loss": 0.7507,
      "step": 35770
    },
    {
      "epoch": 1.51,
      "grad_norm": 5.073725072186133,
      "learning_rate": 5.787461348761039e-06,
      "loss": 0.7379,
      "step": 35775
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.456495687757091,
      "learning_rate": 5.786245561022774e-06,
      "loss": 0.7745,
      "step": 35780
    },
    {
      "epoch": 1.51,
      "grad_norm": 7.007717042346071,
      "learning_rate": 5.785029725616758e-06,
      "loss": 0.7611,
      "step": 35785
    },
    {
      "epoch": 1.51,
      "grad_norm": 6.078397353446753,
      "learning_rate": 5.783813842616703e-06,
      "loss": 0.7829,
      "step": 35790
    },
    {
      "epoch": 1.51,
      "grad_norm": 4.601490639695328,
      "learning_rate": 5.782597912096322e-06,
      "loss": 0.7485,
      "step": 35795
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.6023290843292015,
      "learning_rate": 5.781381934129335e-06,
      "loss": 0.769,
      "step": 35800
    },
    {
      "epoch": 1.52,
      "grad_norm": 14.11802200120585,
      "learning_rate": 5.780165908789463e-06,
      "loss": 0.7777,
      "step": 35805
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.593952283561465,
      "learning_rate": 5.778949836150431e-06,
      "loss": 0.7263,
      "step": 35810
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.867453009854076,
      "learning_rate": 5.777733716285965e-06,
      "loss": 0.7693,
      "step": 35815
    },
    {
      "epoch": 1.52,
      "grad_norm": 16.354286632613253,
      "learning_rate": 5.776517549269793e-06,
      "loss": 0.7407,
      "step": 35820
    },
    {
      "epoch": 1.52,
      "grad_norm": 52.873242025649986,
      "learning_rate": 5.775301335175651e-06,
      "loss": 0.7625,
      "step": 35825
    },
    {
      "epoch": 1.52,
      "grad_norm": 13.794491435493061,
      "learning_rate": 5.7740850740772725e-06,
      "loss": 0.7685,
      "step": 35830
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.796282872941366,
      "learning_rate": 5.772868766048394e-06,
      "loss": 0.7556,
      "step": 35835
    },
    {
      "epoch": 1.52,
      "grad_norm": 14.50886051152214,
      "learning_rate": 5.771652411162763e-06,
      "loss": 0.7381,
      "step": 35840
    },
    {
      "epoch": 1.52,
      "grad_norm": 13.5102747545092,
      "learning_rate": 5.7704360094941145e-06,
      "loss": 0.7293,
      "step": 35845
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.125003338167975,
      "learning_rate": 5.7692195611162035e-06,
      "loss": 0.7566,
      "step": 35850
    },
    {
      "epoch": 1.52,
      "grad_norm": 26.973032001743942,
      "learning_rate": 5.768003066102774e-06,
      "loss": 0.7832,
      "step": 35855
    },
    {
      "epoch": 1.52,
      "grad_norm": 9.180360226734198,
      "learning_rate": 5.766786524527582e-06,
      "loss": 0.793,
      "step": 35860
    },
    {
      "epoch": 1.52,
      "grad_norm": 10.830977219195862,
      "learning_rate": 5.765569936464382e-06,
      "loss": 0.7358,
      "step": 35865
    },
    {
      "epoch": 1.52,
      "grad_norm": 12.52365519570048,
      "learning_rate": 5.764353301986932e-06,
      "loss": 0.7772,
      "step": 35870
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.343703416531583,
      "learning_rate": 5.763136621168992e-06,
      "loss": 0.7344,
      "step": 35875
    },
    {
      "epoch": 1.52,
      "grad_norm": 19.158814711821826,
      "learning_rate": 5.761919894084326e-06,
      "loss": 0.729,
      "step": 35880
    },
    {
      "epoch": 1.52,
      "grad_norm": 13.10645471797121,
      "learning_rate": 5.760703120806702e-06,
      "loss": 0.7296,
      "step": 35885
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.48759836142084,
      "learning_rate": 5.759486301409889e-06,
      "loss": 0.769,
      "step": 35890
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.086018467180757,
      "learning_rate": 5.758269435967657e-06,
      "loss": 0.7409,
      "step": 35895
    },
    {
      "epoch": 1.52,
      "grad_norm": 8.176159238151056,
      "learning_rate": 5.757052524553785e-06,
      "loss": 0.7767,
      "step": 35900
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.629734405044549,
      "learning_rate": 5.7558355672420456e-06,
      "loss": 0.738,
      "step": 35905
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.754163772768,
      "learning_rate": 5.754618564106225e-06,
      "loss": 0.7705,
      "step": 35910
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.227100604205821,
      "learning_rate": 5.753401515220102e-06,
      "loss": 0.725,
      "step": 35915
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.697862327769576,
      "learning_rate": 5.752184420657465e-06,
      "loss": 0.7598,
      "step": 35920
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.942626067055615,
      "learning_rate": 5.750967280492103e-06,
      "loss": 0.7377,
      "step": 35925
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.47493741978674,
      "learning_rate": 5.749750094797804e-06,
      "loss": 0.7357,
      "step": 35930
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.619366415417316,
      "learning_rate": 5.748532863648367e-06,
      "loss": 0.7916,
      "step": 35935
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.49675906704056,
      "learning_rate": 5.747315587117587e-06,
      "loss": 0.7139,
      "step": 35940
    },
    {
      "epoch": 1.52,
      "grad_norm": 12.666261080625173,
      "learning_rate": 5.746098265279265e-06,
      "loss": 0.7243,
      "step": 35945
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.536966332918606,
      "learning_rate": 5.744880898207203e-06,
      "loss": 0.7425,
      "step": 35950
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.9274467738400585,
      "learning_rate": 5.743663485975205e-06,
      "loss": 0.7964,
      "step": 35955
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.153674004720579,
      "learning_rate": 5.7424460286570815e-06,
      "loss": 0.7697,
      "step": 35960
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.330447631159754,
      "learning_rate": 5.741228526326641e-06,
      "loss": 0.767,
      "step": 35965
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.588202054051127,
      "learning_rate": 5.740010979057699e-06,
      "loss": 0.7844,
      "step": 35970
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.534273914711299,
      "learning_rate": 5.738793386924072e-06,
      "loss": 0.781,
      "step": 35975
    },
    {
      "epoch": 1.52,
      "grad_norm": 6.95904690094936,
      "learning_rate": 5.737575749999577e-06,
      "loss": 0.7591,
      "step": 35980
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.895528455288323,
      "learning_rate": 5.736358068358036e-06,
      "loss": 0.7099,
      "step": 35985
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.243089113684096,
      "learning_rate": 5.7351403420732755e-06,
      "loss": 0.7826,
      "step": 35990
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.617223246735131,
      "learning_rate": 5.7339225712191224e-06,
      "loss": 0.7343,
      "step": 35995
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.8746207313840575,
      "learning_rate": 5.7327047558694045e-06,
      "loss": 0.7528,
      "step": 36000
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.937826515880196,
      "learning_rate": 5.731486896097956e-06,
      "loss": 0.7682,
      "step": 36005
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.821692722242406,
      "learning_rate": 5.730268991978613e-06,
      "loss": 0.7982,
      "step": 36010
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.497330473162316,
      "learning_rate": 5.7290510435852094e-06,
      "loss": 0.7536,
      "step": 36015
    },
    {
      "epoch": 1.52,
      "grad_norm": 7.593124973338194,
      "learning_rate": 5.72783305099159e-06,
      "loss": 0.716,
      "step": 36020
    },
    {
      "epoch": 1.52,
      "grad_norm": 5.921043953278889,
      "learning_rate": 5.726615014271595e-06,
      "loss": 0.7794,
      "step": 36025
    },
    {
      "epoch": 1.52,
      "grad_norm": 4.162342563214488,
      "learning_rate": 5.725396933499074e-06,
      "loss": 0.6628,
      "step": 36030
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.547397242211524,
      "learning_rate": 5.724178808747874e-06,
      "loss": 0.7536,
      "step": 36035
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.16295011354217,
      "learning_rate": 5.722960640091843e-06,
      "loss": 0.7627,
      "step": 36040
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.8016290002261535,
      "learning_rate": 5.7217424276048405e-06,
      "loss": 0.7826,
      "step": 36045
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.762170826235397,
      "learning_rate": 5.720524171360719e-06,
      "loss": 0.7415,
      "step": 36050
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.531002839286745,
      "learning_rate": 5.719305871433341e-06,
      "loss": 0.723,
      "step": 36055
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.353437384591896,
      "learning_rate": 5.718087527896566e-06,
      "loss": 0.7358,
      "step": 36060
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.219340645320895,
      "learning_rate": 5.716869140824259e-06,
      "loss": 0.7416,
      "step": 36065
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.904330943635854,
      "learning_rate": 5.715650710290288e-06,
      "loss": 0.7453,
      "step": 36070
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.680002927586169,
      "learning_rate": 5.714432236368522e-06,
      "loss": 0.7335,
      "step": 36075
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.221813272055969,
      "learning_rate": 5.713213719132835e-06,
      "loss": 0.7351,
      "step": 36080
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.582071266389329,
      "learning_rate": 5.711995158657101e-06,
      "loss": 0.7845,
      "step": 36085
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.50926529727831,
      "learning_rate": 5.7107765550151985e-06,
      "loss": 0.7748,
      "step": 36090
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.414969406458174,
      "learning_rate": 5.709557908281006e-06,
      "loss": 0.7819,
      "step": 36095
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.354600689733497,
      "learning_rate": 5.708339218528406e-06,
      "loss": 0.7845,
      "step": 36100
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.025808887933558,
      "learning_rate": 5.707120485831288e-06,
      "loss": 0.7012,
      "step": 36105
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.756287376701307,
      "learning_rate": 5.705901710263537e-06,
      "loss": 0.7851,
      "step": 36110
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.827886235046456,
      "learning_rate": 5.704682891899044e-06,
      "loss": 0.7235,
      "step": 36115
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.668113011226178,
      "learning_rate": 5.703464030811704e-06,
      "loss": 0.7527,
      "step": 36120
    },
    {
      "epoch": 1.53,
      "grad_norm": 10.59265568040324,
      "learning_rate": 5.702245127075411e-06,
      "loss": 0.7311,
      "step": 36125
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.146470669806949,
      "learning_rate": 5.701026180764065e-06,
      "loss": 0.7851,
      "step": 36130
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.832302421405683,
      "learning_rate": 5.699807191951567e-06,
      "loss": 0.7738,
      "step": 36135
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.115130581117767,
      "learning_rate": 5.69858816071182e-06,
      "loss": 0.7432,
      "step": 36140
    },
    {
      "epoch": 1.53,
      "grad_norm": 10.929869080720659,
      "learning_rate": 5.69736908711873e-06,
      "loss": 0.7788,
      "step": 36145
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.4237374532795695,
      "learning_rate": 5.696149971246207e-06,
      "loss": 0.7369,
      "step": 36150
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.821355642582877,
      "learning_rate": 5.694930813168162e-06,
      "loss": 0.7537,
      "step": 36155
    },
    {
      "epoch": 1.53,
      "grad_norm": 11.189306895025428,
      "learning_rate": 5.693711612958508e-06,
      "loss": 0.767,
      "step": 36160
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.78248144737292,
      "learning_rate": 5.692492370691163e-06,
      "loss": 0.7862,
      "step": 36165
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.426321763742699,
      "learning_rate": 5.6912730864400455e-06,
      "loss": 0.7457,
      "step": 36170
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.681999734820031,
      "learning_rate": 5.690053760279076e-06,
      "loss": 0.7556,
      "step": 36175
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.498484256595738,
      "learning_rate": 5.688834392282181e-06,
      "loss": 0.7346,
      "step": 36180
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.013656962446652,
      "learning_rate": 5.687614982523284e-06,
      "loss": 0.742,
      "step": 36185
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.054359516641117,
      "learning_rate": 5.686395531076317e-06,
      "loss": 0.7819,
      "step": 36190
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.035879832579504,
      "learning_rate": 5.685176038015212e-06,
      "loss": 0.7413,
      "step": 36195
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.363547220597939,
      "learning_rate": 5.6839565034139e-06,
      "loss": 0.7503,
      "step": 36200
    },
    {
      "epoch": 1.53,
      "grad_norm": 6.8779848928246485,
      "learning_rate": 5.682736927346322e-06,
      "loss": 0.7763,
      "step": 36205
    },
    {
      "epoch": 1.53,
      "grad_norm": 10.641753574126716,
      "learning_rate": 5.681517309886413e-06,
      "loss": 0.7721,
      "step": 36210
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.569354750994023,
      "learning_rate": 5.680297651108118e-06,
      "loss": 0.7398,
      "step": 36215
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.704886319648279,
      "learning_rate": 5.6790779510853815e-06,
      "loss": 0.7071,
      "step": 36220
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.642769271909682,
      "learning_rate": 5.6778582098921466e-06,
      "loss": 0.7392,
      "step": 36225
    },
    {
      "epoch": 1.53,
      "grad_norm": 10.188283644795725,
      "learning_rate": 5.676638427602366e-06,
      "loss": 0.7401,
      "step": 36230
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.356340993264226,
      "learning_rate": 5.67541860428999e-06,
      "loss": 0.7584,
      "step": 36235
    },
    {
      "epoch": 1.53,
      "grad_norm": 12.863059147962543,
      "learning_rate": 5.674198740028977e-06,
      "loss": 0.782,
      "step": 36240
    },
    {
      "epoch": 1.53,
      "grad_norm": 9.285215615121325,
      "learning_rate": 5.672978834893278e-06,
      "loss": 0.8201,
      "step": 36245
    },
    {
      "epoch": 1.53,
      "grad_norm": 8.05121558073283,
      "learning_rate": 5.671758888956855e-06,
      "loss": 0.7479,
      "step": 36250
    },
    {
      "epoch": 1.53,
      "grad_norm": 7.038312854881935,
      "learning_rate": 5.670538902293669e-06,
      "loss": 0.7801,
      "step": 36255
    },
    {
      "epoch": 1.53,
      "grad_norm": 4.859644511209882,
      "learning_rate": 5.669318874977683e-06,
      "loss": 0.7252,
      "step": 36260
    },
    {
      "epoch": 1.53,
      "grad_norm": 5.309271617820732,
      "learning_rate": 5.6680988070828665e-06,
      "loss": 0.7442,
      "step": 36265
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.99130699160241,
      "learning_rate": 5.666878698683187e-06,
      "loss": 0.8292,
      "step": 36270
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.081780629673708,
      "learning_rate": 5.6656585498526176e-06,
      "loss": 0.7489,
      "step": 36275
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.794065489779067,
      "learning_rate": 5.66443836066513e-06,
      "loss": 0.7764,
      "step": 36280
    },
    {
      "epoch": 1.54,
      "grad_norm": 16.977509881643627,
      "learning_rate": 5.663218131194701e-06,
      "loss": 0.7875,
      "step": 36285
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.144576665501161,
      "learning_rate": 5.661997861515312e-06,
      "loss": 0.7724,
      "step": 36290
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.670751785886699,
      "learning_rate": 5.66077755170094e-06,
      "loss": 0.7555,
      "step": 36295
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.247671311920664,
      "learning_rate": 5.659557201825574e-06,
      "loss": 0.7446,
      "step": 36300
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.192342486706729,
      "learning_rate": 5.658336811963196e-06,
      "loss": 0.7134,
      "step": 36305
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.815642519282774,
      "learning_rate": 5.657116382187797e-06,
      "loss": 0.7561,
      "step": 36310
    },
    {
      "epoch": 1.54,
      "grad_norm": 10.224115556171064,
      "learning_rate": 5.655895912573368e-06,
      "loss": 0.7362,
      "step": 36315
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.264816330795972,
      "learning_rate": 5.6546754031939e-06,
      "loss": 0.7703,
      "step": 36320
    },
    {
      "epoch": 1.54,
      "grad_norm": 10.4430298561932,
      "learning_rate": 5.653454854123393e-06,
      "loss": 0.7336,
      "step": 36325
    },
    {
      "epoch": 1.54,
      "grad_norm": 14.418816036880918,
      "learning_rate": 5.652234265435842e-06,
      "loss": 0.7778,
      "step": 36330
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.244555476837727,
      "learning_rate": 5.65101363720525e-06,
      "loss": 0.7436,
      "step": 36335
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.042267677053026,
      "learning_rate": 5.649792969505617e-06,
      "loss": 0.7758,
      "step": 36340
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.6970479191015375,
      "learning_rate": 5.64857226241095e-06,
      "loss": 0.7654,
      "step": 36345
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.744264625884218,
      "learning_rate": 5.6473515159952585e-06,
      "loss": 0.7713,
      "step": 36350
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.432690146891411,
      "learning_rate": 5.64613073033255e-06,
      "loss": 0.7379,
      "step": 36355
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.916425821797883,
      "learning_rate": 5.64490990549684e-06,
      "loss": 0.7628,
      "step": 36360
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.901548770114443,
      "learning_rate": 5.643689041562141e-06,
      "loss": 0.7758,
      "step": 36365
    },
    {
      "epoch": 1.54,
      "grad_norm": 11.545978616397411,
      "learning_rate": 5.642468138602472e-06,
      "loss": 0.7543,
      "step": 36370
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.766205097303478,
      "learning_rate": 5.641247196691852e-06,
      "loss": 0.7655,
      "step": 36375
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.7761349349455555,
      "learning_rate": 5.640026215904303e-06,
      "loss": 0.7471,
      "step": 36380
    },
    {
      "epoch": 1.54,
      "grad_norm": 9.039754176101685,
      "learning_rate": 5.63880519631385e-06,
      "loss": 0.7601,
      "step": 36385
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.167325564652919,
      "learning_rate": 5.63758413799452e-06,
      "loss": 0.7086,
      "step": 36390
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.641494168042841,
      "learning_rate": 5.6363630410203415e-06,
      "loss": 0.7199,
      "step": 36395
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.796779792245474,
      "learning_rate": 5.635141905465347e-06,
      "loss": 0.7812,
      "step": 36400
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.980531616127717,
      "learning_rate": 5.633920731403569e-06,
      "loss": 0.7511,
      "step": 36405
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.173917776880962,
      "learning_rate": 5.632699518909045e-06,
      "loss": 0.7786,
      "step": 36410
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.0748340996694505,
      "learning_rate": 5.631478268055812e-06,
      "loss": 0.7422,
      "step": 36415
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.531527906834546,
      "learning_rate": 5.6302569789179115e-06,
      "loss": 0.7267,
      "step": 36420
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.421548362470831,
      "learning_rate": 5.629035651569388e-06,
      "loss": 0.7809,
      "step": 36425
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.371329999365018,
      "learning_rate": 5.627814286084284e-06,
      "loss": 0.7634,
      "step": 36430
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.591503494981009,
      "learning_rate": 5.626592882536651e-06,
      "loss": 0.7816,
      "step": 36435
    },
    {
      "epoch": 1.54,
      "grad_norm": 18.435414824542768,
      "learning_rate": 5.625371441000537e-06,
      "loss": 0.7811,
      "step": 36440
    },
    {
      "epoch": 1.54,
      "grad_norm": 10.143455007023272,
      "learning_rate": 5.624149961549992e-06,
      "loss": 0.7538,
      "step": 36445
    },
    {
      "epoch": 1.54,
      "grad_norm": 14.285501295528324,
      "learning_rate": 5.622928444259077e-06,
      "loss": 0.7921,
      "step": 36450
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.233101724093227,
      "learning_rate": 5.6217068892018435e-06,
      "loss": 0.7432,
      "step": 36455
    },
    {
      "epoch": 1.54,
      "grad_norm": 5.852665172259824,
      "learning_rate": 5.620485296452353e-06,
      "loss": 0.8275,
      "step": 36460
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.80568451354192,
      "learning_rate": 5.619263666084667e-06,
      "loss": 0.766,
      "step": 36465
    },
    {
      "epoch": 1.54,
      "grad_norm": 7.240333718505863,
      "learning_rate": 5.618041998172848e-06,
      "loss": 0.778,
      "step": 36470
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.8567276212325945,
      "learning_rate": 5.616820292790963e-06,
      "loss": 0.7552,
      "step": 36475
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.710169902749028,
      "learning_rate": 5.615598550013082e-06,
      "loss": 0.7824,
      "step": 36480
    },
    {
      "epoch": 1.54,
      "grad_norm": 11.247173380305359,
      "learning_rate": 5.614376769913274e-06,
      "loss": 0.7802,
      "step": 36485
    },
    {
      "epoch": 1.54,
      "grad_norm": 8.039763231784095,
      "learning_rate": 5.6131549525656104e-06,
      "loss": 0.7764,
      "step": 36490
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.564663446305456,
      "learning_rate": 5.61193309804417e-06,
      "loss": 0.8137,
      "step": 36495
    },
    {
      "epoch": 1.54,
      "grad_norm": 4.336584212915461,
      "learning_rate": 5.610711206423028e-06,
      "loss": 0.7739,
      "step": 36500
    },
    {
      "epoch": 1.54,
      "grad_norm": 6.234405442418801,
      "learning_rate": 5.6094892777762636e-06,
      "loss": 0.768,
      "step": 36505
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.969296988930027,
      "learning_rate": 5.608267312177961e-06,
      "loss": 0.7673,
      "step": 36510
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.214312746362722,
      "learning_rate": 5.607045309702201e-06,
      "loss": 0.7446,
      "step": 36515
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.482227268583426,
      "learning_rate": 5.605823270423074e-06,
      "loss": 0.7361,
      "step": 36520
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.960761888692762,
      "learning_rate": 5.604601194414666e-06,
      "loss": 0.7529,
      "step": 36525
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.510615187036223,
      "learning_rate": 5.603379081751067e-06,
      "loss": 0.7641,
      "step": 36530
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.309386769968196,
      "learning_rate": 5.602156932506373e-06,
      "loss": 0.7817,
      "step": 36535
    },
    {
      "epoch": 1.55,
      "grad_norm": 11.82336960305453,
      "learning_rate": 5.600934746754677e-06,
      "loss": 0.7117,
      "step": 36540
    },
    {
      "epoch": 1.55,
      "grad_norm": 14.332505091778648,
      "learning_rate": 5.599712524570079e-06,
      "loss": 0.7448,
      "step": 36545
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.589062503407123,
      "learning_rate": 5.598490266026677e-06,
      "loss": 0.772,
      "step": 36550
    },
    {
      "epoch": 1.55,
      "grad_norm": 13.912471931875253,
      "learning_rate": 5.597267971198571e-06,
      "loss": 0.7377,
      "step": 36555
    },
    {
      "epoch": 1.55,
      "grad_norm": 14.817308098016637,
      "learning_rate": 5.59604564015987e-06,
      "loss": 0.7899,
      "step": 36560
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.177076279811699,
      "learning_rate": 5.594823272984675e-06,
      "loss": 0.718,
      "step": 36565
    },
    {
      "epoch": 1.55,
      "grad_norm": 11.832901420938317,
      "learning_rate": 5.593600869747099e-06,
      "loss": 0.7437,
      "step": 36570
    },
    {
      "epoch": 1.55,
      "grad_norm": 19.19514347981489,
      "learning_rate": 5.59237843052125e-06,
      "loss": 0.7732,
      "step": 36575
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.912274216084402,
      "learning_rate": 5.591155955381241e-06,
      "loss": 0.7584,
      "step": 36580
    },
    {
      "epoch": 1.55,
      "grad_norm": 12.474081367444263,
      "learning_rate": 5.58993344440119e-06,
      "loss": 0.764,
      "step": 36585
    },
    {
      "epoch": 1.55,
      "grad_norm": 16.75594749942165,
      "learning_rate": 5.588710897655209e-06,
      "loss": 0.7556,
      "step": 36590
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.164864774329933,
      "learning_rate": 5.587488315217422e-06,
      "loss": 0.7189,
      "step": 36595
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.266079169083353,
      "learning_rate": 5.586265697161947e-06,
      "loss": 0.7449,
      "step": 36600
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.495028947158499,
      "learning_rate": 5.585043043562913e-06,
      "loss": 0.7374,
      "step": 36605
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.467384448965979,
      "learning_rate": 5.583820354494442e-06,
      "loss": 0.7627,
      "step": 36610
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.9600212350243575,
      "learning_rate": 5.58259763003066e-06,
      "loss": 0.7467,
      "step": 36615
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.85516884370735,
      "learning_rate": 5.5813748702457015e-06,
      "loss": 0.7847,
      "step": 36620
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.186771345561652,
      "learning_rate": 5.5801520752136965e-06,
      "loss": 0.768,
      "step": 36625
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.264683175783478,
      "learning_rate": 5.5789292450087805e-06,
      "loss": 0.7426,
      "step": 36630
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.09136167291706,
      "learning_rate": 5.57770637970509e-06,
      "loss": 0.7527,
      "step": 36635
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.480207582642863,
      "learning_rate": 5.576483479376763e-06,
      "loss": 0.7782,
      "step": 36640
    },
    {
      "epoch": 1.55,
      "grad_norm": 9.232047433146006,
      "learning_rate": 5.575260544097941e-06,
      "loss": 0.7613,
      "step": 36645
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.783226481209886,
      "learning_rate": 5.5740375739427665e-06,
      "loss": 0.7327,
      "step": 36650
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.440824695582905,
      "learning_rate": 5.572814568985385e-06,
      "loss": 0.751,
      "step": 36655
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.587148485966971,
      "learning_rate": 5.571591529299942e-06,
      "loss": 0.7349,
      "step": 36660
    },
    {
      "epoch": 1.55,
      "grad_norm": 11.529727946081056,
      "learning_rate": 5.570368454960591e-06,
      "loss": 0.7332,
      "step": 36665
    },
    {
      "epoch": 1.55,
      "grad_norm": 12.257924553311978,
      "learning_rate": 5.569145346041479e-06,
      "loss": 0.7478,
      "step": 36670
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.019248582223086,
      "learning_rate": 5.5679222026167615e-06,
      "loss": 0.7672,
      "step": 36675
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.803416178103751,
      "learning_rate": 5.5666990247605935e-06,
      "loss": 0.7472,
      "step": 36680
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.144868083154078,
      "learning_rate": 5.565475812547133e-06,
      "loss": 0.7656,
      "step": 36685
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.718405158032616,
      "learning_rate": 5.5642525660505386e-06,
      "loss": 0.7811,
      "step": 36690
    },
    {
      "epoch": 1.55,
      "grad_norm": 7.249719790407091,
      "learning_rate": 5.5630292853449755e-06,
      "loss": 0.7625,
      "step": 36695
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.875057832942347,
      "learning_rate": 5.561805970504602e-06,
      "loss": 0.7669,
      "step": 36700
    },
    {
      "epoch": 1.55,
      "grad_norm": 9.412357271744503,
      "learning_rate": 5.56058262160359e-06,
      "loss": 0.7546,
      "step": 36705
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.040148220149561,
      "learning_rate": 5.559359238716105e-06,
      "loss": 0.7552,
      "step": 36710
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.620364459755838,
      "learning_rate": 5.558135821916315e-06,
      "loss": 0.7821,
      "step": 36715
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.3094381143222495,
      "learning_rate": 5.556912371278395e-06,
      "loss": 0.7616,
      "step": 36720
    },
    {
      "epoch": 1.55,
      "grad_norm": 6.2167938283736275,
      "learning_rate": 5.555688886876519e-06,
      "loss": 0.7925,
      "step": 36725
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.520388479915492,
      "learning_rate": 5.554465368784864e-06,
      "loss": 0.7506,
      "step": 36730
    },
    {
      "epoch": 1.55,
      "grad_norm": 8.494528871130239,
      "learning_rate": 5.5532418170776035e-06,
      "loss": 0.7851,
      "step": 36735
    },
    {
      "epoch": 1.55,
      "grad_norm": 5.572486231956009,
      "learning_rate": 5.552018231828924e-06,
      "loss": 0.7661,
      "step": 36740
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.780519966654631,
      "learning_rate": 5.5507946131130045e-06,
      "loss": 0.726,
      "step": 36745
    },
    {
      "epoch": 1.56,
      "grad_norm": 13.976860510585473,
      "learning_rate": 5.5495709610040295e-06,
      "loss": 0.7215,
      "step": 36750
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.19871886315069,
      "learning_rate": 5.548347275576186e-06,
      "loss": 0.776,
      "step": 36755
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.004589988484238,
      "learning_rate": 5.547123556903662e-06,
      "loss": 0.779,
      "step": 36760
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.277347717713167,
      "learning_rate": 5.545899805060649e-06,
      "loss": 0.781,
      "step": 36765
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.56578136290133,
      "learning_rate": 5.544676020121339e-06,
      "loss": 0.7548,
      "step": 36770
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.48614619135727,
      "learning_rate": 5.543452202159926e-06,
      "loss": 0.7486,
      "step": 36775
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.830718026235273,
      "learning_rate": 5.542228351250607e-06,
      "loss": 0.7879,
      "step": 36780
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.312951263409624,
      "learning_rate": 5.541004467467579e-06,
      "loss": 0.7459,
      "step": 36785
    },
    {
      "epoch": 1.56,
      "grad_norm": 22.77551179356512,
      "learning_rate": 5.539780550885046e-06,
      "loss": 0.7134,
      "step": 36790
    },
    {
      "epoch": 1.56,
      "grad_norm": 14.304148161836249,
      "learning_rate": 5.538556601577207e-06,
      "loss": 0.7481,
      "step": 36795
    },
    {
      "epoch": 1.56,
      "grad_norm": 13.651924047565467,
      "learning_rate": 5.5373326196182665e-06,
      "loss": 0.7607,
      "step": 36800
    },
    {
      "epoch": 1.56,
      "grad_norm": 15.777224903436965,
      "learning_rate": 5.536108605082433e-06,
      "loss": 0.7454,
      "step": 36805
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.568584327788012,
      "learning_rate": 5.534884558043915e-06,
      "loss": 0.7162,
      "step": 36810
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.328506333587683,
      "learning_rate": 5.533660478576921e-06,
      "loss": 0.7083,
      "step": 36815
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.341493420058843,
      "learning_rate": 5.532436366755662e-06,
      "loss": 0.7264,
      "step": 36820
    },
    {
      "epoch": 1.56,
      "grad_norm": 9.104313499610381,
      "learning_rate": 5.531212222654358e-06,
      "loss": 0.7818,
      "step": 36825
    },
    {
      "epoch": 1.56,
      "grad_norm": 27.67038211777149,
      "learning_rate": 5.52998804634722e-06,
      "loss": 0.7663,
      "step": 36830
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.19357400529888,
      "learning_rate": 5.528763837908468e-06,
      "loss": 0.7296,
      "step": 36835
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.478316491759568,
      "learning_rate": 5.527539597412322e-06,
      "loss": 0.7805,
      "step": 36840
    },
    {
      "epoch": 1.56,
      "grad_norm": 24.447764618212663,
      "learning_rate": 5.526315324933003e-06,
      "loss": 0.7881,
      "step": 36845
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.4875036111420465,
      "learning_rate": 5.525091020544737e-06,
      "loss": 0.7885,
      "step": 36850
    },
    {
      "epoch": 1.56,
      "grad_norm": 12.387838188757721,
      "learning_rate": 5.52386668432175e-06,
      "loss": 0.7267,
      "step": 36855
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.339878239996746,
      "learning_rate": 5.522642316338268e-06,
      "loss": 0.7527,
      "step": 36860
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.707450650792179,
      "learning_rate": 5.521417916668522e-06,
      "loss": 0.7607,
      "step": 36865
    },
    {
      "epoch": 1.56,
      "grad_norm": 11.741998808300487,
      "learning_rate": 5.5201934853867435e-06,
      "loss": 0.7716,
      "step": 36870
    },
    {
      "epoch": 1.56,
      "grad_norm": 9.698326350076073,
      "learning_rate": 5.518969022567167e-06,
      "loss": 0.7573,
      "step": 36875
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.803692543465708,
      "learning_rate": 5.5177445282840265e-06,
      "loss": 0.7371,
      "step": 36880
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.3719268098684,
      "learning_rate": 5.51652000261156e-06,
      "loss": 0.7564,
      "step": 36885
    },
    {
      "epoch": 1.56,
      "grad_norm": 12.196675314629372,
      "learning_rate": 5.515295445624011e-06,
      "loss": 0.7474,
      "step": 36890
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.713126578903906,
      "learning_rate": 5.514070857395613e-06,
      "loss": 0.7905,
      "step": 36895
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.740431084470808,
      "learning_rate": 5.512846238000614e-06,
      "loss": 0.7316,
      "step": 36900
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.853639860388595,
      "learning_rate": 5.5116215875132574e-06,
      "loss": 0.7161,
      "step": 36905
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.530670146470907,
      "learning_rate": 5.510396906007792e-06,
      "loss": 0.7529,
      "step": 36910
    },
    {
      "epoch": 1.56,
      "grad_norm": 6.035015143263146,
      "learning_rate": 5.5091721935584666e-06,
      "loss": 0.7692,
      "step": 36915
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.147999387556059,
      "learning_rate": 5.507947450239529e-06,
      "loss": 0.7392,
      "step": 36920
    },
    {
      "epoch": 1.56,
      "grad_norm": 5.784650884013642,
      "learning_rate": 5.5067226761252345e-06,
      "loss": 0.8287,
      "step": 36925
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.587609517583636,
      "learning_rate": 5.5054978712898365e-06,
      "loss": 0.7384,
      "step": 36930
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.126016071152624,
      "learning_rate": 5.504273035807592e-06,
      "loss": 0.7464,
      "step": 36935
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.173315600431369,
      "learning_rate": 5.503048169752759e-06,
      "loss": 0.7107,
      "step": 36940
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.086160944977605,
      "learning_rate": 5.501823273199596e-06,
      "loss": 0.7543,
      "step": 36945
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.736721827131469,
      "learning_rate": 5.500598346222367e-06,
      "loss": 0.7332,
      "step": 36950
    },
    {
      "epoch": 1.56,
      "grad_norm": 8.901911838464171,
      "learning_rate": 5.499373388895335e-06,
      "loss": 0.7431,
      "step": 36955
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.381959441030425,
      "learning_rate": 5.498148401292765e-06,
      "loss": 0.7375,
      "step": 36960
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.671123928888369,
      "learning_rate": 5.496923383488926e-06,
      "loss": 0.726,
      "step": 36965
    },
    {
      "epoch": 1.56,
      "grad_norm": 11.333592541522494,
      "learning_rate": 5.495698335558084e-06,
      "loss": 0.7822,
      "step": 36970
    },
    {
      "epoch": 1.56,
      "grad_norm": 7.22204678869541,
      "learning_rate": 5.494473257574514e-06,
      "loss": 0.742,
      "step": 36975
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.527447785608108,
      "learning_rate": 5.493248149612485e-06,
      "loss": 0.7398,
      "step": 36980
    },
    {
      "epoch": 1.57,
      "grad_norm": 10.893179711971822,
      "learning_rate": 5.492023011746275e-06,
      "loss": 0.7549,
      "step": 36985
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.715380391043272,
      "learning_rate": 5.490797844050159e-06,
      "loss": 0.7281,
      "step": 36990
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.946908492612203,
      "learning_rate": 5.489572646598414e-06,
      "loss": 0.7548,
      "step": 36995
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.647822114470655,
      "learning_rate": 5.488347419465323e-06,
      "loss": 0.7374,
      "step": 37000
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.5237322997598035,
      "learning_rate": 5.4871221627251645e-06,
      "loss": 0.7341,
      "step": 37005
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.367895384519147,
      "learning_rate": 5.485896876452227e-06,
      "loss": 0.7747,
      "step": 37010
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.469921205848223,
      "learning_rate": 5.4846715607207915e-06,
      "loss": 0.7705,
      "step": 37015
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.9152224973251855,
      "learning_rate": 5.483446215605146e-06,
      "loss": 0.7125,
      "step": 37020
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.771546057557418,
      "learning_rate": 5.482220841179583e-06,
      "loss": 0.746,
      "step": 37025
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.0988749714523465,
      "learning_rate": 5.480995437518388e-06,
      "loss": 0.7634,
      "step": 37030
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.39238024456054,
      "learning_rate": 5.479770004695859e-06,
      "loss": 0.7688,
      "step": 37035
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.7098113880967265,
      "learning_rate": 5.478544542786286e-06,
      "loss": 0.7485,
      "step": 37040
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.175536373528734,
      "learning_rate": 5.4773190518639675e-06,
      "loss": 0.7435,
      "step": 37045
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.443792539342895,
      "learning_rate": 5.476093532003202e-06,
      "loss": 0.7484,
      "step": 37050
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.75881595350611,
      "learning_rate": 5.474867983278287e-06,
      "loss": 0.7591,
      "step": 37055
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.582404820949536,
      "learning_rate": 5.473642405763525e-06,
      "loss": 0.7538,
      "step": 37060
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.325441087532176,
      "learning_rate": 5.472416799533219e-06,
      "loss": 0.7678,
      "step": 37065
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.721838023340378,
      "learning_rate": 5.471191164661674e-06,
      "loss": 0.7453,
      "step": 37070
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.817735874791188,
      "learning_rate": 5.469965501223198e-06,
      "loss": 0.7452,
      "step": 37075
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.6256209188353195,
      "learning_rate": 5.468739809292095e-06,
      "loss": 0.7714,
      "step": 37080
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.894218377929878,
      "learning_rate": 5.46751408894268e-06,
      "loss": 0.7607,
      "step": 37085
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.8640520559668845,
      "learning_rate": 5.4662883402492624e-06,
      "loss": 0.7284,
      "step": 37090
    },
    {
      "epoch": 1.57,
      "grad_norm": 9.93288853207604,
      "learning_rate": 5.465062563286157e-06,
      "loss": 0.7406,
      "step": 37095
    },
    {
      "epoch": 1.57,
      "grad_norm": 6.523399390325565,
      "learning_rate": 5.463836758127678e-06,
      "loss": 0.7243,
      "step": 37100
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.58465293303299,
      "learning_rate": 5.462610924848142e-06,
      "loss": 0.7515,
      "step": 37105
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.775018568474361,
      "learning_rate": 5.461385063521869e-06,
      "loss": 0.7374,
      "step": 37110
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.105984588538315,
      "learning_rate": 5.460159174223177e-06,
      "loss": 0.8132,
      "step": 37115
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.422627585951194,
      "learning_rate": 5.458933257026392e-06,
      "loss": 0.7376,
      "step": 37120
    },
    {
      "epoch": 1.57,
      "grad_norm": 10.439643363015225,
      "learning_rate": 5.457707312005836e-06,
      "loss": 0.7472,
      "step": 37125
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.649858787735931,
      "learning_rate": 5.4564813392358304e-06,
      "loss": 0.7478,
      "step": 37130
    },
    {
      "epoch": 1.57,
      "grad_norm": 14.425469717219329,
      "learning_rate": 5.455255338790709e-06,
      "loss": 0.7333,
      "step": 37135
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.942382375528479,
      "learning_rate": 5.4540293107447965e-06,
      "loss": 0.7238,
      "step": 37140
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.59034038781036,
      "learning_rate": 5.452803255172424e-06,
      "loss": 0.775,
      "step": 37145
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.586906620111466,
      "learning_rate": 5.4515771721479235e-06,
      "loss": 0.7326,
      "step": 37150
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.000738088083427,
      "learning_rate": 5.45035106174563e-06,
      "loss": 0.7545,
      "step": 37155
    },
    {
      "epoch": 1.57,
      "grad_norm": 31.09949310373615,
      "learning_rate": 5.449124924039878e-06,
      "loss": 0.7566,
      "step": 37160
    },
    {
      "epoch": 1.57,
      "grad_norm": 20.697724272530564,
      "learning_rate": 5.4478987591050035e-06,
      "loss": 0.7572,
      "step": 37165
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.9269519440933,
      "learning_rate": 5.446672567015348e-06,
      "loss": 0.7745,
      "step": 37170
    },
    {
      "epoch": 1.57,
      "grad_norm": 16.864229821019595,
      "learning_rate": 5.44544634784525e-06,
      "loss": 0.7429,
      "step": 37175
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.417397221759656,
      "learning_rate": 5.444220101669053e-06,
      "loss": 0.744,
      "step": 37180
    },
    {
      "epoch": 1.57,
      "grad_norm": 10.43728825453933,
      "learning_rate": 5.4429938285610995e-06,
      "loss": 0.7111,
      "step": 37185
    },
    {
      "epoch": 1.57,
      "grad_norm": 14.265135331748061,
      "learning_rate": 5.441767528595734e-06,
      "loss": 0.7582,
      "step": 37190
    },
    {
      "epoch": 1.57,
      "grad_norm": 5.087609702952805,
      "learning_rate": 5.4405412018473045e-06,
      "loss": 0.7063,
      "step": 37195
    },
    {
      "epoch": 1.57,
      "grad_norm": 8.394425093210955,
      "learning_rate": 5.439314848390159e-06,
      "loss": 0.7444,
      "step": 37200
    },
    {
      "epoch": 1.57,
      "grad_norm": 4.776987664159522,
      "learning_rate": 5.43808846829865e-06,
      "loss": 0.7445,
      "step": 37205
    },
    {
      "epoch": 1.57,
      "grad_norm": 7.030360076516311,
      "learning_rate": 5.436862061647128e-06,
      "loss": 0.7669,
      "step": 37210
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.602799442243262,
      "learning_rate": 5.435635628509945e-06,
      "loss": 0.7506,
      "step": 37215
    },
    {
      "epoch": 1.58,
      "grad_norm": 18.320943870537874,
      "learning_rate": 5.434409168961457e-06,
      "loss": 0.7607,
      "step": 37220
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.269175158454024,
      "learning_rate": 5.43318268307602e-06,
      "loss": 0.7488,
      "step": 37225
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.751540197825144,
      "learning_rate": 5.431956170927995e-06,
      "loss": 0.7432,
      "step": 37230
    },
    {
      "epoch": 1.58,
      "grad_norm": 21.256596588834853,
      "learning_rate": 5.430729632591738e-06,
      "loss": 0.7414,
      "step": 37235
    },
    {
      "epoch": 1.58,
      "grad_norm": 14.155499118333113,
      "learning_rate": 5.429503068141613e-06,
      "loss": 0.717,
      "step": 37240
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.89397437103705,
      "learning_rate": 5.428276477651981e-06,
      "loss": 0.759,
      "step": 37245
    },
    {
      "epoch": 1.58,
      "grad_norm": 8.34298624571846,
      "learning_rate": 5.427049861197208e-06,
      "loss": 0.7348,
      "step": 37250
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.327354589612307,
      "learning_rate": 5.42582321885166e-06,
      "loss": 0.7524,
      "step": 37255
    },
    {
      "epoch": 1.58,
      "grad_norm": 14.62395183877402,
      "learning_rate": 5.424596550689706e-06,
      "loss": 0.7811,
      "step": 37260
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.100585618363656,
      "learning_rate": 5.4233698567857115e-06,
      "loss": 0.7523,
      "step": 37265
    },
    {
      "epoch": 1.58,
      "grad_norm": 14.928983039776778,
      "learning_rate": 5.4221431372140505e-06,
      "loss": 0.7466,
      "step": 37270
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.214963999958787,
      "learning_rate": 5.420916392049095e-06,
      "loss": 0.7161,
      "step": 37275
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.411810849723901,
      "learning_rate": 5.419689621365218e-06,
      "loss": 0.7612,
      "step": 37280
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.983503185711556,
      "learning_rate": 5.4184628252367965e-06,
      "loss": 0.7681,
      "step": 37285
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.21019927678597,
      "learning_rate": 5.417236003738204e-06,
      "loss": 0.7619,
      "step": 37290
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.9375492392208455,
      "learning_rate": 5.416009156943825e-06,
      "loss": 0.7547,
      "step": 37295
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.728254528343914,
      "learning_rate": 5.414782284928034e-06,
      "loss": 0.7768,
      "step": 37300
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.356442019958166,
      "learning_rate": 5.413555387765216e-06,
      "loss": 0.7699,
      "step": 37305
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.294594513123982,
      "learning_rate": 5.412328465529751e-06,
      "loss": 0.7526,
      "step": 37310
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.501263970491084,
      "learning_rate": 5.411101518296028e-06,
      "loss": 0.7642,
      "step": 37315
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.829606427474315,
      "learning_rate": 5.409874546138431e-06,
      "loss": 0.7622,
      "step": 37320
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.19396541593531,
      "learning_rate": 5.408647549131345e-06,
      "loss": 0.741,
      "step": 37325
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.632286932890838,
      "learning_rate": 5.4074205273491645e-06,
      "loss": 0.7923,
      "step": 37330
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.121621608875013,
      "learning_rate": 5.406193480866275e-06,
      "loss": 0.698,
      "step": 37335
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.427187293589908,
      "learning_rate": 5.404966409757073e-06,
      "loss": 0.7292,
      "step": 37340
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.401604642840885,
      "learning_rate": 5.403739314095952e-06,
      "loss": 0.7845,
      "step": 37345
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.640809259029113,
      "learning_rate": 5.402512193957302e-06,
      "loss": 0.7243,
      "step": 37350
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.2337488170726845,
      "learning_rate": 5.401285049415525e-06,
      "loss": 0.7296,
      "step": 37355
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.803299859630813,
      "learning_rate": 5.400057880545017e-06,
      "loss": 0.7251,
      "step": 37360
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.263768078793284,
      "learning_rate": 5.398830687420179e-06,
      "loss": 0.7176,
      "step": 37365
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.146789803540217,
      "learning_rate": 5.3976034701154115e-06,
      "loss": 0.7389,
      "step": 37370
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.798472102905768,
      "learning_rate": 5.396376228705115e-06,
      "loss": 0.7902,
      "step": 37375
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.965530444155731,
      "learning_rate": 5.395148963263699e-06,
      "loss": 0.7246,
      "step": 37380
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.839877668421685,
      "learning_rate": 5.393921673865562e-06,
      "loss": 0.7497,
      "step": 37385
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.227939821119902,
      "learning_rate": 5.392694360585114e-06,
      "loss": 0.7651,
      "step": 37390
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.859012737287445,
      "learning_rate": 5.391467023496765e-06,
      "loss": 0.7445,
      "step": 37395
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.856082471208197,
      "learning_rate": 5.390239662674923e-06,
      "loss": 0.7155,
      "step": 37400
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.315321003108104,
      "learning_rate": 5.3890122781940005e-06,
      "loss": 0.7747,
      "step": 37405
    },
    {
      "epoch": 1.58,
      "grad_norm": 5.083693089221626,
      "learning_rate": 5.387784870128408e-06,
      "loss": 0.7484,
      "step": 37410
    },
    {
      "epoch": 1.58,
      "grad_norm": 14.834778880823702,
      "learning_rate": 5.386557438552562e-06,
      "loss": 0.7513,
      "step": 37415
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.547063559609413,
      "learning_rate": 5.385329983540877e-06,
      "loss": 0.7601,
      "step": 37420
    },
    {
      "epoch": 1.58,
      "grad_norm": 7.828351467269713,
      "learning_rate": 5.384102505167771e-06,
      "loss": 0.7441,
      "step": 37425
    },
    {
      "epoch": 1.58,
      "grad_norm": 4.991494391598989,
      "learning_rate": 5.3828750035076605e-06,
      "loss": 0.7515,
      "step": 37430
    },
    {
      "epoch": 1.58,
      "grad_norm": 10.756184923926918,
      "learning_rate": 5.381647478634966e-06,
      "loss": 0.755,
      "step": 37435
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.01039073193324,
      "learning_rate": 5.38041993062411e-06,
      "loss": 0.7342,
      "step": 37440
    },
    {
      "epoch": 1.58,
      "grad_norm": 6.005893322049624,
      "learning_rate": 5.3791923595495135e-06,
      "loss": 0.7385,
      "step": 37445
    },
    {
      "epoch": 1.58,
      "grad_norm": 9.237466588598005,
      "learning_rate": 5.3779647654856014e-06,
      "loss": 0.7664,
      "step": 37450
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.375668314384845,
      "learning_rate": 5.376737148506799e-06,
      "loss": 0.7058,
      "step": 37455
    },
    {
      "epoch": 1.59,
      "grad_norm": 10.558184279955777,
      "learning_rate": 5.375509508687533e-06,
      "loss": 0.7285,
      "step": 37460
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.542658679631809,
      "learning_rate": 5.374281846102232e-06,
      "loss": 0.7671,
      "step": 37465
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.824039967567152,
      "learning_rate": 5.373054160825324e-06,
      "loss": 0.7366,
      "step": 37470
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.806334703708295,
      "learning_rate": 5.371826452931242e-06,
      "loss": 0.7619,
      "step": 37475
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.135966422980176,
      "learning_rate": 5.370598722494417e-06,
      "loss": 0.7057,
      "step": 37480
    },
    {
      "epoch": 1.59,
      "grad_norm": 13.427610602060648,
      "learning_rate": 5.369370969589284e-06,
      "loss": 0.7456,
      "step": 37485
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.066176785301034,
      "learning_rate": 5.368143194290277e-06,
      "loss": 0.736,
      "step": 37490
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.639154191180423,
      "learning_rate": 5.36691539667183e-06,
      "loss": 0.7467,
      "step": 37495
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.649433836434678,
      "learning_rate": 5.365687576808386e-06,
      "loss": 0.7781,
      "step": 37500
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.631794120537659,
      "learning_rate": 5.36445973477438e-06,
      "loss": 0.7618,
      "step": 37505
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.518038653978764,
      "learning_rate": 5.3632318706442545e-06,
      "loss": 0.7772,
      "step": 37510
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.702757138903261,
      "learning_rate": 5.362003984492451e-06,
      "loss": 0.7681,
      "step": 37515
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.053473663573236,
      "learning_rate": 5.360776076393412e-06,
      "loss": 0.752,
      "step": 37520
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.276327836729093,
      "learning_rate": 5.359548146421582e-06,
      "loss": 0.7582,
      "step": 37525
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.118716133271822,
      "learning_rate": 5.358320194651405e-06,
      "loss": 0.7592,
      "step": 37530
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.528297926027103,
      "learning_rate": 5.3570922211573325e-06,
      "loss": 0.6974,
      "step": 37535
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.593812938615475,
      "learning_rate": 5.35586422601381e-06,
      "loss": 0.8095,
      "step": 37540
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.8775369696974495,
      "learning_rate": 5.354636209295288e-06,
      "loss": 0.758,
      "step": 37545
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.233224925054401,
      "learning_rate": 5.353408171076219e-06,
      "loss": 0.7124,
      "step": 37550
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.077809648511176,
      "learning_rate": 5.35218011143105e-06,
      "loss": 0.7564,
      "step": 37555
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.656496044516427,
      "learning_rate": 5.35095203043424e-06,
      "loss": 0.7514,
      "step": 37560
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.964532200088741,
      "learning_rate": 5.349723928160243e-06,
      "loss": 0.7686,
      "step": 37565
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.785002680947919,
      "learning_rate": 5.3484958046835125e-06,
      "loss": 0.7352,
      "step": 37570
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.886805250032847,
      "learning_rate": 5.3472676600785086e-06,
      "loss": 0.7455,
      "step": 37575
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.064987162246553,
      "learning_rate": 5.346039494419689e-06,
      "loss": 0.7331,
      "step": 37580
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.424624370194576,
      "learning_rate": 5.344811307781515e-06,
      "loss": 0.7529,
      "step": 37585
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.374170531821699,
      "learning_rate": 5.3435831002384475e-06,
      "loss": 0.7376,
      "step": 37590
    },
    {
      "epoch": 1.59,
      "grad_norm": 3.9041495239232895,
      "learning_rate": 5.342354871864946e-06,
      "loss": 0.7465,
      "step": 37595
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.803163403239657,
      "learning_rate": 5.341126622735479e-06,
      "loss": 0.7339,
      "step": 37600
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.480521533068638,
      "learning_rate": 5.33989835292451e-06,
      "loss": 0.7652,
      "step": 37605
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.4214863303846785,
      "learning_rate": 5.338670062506505e-06,
      "loss": 0.7484,
      "step": 37610
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.586590087923795,
      "learning_rate": 5.337441751555932e-06,
      "loss": 0.7517,
      "step": 37615
    },
    {
      "epoch": 1.59,
      "grad_norm": 7.463360755259694,
      "learning_rate": 5.336213420147259e-06,
      "loss": 0.7199,
      "step": 37620
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.475288399969703,
      "learning_rate": 5.334985068354958e-06,
      "loss": 0.7463,
      "step": 37625
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.9987082821957,
      "learning_rate": 5.333756696253496e-06,
      "loss": 0.762,
      "step": 37630
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.162163834297545,
      "learning_rate": 5.33252830391735e-06,
      "loss": 0.7473,
      "step": 37635
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.872435093013567,
      "learning_rate": 5.331299891420992e-06,
      "loss": 0.7027,
      "step": 37640
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.542212780383935,
      "learning_rate": 5.330071458838899e-06,
      "loss": 0.7287,
      "step": 37645
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.562253703132565,
      "learning_rate": 5.328843006245546e-06,
      "loss": 0.7151,
      "step": 37650
    },
    {
      "epoch": 1.59,
      "grad_norm": 4.895724675905981,
      "learning_rate": 5.327614533715408e-06,
      "loss": 0.7159,
      "step": 37655
    },
    {
      "epoch": 1.59,
      "grad_norm": 14.201465476356836,
      "learning_rate": 5.326386041322967e-06,
      "loss": 0.7442,
      "step": 37660
    },
    {
      "epoch": 1.59,
      "grad_norm": 9.161162559764266,
      "learning_rate": 5.325157529142701e-06,
      "loss": 0.7712,
      "step": 37665
    },
    {
      "epoch": 1.59,
      "grad_norm": 8.527358948463428,
      "learning_rate": 5.323928997249093e-06,
      "loss": 0.7303,
      "step": 37670
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.689005078888132,
      "learning_rate": 5.322700445716624e-06,
      "loss": 0.7431,
      "step": 37675
    },
    {
      "epoch": 1.59,
      "grad_norm": 5.699739772856369,
      "learning_rate": 5.321471874619776e-06,
      "loss": 0.7229,
      "step": 37680
    },
    {
      "epoch": 1.59,
      "grad_norm": 6.214568447706823,
      "learning_rate": 5.320243284033036e-06,
      "loss": 0.7247,
      "step": 37685
    },
    {
      "epoch": 1.6,
      "grad_norm": 9.025205590394776,
      "learning_rate": 5.319014674030889e-06,
      "loss": 0.7509,
      "step": 37690
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.561513320039655,
      "learning_rate": 5.317786044687823e-06,
      "loss": 0.7597,
      "step": 37695
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.157538752685827,
      "learning_rate": 5.316557396078324e-06,
      "loss": 0.7117,
      "step": 37700
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.23669726869206,
      "learning_rate": 5.315328728276884e-06,
      "loss": 0.7442,
      "step": 37705
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.548821424513452,
      "learning_rate": 5.314100041357991e-06,
      "loss": 0.7472,
      "step": 37710
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.436883503772997,
      "learning_rate": 5.312871335396137e-06,
      "loss": 0.7304,
      "step": 37715
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.174153402921207,
      "learning_rate": 5.311642610465816e-06,
      "loss": 0.7974,
      "step": 37720
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.523995783694644,
      "learning_rate": 5.310413866641522e-06,
      "loss": 0.7303,
      "step": 37725
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.36651402925244,
      "learning_rate": 5.3091851039977495e-06,
      "loss": 0.7384,
      "step": 37730
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.813853312993525,
      "learning_rate": 5.307956322608995e-06,
      "loss": 0.7347,
      "step": 37735
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.816087668737082,
      "learning_rate": 5.306727522549754e-06,
      "loss": 0.7304,
      "step": 37740
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.965818504706991,
      "learning_rate": 5.305498703894528e-06,
      "loss": 0.7455,
      "step": 37745
    },
    {
      "epoch": 1.6,
      "grad_norm": 10.184874659636407,
      "learning_rate": 5.304269866717816e-06,
      "loss": 0.746,
      "step": 37750
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.236274334477789,
      "learning_rate": 5.303041011094118e-06,
      "loss": 0.6714,
      "step": 37755
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.5494943414679945,
      "learning_rate": 5.301812137097936e-06,
      "loss": 0.7418,
      "step": 37760
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.042245990822691,
      "learning_rate": 5.300583244803773e-06,
      "loss": 0.7423,
      "step": 37765
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.585001465218098,
      "learning_rate": 5.299354334286133e-06,
      "loss": 0.7042,
      "step": 37770
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.2537654455744125,
      "learning_rate": 5.298125405619522e-06,
      "loss": 0.7723,
      "step": 37775
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.937966823473857,
      "learning_rate": 5.296896458878446e-06,
      "loss": 0.7572,
      "step": 37780
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.4112664654020355,
      "learning_rate": 5.295667494137412e-06,
      "loss": 0.7479,
      "step": 37785
    },
    {
      "epoch": 1.6,
      "grad_norm": 10.217407720490653,
      "learning_rate": 5.294438511470929e-06,
      "loss": 0.71,
      "step": 37790
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.196651159654953,
      "learning_rate": 5.293209510953506e-06,
      "loss": 0.7353,
      "step": 37795
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.876056922498158,
      "learning_rate": 5.291980492659654e-06,
      "loss": 0.7797,
      "step": 37800
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.355027706172651,
      "learning_rate": 5.290751456663885e-06,
      "loss": 0.7383,
      "step": 37805
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.444181007814591,
      "learning_rate": 5.289522403040711e-06,
      "loss": 0.744,
      "step": 37810
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.862747358515827,
      "learning_rate": 5.288293331864648e-06,
      "loss": 0.7491,
      "step": 37815
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.436880312505482,
      "learning_rate": 5.28706424321021e-06,
      "loss": 0.7601,
      "step": 37820
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.418785045296565,
      "learning_rate": 5.285835137151909e-06,
      "loss": 0.7405,
      "step": 37825
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.466243589612315,
      "learning_rate": 5.284606013764268e-06,
      "loss": 0.7705,
      "step": 37830
    },
    {
      "epoch": 1.6,
      "grad_norm": 13.006736163739006,
      "learning_rate": 5.283376873121803e-06,
      "loss": 0.7416,
      "step": 37835
    },
    {
      "epoch": 1.6,
      "grad_norm": 10.5033396248192,
      "learning_rate": 5.282147715299031e-06,
      "loss": 0.7239,
      "step": 37840
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.224224050082206,
      "learning_rate": 5.280918540370477e-06,
      "loss": 0.726,
      "step": 37845
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.916785371914052,
      "learning_rate": 5.279689348410656e-06,
      "loss": 0.791,
      "step": 37850
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.9549484259295795,
      "learning_rate": 5.278460139494096e-06,
      "loss": 0.7199,
      "step": 37855
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.5292756530216876,
      "learning_rate": 5.277230913695317e-06,
      "loss": 0.7538,
      "step": 37860
    },
    {
      "epoch": 1.6,
      "grad_norm": 11.714822515629121,
      "learning_rate": 5.276001671088845e-06,
      "loss": 0.7411,
      "step": 37865
    },
    {
      "epoch": 1.6,
      "grad_norm": 11.6102897602213,
      "learning_rate": 5.274772411749205e-06,
      "loss": 0.7295,
      "step": 37870
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.244398201443841,
      "learning_rate": 5.273543135750922e-06,
      "loss": 0.74,
      "step": 37875
    },
    {
      "epoch": 1.6,
      "grad_norm": 5.096831706126591,
      "learning_rate": 5.272313843168525e-06,
      "loss": 0.734,
      "step": 37880
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.926655074334417,
      "learning_rate": 5.2710845340765415e-06,
      "loss": 0.7346,
      "step": 37885
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.488196705088085,
      "learning_rate": 5.269855208549501e-06,
      "loss": 0.7391,
      "step": 37890
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.736495404687314,
      "learning_rate": 5.268625866661935e-06,
      "loss": 0.7561,
      "step": 37895
    },
    {
      "epoch": 1.6,
      "grad_norm": 7.497778266473555,
      "learning_rate": 5.267396508488372e-06,
      "loss": 0.7553,
      "step": 37900
    },
    {
      "epoch": 1.6,
      "grad_norm": 8.688799150542911,
      "learning_rate": 5.266167134103349e-06,
      "loss": 0.7475,
      "step": 37905
    },
    {
      "epoch": 1.6,
      "grad_norm": 12.46596615943758,
      "learning_rate": 5.2649377435813945e-06,
      "loss": 0.7182,
      "step": 37910
    },
    {
      "epoch": 1.6,
      "grad_norm": 6.875630781382229,
      "learning_rate": 5.2637083369970466e-06,
      "loss": 0.7361,
      "step": 37915
    },
    {
      "epoch": 1.6,
      "grad_norm": 4.314581377171161,
      "learning_rate": 5.262478914424839e-06,
      "loss": 0.7261,
      "step": 37920
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.905769432746213,
      "learning_rate": 5.261249475939306e-06,
      "loss": 0.7296,
      "step": 37925
    },
    {
      "epoch": 1.61,
      "grad_norm": 14.423302720315911,
      "learning_rate": 5.260020021614988e-06,
      "loss": 0.7278,
      "step": 37930
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.698815658091996,
      "learning_rate": 5.258790551526421e-06,
      "loss": 0.7097,
      "step": 37935
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.871909615106895,
      "learning_rate": 5.257561065748148e-06,
      "loss": 0.7324,
      "step": 37940
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.425972072814886,
      "learning_rate": 5.256331564354704e-06,
      "loss": 0.74,
      "step": 37945
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.945815598118916,
      "learning_rate": 5.255102047420633e-06,
      "loss": 0.7702,
      "step": 37950
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.806838144095054,
      "learning_rate": 5.253872515020477e-06,
      "loss": 0.7387,
      "step": 37955
    },
    {
      "epoch": 1.61,
      "grad_norm": 17.825496731017804,
      "learning_rate": 5.252642967228777e-06,
      "loss": 0.7324,
      "step": 37960
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.515405220851127,
      "learning_rate": 5.251413404120078e-06,
      "loss": 0.7181,
      "step": 37965
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.861932318933985,
      "learning_rate": 5.250183825768925e-06,
      "loss": 0.7565,
      "step": 37970
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.292486917293608,
      "learning_rate": 5.248954232249865e-06,
      "loss": 0.7045,
      "step": 37975
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.0995285907541925,
      "learning_rate": 5.247724623637441e-06,
      "loss": 0.7475,
      "step": 37980
    },
    {
      "epoch": 1.61,
      "grad_norm": 15.261729559612967,
      "learning_rate": 5.246495000006205e-06,
      "loss": 0.725,
      "step": 37985
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.23125595983438,
      "learning_rate": 5.2452653614307014e-06,
      "loss": 0.7076,
      "step": 37990
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.050972318389208,
      "learning_rate": 5.244035707985481e-06,
      "loss": 0.7425,
      "step": 37995
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.688104495775632,
      "learning_rate": 5.242806039745096e-06,
      "loss": 0.7494,
      "step": 38000
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.208776026998654,
      "learning_rate": 5.241576356784096e-06,
      "loss": 0.7249,
      "step": 38005
    },
    {
      "epoch": 1.61,
      "grad_norm": 10.536042298761876,
      "learning_rate": 5.240346659177033e-06,
      "loss": 0.7233,
      "step": 38010
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.1538416592362335,
      "learning_rate": 5.239116946998459e-06,
      "loss": 0.7711,
      "step": 38015
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.96920591635588,
      "learning_rate": 5.237887220322929e-06,
      "loss": 0.7099,
      "step": 38020
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.265050538591462,
      "learning_rate": 5.236657479224999e-06,
      "loss": 0.7468,
      "step": 38025
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.388958314077751,
      "learning_rate": 5.235427723779222e-06,
      "loss": 0.7112,
      "step": 38030
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.517372700063806,
      "learning_rate": 5.234197954060156e-06,
      "loss": 0.7333,
      "step": 38035
    },
    {
      "epoch": 1.61,
      "grad_norm": 8.324392253768671,
      "learning_rate": 5.2329681701423575e-06,
      "loss": 0.7537,
      "step": 38040
    },
    {
      "epoch": 1.61,
      "grad_norm": 10.09455151632928,
      "learning_rate": 5.2317383721003845e-06,
      "loss": 0.7797,
      "step": 38045
    },
    {
      "epoch": 1.61,
      "grad_norm": 9.428254444130227,
      "learning_rate": 5.230508560008798e-06,
      "loss": 0.7672,
      "step": 38050
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.996490313842462,
      "learning_rate": 5.229278733942155e-06,
      "loss": 0.7222,
      "step": 38055
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.948580857582867,
      "learning_rate": 5.22804889397502e-06,
      "loss": 0.7251,
      "step": 38060
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.494819388927259,
      "learning_rate": 5.22681904018195e-06,
      "loss": 0.734,
      "step": 38065
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.188351150721062,
      "learning_rate": 5.22558917263751e-06,
      "loss": 0.7164,
      "step": 38070
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.775442334496566,
      "learning_rate": 5.224359291416264e-06,
      "loss": 0.7328,
      "step": 38075
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.951136218643271,
      "learning_rate": 5.223129396592774e-06,
      "loss": 0.7173,
      "step": 38080
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.233586879665359,
      "learning_rate": 5.221899488241607e-06,
      "loss": 0.7087,
      "step": 38085
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.050689931576329,
      "learning_rate": 5.220669566437327e-06,
      "loss": 0.7023,
      "step": 38090
    },
    {
      "epoch": 1.61,
      "grad_norm": 11.441282611896956,
      "learning_rate": 5.2194396312545e-06,
      "loss": 0.7673,
      "step": 38095
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.890061726238572,
      "learning_rate": 5.218209682767698e-06,
      "loss": 0.75,
      "step": 38100
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.671385988457021,
      "learning_rate": 5.2169797210514825e-06,
      "loss": 0.7202,
      "step": 38105
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.690109653089311,
      "learning_rate": 5.215749746180427e-06,
      "loss": 0.7376,
      "step": 38110
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.2641787584295825,
      "learning_rate": 5.214519758229099e-06,
      "loss": 0.724,
      "step": 38115
    },
    {
      "epoch": 1.61,
      "grad_norm": 14.614666364112239,
      "learning_rate": 5.213289757272071e-06,
      "loss": 0.7263,
      "step": 38120
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.242702602137728,
      "learning_rate": 5.212059743383914e-06,
      "loss": 0.7284,
      "step": 38125
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.148072719344014,
      "learning_rate": 5.210829716639197e-06,
      "loss": 0.8041,
      "step": 38130
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.247284594665144,
      "learning_rate": 5.209599677112497e-06,
      "loss": 0.7507,
      "step": 38135
    },
    {
      "epoch": 1.61,
      "grad_norm": 7.098537713939876,
      "learning_rate": 5.208369624878386e-06,
      "loss": 0.7774,
      "step": 38140
    },
    {
      "epoch": 1.61,
      "grad_norm": 6.438648231081955,
      "learning_rate": 5.2071395600114375e-06,
      "loss": 0.7511,
      "step": 38145
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.268866208328094,
      "learning_rate": 5.205909482586228e-06,
      "loss": 0.6985,
      "step": 38150
    },
    {
      "epoch": 1.61,
      "grad_norm": 4.902995003229946,
      "learning_rate": 5.204679392677332e-06,
      "loss": 0.7597,
      "step": 38155
    },
    {
      "epoch": 1.61,
      "grad_norm": 5.357762100039163,
      "learning_rate": 5.203449290359329e-06,
      "loss": 0.7099,
      "step": 38160
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.918202985188686,
      "learning_rate": 5.202219175706795e-06,
      "loss": 0.7446,
      "step": 38165
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.192453276337107,
      "learning_rate": 5.200989048794306e-06,
      "loss": 0.7047,
      "step": 38170
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.529710037890714,
      "learning_rate": 5.199758909696445e-06,
      "loss": 0.7169,
      "step": 38175
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.869262658258585,
      "learning_rate": 5.198528758487788e-06,
      "loss": 0.7362,
      "step": 38180
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.4660360564636505,
      "learning_rate": 5.197298595242919e-06,
      "loss": 0.7234,
      "step": 38185
    },
    {
      "epoch": 1.62,
      "grad_norm": 10.356827353203668,
      "learning_rate": 5.196068420036417e-06,
      "loss": 0.683,
      "step": 38190
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.590534639319392,
      "learning_rate": 5.194838232942864e-06,
      "loss": 0.75,
      "step": 38195
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.363151028413926,
      "learning_rate": 5.193608034036843e-06,
      "loss": 0.8102,
      "step": 38200
    },
    {
      "epoch": 1.62,
      "grad_norm": 12.957285686732694,
      "learning_rate": 5.192377823392935e-06,
      "loss": 0.6969,
      "step": 38205
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.606741208167377,
      "learning_rate": 5.191147601085729e-06,
      "loss": 0.7694,
      "step": 38210
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.2494295300462115,
      "learning_rate": 5.189917367189807e-06,
      "loss": 0.7297,
      "step": 38215
    },
    {
      "epoch": 1.62,
      "grad_norm": 9.22458194321671,
      "learning_rate": 5.188687121779753e-06,
      "loss": 0.7275,
      "step": 38220
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.453282643846458,
      "learning_rate": 5.187456864930155e-06,
      "loss": 0.6971,
      "step": 38225
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.700818045679973,
      "learning_rate": 5.186226596715599e-06,
      "loss": 0.7493,
      "step": 38230
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.32433622098943,
      "learning_rate": 5.184996317210674e-06,
      "loss": 0.7384,
      "step": 38235
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.047200545118724,
      "learning_rate": 5.183766026489966e-06,
      "loss": 0.7026,
      "step": 38240
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.207624156426969,
      "learning_rate": 5.182535724628066e-06,
      "loss": 0.7458,
      "step": 38245
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.43006641856987,
      "learning_rate": 5.181305411699563e-06,
      "loss": 0.7303,
      "step": 38250
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.83888487269112,
      "learning_rate": 5.180075087779045e-06,
      "loss": 0.7353,
      "step": 38255
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.824421628005328,
      "learning_rate": 5.178844752941107e-06,
      "loss": 0.7359,
      "step": 38260
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.849450878000392,
      "learning_rate": 5.177614407260338e-06,
      "loss": 0.705,
      "step": 38265
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.772899111878667,
      "learning_rate": 5.17638405081133e-06,
      "loss": 0.7187,
      "step": 38270
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.723866777822633,
      "learning_rate": 5.175153683668678e-06,
      "loss": 0.7254,
      "step": 38275
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.329843141040258,
      "learning_rate": 5.173923305906972e-06,
      "loss": 0.7504,
      "step": 38280
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.059680921165908,
      "learning_rate": 5.17269291760081e-06,
      "loss": 0.7123,
      "step": 38285
    },
    {
      "epoch": 1.62,
      "grad_norm": 13.139054123490984,
      "learning_rate": 5.171462518824785e-06,
      "loss": 0.7514,
      "step": 38290
    },
    {
      "epoch": 1.62,
      "grad_norm": 24.460906690680105,
      "learning_rate": 5.170232109653491e-06,
      "loss": 0.7755,
      "step": 38295
    },
    {
      "epoch": 1.62,
      "grad_norm": 13.225668004013762,
      "learning_rate": 5.169001690161526e-06,
      "loss": 0.7057,
      "step": 38300
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.8211498410596025,
      "learning_rate": 5.167771260423488e-06,
      "loss": 0.709,
      "step": 38305
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.777076331759229,
      "learning_rate": 5.166540820513972e-06,
      "loss": 0.7334,
      "step": 38310
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.895710982857276,
      "learning_rate": 5.165310370507575e-06,
      "loss": 0.7422,
      "step": 38315
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.26165973165786,
      "learning_rate": 5.1640799104788985e-06,
      "loss": 0.7589,
      "step": 38320
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.516287064019683,
      "learning_rate": 5.16284944050254e-06,
      "loss": 0.7663,
      "step": 38325
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.94519302399792,
      "learning_rate": 5.1616189606531e-06,
      "loss": 0.7291,
      "step": 38330
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.942469564670391,
      "learning_rate": 5.160388471005179e-06,
      "loss": 0.753,
      "step": 38335
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.146351340809893,
      "learning_rate": 5.159157971633376e-06,
      "loss": 0.7499,
      "step": 38340
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.896636475563514,
      "learning_rate": 5.157927462612296e-06,
      "loss": 0.704,
      "step": 38345
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.503977712383521,
      "learning_rate": 5.1566969440165375e-06,
      "loss": 0.7228,
      "step": 38350
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.125528059373691,
      "learning_rate": 5.155466415920706e-06,
      "loss": 0.7609,
      "step": 38355
    },
    {
      "epoch": 1.62,
      "grad_norm": 9.645500596115685,
      "learning_rate": 5.154235878399402e-06,
      "loss": 0.7687,
      "step": 38360
    },
    {
      "epoch": 1.62,
      "grad_norm": 8.68835216946066,
      "learning_rate": 5.153005331527232e-06,
      "loss": 0.7288,
      "step": 38365
    },
    {
      "epoch": 1.62,
      "grad_norm": 6.938718902797209,
      "learning_rate": 5.151774775378799e-06,
      "loss": 0.7843,
      "step": 38370
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.063239477426115,
      "learning_rate": 5.150544210028709e-06,
      "loss": 0.7198,
      "step": 38375
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.560870171179861,
      "learning_rate": 5.149313635551567e-06,
      "loss": 0.7269,
      "step": 38380
    },
    {
      "epoch": 1.62,
      "grad_norm": 7.387114697707688,
      "learning_rate": 5.148083052021979e-06,
      "loss": 0.7667,
      "step": 38385
    },
    {
      "epoch": 1.62,
      "grad_norm": 4.571397400733684,
      "learning_rate": 5.146852459514551e-06,
      "loss": 0.7458,
      "step": 38390
    },
    {
      "epoch": 1.62,
      "grad_norm": 5.013995175017584,
      "learning_rate": 5.145621858103892e-06,
      "loss": 0.7828,
      "step": 38395
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.553235113742426,
      "learning_rate": 5.1443912478646076e-06,
      "loss": 0.7424,
      "step": 38400
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.7876537739008365,
      "learning_rate": 5.143160628871307e-06,
      "loss": 0.7537,
      "step": 38405
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.42354296202893,
      "learning_rate": 5.141930001198601e-06,
      "loss": 0.7257,
      "step": 38410
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.07735561077932,
      "learning_rate": 5.140699364921095e-06,
      "loss": 0.7103,
      "step": 38415
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.001674031349458,
      "learning_rate": 5.139468720113403e-06,
      "loss": 0.7512,
      "step": 38420
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.729352659252857,
      "learning_rate": 5.138238066850132e-06,
      "loss": 0.7365,
      "step": 38425
    },
    {
      "epoch": 1.63,
      "grad_norm": 18.94733196076367,
      "learning_rate": 5.137007405205896e-06,
      "loss": 0.7132,
      "step": 38430
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.464003027164705,
      "learning_rate": 5.1357767352553025e-06,
      "loss": 0.7165,
      "step": 38435
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.10485007191571,
      "learning_rate": 5.134546057072965e-06,
      "loss": 0.7438,
      "step": 38440
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.207127812957007,
      "learning_rate": 5.1333153707334985e-06,
      "loss": 0.7342,
      "step": 38445
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.7435453679954955,
      "learning_rate": 5.132084676311512e-06,
      "loss": 0.7257,
      "step": 38450
    },
    {
      "epoch": 1.63,
      "grad_norm": 18.56717101228031,
      "learning_rate": 5.1308539738816225e-06,
      "loss": 0.7253,
      "step": 38455
    },
    {
      "epoch": 1.63,
      "grad_norm": 12.89417137245258,
      "learning_rate": 5.129623263518439e-06,
      "loss": 0.699,
      "step": 38460
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.267103458787332,
      "learning_rate": 5.128392545296582e-06,
      "loss": 0.7318,
      "step": 38465
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.148704075609645,
      "learning_rate": 5.127161819290662e-06,
      "loss": 0.7459,
      "step": 38470
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.699177600899752,
      "learning_rate": 5.125931085575294e-06,
      "loss": 0.7291,
      "step": 38475
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.308206033644535,
      "learning_rate": 5.124700344225098e-06,
      "loss": 0.7228,
      "step": 38480
    },
    {
      "epoch": 1.63,
      "grad_norm": 6.351906069194901,
      "learning_rate": 5.1234695953146855e-06,
      "loss": 0.7604,
      "step": 38485
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.778468303711194,
      "learning_rate": 5.122238838918676e-06,
      "loss": 0.7339,
      "step": 38490
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.6248196285404175,
      "learning_rate": 5.121008075111685e-06,
      "loss": 0.7569,
      "step": 38495
    },
    {
      "epoch": 1.63,
      "grad_norm": 11.706908765142028,
      "learning_rate": 5.119777303968331e-06,
      "loss": 0.7595,
      "step": 38500
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.5344754047486,
      "learning_rate": 5.118546525563232e-06,
      "loss": 0.7209,
      "step": 38505
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.93336636885656,
      "learning_rate": 5.117315739971007e-06,
      "loss": 0.781,
      "step": 38510
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.7111109967346305,
      "learning_rate": 5.116084947266275e-06,
      "loss": 0.7385,
      "step": 38515
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.7206835688398,
      "learning_rate": 5.1148541475236524e-06,
      "loss": 0.7267,
      "step": 38520
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.979254290216709,
      "learning_rate": 5.113623340817763e-06,
      "loss": 0.7077,
      "step": 38525
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.3100144186804705,
      "learning_rate": 5.112392527223224e-06,
      "loss": 0.7404,
      "step": 38530
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.694691552937787,
      "learning_rate": 5.111161706814657e-06,
      "loss": 0.7404,
      "step": 38535
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.417069111162206,
      "learning_rate": 5.109930879666683e-06,
      "loss": 0.7569,
      "step": 38540
    },
    {
      "epoch": 1.63,
      "grad_norm": 8.143259674468673,
      "learning_rate": 5.1087000458539235e-06,
      "loss": 0.7547,
      "step": 38545
    },
    {
      "epoch": 1.63,
      "grad_norm": 12.147930581744653,
      "learning_rate": 5.107469205451001e-06,
      "loss": 0.7441,
      "step": 38550
    },
    {
      "epoch": 1.63,
      "grad_norm": 24.780915801452174,
      "learning_rate": 5.106238358532538e-06,
      "loss": 0.7271,
      "step": 38555
    },
    {
      "epoch": 1.63,
      "grad_norm": 13.954128178077408,
      "learning_rate": 5.105007505173155e-06,
      "loss": 0.7158,
      "step": 38560
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.604538578681041,
      "learning_rate": 5.1037766454474765e-06,
      "loss": 0.7011,
      "step": 38565
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.722472459779395,
      "learning_rate": 5.102545779430126e-06,
      "loss": 0.7613,
      "step": 38570
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.817638966081922,
      "learning_rate": 5.101314907195727e-06,
      "loss": 0.7076,
      "step": 38575
    },
    {
      "epoch": 1.63,
      "grad_norm": 4.462938828105389,
      "learning_rate": 5.100084028818904e-06,
      "loss": 0.748,
      "step": 38580
    },
    {
      "epoch": 1.63,
      "grad_norm": 9.868155733504105,
      "learning_rate": 5.098853144374279e-06,
      "loss": 0.7665,
      "step": 38585
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.851905751315776,
      "learning_rate": 5.0976222539364824e-06,
      "loss": 0.7321,
      "step": 38590
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.22019204904882,
      "learning_rate": 5.096391357580134e-06,
      "loss": 0.7523,
      "step": 38595
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.708833159397612,
      "learning_rate": 5.095160455379861e-06,
      "loss": 0.7406,
      "step": 38600
    },
    {
      "epoch": 1.63,
      "grad_norm": 5.843196684451233,
      "learning_rate": 5.09392954741029e-06,
      "loss": 0.6938,
      "step": 38605
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.061809364327924,
      "learning_rate": 5.09269863374605e-06,
      "loss": 0.7154,
      "step": 38610
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.729138490638303,
      "learning_rate": 5.0914677144617616e-06,
      "loss": 0.7248,
      "step": 38615
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.544875272525832,
      "learning_rate": 5.090236789632056e-06,
      "loss": 0.7064,
      "step": 38620
    },
    {
      "epoch": 1.63,
      "grad_norm": 10.947231427653804,
      "learning_rate": 5.089005859331561e-06,
      "loss": 0.738,
      "step": 38625
    },
    {
      "epoch": 1.63,
      "grad_norm": 7.169062010462444,
      "learning_rate": 5.087774923634901e-06,
      "loss": 0.7374,
      "step": 38630
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.6152549406176275,
      "learning_rate": 5.086543982616706e-06,
      "loss": 0.7261,
      "step": 38635
    },
    {
      "epoch": 1.64,
      "grad_norm": 10.91028501616054,
      "learning_rate": 5.085313036351606e-06,
      "loss": 0.723,
      "step": 38640
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.67682459044136,
      "learning_rate": 5.0840820849142255e-06,
      "loss": 0.7307,
      "step": 38645
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.821486551977015,
      "learning_rate": 5.082851128379198e-06,
      "loss": 0.7239,
      "step": 38650
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.5710057581430075,
      "learning_rate": 5.08162016682115e-06,
      "loss": 0.7385,
      "step": 38655
    },
    {
      "epoch": 1.64,
      "grad_norm": 10.005494273260792,
      "learning_rate": 5.080389200314711e-06,
      "loss": 0.7606,
      "step": 38660
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.410629991613898,
      "learning_rate": 5.079158228934513e-06,
      "loss": 0.7249,
      "step": 38665
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.413364675518424,
      "learning_rate": 5.0779272527551835e-06,
      "loss": 0.7097,
      "step": 38670
    },
    {
      "epoch": 1.64,
      "grad_norm": 10.187205683441405,
      "learning_rate": 5.076696271851357e-06,
      "loss": 0.7458,
      "step": 38675
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.392312320910289,
      "learning_rate": 5.075465286297658e-06,
      "loss": 0.7296,
      "step": 38680
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.512338703056201,
      "learning_rate": 5.074234296168722e-06,
      "loss": 0.7638,
      "step": 38685
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.024975186374453,
      "learning_rate": 5.073003301539179e-06,
      "loss": 0.6939,
      "step": 38690
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.707456538155756,
      "learning_rate": 5.071772302483662e-06,
      "loss": 0.6945,
      "step": 38695
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.027207596013906,
      "learning_rate": 5.070541299076801e-06,
      "loss": 0.7412,
      "step": 38700
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.648928578291132,
      "learning_rate": 5.069310291393228e-06,
      "loss": 0.727,
      "step": 38705
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.7591575323394615,
      "learning_rate": 5.0680792795075774e-06,
      "loss": 0.6787,
      "step": 38710
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.060886596846138,
      "learning_rate": 5.066848263494481e-06,
      "loss": 0.7224,
      "step": 38715
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.155569464062171,
      "learning_rate": 5.06561724342857e-06,
      "loss": 0.7086,
      "step": 38720
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.472648710194574,
      "learning_rate": 5.0643862193844795e-06,
      "loss": 0.7181,
      "step": 38725
    },
    {
      "epoch": 1.64,
      "grad_norm": 19.17727010110648,
      "learning_rate": 5.063155191436842e-06,
      "loss": 0.7455,
      "step": 38730
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.7514713197903005,
      "learning_rate": 5.061924159660291e-06,
      "loss": 0.742,
      "step": 38735
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.786379675127142,
      "learning_rate": 5.060693124129461e-06,
      "loss": 0.7212,
      "step": 38740
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.225353448777373,
      "learning_rate": 5.059462084918985e-06,
      "loss": 0.6941,
      "step": 38745
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.92921086634663,
      "learning_rate": 5.058231042103499e-06,
      "loss": 0.7104,
      "step": 38750
    },
    {
      "epoch": 1.64,
      "grad_norm": 15.895572467852839,
      "learning_rate": 5.0569999957576355e-06,
      "loss": 0.7477,
      "step": 38755
    },
    {
      "epoch": 1.64,
      "grad_norm": 8.54227981184666,
      "learning_rate": 5.055768945956031e-06,
      "loss": 0.736,
      "step": 38760
    },
    {
      "epoch": 1.64,
      "grad_norm": 17.389323403061443,
      "learning_rate": 5.054537892773319e-06,
      "loss": 0.7363,
      "step": 38765
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.1832360109612345,
      "learning_rate": 5.053306836284136e-06,
      "loss": 0.7095,
      "step": 38770
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.507847685104999,
      "learning_rate": 5.052075776563116e-06,
      "loss": 0.7349,
      "step": 38775
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.574810514027077,
      "learning_rate": 5.050844713684894e-06,
      "loss": 0.7639,
      "step": 38780
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.442106607597708,
      "learning_rate": 5.049613647724109e-06,
      "loss": 0.6987,
      "step": 38785
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.0355019388825975,
      "learning_rate": 5.0483825787553915e-06,
      "loss": 0.737,
      "step": 38790
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.681679663465263,
      "learning_rate": 5.047151506853384e-06,
      "loss": 0.7525,
      "step": 38795
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.41609564935384,
      "learning_rate": 5.045920432092718e-06,
      "loss": 0.7196,
      "step": 38800
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.096175048628023,
      "learning_rate": 5.0446893545480315e-06,
      "loss": 0.7239,
      "step": 38805
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.458210694000894,
      "learning_rate": 5.043458274293961e-06,
      "loss": 0.7155,
      "step": 38810
    },
    {
      "epoch": 1.64,
      "grad_norm": 7.151649143661842,
      "learning_rate": 5.042227191405143e-06,
      "loss": 0.6856,
      "step": 38815
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.819867556109795,
      "learning_rate": 5.040996105956217e-06,
      "loss": 0.7484,
      "step": 38820
    },
    {
      "epoch": 1.64,
      "grad_norm": 13.803193459452087,
      "learning_rate": 5.039765018021816e-06,
      "loss": 0.7409,
      "step": 38825
    },
    {
      "epoch": 1.64,
      "grad_norm": 5.57802568665564,
      "learning_rate": 5.038533927676578e-06,
      "loss": 0.7428,
      "step": 38830
    },
    {
      "epoch": 1.64,
      "grad_norm": 10.979246592257468,
      "learning_rate": 5.037302834995145e-06,
      "loss": 0.7076,
      "step": 38835
    },
    {
      "epoch": 1.64,
      "grad_norm": 15.00731497398954,
      "learning_rate": 5.036071740052147e-06,
      "loss": 0.7278,
      "step": 38840
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.659352886272187,
      "learning_rate": 5.034840642922229e-06,
      "loss": 0.7542,
      "step": 38845
    },
    {
      "epoch": 1.64,
      "grad_norm": 9.62919838009354,
      "learning_rate": 5.033609543680023e-06,
      "loss": 0.7113,
      "step": 38850
    },
    {
      "epoch": 1.64,
      "grad_norm": 29.68659959981478,
      "learning_rate": 5.0323784424001695e-06,
      "loss": 0.7319,
      "step": 38855
    },
    {
      "epoch": 1.64,
      "grad_norm": 22.718921663937778,
      "learning_rate": 5.031147339157308e-06,
      "loss": 0.6668,
      "step": 38860
    },
    {
      "epoch": 1.64,
      "grad_norm": 6.583284880052477,
      "learning_rate": 5.0299162340260735e-06,
      "loss": 0.7271,
      "step": 38865
    },
    {
      "epoch": 1.65,
      "grad_norm": 34.708676010055925,
      "learning_rate": 5.0286851270811065e-06,
      "loss": 0.7208,
      "step": 38870
    },
    {
      "epoch": 1.65,
      "grad_norm": 15.969158357696985,
      "learning_rate": 5.0274540183970446e-06,
      "loss": 0.7454,
      "step": 38875
    },
    {
      "epoch": 1.65,
      "grad_norm": 27.920971463055974,
      "learning_rate": 5.026222908048527e-06,
      "loss": 0.7622,
      "step": 38880
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.810871778982594,
      "learning_rate": 5.024991796110193e-06,
      "loss": 0.7124,
      "step": 38885
    },
    {
      "epoch": 1.65,
      "grad_norm": 12.50487040816802,
      "learning_rate": 5.023760682656676e-06,
      "loss": 0.7434,
      "step": 38890
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.911366316988982,
      "learning_rate": 5.022529567762622e-06,
      "loss": 0.7158,
      "step": 38895
    },
    {
      "epoch": 1.65,
      "grad_norm": 12.35380177421443,
      "learning_rate": 5.021298451502665e-06,
      "loss": 0.7578,
      "step": 38900
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.045920455908149,
      "learning_rate": 5.020067333951447e-06,
      "loss": 0.7923,
      "step": 38905
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.046330821111917,
      "learning_rate": 5.018836215183605e-06,
      "loss": 0.7464,
      "step": 38910
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.733348969364037,
      "learning_rate": 5.01760509527378e-06,
      "loss": 0.6798,
      "step": 38915
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.029126130778101,
      "learning_rate": 5.016373974296609e-06,
      "loss": 0.7184,
      "step": 38920
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.708077686296743,
      "learning_rate": 5.015142852326731e-06,
      "loss": 0.7517,
      "step": 38925
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.867181842421245,
      "learning_rate": 5.0139117294387885e-06,
      "loss": 0.7234,
      "step": 38930
    },
    {
      "epoch": 1.65,
      "grad_norm": 15.224036817365553,
      "learning_rate": 5.0126806057074175e-06,
      "loss": 0.7098,
      "step": 38935
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.936656548111875,
      "learning_rate": 5.011449481207259e-06,
      "loss": 0.781,
      "step": 38940
    },
    {
      "epoch": 1.65,
      "grad_norm": 13.211384679672173,
      "learning_rate": 5.010218356012953e-06,
      "loss": 0.7278,
      "step": 38945
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.4169150718912675,
      "learning_rate": 5.008987230199136e-06,
      "loss": 0.7556,
      "step": 38950
    },
    {
      "epoch": 1.65,
      "grad_norm": 19.025581752349648,
      "learning_rate": 5.007756103840451e-06,
      "loss": 0.7185,
      "step": 38955
    },
    {
      "epoch": 1.65,
      "grad_norm": 12.317981852328783,
      "learning_rate": 5.006524977011536e-06,
      "loss": 0.7301,
      "step": 38960
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.732610992722464,
      "learning_rate": 5.005293849787032e-06,
      "loss": 0.7152,
      "step": 38965
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.744911670654091,
      "learning_rate": 5.004062722241577e-06,
      "loss": 0.7225,
      "step": 38970
    },
    {
      "epoch": 1.65,
      "grad_norm": 10.214394631683735,
      "learning_rate": 5.00283159444981e-06,
      "loss": 0.7108,
      "step": 38975
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.411985973658635,
      "learning_rate": 5.0016004664863726e-06,
      "loss": 0.7054,
      "step": 38980
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.516479047345034,
      "learning_rate": 5.000369338425904e-06,
      "loss": 0.7911,
      "step": 38985
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.259185252382097,
      "learning_rate": 4.999138210343043e-06,
      "loss": 0.7097,
      "step": 38990
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.7664153334286175,
      "learning_rate": 4.99790708231243e-06,
      "loss": 0.6867,
      "step": 38995
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.418664024187641,
      "learning_rate": 4.996675954408703e-06,
      "loss": 0.7566,
      "step": 39000
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.684163844704137,
      "learning_rate": 4.995444826706505e-06,
      "loss": 0.7507,
      "step": 39005
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.367469543189487,
      "learning_rate": 4.994213699280473e-06,
      "loss": 0.7564,
      "step": 39010
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.894946726685241,
      "learning_rate": 4.992982572205248e-06,
      "loss": 0.7026,
      "step": 39015
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.322554527229888,
      "learning_rate": 4.991751445555468e-06,
      "loss": 0.7298,
      "step": 39020
    },
    {
      "epoch": 1.65,
      "grad_norm": 10.237237565124225,
      "learning_rate": 4.990520319405774e-06,
      "loss": 0.7274,
      "step": 39025
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.291115550015801,
      "learning_rate": 4.989289193830805e-06,
      "loss": 0.7115,
      "step": 39030
    },
    {
      "epoch": 1.65,
      "grad_norm": 8.642118633755736,
      "learning_rate": 4.988058068905202e-06,
      "loss": 0.7116,
      "step": 39035
    },
    {
      "epoch": 1.65,
      "grad_norm": 12.411211068878652,
      "learning_rate": 4.986826944703603e-06,
      "loss": 0.6878,
      "step": 39040
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.40767518200429,
      "learning_rate": 4.985595821300645e-06,
      "loss": 0.7228,
      "step": 39045
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.229935722078179,
      "learning_rate": 4.984364698770972e-06,
      "loss": 0.7251,
      "step": 39050
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.307865294263673,
      "learning_rate": 4.983133577189221e-06,
      "loss": 0.7157,
      "step": 39055
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.522544810074576,
      "learning_rate": 4.981902456630031e-06,
      "loss": 0.7121,
      "step": 39060
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.414956432548025,
      "learning_rate": 4.980671337168042e-06,
      "loss": 0.7081,
      "step": 39065
    },
    {
      "epoch": 1.65,
      "grad_norm": 6.829418573053323,
      "learning_rate": 4.979440218877892e-06,
      "loss": 0.7757,
      "step": 39070
    },
    {
      "epoch": 1.65,
      "grad_norm": 17.100654891590057,
      "learning_rate": 4.978209101834222e-06,
      "loss": 0.7311,
      "step": 39075
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.431005134202259,
      "learning_rate": 4.97697798611167e-06,
      "loss": 0.7256,
      "step": 39080
    },
    {
      "epoch": 1.65,
      "grad_norm": 13.696994487838916,
      "learning_rate": 4.975746871784874e-06,
      "loss": 0.7508,
      "step": 39085
    },
    {
      "epoch": 1.65,
      "grad_norm": 14.161268596714205,
      "learning_rate": 4.9745157589284745e-06,
      "loss": 0.7231,
      "step": 39090
    },
    {
      "epoch": 1.65,
      "grad_norm": 5.352870700421167,
      "learning_rate": 4.9732846476171094e-06,
      "loss": 0.735,
      "step": 39095
    },
    {
      "epoch": 1.65,
      "grad_norm": 9.547002155121941,
      "learning_rate": 4.972053537925418e-06,
      "loss": 0.7196,
      "step": 39100
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.9030813411944125,
      "learning_rate": 4.970822429928037e-06,
      "loss": 0.7626,
      "step": 39105
    },
    {
      "epoch": 1.66,
      "grad_norm": 11.560078119663775,
      "learning_rate": 4.969591323699607e-06,
      "loss": 0.7205,
      "step": 39110
    },
    {
      "epoch": 1.66,
      "grad_norm": 3.9546946218618078,
      "learning_rate": 4.9683602193147654e-06,
      "loss": 0.7055,
      "step": 39115
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.182221932789773,
      "learning_rate": 4.967129116848149e-06,
      "loss": 0.7228,
      "step": 39120
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.431510659738178,
      "learning_rate": 4.965898016374399e-06,
      "loss": 0.7037,
      "step": 39125
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.751674381278151,
      "learning_rate": 4.964666917968152e-06,
      "loss": 0.7312,
      "step": 39130
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.723498859834188,
      "learning_rate": 4.9634358217040445e-06,
      "loss": 0.757,
      "step": 39135
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.49013897309288,
      "learning_rate": 4.962204727656717e-06,
      "loss": 0.7199,
      "step": 39140
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.59015418272288,
      "learning_rate": 4.960973635900804e-06,
      "loss": 0.7022,
      "step": 39145
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.595045530551842,
      "learning_rate": 4.959742546510946e-06,
      "loss": 0.7503,
      "step": 39150
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.309591454294607,
      "learning_rate": 4.958511459561778e-06,
      "loss": 0.6915,
      "step": 39155
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.699324864320143,
      "learning_rate": 4.957280375127937e-06,
      "loss": 0.733,
      "step": 39160
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.296896202533893,
      "learning_rate": 4.956049293284062e-06,
      "loss": 0.7018,
      "step": 39165
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.282659105772536,
      "learning_rate": 4.954818214104789e-06,
      "loss": 0.6878,
      "step": 39170
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.660140119309361,
      "learning_rate": 4.9535871376647546e-06,
      "loss": 0.731,
      "step": 39175
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.693880405776472,
      "learning_rate": 4.952356064038595e-06,
      "loss": 0.732,
      "step": 39180
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.141226340867173,
      "learning_rate": 4.951124993300947e-06,
      "loss": 0.7091,
      "step": 39185
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.238699535472709,
      "learning_rate": 4.949893925526448e-06,
      "loss": 0.72,
      "step": 39190
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.409368649569806,
      "learning_rate": 4.9486628607897304e-06,
      "loss": 0.7547,
      "step": 39195
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.090632741482874,
      "learning_rate": 4.947431799165436e-06,
      "loss": 0.7536,
      "step": 39200
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.960384708621088,
      "learning_rate": 4.946200740728193e-06,
      "loss": 0.7092,
      "step": 39205
    },
    {
      "epoch": 1.66,
      "grad_norm": 13.668174317046393,
      "learning_rate": 4.9449696855526425e-06,
      "loss": 0.7431,
      "step": 39210
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.209868264852339,
      "learning_rate": 4.9437386337134165e-06,
      "loss": 0.6946,
      "step": 39215
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.768132793817733,
      "learning_rate": 4.942507585285152e-06,
      "loss": 0.7295,
      "step": 39220
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.589142410257591,
      "learning_rate": 4.941276540342485e-06,
      "loss": 0.7153,
      "step": 39225
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.93283486789867,
      "learning_rate": 4.9400454989600456e-06,
      "loss": 0.7498,
      "step": 39230
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.458084878919453,
      "learning_rate": 4.938814461212472e-06,
      "loss": 0.7639,
      "step": 39235
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.275800369154017,
      "learning_rate": 4.937583427174397e-06,
      "loss": 0.7189,
      "step": 39240
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.333813191696346,
      "learning_rate": 4.936352396920457e-06,
      "loss": 0.7347,
      "step": 39245
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.784280836713963,
      "learning_rate": 4.935121370525281e-06,
      "loss": 0.7298,
      "step": 39250
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.472661722034233,
      "learning_rate": 4.933890348063507e-06,
      "loss": 0.7376,
      "step": 39255
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.959245294242832,
      "learning_rate": 4.932659329609766e-06,
      "loss": 0.7467,
      "step": 39260
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.853706487299998,
      "learning_rate": 4.931428315238691e-06,
      "loss": 0.7201,
      "step": 39265
    },
    {
      "epoch": 1.66,
      "grad_norm": 8.81919366211316,
      "learning_rate": 4.930197305024915e-06,
      "loss": 0.7198,
      "step": 39270
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.204611284738818,
      "learning_rate": 4.928966299043072e-06,
      "loss": 0.7467,
      "step": 39275
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.056566342560484,
      "learning_rate": 4.927735297367792e-06,
      "loss": 0.7232,
      "step": 39280
    },
    {
      "epoch": 1.66,
      "grad_norm": 12.032987357069556,
      "learning_rate": 4.926504300073709e-06,
      "loss": 0.7312,
      "step": 39285
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.480974416592697,
      "learning_rate": 4.925273307235454e-06,
      "loss": 0.69,
      "step": 39290
    },
    {
      "epoch": 1.66,
      "grad_norm": 13.862688240696299,
      "learning_rate": 4.924042318927656e-06,
      "loss": 0.711,
      "step": 39295
    },
    {
      "epoch": 1.66,
      "grad_norm": 7.048897852494733,
      "learning_rate": 4.922811335224951e-06,
      "loss": 0.7429,
      "step": 39300
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.908037277333349,
      "learning_rate": 4.921580356201966e-06,
      "loss": 0.7406,
      "step": 39305
    },
    {
      "epoch": 1.66,
      "grad_norm": 4.9227218040384795,
      "learning_rate": 4.920349381933333e-06,
      "loss": 0.7098,
      "step": 39310
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.58803533390825,
      "learning_rate": 4.919118412493683e-06,
      "loss": 0.7395,
      "step": 39315
    },
    {
      "epoch": 1.66,
      "grad_norm": 14.716148994417422,
      "learning_rate": 4.917887447957645e-06,
      "loss": 0.698,
      "step": 39320
    },
    {
      "epoch": 1.66,
      "grad_norm": 10.682936779010463,
      "learning_rate": 4.91665648839985e-06,
      "loss": 0.7289,
      "step": 39325
    },
    {
      "epoch": 1.66,
      "grad_norm": 5.268157920334467,
      "learning_rate": 4.915425533894925e-06,
      "loss": 0.6791,
      "step": 39330
    },
    {
      "epoch": 1.66,
      "grad_norm": 10.329798296391582,
      "learning_rate": 4.914194584517502e-06,
      "loss": 0.7317,
      "step": 39335
    },
    {
      "epoch": 1.66,
      "grad_norm": 6.5981460657342845,
      "learning_rate": 4.9129636403422086e-06,
      "loss": 0.7316,
      "step": 39340
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.090351394412334,
      "learning_rate": 4.911732701443673e-06,
      "loss": 0.715,
      "step": 39345
    },
    {
      "epoch": 1.67,
      "grad_norm": 21.277789314270045,
      "learning_rate": 4.910501767896524e-06,
      "loss": 0.771,
      "step": 39350
    },
    {
      "epoch": 1.67,
      "grad_norm": 10.221220727549627,
      "learning_rate": 4.909270839775388e-06,
      "loss": 0.7229,
      "step": 39355
    },
    {
      "epoch": 1.67,
      "grad_norm": 10.009485056935914,
      "learning_rate": 4.908039917154897e-06,
      "loss": 0.7358,
      "step": 39360
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.703508019823032,
      "learning_rate": 4.906809000109672e-06,
      "loss": 0.7203,
      "step": 39365
    },
    {
      "epoch": 1.67,
      "grad_norm": 11.042882271869123,
      "learning_rate": 4.905578088714344e-06,
      "loss": 0.7013,
      "step": 39370
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.131777378595308,
      "learning_rate": 4.904347183043537e-06,
      "loss": 0.7293,
      "step": 39375
    },
    {
      "epoch": 1.67,
      "grad_norm": 11.986949587345036,
      "learning_rate": 4.9031162831718795e-06,
      "loss": 0.7124,
      "step": 39380
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.590058305911143,
      "learning_rate": 4.9018853891739965e-06,
      "loss": 0.6811,
      "step": 39385
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.316860790830962,
      "learning_rate": 4.9006545011245125e-06,
      "loss": 0.7493,
      "step": 39390
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.72287251393382,
      "learning_rate": 4.8994236190980545e-06,
      "loss": 0.7174,
      "step": 39395
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.854920710539936,
      "learning_rate": 4.898192743169245e-06,
      "loss": 0.7014,
      "step": 39400
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.728179292396962,
      "learning_rate": 4.896961873412711e-06,
      "loss": 0.7076,
      "step": 39405
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.259499526374727,
      "learning_rate": 4.895731009903075e-06,
      "loss": 0.7227,
      "step": 39410
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.491012071798648,
      "learning_rate": 4.8945001527149585e-06,
      "loss": 0.7395,
      "step": 39415
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.581669912223495,
      "learning_rate": 4.8932693019229895e-06,
      "loss": 0.7019,
      "step": 39420
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.3564090777258695,
      "learning_rate": 4.892038457601786e-06,
      "loss": 0.7311,
      "step": 39425
    },
    {
      "epoch": 1.67,
      "grad_norm": 12.427237124117909,
      "learning_rate": 4.890807619825975e-06,
      "loss": 0.7517,
      "step": 39430
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.00105854768971,
      "learning_rate": 4.8895767886701765e-06,
      "loss": 0.7061,
      "step": 39435
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.239250098341716,
      "learning_rate": 4.88834596420901e-06,
      "loss": 0.7166,
      "step": 39440
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.070223604284234,
      "learning_rate": 4.887115146517101e-06,
      "loss": 0.699,
      "step": 39445
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.760926303968526,
      "learning_rate": 4.885884335669066e-06,
      "loss": 0.7095,
      "step": 39450
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.894969297362118,
      "learning_rate": 4.884653531739529e-06,
      "loss": 0.7275,
      "step": 39455
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.423920306033041,
      "learning_rate": 4.883422734803109e-06,
      "loss": 0.7167,
      "step": 39460
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.30283957107836,
      "learning_rate": 4.882191944934424e-06,
      "loss": 0.7234,
      "step": 39465
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.114078170419382,
      "learning_rate": 4.8809611622080955e-06,
      "loss": 0.7554,
      "step": 39470
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.570009689911861,
      "learning_rate": 4.8797303866987395e-06,
      "loss": 0.7097,
      "step": 39475
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.13467054813339,
      "learning_rate": 4.8784996184809776e-06,
      "loss": 0.728,
      "step": 39480
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.694407376195911,
      "learning_rate": 4.877268857629425e-06,
      "loss": 0.762,
      "step": 39485
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.658332704738483,
      "learning_rate": 4.8760381042187e-06,
      "loss": 0.7053,
      "step": 39490
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.97803009345767,
      "learning_rate": 4.87480735832342e-06,
      "loss": 0.7013,
      "step": 39495
    },
    {
      "epoch": 1.67,
      "grad_norm": 10.12072431355536,
      "learning_rate": 4.873576620018201e-06,
      "loss": 0.712,
      "step": 39500
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.419425943375269,
      "learning_rate": 4.872345889377659e-06,
      "loss": 0.6834,
      "step": 39505
    },
    {
      "epoch": 1.67,
      "grad_norm": 7.293104090786496,
      "learning_rate": 4.87111516647641e-06,
      "loss": 0.6874,
      "step": 39510
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.941988238994929,
      "learning_rate": 4.869884451389068e-06,
      "loss": 0.7419,
      "step": 39515
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.253452100644815,
      "learning_rate": 4.86865374419025e-06,
      "loss": 0.6969,
      "step": 39520
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.262419430038404,
      "learning_rate": 4.867423044954569e-06,
      "loss": 0.7412,
      "step": 39525
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.460377314094931,
      "learning_rate": 4.8661923537566365e-06,
      "loss": 0.7011,
      "step": 39530
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.712015111269397,
      "learning_rate": 4.864961670671068e-06,
      "loss": 0.6956,
      "step": 39535
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.966923424402484,
      "learning_rate": 4.863730995772476e-06,
      "loss": 0.7222,
      "step": 39540
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.59073988007475,
      "learning_rate": 4.862500329135471e-06,
      "loss": 0.6867,
      "step": 39545
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.46026171760473,
      "learning_rate": 4.861269670834668e-06,
      "loss": 0.7002,
      "step": 39550
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.2631493064104635,
      "learning_rate": 4.860039020944674e-06,
      "loss": 0.7512,
      "step": 39555
    },
    {
      "epoch": 1.67,
      "grad_norm": 9.416840917886876,
      "learning_rate": 4.858808379540103e-06,
      "loss": 0.7044,
      "step": 39560
    },
    {
      "epoch": 1.67,
      "grad_norm": 6.351489257132697,
      "learning_rate": 4.857577746695564e-06,
      "loss": 0.7312,
      "step": 39565
    },
    {
      "epoch": 1.67,
      "grad_norm": 8.825593039475388,
      "learning_rate": 4.856347122485666e-06,
      "loss": 0.6902,
      "step": 39570
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.794633943493452,
      "learning_rate": 4.85511650698502e-06,
      "loss": 0.7127,
      "step": 39575
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.177521844398603,
      "learning_rate": 4.853885900268234e-06,
      "loss": 0.7398,
      "step": 39580
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.493641002621619,
      "learning_rate": 4.852655302409913e-06,
      "loss": 0.724,
      "step": 39585
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.3382402083022065,
      "learning_rate": 4.851424713484669e-06,
      "loss": 0.7383,
      "step": 39590
    },
    {
      "epoch": 1.68,
      "grad_norm": 19.71317207737549,
      "learning_rate": 4.850194133567107e-06,
      "loss": 0.7353,
      "step": 39595
    },
    {
      "epoch": 1.68,
      "grad_norm": 17.52905487377739,
      "learning_rate": 4.8489635627318325e-06,
      "loss": 0.7245,
      "step": 39600
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.14286226907859,
      "learning_rate": 4.847733001053453e-06,
      "loss": 0.6979,
      "step": 39605
    },
    {
      "epoch": 1.68,
      "grad_norm": 11.522905329586116,
      "learning_rate": 4.846502448606572e-06,
      "loss": 0.7393,
      "step": 39610
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.40470777967269,
      "learning_rate": 4.845271905465796e-06,
      "loss": 0.6901,
      "step": 39615
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.47893831147272,
      "learning_rate": 4.844041371705729e-06,
      "loss": 0.7544,
      "step": 39620
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.943106389762857,
      "learning_rate": 4.842810847400974e-06,
      "loss": 0.7191,
      "step": 39625
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.756408825568372,
      "learning_rate": 4.841580332626135e-06,
      "loss": 0.7082,
      "step": 39630
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.315338751477616,
      "learning_rate": 4.840349827455812e-06,
      "loss": 0.7228,
      "step": 39635
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.401464589143955,
      "learning_rate": 4.83911933196461e-06,
      "loss": 0.7146,
      "step": 39640
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.485151629827089,
      "learning_rate": 4.83788884622713e-06,
      "loss": 0.7651,
      "step": 39645
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.114077904628738,
      "learning_rate": 4.8366583703179696e-06,
      "loss": 0.7168,
      "step": 39650
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.145564990854587,
      "learning_rate": 4.835427904311733e-06,
      "loss": 0.7105,
      "step": 39655
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.561224035126282,
      "learning_rate": 4.834197448283016e-06,
      "loss": 0.7441,
      "step": 39660
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.0698409652583845,
      "learning_rate": 4.83296700230642e-06,
      "loss": 0.69,
      "step": 39665
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.601807455960646,
      "learning_rate": 4.831736566456543e-06,
      "loss": 0.7276,
      "step": 39670
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.624826856869231,
      "learning_rate": 4.830506140807982e-06,
      "loss": 0.7135,
      "step": 39675
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.848990327930977,
      "learning_rate": 4.829275725435335e-06,
      "loss": 0.7073,
      "step": 39680
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.437641584231019,
      "learning_rate": 4.828045320413198e-06,
      "loss": 0.7009,
      "step": 39685
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.379580425533575,
      "learning_rate": 4.826814925816167e-06,
      "loss": 0.6975,
      "step": 39690
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.831021137783145,
      "learning_rate": 4.825584541718835e-06,
      "loss": 0.7528,
      "step": 39695
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.578861034865445,
      "learning_rate": 4.8243541681958e-06,
      "loss": 0.6722,
      "step": 39700
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.291883998533101,
      "learning_rate": 4.8231238053216534e-06,
      "loss": 0.6925,
      "step": 39705
    },
    {
      "epoch": 1.68,
      "grad_norm": 7.360668821145277,
      "learning_rate": 4.821893453170991e-06,
      "loss": 0.688,
      "step": 39710
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.406708146098044,
      "learning_rate": 4.820663111818402e-06,
      "loss": 0.7119,
      "step": 39715
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.521141169289952,
      "learning_rate": 4.819432781338481e-06,
      "loss": 0.7396,
      "step": 39720
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.283070472792619,
      "learning_rate": 4.818202461805819e-06,
      "loss": 0.7734,
      "step": 39725
    },
    {
      "epoch": 1.68,
      "grad_norm": 11.634371063328796,
      "learning_rate": 4.816972153295006e-06,
      "loss": 0.6897,
      "step": 39730
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.129365848161817,
      "learning_rate": 4.815741855880632e-06,
      "loss": 0.732,
      "step": 39735
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.686810705572237,
      "learning_rate": 4.814511569637288e-06,
      "loss": 0.7025,
      "step": 39740
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.356507928196501,
      "learning_rate": 4.813281294639559e-06,
      "loss": 0.6915,
      "step": 39745
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.483271930102498,
      "learning_rate": 4.812051030962036e-06,
      "loss": 0.7392,
      "step": 39750
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.1640249320992,
      "learning_rate": 4.810820778679305e-06,
      "loss": 0.7636,
      "step": 39755
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.4631727948873365,
      "learning_rate": 4.8095905378659535e-06,
      "loss": 0.7272,
      "step": 39760
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.129178105104451,
      "learning_rate": 4.808360308596568e-06,
      "loss": 0.6786,
      "step": 39765
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.24952624696542,
      "learning_rate": 4.80713009094573e-06,
      "loss": 0.7135,
      "step": 39770
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.461017439161896,
      "learning_rate": 4.8058998849880275e-06,
      "loss": 0.7123,
      "step": 39775
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.854926805542565,
      "learning_rate": 4.804669690798043e-06,
      "loss": 0.717,
      "step": 39780
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.179146733034479,
      "learning_rate": 4.803439508450361e-06,
      "loss": 0.6931,
      "step": 39785
    },
    {
      "epoch": 1.68,
      "grad_norm": 5.040728834183274,
      "learning_rate": 4.8022093380195615e-06,
      "loss": 0.6892,
      "step": 39790
    },
    {
      "epoch": 1.68,
      "grad_norm": 6.425779939277026,
      "learning_rate": 4.800979179580227e-06,
      "loss": 0.716,
      "step": 39795
    },
    {
      "epoch": 1.68,
      "grad_norm": 4.249803249537637,
      "learning_rate": 4.79974903320694e-06,
      "loss": 0.688,
      "step": 39800
    },
    {
      "epoch": 1.68,
      "grad_norm": 9.369177953038742,
      "learning_rate": 4.798518898974277e-06,
      "loss": 0.701,
      "step": 39805
    },
    {
      "epoch": 1.68,
      "grad_norm": 8.63859999371445,
      "learning_rate": 4.797288776956822e-06,
      "loss": 0.7096,
      "step": 39810
    },
    {
      "epoch": 1.69,
      "grad_norm": 10.550037524415822,
      "learning_rate": 4.79605866722915e-06,
      "loss": 0.7181,
      "step": 39815
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.951470479398299,
      "learning_rate": 4.7948285698658405e-06,
      "loss": 0.7087,
      "step": 39820
    },
    {
      "epoch": 1.69,
      "grad_norm": 11.595479918201391,
      "learning_rate": 4.7935984849414705e-06,
      "loss": 0.7452,
      "step": 39825
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.6329078374578705,
      "learning_rate": 4.792368412530616e-06,
      "loss": 0.7389,
      "step": 39830
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.095987535558149,
      "learning_rate": 4.791138352707854e-06,
      "loss": 0.7478,
      "step": 39835
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.274888306313262,
      "learning_rate": 4.789908305547759e-06,
      "loss": 0.6992,
      "step": 39840
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.35511128294863,
      "learning_rate": 4.788678271124904e-06,
      "loss": 0.7184,
      "step": 39845
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.431734428530755,
      "learning_rate": 4.787448249513863e-06,
      "loss": 0.744,
      "step": 39850
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.7091540250472255,
      "learning_rate": 4.786218240789208e-06,
      "loss": 0.7274,
      "step": 39855
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.867142880139954,
      "learning_rate": 4.784988245025512e-06,
      "loss": 0.7613,
      "step": 39860
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.280060261176426,
      "learning_rate": 4.7837582622973436e-06,
      "loss": 0.7267,
      "step": 39865
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.942731454481499,
      "learning_rate": 4.7825282926792765e-06,
      "loss": 0.708,
      "step": 39870
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.0298091550005815,
      "learning_rate": 4.781298336245879e-06,
      "loss": 0.7299,
      "step": 39875
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.899510424477942,
      "learning_rate": 4.780068393071717e-06,
      "loss": 0.6671,
      "step": 39880
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.5852446082870175,
      "learning_rate": 4.778838463231361e-06,
      "loss": 0.7021,
      "step": 39885
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.093748580642361,
      "learning_rate": 4.777608546799378e-06,
      "loss": 0.7335,
      "step": 39890
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.192362579982167,
      "learning_rate": 4.776378643850335e-06,
      "loss": 0.6636,
      "step": 39895
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.033108171641508,
      "learning_rate": 4.775148754458795e-06,
      "loss": 0.7086,
      "step": 39900
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.275264768661221,
      "learning_rate": 4.7739188786993225e-06,
      "loss": 0.701,
      "step": 39905
    },
    {
      "epoch": 1.69,
      "grad_norm": 6.885510907876548,
      "learning_rate": 4.772689016646484e-06,
      "loss": 0.6726,
      "step": 39910
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.558695783358456,
      "learning_rate": 4.77145916837484e-06,
      "loss": 0.7225,
      "step": 39915
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.217895765645389,
      "learning_rate": 4.770229333958955e-06,
      "loss": 0.7182,
      "step": 39920
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.759919808358852,
      "learning_rate": 4.768999513473387e-06,
      "loss": 0.7038,
      "step": 39925
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.404276679649746,
      "learning_rate": 4.767769706992698e-06,
      "loss": 0.7196,
      "step": 39930
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.974461983498978,
      "learning_rate": 4.766539914591449e-06,
      "loss": 0.7524,
      "step": 39935
    },
    {
      "epoch": 1.69,
      "grad_norm": 3.9847870538195918,
      "learning_rate": 4.7653101363441955e-06,
      "loss": 0.7003,
      "step": 39940
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.05523154703765,
      "learning_rate": 4.764080372325499e-06,
      "loss": 0.754,
      "step": 39945
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.210498620113864,
      "learning_rate": 4.762850622609914e-06,
      "loss": 0.7019,
      "step": 39950
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.861466082402762,
      "learning_rate": 4.761620887271996e-06,
      "loss": 0.6973,
      "step": 39955
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.101914122790328,
      "learning_rate": 4.760391166386303e-06,
      "loss": 0.7179,
      "step": 39960
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.648802924640338,
      "learning_rate": 4.759161460027387e-06,
      "loss": 0.7252,
      "step": 39965
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.123606777815109,
      "learning_rate": 4.757931768269803e-06,
      "loss": 0.6936,
      "step": 39970
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.756488595971078,
      "learning_rate": 4.756702091188102e-06,
      "loss": 0.721,
      "step": 39975
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.332662992986555,
      "learning_rate": 4.755472428856836e-06,
      "loss": 0.7271,
      "step": 39980
    },
    {
      "epoch": 1.69,
      "grad_norm": 8.559651767187804,
      "learning_rate": 4.7542427813505575e-06,
      "loss": 0.7427,
      "step": 39985
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.180718016379323,
      "learning_rate": 4.753013148743815e-06,
      "loss": 0.7406,
      "step": 39990
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.217809733792976,
      "learning_rate": 4.7517835311111575e-06,
      "loss": 0.7086,
      "step": 39995
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.745238403338596,
      "learning_rate": 4.750553928527134e-06,
      "loss": 0.6998,
      "step": 40000
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.324525218520384,
      "learning_rate": 4.74932434106629e-06,
      "loss": 0.6864,
      "step": 40005
    },
    {
      "epoch": 1.69,
      "grad_norm": 4.891762423544337,
      "learning_rate": 4.748094768803175e-06,
      "loss": 0.7095,
      "step": 40010
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.027405626283015,
      "learning_rate": 4.746865211812331e-06,
      "loss": 0.7473,
      "step": 40015
    },
    {
      "epoch": 1.69,
      "grad_norm": 7.488878478823422,
      "learning_rate": 4.745635670168304e-06,
      "loss": 0.7583,
      "step": 40020
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.8943238792251,
      "learning_rate": 4.744406143945636e-06,
      "loss": 0.6804,
      "step": 40025
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.976427902553,
      "learning_rate": 4.743176633218871e-06,
      "loss": 0.7185,
      "step": 40030
    },
    {
      "epoch": 1.69,
      "grad_norm": 14.360049235549553,
      "learning_rate": 4.741947138062549e-06,
      "loss": 0.6896,
      "step": 40035
    },
    {
      "epoch": 1.69,
      "grad_norm": 5.210505362779091,
      "learning_rate": 4.740717658551214e-06,
      "loss": 0.6848,
      "step": 40040
    },
    {
      "epoch": 1.69,
      "grad_norm": 10.888739064856459,
      "learning_rate": 4.7394881947594026e-06,
      "loss": 0.6733,
      "step": 40045
    },
    {
      "epoch": 1.69,
      "grad_norm": 9.480799646652288,
      "learning_rate": 4.738258746761654e-06,
      "loss": 0.7187,
      "step": 40050
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.244256540168746,
      "learning_rate": 4.737029314632507e-06,
      "loss": 0.7168,
      "step": 40055
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.070106624832603,
      "learning_rate": 4.735799898446498e-06,
      "loss": 0.7056,
      "step": 40060
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.54451239076145,
      "learning_rate": 4.734570498278164e-06,
      "loss": 0.6979,
      "step": 40065
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.4103768862418455,
      "learning_rate": 4.733341114202038e-06,
      "loss": 0.7191,
      "step": 40070
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.37334694015689,
      "learning_rate": 4.732111746292654e-06,
      "loss": 0.7318,
      "step": 40075
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.284819691901522,
      "learning_rate": 4.730882394624547e-06,
      "loss": 0.6985,
      "step": 40080
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.879646331275877,
      "learning_rate": 4.7296530592722464e-06,
      "loss": 0.7274,
      "step": 40085
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.413134281956985,
      "learning_rate": 4.7284237403102864e-06,
      "loss": 0.7219,
      "step": 40090
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.627777403998653,
      "learning_rate": 4.727194437813194e-06,
      "loss": 0.6732,
      "step": 40095
    },
    {
      "epoch": 1.7,
      "grad_norm": 7.022429373693186,
      "learning_rate": 4.725965151855499e-06,
      "loss": 0.717,
      "step": 40100
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.161534552545788,
      "learning_rate": 4.72473588251173e-06,
      "loss": 0.718,
      "step": 40105
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.222594908998444,
      "learning_rate": 4.723506629856413e-06,
      "loss": 0.7452,
      "step": 40110
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.164243188413465,
      "learning_rate": 4.722277393964075e-06,
      "loss": 0.6857,
      "step": 40115
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.911601423939285,
      "learning_rate": 4.721048174909241e-06,
      "loss": 0.7031,
      "step": 40120
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.303066566934856,
      "learning_rate": 4.719818972766434e-06,
      "loss": 0.6681,
      "step": 40125
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.73612148922154,
      "learning_rate": 4.718589787610178e-06,
      "loss": 0.7469,
      "step": 40130
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.392500709570763,
      "learning_rate": 4.717360619514994e-06,
      "loss": 0.709,
      "step": 40135
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.327038120983462,
      "learning_rate": 4.716131468555402e-06,
      "loss": 0.7211,
      "step": 40140
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.608148913910321,
      "learning_rate": 4.7149023348059245e-06,
      "loss": 0.7105,
      "step": 40145
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.5791389418503465,
      "learning_rate": 4.713673218341077e-06,
      "loss": 0.7067,
      "step": 40150
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.9580663846247655,
      "learning_rate": 4.712444119235381e-06,
      "loss": 0.7043,
      "step": 40155
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.5917155931653015,
      "learning_rate": 4.71121503756335e-06,
      "loss": 0.6919,
      "step": 40160
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.16942288658068,
      "learning_rate": 4.709985973399499e-06,
      "loss": 0.7134,
      "step": 40165
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.608652546049841,
      "learning_rate": 4.708756926818348e-06,
      "loss": 0.698,
      "step": 40170
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.745420255080977,
      "learning_rate": 4.707527897894404e-06,
      "loss": 0.7095,
      "step": 40175
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.887335736110537,
      "learning_rate": 4.706298886702183e-06,
      "loss": 0.6958,
      "step": 40180
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.485722528390209,
      "learning_rate": 4.705069893316194e-06,
      "loss": 0.7338,
      "step": 40185
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.552133281193531,
      "learning_rate": 4.70384091781095e-06,
      "loss": 0.7142,
      "step": 40190
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.302154707740554,
      "learning_rate": 4.702611960260957e-06,
      "loss": 0.6743,
      "step": 40195
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.522868850517307,
      "learning_rate": 4.701383020740727e-06,
      "loss": 0.7358,
      "step": 40200
    },
    {
      "epoch": 1.7,
      "grad_norm": 12.653221367737457,
      "learning_rate": 4.700154099324764e-06,
      "loss": 0.696,
      "step": 40205
    },
    {
      "epoch": 1.7,
      "grad_norm": 10.998142257772557,
      "learning_rate": 4.698925196087573e-06,
      "loss": 0.7075,
      "step": 40210
    },
    {
      "epoch": 1.7,
      "grad_norm": 6.408545515555583,
      "learning_rate": 4.697696311103663e-06,
      "loss": 0.7205,
      "step": 40215
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.825649986232741,
      "learning_rate": 4.696467444447534e-06,
      "loss": 0.683,
      "step": 40220
    },
    {
      "epoch": 1.7,
      "grad_norm": 11.771799509366787,
      "learning_rate": 4.69523859619369e-06,
      "loss": 0.7399,
      "step": 40225
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.796182479290588,
      "learning_rate": 4.694009766416633e-06,
      "loss": 0.7237,
      "step": 40230
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.663438462396627,
      "learning_rate": 4.692780955190861e-06,
      "loss": 0.7054,
      "step": 40235
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.348336445676226,
      "learning_rate": 4.691552162590876e-06,
      "loss": 0.7062,
      "step": 40240
    },
    {
      "epoch": 1.7,
      "grad_norm": 5.866163268084782,
      "learning_rate": 4.690323388691173e-06,
      "loss": 0.7178,
      "step": 40245
    },
    {
      "epoch": 1.7,
      "grad_norm": 15.66581819928323,
      "learning_rate": 4.689094633566253e-06,
      "loss": 0.722,
      "step": 40250
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.9008825748689535,
      "learning_rate": 4.687865897290609e-06,
      "loss": 0.7434,
      "step": 40255
    },
    {
      "epoch": 1.7,
      "grad_norm": 12.788330379681447,
      "learning_rate": 4.686637179938736e-06,
      "loss": 0.6936,
      "step": 40260
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.528206323309659,
      "learning_rate": 4.685408481585128e-06,
      "loss": 0.7162,
      "step": 40265
    },
    {
      "epoch": 1.7,
      "grad_norm": 9.395030726229628,
      "learning_rate": 4.684179802304276e-06,
      "loss": 0.7015,
      "step": 40270
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.228246568916481,
      "learning_rate": 4.682951142170673e-06,
      "loss": 0.6738,
      "step": 40275
    },
    {
      "epoch": 1.7,
      "grad_norm": 13.89603387132787,
      "learning_rate": 4.681722501258808e-06,
      "loss": 0.7307,
      "step": 40280
    },
    {
      "epoch": 1.7,
      "grad_norm": 8.846280609819637,
      "learning_rate": 4.680493879643171e-06,
      "loss": 0.6938,
      "step": 40285
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.229496517600412,
      "learning_rate": 4.679265277398248e-06,
      "loss": 0.7186,
      "step": 40290
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.264446525603521,
      "learning_rate": 4.6780366945985265e-06,
      "loss": 0.6943,
      "step": 40295
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.812768525411609,
      "learning_rate": 4.6768081313184935e-06,
      "loss": 0.7278,
      "step": 40300
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.67039790149491,
      "learning_rate": 4.675579587632631e-06,
      "loss": 0.7026,
      "step": 40305
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.098840076953424,
      "learning_rate": 4.674351063615421e-06,
      "loss": 0.7088,
      "step": 40310
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.757394437001381,
      "learning_rate": 4.673122559341348e-06,
      "loss": 0.7291,
      "step": 40315
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.767413468165906,
      "learning_rate": 4.67189407488489e-06,
      "loss": 0.6992,
      "step": 40320
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.593469047111174,
      "learning_rate": 4.6706656103205295e-06,
      "loss": 0.6824,
      "step": 40325
    },
    {
      "epoch": 1.71,
      "grad_norm": 10.180915464428734,
      "learning_rate": 4.6694371657227425e-06,
      "loss": 0.7234,
      "step": 40330
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.082112532173597,
      "learning_rate": 4.668208741166007e-06,
      "loss": 0.7265,
      "step": 40335
    },
    {
      "epoch": 1.71,
      "grad_norm": 18.246601260167385,
      "learning_rate": 4.666980336724797e-06,
      "loss": 0.6961,
      "step": 40340
    },
    {
      "epoch": 1.71,
      "grad_norm": 11.592899768537889,
      "learning_rate": 4.665751952473589e-06,
      "loss": 0.6947,
      "step": 40345
    },
    {
      "epoch": 1.71,
      "grad_norm": 11.207575120032372,
      "learning_rate": 4.6645235884868555e-06,
      "loss": 0.7228,
      "step": 40350
    },
    {
      "epoch": 1.71,
      "grad_norm": 31.312642096407398,
      "learning_rate": 4.6632952448390675e-06,
      "loss": 0.6797,
      "step": 40355
    },
    {
      "epoch": 1.71,
      "grad_norm": 14.802131875737823,
      "learning_rate": 4.662066921604699e-06,
      "loss": 0.7206,
      "step": 40360
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.0488991483100625,
      "learning_rate": 4.660838618858216e-06,
      "loss": 0.7286,
      "step": 40365
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.203618222717946,
      "learning_rate": 4.659610336674089e-06,
      "loss": 0.6963,
      "step": 40370
    },
    {
      "epoch": 1.71,
      "grad_norm": 10.787936082574443,
      "learning_rate": 4.658382075126785e-06,
      "loss": 0.6964,
      "step": 40375
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.421266854415682,
      "learning_rate": 4.657153834290769e-06,
      "loss": 0.7678,
      "step": 40380
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.409093234752213,
      "learning_rate": 4.655925614240506e-06,
      "loss": 0.711,
      "step": 40385
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.371735944911331,
      "learning_rate": 4.65469741505046e-06,
      "loss": 0.7147,
      "step": 40390
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.316387419806424,
      "learning_rate": 4.653469236795092e-06,
      "loss": 0.7094,
      "step": 40395
    },
    {
      "epoch": 1.71,
      "grad_norm": 10.017469850902847,
      "learning_rate": 4.652241079548864e-06,
      "loss": 0.6888,
      "step": 40400
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.062566063409066,
      "learning_rate": 4.651012943386234e-06,
      "loss": 0.6867,
      "step": 40405
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.3979896850548235,
      "learning_rate": 4.649784828381662e-06,
      "loss": 0.7057,
      "step": 40410
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.162163234031683,
      "learning_rate": 4.648556734609604e-06,
      "loss": 0.6761,
      "step": 40415
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.90467689424634,
      "learning_rate": 4.647328662144518e-06,
      "loss": 0.7552,
      "step": 40420
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.252126655323955,
      "learning_rate": 4.646100611060854e-06,
      "loss": 0.7123,
      "step": 40425
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.300645138228368,
      "learning_rate": 4.644872581433068e-06,
      "loss": 0.7237,
      "step": 40430
    },
    {
      "epoch": 1.71,
      "grad_norm": 7.832616414623262,
      "learning_rate": 4.643644573335612e-06,
      "loss": 0.6949,
      "step": 40435
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.20222271768449,
      "learning_rate": 4.6424165868429355e-06,
      "loss": 0.7091,
      "step": 40440
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.994200385786906,
      "learning_rate": 4.641188622029488e-06,
      "loss": 0.7043,
      "step": 40445
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.221544947824488,
      "learning_rate": 4.6399606789697175e-06,
      "loss": 0.7461,
      "step": 40450
    },
    {
      "epoch": 1.71,
      "grad_norm": 5.015290124780768,
      "learning_rate": 4.638732757738069e-06,
      "loss": 0.687,
      "step": 40455
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.130280514478898,
      "learning_rate": 4.637504858408991e-06,
      "loss": 0.7403,
      "step": 40460
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.290461657371866,
      "learning_rate": 4.636276981056927e-06,
      "loss": 0.6981,
      "step": 40465
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.754473736695614,
      "learning_rate": 4.635049125756315e-06,
      "loss": 0.6769,
      "step": 40470
    },
    {
      "epoch": 1.71,
      "grad_norm": 11.71146981448402,
      "learning_rate": 4.633821292581601e-06,
      "loss": 0.6909,
      "step": 40475
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.972559461512804,
      "learning_rate": 4.632593481607223e-06,
      "loss": 0.7271,
      "step": 40480
    },
    {
      "epoch": 1.71,
      "grad_norm": 8.25667688724223,
      "learning_rate": 4.631365692907621e-06,
      "loss": 0.7151,
      "step": 40485
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.356874540646292,
      "learning_rate": 4.630137926557231e-06,
      "loss": 0.7142,
      "step": 40490
    },
    {
      "epoch": 1.71,
      "grad_norm": 6.293168262553135,
      "learning_rate": 4.62891018263049e-06,
      "loss": 0.7384,
      "step": 40495
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.483999868269908,
      "learning_rate": 4.62768246120183e-06,
      "loss": 0.708,
      "step": 40500
    },
    {
      "epoch": 1.71,
      "grad_norm": 4.194548122553793,
      "learning_rate": 4.6264547623456865e-06,
      "loss": 0.6888,
      "step": 40505
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.42388482192117,
      "learning_rate": 4.6252270861364905e-06,
      "loss": 0.7061,
      "step": 40510
    },
    {
      "epoch": 1.71,
      "grad_norm": 10.140510610854582,
      "learning_rate": 4.62399943264867e-06,
      "loss": 0.6982,
      "step": 40515
    },
    {
      "epoch": 1.71,
      "grad_norm": 9.740542082420419,
      "learning_rate": 4.622771801956659e-06,
      "loss": 0.6721,
      "step": 40520
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.337390238200994,
      "learning_rate": 4.62154419413488e-06,
      "loss": 0.7091,
      "step": 40525
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.050203097910411,
      "learning_rate": 4.620316609257765e-06,
      "loss": 0.7023,
      "step": 40530
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.238168789705431,
      "learning_rate": 4.619089047399734e-06,
      "loss": 0.7101,
      "step": 40535
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.165346306193902,
      "learning_rate": 4.617861508635211e-06,
      "loss": 0.6726,
      "step": 40540
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.278345247822616,
      "learning_rate": 4.61663399303862e-06,
      "loss": 0.736,
      "step": 40545
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.220449471327732,
      "learning_rate": 4.61540650068438e-06,
      "loss": 0.7167,
      "step": 40550
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.811219220398779,
      "learning_rate": 4.614179031646912e-06,
      "loss": 0.6907,
      "step": 40555
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.675247525986704,
      "learning_rate": 4.612951586000634e-06,
      "loss": 0.7089,
      "step": 40560
    },
    {
      "epoch": 1.72,
      "grad_norm": 9.225890302560444,
      "learning_rate": 4.611724163819958e-06,
      "loss": 0.7077,
      "step": 40565
    },
    {
      "epoch": 1.72,
      "grad_norm": 17.51088320925971,
      "learning_rate": 4.610496765179305e-06,
      "loss": 0.7444,
      "step": 40570
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.031438583548248,
      "learning_rate": 4.609269390153084e-06,
      "loss": 0.6655,
      "step": 40575
    },
    {
      "epoch": 1.72,
      "grad_norm": 12.677225472810388,
      "learning_rate": 4.6080420388157105e-06,
      "loss": 0.6794,
      "step": 40580
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.988997089062789,
      "learning_rate": 4.6068147112415935e-06,
      "loss": 0.6919,
      "step": 40585
    },
    {
      "epoch": 1.72,
      "grad_norm": 15.439196307514528,
      "learning_rate": 4.605587407505141e-06,
      "loss": 0.6795,
      "step": 40590
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.5806037455168385,
      "learning_rate": 4.604360127680764e-06,
      "loss": 0.6972,
      "step": 40595
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.17218698260249,
      "learning_rate": 4.603132871842865e-06,
      "loss": 0.6798,
      "step": 40600
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.241287560036608,
      "learning_rate": 4.601905640065853e-06,
      "loss": 0.6988,
      "step": 40605
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.336714082651554,
      "learning_rate": 4.600678432424128e-06,
      "loss": 0.6959,
      "step": 40610
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.325714038513943,
      "learning_rate": 4.599451248992093e-06,
      "loss": 0.7076,
      "step": 40615
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.309510793451503,
      "learning_rate": 4.598224089844151e-06,
      "loss": 0.7309,
      "step": 40620
    },
    {
      "epoch": 1.72,
      "grad_norm": 14.216833003213026,
      "learning_rate": 4.596996955054697e-06,
      "loss": 0.6979,
      "step": 40625
    },
    {
      "epoch": 1.72,
      "grad_norm": 18.367344923241486,
      "learning_rate": 4.595769844698132e-06,
      "loss": 0.7259,
      "step": 40630
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.838433991858137,
      "learning_rate": 4.59454275884885e-06,
      "loss": 0.7206,
      "step": 40635
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.293420884429469,
      "learning_rate": 4.593315697581245e-06,
      "loss": 0.7331,
      "step": 40640
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.906156899267601,
      "learning_rate": 4.592088660969713e-06,
      "loss": 0.6803,
      "step": 40645
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.595315555443005,
      "learning_rate": 4.590861649088643e-06,
      "loss": 0.6757,
      "step": 40650
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.153649506587127,
      "learning_rate": 4.589634662012429e-06,
      "loss": 0.6789,
      "step": 40655
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.707493079234177,
      "learning_rate": 4.588407699815454e-06,
      "loss": 0.7225,
      "step": 40660
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.903997074231319,
      "learning_rate": 4.587180762572108e-06,
      "loss": 0.711,
      "step": 40665
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.3011501092994,
      "learning_rate": 4.585953850356776e-06,
      "loss": 0.707,
      "step": 40670
    },
    {
      "epoch": 1.72,
      "grad_norm": 18.0811366856455,
      "learning_rate": 4.584726963243843e-06,
      "loss": 0.7669,
      "step": 40675
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.696937386490977,
      "learning_rate": 4.583500101307691e-06,
      "loss": 0.6961,
      "step": 40680
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.635201377707462,
      "learning_rate": 4.5822732646227e-06,
      "loss": 0.7318,
      "step": 40685
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.313409920222185,
      "learning_rate": 4.5810464532632524e-06,
      "loss": 0.7465,
      "step": 40690
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.359442707543011,
      "learning_rate": 4.579819667303722e-06,
      "loss": 0.6898,
      "step": 40695
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.368049123471514,
      "learning_rate": 4.578592906818489e-06,
      "loss": 0.6818,
      "step": 40700
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.161906066427198,
      "learning_rate": 4.577366171881927e-06,
      "loss": 0.7386,
      "step": 40705
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.925707971255135,
      "learning_rate": 4.576139462568408e-06,
      "loss": 0.7412,
      "step": 40710
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.682431489538206,
      "learning_rate": 4.574912778952305e-06,
      "loss": 0.6973,
      "step": 40715
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.112265387489211,
      "learning_rate": 4.573686121107988e-06,
      "loss": 0.6969,
      "step": 40720
    },
    {
      "epoch": 1.72,
      "grad_norm": 7.5451346519498435,
      "learning_rate": 4.572459489109827e-06,
      "loss": 0.7124,
      "step": 40725
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.696779053294159,
      "learning_rate": 4.571232883032187e-06,
      "loss": 0.7632,
      "step": 40730
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.185652694528235,
      "learning_rate": 4.570006302949434e-06,
      "loss": 0.6959,
      "step": 40735
    },
    {
      "epoch": 1.72,
      "grad_norm": 4.653334686800743,
      "learning_rate": 4.5687797489359324e-06,
      "loss": 0.728,
      "step": 40740
    },
    {
      "epoch": 1.72,
      "grad_norm": 5.683994158080438,
      "learning_rate": 4.567553221066045e-06,
      "loss": 0.689,
      "step": 40745
    },
    {
      "epoch": 1.72,
      "grad_norm": 6.515685119949525,
      "learning_rate": 4.566326719414131e-06,
      "loss": 0.7024,
      "step": 40750
    },
    {
      "epoch": 1.72,
      "grad_norm": 8.282299888356041,
      "learning_rate": 4.565100244054552e-06,
      "loss": 0.7331,
      "step": 40755
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.868601929400915,
      "learning_rate": 4.563873795061662e-06,
      "loss": 0.7003,
      "step": 40760
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.257700709836006,
      "learning_rate": 4.562647372509821e-06,
      "loss": 0.7154,
      "step": 40765
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.5806258142587195,
      "learning_rate": 4.561420976473381e-06,
      "loss": 0.7035,
      "step": 40770
    },
    {
      "epoch": 1.73,
      "grad_norm": 11.255740564145732,
      "learning_rate": 4.560194607026695e-06,
      "loss": 0.6957,
      "step": 40775
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.214913681768481,
      "learning_rate": 4.558968264244115e-06,
      "loss": 0.7077,
      "step": 40780
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.912957486619837,
      "learning_rate": 4.557741948199989e-06,
      "loss": 0.6667,
      "step": 40785
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.984624281071443,
      "learning_rate": 4.556515658968668e-06,
      "loss": 0.7094,
      "step": 40790
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.335371484562447,
      "learning_rate": 4.555289396624495e-06,
      "loss": 0.6877,
      "step": 40795
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.739566591570247,
      "learning_rate": 4.554063161241816e-06,
      "loss": 0.7047,
      "step": 40800
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.777488957227169,
      "learning_rate": 4.552836952894974e-06,
      "loss": 0.7277,
      "step": 40805
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.593148297552575,
      "learning_rate": 4.55161077165831e-06,
      "loss": 0.7152,
      "step": 40810
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.539526122500462,
      "learning_rate": 4.550384617606165e-06,
      "loss": 0.745,
      "step": 40815
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.859223225750056,
      "learning_rate": 4.549158490812878e-06,
      "loss": 0.6885,
      "step": 40820
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.701733611189258,
      "learning_rate": 4.54793239135278e-06,
      "loss": 0.7104,
      "step": 40825
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.935317457453757,
      "learning_rate": 4.546706319300211e-06,
      "loss": 0.7211,
      "step": 40830
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.271784197889793,
      "learning_rate": 4.545480274729503e-06,
      "loss": 0.7043,
      "step": 40835
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.008213123870857,
      "learning_rate": 4.5442542577149875e-06,
      "loss": 0.6835,
      "step": 40840
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.213446684695098,
      "learning_rate": 4.5430282683309925e-06,
      "loss": 0.6895,
      "step": 40845
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.115450079178496,
      "learning_rate": 4.541802306651849e-06,
      "loss": 0.6631,
      "step": 40850
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.5697187845498615,
      "learning_rate": 4.540576372751879e-06,
      "loss": 0.6993,
      "step": 40855
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.168700241834609,
      "learning_rate": 4.539350466705414e-06,
      "loss": 0.6785,
      "step": 40860
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.43332745961926,
      "learning_rate": 4.5381245885867715e-06,
      "loss": 0.6951,
      "step": 40865
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.383333325342679,
      "learning_rate": 4.536898738470274e-06,
      "loss": 0.712,
      "step": 40870
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.74270911697505,
      "learning_rate": 4.535672916430244e-06,
      "loss": 0.7038,
      "step": 40875
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.510567056136337,
      "learning_rate": 4.5344471225409955e-06,
      "loss": 0.6934,
      "step": 40880
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.071575026211772,
      "learning_rate": 4.533221356876848e-06,
      "loss": 0.6985,
      "step": 40885
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.848061227840281,
      "learning_rate": 4.531995619512114e-06,
      "loss": 0.6734,
      "step": 40890
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.549032250938532,
      "learning_rate": 4.530769910521106e-06,
      "loss": 0.6804,
      "step": 40895
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.576719701028109,
      "learning_rate": 4.529544229978138e-06,
      "loss": 0.6786,
      "step": 40900
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.510570651790196,
      "learning_rate": 4.528318577957516e-06,
      "loss": 0.6736,
      "step": 40905
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.807543214672011,
      "learning_rate": 4.52709295453355e-06,
      "loss": 0.7002,
      "step": 40910
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.291479651011716,
      "learning_rate": 4.525867359780546e-06,
      "loss": 0.7004,
      "step": 40915
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.479014636639145,
      "learning_rate": 4.524641793772806e-06,
      "loss": 0.7219,
      "step": 40920
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.299863400762476,
      "learning_rate": 4.523416256584634e-06,
      "loss": 0.6748,
      "step": 40925
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.78407426375857,
      "learning_rate": 4.52219074829033e-06,
      "loss": 0.7149,
      "step": 40930
    },
    {
      "epoch": 1.73,
      "grad_norm": 10.122704198058736,
      "learning_rate": 4.520965268964196e-06,
      "loss": 0.6858,
      "step": 40935
    },
    {
      "epoch": 1.73,
      "grad_norm": 11.098143264324804,
      "learning_rate": 4.519739818680525e-06,
      "loss": 0.6954,
      "step": 40940
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.367903757270598,
      "learning_rate": 4.518514397513613e-06,
      "loss": 0.7089,
      "step": 40945
    },
    {
      "epoch": 1.73,
      "grad_norm": 22.725613467330067,
      "learning_rate": 4.517289005537757e-06,
      "loss": 0.6728,
      "step": 40950
    },
    {
      "epoch": 1.73,
      "grad_norm": 7.816836361598131,
      "learning_rate": 4.516063642827245e-06,
      "loss": 0.6951,
      "step": 40955
    },
    {
      "epoch": 1.73,
      "grad_norm": 8.66108065631532,
      "learning_rate": 4.514838309456371e-06,
      "loss": 0.7259,
      "step": 40960
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.330762698203375,
      "learning_rate": 4.513613005499419e-06,
      "loss": 0.7139,
      "step": 40965
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.725894270411569,
      "learning_rate": 4.5123877310306785e-06,
      "loss": 0.6975,
      "step": 40970
    },
    {
      "epoch": 1.73,
      "grad_norm": 9.425054397753808,
      "learning_rate": 4.511162486124434e-06,
      "loss": 0.6841,
      "step": 40975
    },
    {
      "epoch": 1.73,
      "grad_norm": 4.025001568247798,
      "learning_rate": 4.509937270854969e-06,
      "loss": 0.6711,
      "step": 40980
    },
    {
      "epoch": 1.73,
      "grad_norm": 12.54357927818446,
      "learning_rate": 4.508712085296563e-06,
      "loss": 0.7224,
      "step": 40985
    },
    {
      "epoch": 1.73,
      "grad_norm": 6.4010894105050244,
      "learning_rate": 4.507486929523494e-06,
      "loss": 0.7212,
      "step": 40990
    },
    {
      "epoch": 1.73,
      "grad_norm": 5.862641347189652,
      "learning_rate": 4.506261803610045e-06,
      "loss": 0.721,
      "step": 40995
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.399701280584715,
      "learning_rate": 4.505036707630486e-06,
      "loss": 0.7054,
      "step": 41000
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.162720720048479,
      "learning_rate": 4.503811641659094e-06,
      "loss": 0.711,
      "step": 41005
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.418193931178021,
      "learning_rate": 4.502586605770141e-06,
      "loss": 0.7297,
      "step": 41010
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.90839659657626,
      "learning_rate": 4.5013616000378965e-06,
      "loss": 0.7299,
      "step": 41015
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.0669579051267934,
      "learning_rate": 4.50013662453663e-06,
      "loss": 0.7261,
      "step": 41020
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.169493863682579,
      "learning_rate": 4.498911679340608e-06,
      "loss": 0.6801,
      "step": 41025
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.307078054410746,
      "learning_rate": 4.497686764524093e-06,
      "loss": 0.6934,
      "step": 41030
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.255976975057522,
      "learning_rate": 4.496461880161352e-06,
      "loss": 0.6885,
      "step": 41035
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.953837583434858,
      "learning_rate": 4.495237026326642e-06,
      "loss": 0.6641,
      "step": 41040
    },
    {
      "epoch": 1.74,
      "grad_norm": 9.832547129702194,
      "learning_rate": 4.494012203094226e-06,
      "loss": 0.6928,
      "step": 41045
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.075160721166977,
      "learning_rate": 4.492787410538359e-06,
      "loss": 0.7237,
      "step": 41050
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.4175850901384255,
      "learning_rate": 4.491562648733296e-06,
      "loss": 0.7334,
      "step": 41055
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.616093129241495,
      "learning_rate": 4.490337917753294e-06,
      "loss": 0.7259,
      "step": 41060
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.749246833602826,
      "learning_rate": 4.489113217672601e-06,
      "loss": 0.7311,
      "step": 41065
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.955244291903903,
      "learning_rate": 4.48788854856547e-06,
      "loss": 0.715,
      "step": 41070
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.239642515167693,
      "learning_rate": 4.486663910506148e-06,
      "loss": 0.7216,
      "step": 41075
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.038770190952199,
      "learning_rate": 4.4854393035688794e-06,
      "loss": 0.7101,
      "step": 41080
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.609423069896483,
      "learning_rate": 4.484214727827913e-06,
      "loss": 0.6984,
      "step": 41085
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.611287757428572,
      "learning_rate": 4.482990183357485e-06,
      "loss": 0.6969,
      "step": 41090
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.242686371183405,
      "learning_rate": 4.481765670231842e-06,
      "loss": 0.6848,
      "step": 41095
    },
    {
      "epoch": 1.74,
      "grad_norm": 11.98686357214418,
      "learning_rate": 4.4805411885252185e-06,
      "loss": 0.6808,
      "step": 41100
    },
    {
      "epoch": 1.74,
      "grad_norm": 8.57281245912513,
      "learning_rate": 4.479316738311852e-06,
      "loss": 0.7113,
      "step": 41105
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.530923429144713,
      "learning_rate": 4.47809231966598e-06,
      "loss": 0.72,
      "step": 41110
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.811481825265242,
      "learning_rate": 4.4768679326618305e-06,
      "loss": 0.6888,
      "step": 41115
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.272928658675119,
      "learning_rate": 4.47564357737364e-06,
      "loss": 0.7013,
      "step": 41120
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.850833580960397,
      "learning_rate": 4.474419253875634e-06,
      "loss": 0.6752,
      "step": 41125
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.4410982327331645,
      "learning_rate": 4.4731949622420405e-06,
      "loss": 0.7232,
      "step": 41130
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.273107944058135,
      "learning_rate": 4.4719707025470845e-06,
      "loss": 0.6574,
      "step": 41135
    },
    {
      "epoch": 1.74,
      "grad_norm": 13.901110685840688,
      "learning_rate": 4.470746474864991e-06,
      "loss": 0.701,
      "step": 41140
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.598665791509871,
      "learning_rate": 4.469522279269978e-06,
      "loss": 0.6955,
      "step": 41145
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.67130034064445,
      "learning_rate": 4.468298115836266e-06,
      "loss": 0.7239,
      "step": 41150
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.023251418608654,
      "learning_rate": 4.4670739846380755e-06,
      "loss": 0.6773,
      "step": 41155
    },
    {
      "epoch": 1.74,
      "grad_norm": 17.671309476789585,
      "learning_rate": 4.465849885749617e-06,
      "loss": 0.7023,
      "step": 41160
    },
    {
      "epoch": 1.74,
      "grad_norm": 19.21573675240208,
      "learning_rate": 4.464625819245108e-06,
      "loss": 0.7143,
      "step": 41165
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.667298144271295,
      "learning_rate": 4.4634017851987575e-06,
      "loss": 0.6741,
      "step": 41170
    },
    {
      "epoch": 1.74,
      "grad_norm": 14.460748625139859,
      "learning_rate": 4.462177783684776e-06,
      "loss": 0.6641,
      "step": 41175
    },
    {
      "epoch": 1.74,
      "grad_norm": 15.645927793533927,
      "learning_rate": 4.460953814777373e-06,
      "loss": 0.6891,
      "step": 41180
    },
    {
      "epoch": 1.74,
      "grad_norm": 11.057490905392672,
      "learning_rate": 4.45972987855075e-06,
      "loss": 0.7125,
      "step": 41185
    },
    {
      "epoch": 1.74,
      "grad_norm": 10.789285281800444,
      "learning_rate": 4.458505975079115e-06,
      "loss": 0.7229,
      "step": 41190
    },
    {
      "epoch": 1.74,
      "grad_norm": 7.853955356774067,
      "learning_rate": 4.4572821044366675e-06,
      "loss": 0.7051,
      "step": 41195
    },
    {
      "epoch": 1.74,
      "grad_norm": 6.254016092310586,
      "learning_rate": 4.456058266697607e-06,
      "loss": 0.6897,
      "step": 41200
    },
    {
      "epoch": 1.74,
      "grad_norm": 12.657675526416494,
      "learning_rate": 4.454834461936132e-06,
      "loss": 0.721,
      "step": 41205
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.324148505672216,
      "learning_rate": 4.453610690226436e-06,
      "loss": 0.7158,
      "step": 41210
    },
    {
      "epoch": 1.74,
      "grad_norm": 18.312927544035162,
      "learning_rate": 4.4523869516427164e-06,
      "loss": 0.6942,
      "step": 41215
    },
    {
      "epoch": 1.74,
      "grad_norm": 5.699636477499295,
      "learning_rate": 4.451163246259162e-06,
      "loss": 0.6575,
      "step": 41220
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.401382496816984,
      "learning_rate": 4.449939574149963e-06,
      "loss": 0.7004,
      "step": 41225
    },
    {
      "epoch": 1.74,
      "grad_norm": 4.776999971028778,
      "learning_rate": 4.448715935389309e-06,
      "loss": 0.7341,
      "step": 41230
    },
    {
      "epoch": 1.75,
      "grad_norm": 10.184943873675213,
      "learning_rate": 4.447492330051382e-06,
      "loss": 0.713,
      "step": 41235
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.407191375103588,
      "learning_rate": 4.446268758210369e-06,
      "loss": 0.7119,
      "step": 41240
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.523253447246215,
      "learning_rate": 4.44504521994045e-06,
      "loss": 0.722,
      "step": 41245
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.1688406822208135,
      "learning_rate": 4.443821715315803e-06,
      "loss": 0.6904,
      "step": 41250
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.715719931227288,
      "learning_rate": 4.442598244410609e-06,
      "loss": 0.7005,
      "step": 41255
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.353349145935558,
      "learning_rate": 4.44137480729904e-06,
      "loss": 0.6897,
      "step": 41260
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.030706416853716,
      "learning_rate": 4.440151404055272e-06,
      "loss": 0.7239,
      "step": 41265
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.321695796013148,
      "learning_rate": 4.4389280347534754e-06,
      "loss": 0.6988,
      "step": 41270
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.3983322149477475,
      "learning_rate": 4.437704699467819e-06,
      "loss": 0.6902,
      "step": 41275
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.225537836446942,
      "learning_rate": 4.4364813982724705e-06,
      "loss": 0.6905,
      "step": 41280
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.188808089580969,
      "learning_rate": 4.435258131241595e-06,
      "loss": 0.6843,
      "step": 41285
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.1482019008955815,
      "learning_rate": 4.434034898449355e-06,
      "loss": 0.6927,
      "step": 41290
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.973390228071283,
      "learning_rate": 4.432811699969913e-06,
      "loss": 0.6978,
      "step": 41295
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.4529929114909805,
      "learning_rate": 4.431588535877425e-06,
      "loss": 0.6496,
      "step": 41300
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.6656455899527085,
      "learning_rate": 4.430365406246053e-06,
      "loss": 0.7164,
      "step": 41305
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.971301505837978,
      "learning_rate": 4.429142311149947e-06,
      "loss": 0.708,
      "step": 41310
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.06910380127235,
      "learning_rate": 4.42791925066326e-06,
      "loss": 0.7298,
      "step": 41315
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.531469998738722,
      "learning_rate": 4.426696224860144e-06,
      "loss": 0.6812,
      "step": 41320
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.459097723260169,
      "learning_rate": 4.425473233814749e-06,
      "loss": 0.7048,
      "step": 41325
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.13636657777727,
      "learning_rate": 4.424250277601218e-06,
      "loss": 0.7104,
      "step": 41330
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.640087489898789,
      "learning_rate": 4.423027356293696e-06,
      "loss": 0.7005,
      "step": 41335
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.317660024328741,
      "learning_rate": 4.421804469966328e-06,
      "loss": 0.7149,
      "step": 41340
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.333737575520007,
      "learning_rate": 4.4205816186932496e-06,
      "loss": 0.6837,
      "step": 41345
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.437089497856815,
      "learning_rate": 4.419358802548602e-06,
      "loss": 0.6659,
      "step": 41350
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.116117130564452,
      "learning_rate": 4.41813602160652e-06,
      "loss": 0.6979,
      "step": 41355
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.101355851296357,
      "learning_rate": 4.416913275941137e-06,
      "loss": 0.6496,
      "step": 41360
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.240924472928226,
      "learning_rate": 4.4156905656265845e-06,
      "loss": 0.6745,
      "step": 41365
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.639549609725373,
      "learning_rate": 4.414467890736991e-06,
      "loss": 0.6949,
      "step": 41370
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.183726622677437,
      "learning_rate": 4.413245251346485e-06,
      "loss": 0.7528,
      "step": 41375
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.183016943714907,
      "learning_rate": 4.412022647529192e-06,
      "loss": 0.6911,
      "step": 41380
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.696687503694703,
      "learning_rate": 4.410800079359233e-06,
      "loss": 0.6985,
      "step": 41385
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.514664715052782,
      "learning_rate": 4.4095775469107296e-06,
      "loss": 0.6822,
      "step": 41390
    },
    {
      "epoch": 1.75,
      "grad_norm": 9.576376008610477,
      "learning_rate": 4.408355050257801e-06,
      "loss": 0.7156,
      "step": 41395
    },
    {
      "epoch": 1.75,
      "grad_norm": 7.615669366833428,
      "learning_rate": 4.407132589474563e-06,
      "loss": 0.701,
      "step": 41400
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.803455207254274,
      "learning_rate": 4.4059101646351295e-06,
      "loss": 0.6718,
      "step": 41405
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.78459997879738,
      "learning_rate": 4.404687775813612e-06,
      "loss": 0.662,
      "step": 41410
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.926111065450465,
      "learning_rate": 4.403465423084122e-06,
      "loss": 0.6695,
      "step": 41415
    },
    {
      "epoch": 1.75,
      "grad_norm": 8.054707762876705,
      "learning_rate": 4.4022431065207654e-06,
      "loss": 0.6881,
      "step": 41420
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.9626230718184905,
      "learning_rate": 4.40102082619765e-06,
      "loss": 0.7088,
      "step": 41425
    },
    {
      "epoch": 1.75,
      "grad_norm": 12.7569359141551,
      "learning_rate": 4.3997985821888764e-06,
      "loss": 0.6712,
      "step": 41430
    },
    {
      "epoch": 1.75,
      "grad_norm": 5.608489758689082,
      "learning_rate": 4.398576374568547e-06,
      "loss": 0.7206,
      "step": 41435
    },
    {
      "epoch": 1.75,
      "grad_norm": 10.74472222385305,
      "learning_rate": 4.397354203410761e-06,
      "loss": 0.7072,
      "step": 41440
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.217847917707725,
      "learning_rate": 4.396132068789613e-06,
      "loss": 0.6887,
      "step": 41445
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.615819081892022,
      "learning_rate": 4.3949099707792004e-06,
      "loss": 0.6812,
      "step": 41450
    },
    {
      "epoch": 1.75,
      "grad_norm": 6.3576411326308016,
      "learning_rate": 4.393687909453614e-06,
      "loss": 0.6962,
      "step": 41455
    },
    {
      "epoch": 1.75,
      "grad_norm": 4.381837709432909,
      "learning_rate": 4.392465884886942e-06,
      "loss": 0.6629,
      "step": 41460
    },
    {
      "epoch": 1.75,
      "grad_norm": 10.752500399141914,
      "learning_rate": 4.391243897153276e-06,
      "loss": 0.6688,
      "step": 41465
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.111570643144335,
      "learning_rate": 4.3900219463267e-06,
      "loss": 0.706,
      "step": 41470
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.146414717517184,
      "learning_rate": 4.3888000324812955e-06,
      "loss": 0.6998,
      "step": 41475
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.181769277666174,
      "learning_rate": 4.387578155691143e-06,
      "loss": 0.6951,
      "step": 41480
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.457317139763903,
      "learning_rate": 4.386356316030325e-06,
      "loss": 0.7038,
      "step": 41485
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.3381424911736355,
      "learning_rate": 4.385134513572914e-06,
      "loss": 0.7007,
      "step": 41490
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.578728868410613,
      "learning_rate": 4.383912748392988e-06,
      "loss": 0.6626,
      "step": 41495
    },
    {
      "epoch": 1.76,
      "grad_norm": 20.444096498890847,
      "learning_rate": 4.382691020564617e-06,
      "loss": 0.7567,
      "step": 41500
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.47502861388027,
      "learning_rate": 4.38146933016187e-06,
      "loss": 0.698,
      "step": 41505
    },
    {
      "epoch": 1.76,
      "grad_norm": 7.741848196621235,
      "learning_rate": 4.380247677258818e-06,
      "loss": 0.7276,
      "step": 41510
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.119542702248495,
      "learning_rate": 4.379026061929521e-06,
      "loss": 0.6925,
      "step": 41515
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.503191406435533,
      "learning_rate": 4.377804484248046e-06,
      "loss": 0.7031,
      "step": 41520
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.46876410824929,
      "learning_rate": 4.376582944288453e-06,
      "loss": 0.6366,
      "step": 41525
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.383836475755269,
      "learning_rate": 4.375361442124799e-06,
      "loss": 0.6506,
      "step": 41530
    },
    {
      "epoch": 1.76,
      "grad_norm": 9.274656290398722,
      "learning_rate": 4.374139977831141e-06,
      "loss": 0.7038,
      "step": 41535
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.134286668807976,
      "learning_rate": 4.372918551481532e-06,
      "loss": 0.6657,
      "step": 41540
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.397411484274887,
      "learning_rate": 4.371697163150025e-06,
      "loss": 0.6471,
      "step": 41545
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.245770500846576,
      "learning_rate": 4.370475812910669e-06,
      "loss": 0.6983,
      "step": 41550
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.304664905376411,
      "learning_rate": 4.369254500837508e-06,
      "loss": 0.6831,
      "step": 41555
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.994402427718099,
      "learning_rate": 4.36803322700459e-06,
      "loss": 0.6873,
      "step": 41560
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.39437785142945,
      "learning_rate": 4.366811991485955e-06,
      "loss": 0.7108,
      "step": 41565
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.329835864751196,
      "learning_rate": 4.365590794355645e-06,
      "loss": 0.6978,
      "step": 41570
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.993750631379819,
      "learning_rate": 4.364369635687696e-06,
      "loss": 0.6681,
      "step": 41575
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.911751070974839,
      "learning_rate": 4.363148515556143e-06,
      "loss": 0.6779,
      "step": 41580
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.444657954570195,
      "learning_rate": 4.361927434035021e-06,
      "loss": 0.7388,
      "step": 41585
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.413420115750064,
      "learning_rate": 4.360706391198359e-06,
      "loss": 0.7059,
      "step": 41590
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.7506309876311,
      "learning_rate": 4.359485387120184e-06,
      "loss": 0.7236,
      "step": 41595
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.33538576420163,
      "learning_rate": 4.358264421874524e-06,
      "loss": 0.7279,
      "step": 41600
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.303221688509321,
      "learning_rate": 4.357043495535401e-06,
      "loss": 0.6847,
      "step": 41605
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.756077045874945,
      "learning_rate": 4.355822608176839e-06,
      "loss": 0.6866,
      "step": 41610
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.42686479470545,
      "learning_rate": 4.354601759872854e-06,
      "loss": 0.6803,
      "step": 41615
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.124553141007188,
      "learning_rate": 4.3533809506974624e-06,
      "loss": 0.7105,
      "step": 41620
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.933124820547007,
      "learning_rate": 4.3521601807246795e-06,
      "loss": 0.6977,
      "step": 41625
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.112338062158038,
      "learning_rate": 4.350939450028519e-06,
      "loss": 0.6691,
      "step": 41630
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.169957823657374,
      "learning_rate": 4.349718758682984e-06,
      "loss": 0.7036,
      "step": 41635
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.339562749315671,
      "learning_rate": 4.348498106762087e-06,
      "loss": 0.6688,
      "step": 41640
    },
    {
      "epoch": 1.76,
      "grad_norm": 10.337215799821285,
      "learning_rate": 4.34727749433983e-06,
      "loss": 0.7371,
      "step": 41645
    },
    {
      "epoch": 1.76,
      "grad_norm": 12.984369647833498,
      "learning_rate": 4.346056921490216e-06,
      "loss": 0.693,
      "step": 41650
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.683125618128475,
      "learning_rate": 4.3448363882872456e-06,
      "loss": 0.6708,
      "step": 41655
    },
    {
      "epoch": 1.76,
      "grad_norm": 11.01406555218198,
      "learning_rate": 4.343615894804915e-06,
      "loss": 0.7226,
      "step": 41660
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.651445346898479,
      "learning_rate": 4.3423954411172175e-06,
      "loss": 0.7204,
      "step": 41665
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.845400441709653,
      "learning_rate": 4.341175027298148e-06,
      "loss": 0.6808,
      "step": 41670
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.576923633832933,
      "learning_rate": 4.339954653421697e-06,
      "loss": 0.6844,
      "step": 41675
    },
    {
      "epoch": 1.76,
      "grad_norm": 5.525879128160491,
      "learning_rate": 4.338734319561851e-06,
      "loss": 0.6981,
      "step": 41680
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.511082053619075,
      "learning_rate": 4.337514025792595e-06,
      "loss": 0.7227,
      "step": 41685
    },
    {
      "epoch": 1.76,
      "grad_norm": 4.235310244332541,
      "learning_rate": 4.336293772187911e-06,
      "loss": 0.7018,
      "step": 41690
    },
    {
      "epoch": 1.76,
      "grad_norm": 6.77417592961108,
      "learning_rate": 4.3350735588217814e-06,
      "loss": 0.6807,
      "step": 41695
    },
    {
      "epoch": 1.76,
      "grad_norm": 9.124046805727305,
      "learning_rate": 4.333853385768183e-06,
      "loss": 0.6757,
      "step": 41700
    },
    {
      "epoch": 1.77,
      "grad_norm": 9.19145591510403,
      "learning_rate": 4.3326332531010934e-06,
      "loss": 0.7487,
      "step": 41705
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.134822808298634,
      "learning_rate": 4.331413160894483e-06,
      "loss": 0.6891,
      "step": 41710
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.000803596675463,
      "learning_rate": 4.330193109222322e-06,
      "loss": 0.6827,
      "step": 41715
    },
    {
      "epoch": 1.77,
      "grad_norm": 26.28811974164926,
      "learning_rate": 4.3289730981585815e-06,
      "loss": 0.6878,
      "step": 41720
    },
    {
      "epoch": 1.77,
      "grad_norm": 13.687295803189956,
      "learning_rate": 4.327753127777224e-06,
      "loss": 0.6814,
      "step": 41725
    },
    {
      "epoch": 1.77,
      "grad_norm": 8.649028554761147,
      "learning_rate": 4.3265331981522156e-06,
      "loss": 0.6652,
      "step": 41730
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.856391345297885,
      "learning_rate": 4.325313309357515e-06,
      "loss": 0.677,
      "step": 41735
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.6117888242585074,
      "learning_rate": 4.324093461467081e-06,
      "loss": 0.6944,
      "step": 41740
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.693178156978733,
      "learning_rate": 4.322873654554869e-06,
      "loss": 0.6818,
      "step": 41745
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.5912449123003976,
      "learning_rate": 4.321653888694833e-06,
      "loss": 0.686,
      "step": 41750
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.522191686228286,
      "learning_rate": 4.320434163960925e-06,
      "loss": 0.6938,
      "step": 41755
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.6925472814253055,
      "learning_rate": 4.319214480427091e-06,
      "loss": 0.6799,
      "step": 41760
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.789851216435146,
      "learning_rate": 4.317994838167276e-06,
      "loss": 0.6858,
      "step": 41765
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.586793691870011,
      "learning_rate": 4.316775237255428e-06,
      "loss": 0.6823,
      "step": 41770
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.723823950520834,
      "learning_rate": 4.315555677765483e-06,
      "loss": 0.6999,
      "step": 41775
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.445372470781785,
      "learning_rate": 4.314336159771382e-06,
      "loss": 0.7092,
      "step": 41780
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.394353726974539,
      "learning_rate": 4.31311668334706e-06,
      "loss": 0.6831,
      "step": 41785
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.018983060982324,
      "learning_rate": 4.311897248566451e-06,
      "loss": 0.7202,
      "step": 41790
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.263328794580536,
      "learning_rate": 4.310677855503483e-06,
      "loss": 0.727,
      "step": 41795
    },
    {
      "epoch": 1.77,
      "grad_norm": 14.392740548779859,
      "learning_rate": 4.309458504232087e-06,
      "loss": 0.7236,
      "step": 41800
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.288968009324223,
      "learning_rate": 4.3082391948261885e-06,
      "loss": 0.7049,
      "step": 41805
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.3329306659887665,
      "learning_rate": 4.307019927359708e-06,
      "loss": 0.661,
      "step": 41810
    },
    {
      "epoch": 1.77,
      "grad_norm": 15.928413824814557,
      "learning_rate": 4.30580070190657e-06,
      "loss": 0.7232,
      "step": 41815
    },
    {
      "epoch": 1.77,
      "grad_norm": 7.83135742146612,
      "learning_rate": 4.304581518540689e-06,
      "loss": 0.6546,
      "step": 41820
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.009168171582179,
      "learning_rate": 4.303362377335984e-06,
      "loss": 0.693,
      "step": 41825
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.073120865542738,
      "learning_rate": 4.302143278366365e-06,
      "loss": 0.672,
      "step": 41830
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.5140358420284485,
      "learning_rate": 4.300924221705743e-06,
      "loss": 0.7033,
      "step": 41835
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.676453752497495,
      "learning_rate": 4.299705207428027e-06,
      "loss": 0.6792,
      "step": 41840
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.462030509091536,
      "learning_rate": 4.298486235607121e-06,
      "loss": 0.6854,
      "step": 41845
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.810799553620261,
      "learning_rate": 4.2972673063169285e-06,
      "loss": 0.6486,
      "step": 41850
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.687485354310291,
      "learning_rate": 4.296048419631349e-06,
      "loss": 0.6905,
      "step": 41855
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.940863919819981,
      "learning_rate": 4.29482957562428e-06,
      "loss": 0.6779,
      "step": 41860
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.7574506375981285,
      "learning_rate": 4.293610774369617e-06,
      "loss": 0.6892,
      "step": 41865
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.859548398044227,
      "learning_rate": 4.292392015941253e-06,
      "loss": 0.6961,
      "step": 41870
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.819842460526834,
      "learning_rate": 4.291173300413075e-06,
      "loss": 0.7269,
      "step": 41875
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.490555457291418,
      "learning_rate": 4.289954627858973e-06,
      "loss": 0.6737,
      "step": 41880
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.863252744797067,
      "learning_rate": 4.2887359983528295e-06,
      "loss": 0.6704,
      "step": 41885
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.432766473447019,
      "learning_rate": 4.287517411968529e-06,
      "loss": 0.7065,
      "step": 41890
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.811268726667623,
      "learning_rate": 4.2862988687799484e-06,
      "loss": 0.6977,
      "step": 41895
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.471372115842373,
      "learning_rate": 4.285080368860964e-06,
      "loss": 0.7074,
      "step": 41900
    },
    {
      "epoch": 1.77,
      "grad_norm": 6.064357412941498,
      "learning_rate": 4.283861912285452e-06,
      "loss": 0.6836,
      "step": 41905
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.737171949375136,
      "learning_rate": 4.282643499127282e-06,
      "loss": 0.7065,
      "step": 41910
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.288568100906577,
      "learning_rate": 4.2814251294603245e-06,
      "loss": 0.7151,
      "step": 41915
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.220457335673189,
      "learning_rate": 4.280206803358445e-06,
      "loss": 0.6631,
      "step": 41920
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.575144700356605,
      "learning_rate": 4.278988520895505e-06,
      "loss": 0.6862,
      "step": 41925
    },
    {
      "epoch": 1.77,
      "grad_norm": 5.005245709547172,
      "learning_rate": 4.277770282145369e-06,
      "loss": 0.7374,
      "step": 41930
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.299086854987855,
      "learning_rate": 4.276552087181892e-06,
      "loss": 0.6783,
      "step": 41935
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.5816918177444865,
      "learning_rate": 4.275333936078932e-06,
      "loss": 0.6681,
      "step": 41940
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.206306678317147,
      "learning_rate": 4.2741158289103416e-06,
      "loss": 0.6974,
      "step": 41945
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.373012306092978,
      "learning_rate": 4.27289776574997e-06,
      "loss": 0.7036,
      "step": 41950
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.269530211442046,
      "learning_rate": 4.271679746671664e-06,
      "loss": 0.718,
      "step": 41955
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.679785523190371,
      "learning_rate": 4.2704617717492705e-06,
      "loss": 0.7078,
      "step": 41960
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.631037808795886,
      "learning_rate": 4.269243841056631e-06,
      "loss": 0.677,
      "step": 41965
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.121768362462444,
      "learning_rate": 4.268025954667584e-06,
      "loss": 0.6861,
      "step": 41970
    },
    {
      "epoch": 1.78,
      "grad_norm": 10.513819952364718,
      "learning_rate": 4.266808112655968e-06,
      "loss": 0.7119,
      "step": 41975
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.281636669906912,
      "learning_rate": 4.265590315095615e-06,
      "loss": 0.7243,
      "step": 41980
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.639357720002547,
      "learning_rate": 4.264372562060359e-06,
      "loss": 0.6819,
      "step": 41985
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.321326010370948,
      "learning_rate": 4.263154853624028e-06,
      "loss": 0.6715,
      "step": 41990
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.537412296080619,
      "learning_rate": 4.261937189860447e-06,
      "loss": 0.7035,
      "step": 41995
    },
    {
      "epoch": 1.78,
      "grad_norm": 3.9775277436946994,
      "learning_rate": 4.260719570843441e-06,
      "loss": 0.6385,
      "step": 42000
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.998980405852894,
      "learning_rate": 4.25950199664683e-06,
      "loss": 0.6954,
      "step": 42005
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.21519075230146,
      "learning_rate": 4.258284467344431e-06,
      "loss": 0.669,
      "step": 42010
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.804256399748408,
      "learning_rate": 4.257066983010061e-06,
      "loss": 0.709,
      "step": 42015
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.702268189519116,
      "learning_rate": 4.255849543717529e-06,
      "loss": 0.6648,
      "step": 42020
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.332881912935103,
      "learning_rate": 4.254632149540649e-06,
      "loss": 0.7021,
      "step": 42025
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.074398822331957,
      "learning_rate": 4.253414800553225e-06,
      "loss": 0.6661,
      "step": 42030
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.531264449761033,
      "learning_rate": 4.252197496829064e-06,
      "loss": 0.6817,
      "step": 42035
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.731946287279799,
      "learning_rate": 4.250980238441967e-06,
      "loss": 0.6703,
      "step": 42040
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.005358749709967,
      "learning_rate": 4.249763025465729e-06,
      "loss": 0.6715,
      "step": 42045
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.309545467515519,
      "learning_rate": 4.2485458579741504e-06,
      "loss": 0.6743,
      "step": 42050
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.5921692257259235,
      "learning_rate": 4.247328736041022e-06,
      "loss": 0.6833,
      "step": 42055
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.207482283506148,
      "learning_rate": 4.2461116597401365e-06,
      "loss": 0.7019,
      "step": 42060
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.81628216055002,
      "learning_rate": 4.24489462914528e-06,
      "loss": 0.683,
      "step": 42065
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.206246918770953,
      "learning_rate": 4.243677644330237e-06,
      "loss": 0.7101,
      "step": 42070
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.875733211150892,
      "learning_rate": 4.242460705368792e-06,
      "loss": 0.6966,
      "step": 42075
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.014781819657245,
      "learning_rate": 4.241243812334722e-06,
      "loss": 0.6427,
      "step": 42080
    },
    {
      "epoch": 1.78,
      "grad_norm": 13.81215859906448,
      "learning_rate": 4.240026965301806e-06,
      "loss": 0.7045,
      "step": 42085
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.53422817954452,
      "learning_rate": 4.238810164343816e-06,
      "loss": 0.7183,
      "step": 42090
    },
    {
      "epoch": 1.78,
      "grad_norm": 19.773675846800426,
      "learning_rate": 4.237593409534522e-06,
      "loss": 0.6958,
      "step": 42095
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.030200045497101,
      "learning_rate": 4.236376700947696e-06,
      "loss": 0.6603,
      "step": 42100
    },
    {
      "epoch": 1.78,
      "grad_norm": 8.65824640530128,
      "learning_rate": 4.235160038657099e-06,
      "loss": 0.6604,
      "step": 42105
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.885639185714243,
      "learning_rate": 4.233943422736498e-06,
      "loss": 0.7285,
      "step": 42110
    },
    {
      "epoch": 1.78,
      "grad_norm": 10.270268632635593,
      "learning_rate": 4.232726853259651e-06,
      "loss": 0.683,
      "step": 42115
    },
    {
      "epoch": 1.78,
      "grad_norm": 9.545263576809146,
      "learning_rate": 4.231510330300315e-06,
      "loss": 0.7042,
      "step": 42120
    },
    {
      "epoch": 1.78,
      "grad_norm": 9.130343490756728,
      "learning_rate": 4.2302938539322415e-06,
      "loss": 0.6602,
      "step": 42125
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.253400470017544,
      "learning_rate": 4.229077424229187e-06,
      "loss": 0.7072,
      "step": 42130
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.746128899357891,
      "learning_rate": 4.227861041264896e-06,
      "loss": 0.6675,
      "step": 42135
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.960438129249485,
      "learning_rate": 4.2266447051131155e-06,
      "loss": 0.7239,
      "step": 42140
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.976974691785344,
      "learning_rate": 4.225428415847589e-06,
      "loss": 0.7006,
      "step": 42145
    },
    {
      "epoch": 1.78,
      "grad_norm": 5.120069175085671,
      "learning_rate": 4.224212173542056e-06,
      "loss": 0.7032,
      "step": 42150
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.72126767478556,
      "learning_rate": 4.222995978270254e-06,
      "loss": 0.6733,
      "step": 42155
    },
    {
      "epoch": 1.78,
      "grad_norm": 7.5826021749362305,
      "learning_rate": 4.221779830105917e-06,
      "loss": 0.7106,
      "step": 42160
    },
    {
      "epoch": 1.78,
      "grad_norm": 4.978620790444742,
      "learning_rate": 4.220563729122776e-06,
      "loss": 0.7018,
      "step": 42165
    },
    {
      "epoch": 1.78,
      "grad_norm": 9.401651812723802,
      "learning_rate": 4.219347675394561e-06,
      "loss": 0.7191,
      "step": 42170
    },
    {
      "epoch": 1.78,
      "grad_norm": 6.14442799549802,
      "learning_rate": 4.218131668994997e-06,
      "loss": 0.7023,
      "step": 42175
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.8815002277518875,
      "learning_rate": 4.216915709997806e-06,
      "loss": 0.6612,
      "step": 42180
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.441470440867683,
      "learning_rate": 4.21569979847671e-06,
      "loss": 0.6899,
      "step": 42185
    },
    {
      "epoch": 1.79,
      "grad_norm": 13.741208315742687,
      "learning_rate": 4.2144839345054224e-06,
      "loss": 0.7296,
      "step": 42190
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.039423664982864,
      "learning_rate": 4.213268118157663e-06,
      "loss": 0.706,
      "step": 42195
    },
    {
      "epoch": 1.79,
      "grad_norm": 11.480485569013796,
      "learning_rate": 4.2120523495071395e-06,
      "loss": 0.6494,
      "step": 42200
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.90351864993707,
      "learning_rate": 4.21083662862756e-06,
      "loss": 0.6936,
      "step": 42205
    },
    {
      "epoch": 1.79,
      "grad_norm": 10.664897595372203,
      "learning_rate": 4.209620955592632e-06,
      "loss": 0.6577,
      "step": 42210
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.755547350467756,
      "learning_rate": 4.208405330476056e-06,
      "loss": 0.6862,
      "step": 42215
    },
    {
      "epoch": 1.79,
      "grad_norm": 10.955391716986584,
      "learning_rate": 4.207189753351535e-06,
      "loss": 0.671,
      "step": 42220
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.943061193500395,
      "learning_rate": 4.205974224292764e-06,
      "loss": 0.6708,
      "step": 42225
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.045553013170885,
      "learning_rate": 4.2047587433734356e-06,
      "loss": 0.717,
      "step": 42230
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.840505266032793,
      "learning_rate": 4.203543310667243e-06,
      "loss": 0.6746,
      "step": 42235
    },
    {
      "epoch": 1.79,
      "grad_norm": 10.795634284051419,
      "learning_rate": 4.202327926247873e-06,
      "loss": 0.6936,
      "step": 42240
    },
    {
      "epoch": 1.79,
      "grad_norm": 11.753644661513787,
      "learning_rate": 4.201112590189011e-06,
      "loss": 0.6807,
      "step": 42245
    },
    {
      "epoch": 1.79,
      "grad_norm": 12.240200134571912,
      "learning_rate": 4.19989730256434e-06,
      "loss": 0.6586,
      "step": 42250
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.076927703189254,
      "learning_rate": 4.1986820634475375e-06,
      "loss": 0.7056,
      "step": 42255
    },
    {
      "epoch": 1.79,
      "grad_norm": 11.94973996266796,
      "learning_rate": 4.1974668729122834e-06,
      "loss": 0.7293,
      "step": 42260
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.360933031483773,
      "learning_rate": 4.196251731032247e-06,
      "loss": 0.6841,
      "step": 42265
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.079093125525384,
      "learning_rate": 4.195036637881102e-06,
      "loss": 0.707,
      "step": 42270
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.494899354348531,
      "learning_rate": 4.193821593532515e-06,
      "loss": 0.6813,
      "step": 42275
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.022060179574725,
      "learning_rate": 4.1926065980601486e-06,
      "loss": 0.6812,
      "step": 42280
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.140898430755351,
      "learning_rate": 4.191391651537666e-06,
      "loss": 0.6815,
      "step": 42285
    },
    {
      "epoch": 1.79,
      "grad_norm": 11.22738000981659,
      "learning_rate": 4.190176754038726e-06,
      "loss": 0.6835,
      "step": 42290
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.892685114888502,
      "learning_rate": 4.188961905636984e-06,
      "loss": 0.6789,
      "step": 42295
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.1555990938047564,
      "learning_rate": 4.187747106406092e-06,
      "loss": 0.6931,
      "step": 42300
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.229148641640653,
      "learning_rate": 4.186532356419701e-06,
      "loss": 0.6846,
      "step": 42305
    },
    {
      "epoch": 1.79,
      "grad_norm": 10.700529805122455,
      "learning_rate": 4.185317655751457e-06,
      "loss": 0.6994,
      "step": 42310
    },
    {
      "epoch": 1.79,
      "grad_norm": 3.990729889829295,
      "learning_rate": 4.184103004475004e-06,
      "loss": 0.6611,
      "step": 42315
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.96171242645658,
      "learning_rate": 4.182888402663983e-06,
      "loss": 0.686,
      "step": 42320
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.477819894215142,
      "learning_rate": 4.18167385039203e-06,
      "loss": 0.6837,
      "step": 42325
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.343856673608614,
      "learning_rate": 4.180459347732782e-06,
      "loss": 0.6719,
      "step": 42330
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.015435266272118,
      "learning_rate": 4.179244894759868e-06,
      "loss": 0.678,
      "step": 42335
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.320618436407309,
      "learning_rate": 4.178030491546921e-06,
      "loss": 0.6832,
      "step": 42340
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.767237463003962,
      "learning_rate": 4.176816138167563e-06,
      "loss": 0.6778,
      "step": 42345
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.161767250032055,
      "learning_rate": 4.1756018346954166e-06,
      "loss": 0.671,
      "step": 42350
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.369859810768393,
      "learning_rate": 4.174387581204104e-06,
      "loss": 0.6783,
      "step": 42355
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.8489460281304435,
      "learning_rate": 4.17317337776724e-06,
      "loss": 0.6848,
      "step": 42360
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.584790571671452,
      "learning_rate": 4.171959224458439e-06,
      "loss": 0.6649,
      "step": 42365
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.258603398563617,
      "learning_rate": 4.170745121351311e-06,
      "loss": 0.7257,
      "step": 42370
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.492669345379982,
      "learning_rate": 4.1695310685194625e-06,
      "loss": 0.6794,
      "step": 42375
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.449003496315726,
      "learning_rate": 4.168317066036501e-06,
      "loss": 0.6958,
      "step": 42380
    },
    {
      "epoch": 1.79,
      "grad_norm": 9.835507540706411,
      "learning_rate": 4.167103113976024e-06,
      "loss": 0.6741,
      "step": 42385
    },
    {
      "epoch": 1.79,
      "grad_norm": 8.32568488237581,
      "learning_rate": 4.165889212411633e-06,
      "loss": 0.6725,
      "step": 42390
    },
    {
      "epoch": 1.79,
      "grad_norm": 5.8609541056138825,
      "learning_rate": 4.164675361416922e-06,
      "loss": 0.7071,
      "step": 42395
    },
    {
      "epoch": 1.79,
      "grad_norm": 4.342518761506975,
      "learning_rate": 4.163461561065484e-06,
      "loss": 0.6874,
      "step": 42400
    },
    {
      "epoch": 1.79,
      "grad_norm": 6.9037508711612485,
      "learning_rate": 4.1622478114309065e-06,
      "loss": 0.6895,
      "step": 42405
    },
    {
      "epoch": 1.79,
      "grad_norm": 7.442259485284025,
      "learning_rate": 4.161034112586776e-06,
      "loss": 0.6978,
      "step": 42410
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.055801620382749,
      "learning_rate": 4.159820464606677e-06,
      "loss": 0.6781,
      "step": 42415
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.3873154365969205,
      "learning_rate": 4.158606867564189e-06,
      "loss": 0.6823,
      "step": 42420
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.683709204844898,
      "learning_rate": 4.157393321532886e-06,
      "loss": 0.7027,
      "step": 42425
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.944155811403085,
      "learning_rate": 4.156179826586346e-06,
      "loss": 0.6951,
      "step": 42430
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.686202837427037,
      "learning_rate": 4.154966382798138e-06,
      "loss": 0.6836,
      "step": 42435
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.135823246484916,
      "learning_rate": 4.153752990241829e-06,
      "loss": 0.6663,
      "step": 42440
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.196341787119352,
      "learning_rate": 4.152539648990982e-06,
      "loss": 0.6704,
      "step": 42445
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.765125241332584,
      "learning_rate": 4.1513263591191606e-06,
      "loss": 0.7176,
      "step": 42450
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.881593694758747,
      "learning_rate": 4.1501131206999216e-06,
      "loss": 0.6802,
      "step": 42455
    },
    {
      "epoch": 1.8,
      "grad_norm": 14.225096969964921,
      "learning_rate": 4.148899933806821e-06,
      "loss": 0.6609,
      "step": 42460
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.350332188253708,
      "learning_rate": 4.14768679851341e-06,
      "loss": 0.6266,
      "step": 42465
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.869290750545613,
      "learning_rate": 4.146473714893238e-06,
      "loss": 0.7065,
      "step": 42470
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.201482551297676,
      "learning_rate": 4.145260683019851e-06,
      "loss": 0.6707,
      "step": 42475
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.641458278395227,
      "learning_rate": 4.144047702966792e-06,
      "loss": 0.6744,
      "step": 42480
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.82207023762934,
      "learning_rate": 4.142834774807597e-06,
      "loss": 0.6877,
      "step": 42485
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.2165422331046845,
      "learning_rate": 4.141621898615806e-06,
      "loss": 0.6964,
      "step": 42490
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.110702536763148,
      "learning_rate": 4.140409074464951e-06,
      "loss": 0.709,
      "step": 42495
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.103655875020122,
      "learning_rate": 4.139196302428561e-06,
      "loss": 0.6727,
      "step": 42500
    },
    {
      "epoch": 1.8,
      "grad_norm": 8.314640417506359,
      "learning_rate": 4.137983582580165e-06,
      "loss": 0.6854,
      "step": 42505
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.636320212020477,
      "learning_rate": 4.136770914993284e-06,
      "loss": 0.6591,
      "step": 42510
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.041677791311336,
      "learning_rate": 4.13555829974144e-06,
      "loss": 0.6663,
      "step": 42515
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.913731597963763,
      "learning_rate": 4.13434573689815e-06,
      "loss": 0.6548,
      "step": 42520
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.256590195802156,
      "learning_rate": 4.133133226536929e-06,
      "loss": 0.6547,
      "step": 42525
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.315666503135458,
      "learning_rate": 4.131920768731286e-06,
      "loss": 0.6528,
      "step": 42530
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.632496180714526,
      "learning_rate": 4.1307083635547296e-06,
      "loss": 0.7262,
      "step": 42535
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.815366664152842,
      "learning_rate": 4.1294960110807655e-06,
      "loss": 0.697,
      "step": 42540
    },
    {
      "epoch": 1.8,
      "grad_norm": 6.5961466485856,
      "learning_rate": 4.128283711382893e-06,
      "loss": 0.6659,
      "step": 42545
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.565851561706001,
      "learning_rate": 4.127071464534612e-06,
      "loss": 0.6818,
      "step": 42550
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.257829391267469,
      "learning_rate": 4.125859270609418e-06,
      "loss": 0.6916,
      "step": 42555
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.78752117274731,
      "learning_rate": 4.124647129680799e-06,
      "loss": 0.7118,
      "step": 42560
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.552219681399091,
      "learning_rate": 4.123435041822248e-06,
      "loss": 0.6533,
      "step": 42565
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.6464112733559775,
      "learning_rate": 4.1222230071072465e-06,
      "loss": 0.6971,
      "step": 42570
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.406294171726127,
      "learning_rate": 4.121011025609281e-06,
      "loss": 0.6694,
      "step": 42575
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.266505689055188,
      "learning_rate": 4.119799097401827e-06,
      "loss": 0.6923,
      "step": 42580
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.269559038937698,
      "learning_rate": 4.11858722255836e-06,
      "loss": 0.6775,
      "step": 42585
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.248212633041329,
      "learning_rate": 4.117375401152354e-06,
      "loss": 0.7157,
      "step": 42590
    },
    {
      "epoch": 1.8,
      "grad_norm": 12.346544454992358,
      "learning_rate": 4.116163633257277e-06,
      "loss": 0.6549,
      "step": 42595
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.463070212937211,
      "learning_rate": 4.114951918946599e-06,
      "loss": 0.65,
      "step": 42600
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.493145910568353,
      "learning_rate": 4.113740258293776e-06,
      "loss": 0.7506,
      "step": 42605
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.272638968208005,
      "learning_rate": 4.112528651372271e-06,
      "loss": 0.6304,
      "step": 42610
    },
    {
      "epoch": 1.8,
      "grad_norm": 7.379020612799964,
      "learning_rate": 4.111317098255539e-06,
      "loss": 0.658,
      "step": 42615
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.84449738803698,
      "learning_rate": 4.110105599017035e-06,
      "loss": 0.6819,
      "step": 42620
    },
    {
      "epoch": 1.8,
      "grad_norm": 9.630285092050714,
      "learning_rate": 4.108894153730207e-06,
      "loss": 0.6707,
      "step": 42625
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.238051576522061,
      "learning_rate": 4.1076827624685e-06,
      "loss": 0.678,
      "step": 42630
    },
    {
      "epoch": 1.8,
      "grad_norm": 4.754876266528547,
      "learning_rate": 4.106471425305361e-06,
      "loss": 0.7228,
      "step": 42635
    },
    {
      "epoch": 1.8,
      "grad_norm": 5.109287206768923,
      "learning_rate": 4.1052601423142254e-06,
      "loss": 0.7104,
      "step": 42640
    },
    {
      "epoch": 1.8,
      "grad_norm": 3.9729781728225624,
      "learning_rate": 4.104048913568533e-06,
      "loss": 0.693,
      "step": 42645
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.789082408588649,
      "learning_rate": 4.102837739141716e-06,
      "loss": 0.6758,
      "step": 42650
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.015963063830426,
      "learning_rate": 4.101626619107203e-06,
      "loss": 0.6664,
      "step": 42655
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.168783415548996,
      "learning_rate": 4.100415553538423e-06,
      "loss": 0.7004,
      "step": 42660
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.744637980304332,
      "learning_rate": 4.099204542508798e-06,
      "loss": 0.6847,
      "step": 42665
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.490416352488437,
      "learning_rate": 4.097993586091748e-06,
      "loss": 0.6569,
      "step": 42670
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.61737599647957,
      "learning_rate": 4.096782684360691e-06,
      "loss": 0.69,
      "step": 42675
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.987011273728578,
      "learning_rate": 4.095571837389038e-06,
      "loss": 0.6954,
      "step": 42680
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.9241795801314865,
      "learning_rate": 4.0943610452502e-06,
      "loss": 0.6537,
      "step": 42685
    },
    {
      "epoch": 1.81,
      "grad_norm": 11.97891480117837,
      "learning_rate": 4.093150308017585e-06,
      "loss": 0.6662,
      "step": 42690
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.49038262641353,
      "learning_rate": 4.091939625764595e-06,
      "loss": 0.7031,
      "step": 42695
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.867259272331799,
      "learning_rate": 4.090728998564632e-06,
      "loss": 0.6944,
      "step": 42700
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.507223375357221,
      "learning_rate": 4.08951842649109e-06,
      "loss": 0.6657,
      "step": 42705
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.1195692872820855,
      "learning_rate": 4.088307909617365e-06,
      "loss": 0.6778,
      "step": 42710
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.739349229460742,
      "learning_rate": 4.087097448016845e-06,
      "loss": 0.6637,
      "step": 42715
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.757396881454455,
      "learning_rate": 4.085887041762919e-06,
      "loss": 0.6634,
      "step": 42720
    },
    {
      "epoch": 1.81,
      "grad_norm": 7.662222596964793,
      "learning_rate": 4.084676690928969e-06,
      "loss": 0.7626,
      "step": 42725
    },
    {
      "epoch": 1.81,
      "grad_norm": 18.697364622642063,
      "learning_rate": 4.083466395588374e-06,
      "loss": 0.6745,
      "step": 42730
    },
    {
      "epoch": 1.81,
      "grad_norm": 25.430694875802114,
      "learning_rate": 4.082256155814512e-06,
      "loss": 0.673,
      "step": 42735
    },
    {
      "epoch": 1.81,
      "grad_norm": 11.785301143460835,
      "learning_rate": 4.081045971680756e-06,
      "loss": 0.6699,
      "step": 42740
    },
    {
      "epoch": 1.81,
      "grad_norm": 12.786413668671589,
      "learning_rate": 4.079835843260475e-06,
      "loss": 0.6817,
      "step": 42745
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.201387530922869,
      "learning_rate": 4.0786257706270385e-06,
      "loss": 0.7032,
      "step": 42750
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.429761470594315,
      "learning_rate": 4.077415753853806e-06,
      "loss": 0.6685,
      "step": 42755
    },
    {
      "epoch": 1.81,
      "grad_norm": 9.882651472248737,
      "learning_rate": 4.0762057930141406e-06,
      "loss": 0.683,
      "step": 42760
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.456997518830199,
      "learning_rate": 4.074995888181395e-06,
      "loss": 0.6675,
      "step": 42765
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.445490356441053,
      "learning_rate": 4.073786039428925e-06,
      "loss": 0.6734,
      "step": 42770
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.587519992125487,
      "learning_rate": 4.072576246830078e-06,
      "loss": 0.6984,
      "step": 42775
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.035548520216146,
      "learning_rate": 4.071366510458202e-06,
      "loss": 0.68,
      "step": 42780
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.628397413006825,
      "learning_rate": 4.07015683038664e-06,
      "loss": 0.6383,
      "step": 42785
    },
    {
      "epoch": 1.81,
      "grad_norm": 3.9451838614915125,
      "learning_rate": 4.068947206688728e-06,
      "loss": 0.6726,
      "step": 42790
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.403847273027461,
      "learning_rate": 4.067737639437807e-06,
      "loss": 0.6576,
      "step": 42795
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.480057100090645,
      "learning_rate": 4.066528128707204e-06,
      "loss": 0.7262,
      "step": 42800
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.599297243797112,
      "learning_rate": 4.0653186745702535e-06,
      "loss": 0.7005,
      "step": 42805
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.7367784025859105,
      "learning_rate": 4.064109277100278e-06,
      "loss": 0.6824,
      "step": 42810
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.735311627778898,
      "learning_rate": 4.062899936370599e-06,
      "loss": 0.6903,
      "step": 42815
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.05287634308941,
      "learning_rate": 4.061690652454538e-06,
      "loss": 0.6592,
      "step": 42820
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.964244157658633,
      "learning_rate": 4.060481425425408e-06,
      "loss": 0.6846,
      "step": 42825
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.380740471536525,
      "learning_rate": 4.059272255356522e-06,
      "loss": 0.6917,
      "step": 42830
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.204287396844997,
      "learning_rate": 4.058063142321189e-06,
      "loss": 0.7035,
      "step": 42835
    },
    {
      "epoch": 1.81,
      "grad_norm": 8.181019888144954,
      "learning_rate": 4.0568540863927105e-06,
      "loss": 0.6787,
      "step": 42840
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.49555499702104,
      "learning_rate": 4.055645087644393e-06,
      "loss": 0.6609,
      "step": 42845
    },
    {
      "epoch": 1.81,
      "grad_norm": 13.346615494352267,
      "learning_rate": 4.054436146149531e-06,
      "loss": 0.6923,
      "step": 42850
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.785637431083596,
      "learning_rate": 4.053227261981422e-06,
      "loss": 0.6811,
      "step": 42855
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.678085778572287,
      "learning_rate": 4.0520184352133544e-06,
      "loss": 0.6507,
      "step": 42860
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.5644814219539,
      "learning_rate": 4.050809665918615e-06,
      "loss": 0.7023,
      "step": 42865
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.75551067191254,
      "learning_rate": 4.049600954170492e-06,
      "loss": 0.7079,
      "step": 42870
    },
    {
      "epoch": 1.81,
      "grad_norm": 4.476576985681184,
      "learning_rate": 4.0483923000422615e-06,
      "loss": 0.68,
      "step": 42875
    },
    {
      "epoch": 1.81,
      "grad_norm": 5.113299853337574,
      "learning_rate": 4.047183703607204e-06,
      "loss": 0.7218,
      "step": 42880
    },
    {
      "epoch": 1.81,
      "grad_norm": 6.686470891498753,
      "learning_rate": 4.045975164938592e-06,
      "loss": 0.6609,
      "step": 42885
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.31719190751155,
      "learning_rate": 4.0447666841096955e-06,
      "loss": 0.6567,
      "step": 42890
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.446139312597013,
      "learning_rate": 4.0435582611937815e-06,
      "loss": 0.7131,
      "step": 42895
    },
    {
      "epoch": 1.82,
      "grad_norm": 10.955837551796545,
      "learning_rate": 4.042349896264112e-06,
      "loss": 0.6462,
      "step": 42900
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.24887101546627,
      "learning_rate": 4.041141589393948e-06,
      "loss": 0.6875,
      "step": 42905
    },
    {
      "epoch": 1.82,
      "grad_norm": 11.72253193493209,
      "learning_rate": 4.039933340656546e-06,
      "loss": 0.6963,
      "step": 42910
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.424768905837444,
      "learning_rate": 4.038725150125156e-06,
      "loss": 0.6916,
      "step": 42915
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.381023314025799,
      "learning_rate": 4.0375170178730285e-06,
      "loss": 0.6692,
      "step": 42920
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.36596692451909,
      "learning_rate": 4.036308943973412e-06,
      "loss": 0.7168,
      "step": 42925
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.773857226520101,
      "learning_rate": 4.035100928499543e-06,
      "loss": 0.6344,
      "step": 42930
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.048114274200787,
      "learning_rate": 4.033892971524663e-06,
      "loss": 0.6491,
      "step": 42935
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.9224788482671915,
      "learning_rate": 4.0326850731220065e-06,
      "loss": 0.63,
      "step": 42940
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.959831529994112,
      "learning_rate": 4.031477233364804e-06,
      "loss": 0.69,
      "step": 42945
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.139299680929505,
      "learning_rate": 4.030269452326284e-06,
      "loss": 0.6691,
      "step": 42950
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.4284829007798585,
      "learning_rate": 4.029061730079672e-06,
      "loss": 0.682,
      "step": 42955
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.517727119276334,
      "learning_rate": 4.027854066698186e-06,
      "loss": 0.6744,
      "step": 42960
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.920386661228197,
      "learning_rate": 4.026646462255045e-06,
      "loss": 0.6657,
      "step": 42965
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.200026848460221,
      "learning_rate": 4.025438916823461e-06,
      "loss": 0.6633,
      "step": 42970
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.4373524987347,
      "learning_rate": 4.024231430476645e-06,
      "loss": 0.6675,
      "step": 42975
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.223707927261378,
      "learning_rate": 4.023024003287804e-06,
      "loss": 0.7131,
      "step": 42980
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.8532090677127595,
      "learning_rate": 4.021816635330139e-06,
      "loss": 0.6608,
      "step": 42985
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.928582682343386,
      "learning_rate": 4.020609326676852e-06,
      "loss": 0.6995,
      "step": 42990
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.911822876874502,
      "learning_rate": 4.0194020774011334e-06,
      "loss": 0.6669,
      "step": 42995
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.197969201106623,
      "learning_rate": 4.018194887576181e-06,
      "loss": 0.6882,
      "step": 43000
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.395820069797729,
      "learning_rate": 4.01698775727518e-06,
      "loss": 0.7191,
      "step": 43005
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.805631581243783,
      "learning_rate": 4.0157806865713154e-06,
      "loss": 0.6535,
      "step": 43010
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.077682574791148,
      "learning_rate": 4.014573675537769e-06,
      "loss": 0.7014,
      "step": 43015
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.724832635019642,
      "learning_rate": 4.013366724247717e-06,
      "loss": 0.7108,
      "step": 43020
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.82672469863374,
      "learning_rate": 4.012159832774336e-06,
      "loss": 0.6708,
      "step": 43025
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.936812031625569,
      "learning_rate": 4.010953001190794e-06,
      "loss": 0.7028,
      "step": 43030
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.296884618850802,
      "learning_rate": 4.009746229570256e-06,
      "loss": 0.6843,
      "step": 43035
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.000845693940138,
      "learning_rate": 4.0085395179858895e-06,
      "loss": 0.6815,
      "step": 43040
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.2912817379875206,
      "learning_rate": 4.007332866510852e-06,
      "loss": 0.6436,
      "step": 43045
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.217812078661837,
      "learning_rate": 4.0061262752182975e-06,
      "loss": 0.6549,
      "step": 43050
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.4760541301615415,
      "learning_rate": 4.004919744181379e-06,
      "loss": 0.6672,
      "step": 43055
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.287533142374523,
      "learning_rate": 4.003713273473246e-06,
      "loss": 0.6656,
      "step": 43060
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.592831254266362,
      "learning_rate": 4.002506863167044e-06,
      "loss": 0.6804,
      "step": 43065
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.456323013845729,
      "learning_rate": 4.001300513335911e-06,
      "loss": 0.6713,
      "step": 43070
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.649686742004004,
      "learning_rate": 4.0000942240529856e-06,
      "loss": 0.6639,
      "step": 43075
    },
    {
      "epoch": 1.82,
      "grad_norm": 8.147971227156999,
      "learning_rate": 3.9988879953914035e-06,
      "loss": 0.6982,
      "step": 43080
    },
    {
      "epoch": 1.82,
      "grad_norm": 9.02233394372782,
      "learning_rate": 3.997681827424294e-06,
      "loss": 0.6887,
      "step": 43085
    },
    {
      "epoch": 1.82,
      "grad_norm": 6.04463373169278,
      "learning_rate": 3.996475720224782e-06,
      "loss": 0.7025,
      "step": 43090
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.210965913924422,
      "learning_rate": 3.9952696738659905e-06,
      "loss": 0.6874,
      "step": 43095
    },
    {
      "epoch": 1.82,
      "grad_norm": 5.689904098285844,
      "learning_rate": 3.99406368842104e-06,
      "loss": 0.6906,
      "step": 43100
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.590081615381318,
      "learning_rate": 3.992857763963043e-06,
      "loss": 0.6603,
      "step": 43105
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.009598935772262,
      "learning_rate": 3.991651900565116e-06,
      "loss": 0.6476,
      "step": 43110
    },
    {
      "epoch": 1.82,
      "grad_norm": 4.3650277809949936,
      "learning_rate": 3.990446098300362e-06,
      "loss": 0.6724,
      "step": 43115
    },
    {
      "epoch": 1.82,
      "grad_norm": 7.460492272736713,
      "learning_rate": 3.989240357241888e-06,
      "loss": 0.6995,
      "step": 43120
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.5243059050061225,
      "learning_rate": 3.9880346774627935e-06,
      "loss": 0.6766,
      "step": 43125
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.386004474443567,
      "learning_rate": 3.986829059036176e-06,
      "loss": 0.7087,
      "step": 43130
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.777158511017598,
      "learning_rate": 3.985623502035128e-06,
      "loss": 0.6918,
      "step": 43135
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.226336800798119,
      "learning_rate": 3.9844180065327395e-06,
      "loss": 0.6321,
      "step": 43140
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.376727518405943,
      "learning_rate": 3.983212572602095e-06,
      "loss": 0.6707,
      "step": 43145
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.511898541315905,
      "learning_rate": 3.982007200316279e-06,
      "loss": 0.7057,
      "step": 43150
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.648768887154723,
      "learning_rate": 3.980801889748366e-06,
      "loss": 0.6536,
      "step": 43155
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.634724018784112,
      "learning_rate": 3.979596640971433e-06,
      "loss": 0.6667,
      "step": 43160
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.420363142647904,
      "learning_rate": 3.97839145405855e-06,
      "loss": 0.6986,
      "step": 43165
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.2415089473267376,
      "learning_rate": 3.9771863290827836e-06,
      "loss": 0.6753,
      "step": 43170
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.747292937693537,
      "learning_rate": 3.975981266117197e-06,
      "loss": 0.6609,
      "step": 43175
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.70045843305413,
      "learning_rate": 3.9747762652348494e-06,
      "loss": 0.6858,
      "step": 43180
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.332969479247198,
      "learning_rate": 3.973571326508799e-06,
      "loss": 0.644,
      "step": 43185
    },
    {
      "epoch": 1.83,
      "grad_norm": 10.005201811459234,
      "learning_rate": 3.972366450012095e-06,
      "loss": 0.6785,
      "step": 43190
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.62608518933242,
      "learning_rate": 3.971161635817785e-06,
      "loss": 0.6739,
      "step": 43195
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.28027197222515,
      "learning_rate": 3.969956883998916e-06,
      "loss": 0.6836,
      "step": 43200
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.090102103521136,
      "learning_rate": 3.968752194628525e-06,
      "loss": 0.6756,
      "step": 43205
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.655699841077139,
      "learning_rate": 3.967547567779652e-06,
      "loss": 0.6905,
      "step": 43210
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.669396335251718,
      "learning_rate": 3.966343003525329e-06,
      "loss": 0.663,
      "step": 43215
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.479655237704938,
      "learning_rate": 3.965138501938584e-06,
      "loss": 0.6641,
      "step": 43220
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.332540527831468,
      "learning_rate": 3.963934063092445e-06,
      "loss": 0.6408,
      "step": 43225
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.188852296245308,
      "learning_rate": 3.962729687059929e-06,
      "loss": 0.6873,
      "step": 43230
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.517221854255131,
      "learning_rate": 3.96152537391406e-06,
      "loss": 0.6708,
      "step": 43235
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.433775859643354,
      "learning_rate": 3.9603211237278475e-06,
      "loss": 0.6552,
      "step": 43240
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.860964607100632,
      "learning_rate": 3.959116936574304e-06,
      "loss": 0.719,
      "step": 43245
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.523789542782692,
      "learning_rate": 3.957912812526433e-06,
      "loss": 0.6333,
      "step": 43250
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.517129032993037,
      "learning_rate": 3.956708751657239e-06,
      "loss": 0.6723,
      "step": 43255
    },
    {
      "epoch": 1.83,
      "grad_norm": 19.522112154133893,
      "learning_rate": 3.9555047540397205e-06,
      "loss": 0.6627,
      "step": 43260
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.1641965941204075,
      "learning_rate": 3.954300819746871e-06,
      "loss": 0.6737,
      "step": 43265
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.60216412896515,
      "learning_rate": 3.953096948851684e-06,
      "loss": 0.6794,
      "step": 43270
    },
    {
      "epoch": 1.83,
      "grad_norm": 9.899998609851128,
      "learning_rate": 3.951893141427143e-06,
      "loss": 0.7121,
      "step": 43275
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.081734460823887,
      "learning_rate": 3.950689397546236e-06,
      "loss": 0.6708,
      "step": 43280
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.768901156262409,
      "learning_rate": 3.94948571728194e-06,
      "loss": 0.6351,
      "step": 43285
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.768818452468032,
      "learning_rate": 3.948282100707228e-06,
      "loss": 0.6552,
      "step": 43290
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.916892897619035,
      "learning_rate": 3.947078547895077e-06,
      "loss": 0.6628,
      "step": 43295
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.086628922300244,
      "learning_rate": 3.94587505891845e-06,
      "loss": 0.6966,
      "step": 43300
    },
    {
      "epoch": 1.83,
      "grad_norm": 7.269542017158951,
      "learning_rate": 3.944671633850315e-06,
      "loss": 0.6754,
      "step": 43305
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.1233819927685715,
      "learning_rate": 3.9434682727636295e-06,
      "loss": 0.6616,
      "step": 43310
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.184346981362648,
      "learning_rate": 3.9422649757313495e-06,
      "loss": 0.6528,
      "step": 43315
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.262895066599015,
      "learning_rate": 3.9410617428264295e-06,
      "loss": 0.635,
      "step": 43320
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.071039072377434,
      "learning_rate": 3.939858574121818e-06,
      "loss": 0.6945,
      "step": 43325
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.347479955240949,
      "learning_rate": 3.938655469690456e-06,
      "loss": 0.7107,
      "step": 43330
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.6231770735983595,
      "learning_rate": 3.937452429605288e-06,
      "loss": 0.633,
      "step": 43335
    },
    {
      "epoch": 1.83,
      "grad_norm": 4.2725807508336,
      "learning_rate": 3.936249453939248e-06,
      "loss": 0.6712,
      "step": 43340
    },
    {
      "epoch": 1.83,
      "grad_norm": 8.144957472916598,
      "learning_rate": 3.9350465427652715e-06,
      "loss": 0.6687,
      "step": 43345
    },
    {
      "epoch": 1.83,
      "grad_norm": 6.118678199907555,
      "learning_rate": 3.933843696156287e-06,
      "loss": 0.6556,
      "step": 43350
    },
    {
      "epoch": 1.83,
      "grad_norm": 5.85161572709164,
      "learning_rate": 3.932640914185217e-06,
      "loss": 0.6659,
      "step": 43355
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.457813280540542,
      "learning_rate": 3.931438196924986e-06,
      "loss": 0.6673,
      "step": 43360
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.209759208412717,
      "learning_rate": 3.930235544448508e-06,
      "loss": 0.7048,
      "step": 43365
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.454293060768661,
      "learning_rate": 3.929032956828698e-06,
      "loss": 0.6994,
      "step": 43370
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.216516722951402,
      "learning_rate": 3.927830434138466e-06,
      "loss": 0.6769,
      "step": 43375
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.560956924148917,
      "learning_rate": 3.926627976450715e-06,
      "loss": 0.6781,
      "step": 43380
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.077301188210084,
      "learning_rate": 3.925425583838351e-06,
      "loss": 0.666,
      "step": 43385
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.340286980880687,
      "learning_rate": 3.924223256374266e-06,
      "loss": 0.6578,
      "step": 43390
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.683872627804867,
      "learning_rate": 3.923020994131357e-06,
      "loss": 0.6669,
      "step": 43395
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.835574324865804,
      "learning_rate": 3.9218187971825125e-06,
      "loss": 0.6834,
      "step": 43400
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.6837097572560955,
      "learning_rate": 3.920616665600618e-06,
      "loss": 0.6454,
      "step": 43405
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.2785676195646305,
      "learning_rate": 3.9194145994585585e-06,
      "loss": 0.6662,
      "step": 43410
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.007243222649783,
      "learning_rate": 3.918212598829207e-06,
      "loss": 0.6721,
      "step": 43415
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.7932671210252,
      "learning_rate": 3.9170106637854394e-06,
      "loss": 0.6537,
      "step": 43420
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.8618005123900705,
      "learning_rate": 3.915808794400124e-06,
      "loss": 0.6424,
      "step": 43425
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.871929239660475,
      "learning_rate": 3.914606990746129e-06,
      "loss": 0.716,
      "step": 43430
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.5344097898610505,
      "learning_rate": 3.913405252896314e-06,
      "loss": 0.6954,
      "step": 43435
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.75999473005663,
      "learning_rate": 3.912203580923539e-06,
      "loss": 0.6686,
      "step": 43440
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.1841573787862325,
      "learning_rate": 3.911001974900656e-06,
      "loss": 0.6782,
      "step": 43445
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.971867231279984,
      "learning_rate": 3.9098004349005145e-06,
      "loss": 0.6682,
      "step": 43450
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.428575247710614,
      "learning_rate": 3.908598960995963e-06,
      "loss": 0.656,
      "step": 43455
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.208412098841864,
      "learning_rate": 3.907397553259841e-06,
      "loss": 0.6449,
      "step": 43460
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.450333295555813,
      "learning_rate": 3.906196211764987e-06,
      "loss": 0.6913,
      "step": 43465
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.903947897866068,
      "learning_rate": 3.904994936584235e-06,
      "loss": 0.6692,
      "step": 43470
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.954170114003696,
      "learning_rate": 3.903793727790414e-06,
      "loss": 0.702,
      "step": 43475
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.900158794673702,
      "learning_rate": 3.902592585456351e-06,
      "loss": 0.6546,
      "step": 43480
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.49567421163392,
      "learning_rate": 3.901391509654866e-06,
      "loss": 0.7139,
      "step": 43485
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.584744835739641,
      "learning_rate": 3.9001905004587795e-06,
      "loss": 0.68,
      "step": 43490
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.144948128211909,
      "learning_rate": 3.898989557940902e-06,
      "loss": 0.6618,
      "step": 43495
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.643356151682625,
      "learning_rate": 3.897788682174044e-06,
      "loss": 0.6659,
      "step": 43500
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.580475619039265,
      "learning_rate": 3.8965878732310135e-06,
      "loss": 0.6576,
      "step": 43505
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.523542944231962,
      "learning_rate": 3.8953871311846075e-06,
      "loss": 0.6855,
      "step": 43510
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.120210122480535,
      "learning_rate": 3.894186456107628e-06,
      "loss": 0.681,
      "step": 43515
    },
    {
      "epoch": 1.84,
      "grad_norm": 6.773309070503352,
      "learning_rate": 3.892985848072866e-06,
      "loss": 0.649,
      "step": 43520
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.667611663644984,
      "learning_rate": 3.8917853071531104e-06,
      "loss": 0.7167,
      "step": 43525
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.517046760241202,
      "learning_rate": 3.890584833421148e-06,
      "loss": 0.6888,
      "step": 43530
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.514075944504898,
      "learning_rate": 3.889384426949758e-06,
      "loss": 0.6695,
      "step": 43535
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.237499043077662,
      "learning_rate": 3.888184087811721e-06,
      "loss": 0.6532,
      "step": 43540
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.692564774375484,
      "learning_rate": 3.8869838160798066e-06,
      "loss": 0.666,
      "step": 43545
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.875781791813479,
      "learning_rate": 3.885783611826785e-06,
      "loss": 0.664,
      "step": 43550
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.474151908840804,
      "learning_rate": 3.8845834751254205e-06,
      "loss": 0.6678,
      "step": 43555
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.552780103830502,
      "learning_rate": 3.883383406048474e-06,
      "loss": 0.673,
      "step": 43560
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.408244640195775,
      "learning_rate": 3.882183404668703e-06,
      "loss": 0.6908,
      "step": 43565
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.447227750290457,
      "learning_rate": 3.8809834710588625e-06,
      "loss": 0.7058,
      "step": 43570
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.5794096861507745,
      "learning_rate": 3.8797836052916955e-06,
      "loss": 0.6479,
      "step": 43575
    },
    {
      "epoch": 1.84,
      "grad_norm": 9.686855068802615,
      "learning_rate": 3.878583807439947e-06,
      "loss": 0.675,
      "step": 43580
    },
    {
      "epoch": 1.84,
      "grad_norm": 4.766168367206868,
      "learning_rate": 3.877384077576362e-06,
      "loss": 0.657,
      "step": 43585
    },
    {
      "epoch": 1.84,
      "grad_norm": 7.721907734351429,
      "learning_rate": 3.8761844157736724e-06,
      "loss": 0.7091,
      "step": 43590
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.027681405503157,
      "learning_rate": 3.874984822104611e-06,
      "loss": 0.6949,
      "step": 43595
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.755042645828961,
      "learning_rate": 3.873785296641906e-06,
      "loss": 0.6486,
      "step": 43600
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.488500755511872,
      "learning_rate": 3.87258583945828e-06,
      "loss": 0.6395,
      "step": 43605
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.276374043897645,
      "learning_rate": 3.871386450626455e-06,
      "loss": 0.6966,
      "step": 43610
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.534321025774247,
      "learning_rate": 3.870187130219146e-06,
      "loss": 0.7133,
      "step": 43615
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.352187711296351,
      "learning_rate": 3.868987878309062e-06,
      "loss": 0.6492,
      "step": 43620
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.639115451362402,
      "learning_rate": 3.867788694968911e-06,
      "loss": 0.6417,
      "step": 43625
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.31656874232237,
      "learning_rate": 3.866589580271398e-06,
      "loss": 0.6604,
      "step": 43630
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.536399098647603,
      "learning_rate": 3.865390534289219e-06,
      "loss": 0.6356,
      "step": 43635
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.448698959558375,
      "learning_rate": 3.864191557095071e-06,
      "loss": 0.6585,
      "step": 43640
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.062678890480077,
      "learning_rate": 3.862992648761641e-06,
      "loss": 0.6616,
      "step": 43645
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.032453399006003,
      "learning_rate": 3.8617938093616194e-06,
      "loss": 0.6656,
      "step": 43650
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.446687201332551,
      "learning_rate": 3.860595038967686e-06,
      "loss": 0.67,
      "step": 43655
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.647943885302295,
      "learning_rate": 3.859396337652518e-06,
      "loss": 0.7045,
      "step": 43660
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.782269670955087,
      "learning_rate": 3.8581977054887924e-06,
      "loss": 0.6533,
      "step": 43665
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.865917464806391,
      "learning_rate": 3.856999142549175e-06,
      "loss": 0.6553,
      "step": 43670
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.109111346277947,
      "learning_rate": 3.855800648906334e-06,
      "loss": 0.6462,
      "step": 43675
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.570900878539706,
      "learning_rate": 3.854602224632928e-06,
      "loss": 0.6016,
      "step": 43680
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.764911045235446,
      "learning_rate": 3.853403869801616e-06,
      "loss": 0.6391,
      "step": 43685
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.394139698808642,
      "learning_rate": 3.85220558448505e-06,
      "loss": 0.6575,
      "step": 43690
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.770516318324496,
      "learning_rate": 3.851007368755878e-06,
      "loss": 0.6773,
      "step": 43695
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.790336293660406,
      "learning_rate": 3.849809222686745e-06,
      "loss": 0.6344,
      "step": 43700
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.741139450375439,
      "learning_rate": 3.848611146350292e-06,
      "loss": 0.6501,
      "step": 43705
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.439603363694406,
      "learning_rate": 3.847413139819151e-06,
      "loss": 0.6503,
      "step": 43710
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.349836624360438,
      "learning_rate": 3.846215203165959e-06,
      "loss": 0.6303,
      "step": 43715
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.213788028895998,
      "learning_rate": 3.845017336463339e-06,
      "loss": 0.6586,
      "step": 43720
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.9169042540232635,
      "learning_rate": 3.8438195397839165e-06,
      "loss": 0.6541,
      "step": 43725
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.043401929949646,
      "learning_rate": 3.842621813200309e-06,
      "loss": 0.698,
      "step": 43730
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.429431065122614,
      "learning_rate": 3.8414241567851345e-06,
      "loss": 0.7068,
      "step": 43735
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.470004504089073,
      "learning_rate": 3.840226570610998e-06,
      "loss": 0.6766,
      "step": 43740
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.391979936181031,
      "learning_rate": 3.839029054750509e-06,
      "loss": 0.6891,
      "step": 43745
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.5788332917059815,
      "learning_rate": 3.837831609276269e-06,
      "loss": 0.6633,
      "step": 43750
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.850638994402555,
      "learning_rate": 3.8366342342608736e-06,
      "loss": 0.6462,
      "step": 43755
    },
    {
      "epoch": 1.85,
      "grad_norm": 10.51537997514676,
      "learning_rate": 3.835436929776919e-06,
      "loss": 0.7082,
      "step": 43760
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.498857852806296,
      "learning_rate": 3.834239695896993e-06,
      "loss": 0.6733,
      "step": 43765
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.8515937201451464,
      "learning_rate": 3.83304253269368e-06,
      "loss": 0.6736,
      "step": 43770
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.2572020506823565,
      "learning_rate": 3.831845440239561e-06,
      "loss": 0.6869,
      "step": 43775
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.020769759559667,
      "learning_rate": 3.830648418607212e-06,
      "loss": 0.6398,
      "step": 43780
    },
    {
      "epoch": 1.85,
      "grad_norm": 4.519128285080232,
      "learning_rate": 3.8294514678692055e-06,
      "loss": 0.6847,
      "step": 43785
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.44723560874819,
      "learning_rate": 3.828254588098107e-06,
      "loss": 0.669,
      "step": 43790
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.467611578216941,
      "learning_rate": 3.827057779366483e-06,
      "loss": 0.6415,
      "step": 43795
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.36134911507406,
      "learning_rate": 3.825861041746891e-06,
      "loss": 0.67,
      "step": 43800
    },
    {
      "epoch": 1.85,
      "grad_norm": 8.740652603841294,
      "learning_rate": 3.824664375311884e-06,
      "loss": 0.6809,
      "step": 43805
    },
    {
      "epoch": 1.85,
      "grad_norm": 6.1788167619479655,
      "learning_rate": 3.8234677801340145e-06,
      "loss": 0.6629,
      "step": 43810
    },
    {
      "epoch": 1.85,
      "grad_norm": 12.6951043969471,
      "learning_rate": 3.822271256285828e-06,
      "loss": 0.6762,
      "step": 43815
    },
    {
      "epoch": 1.85,
      "grad_norm": 5.095298892695629,
      "learning_rate": 3.821074803839866e-06,
      "loss": 0.6631,
      "step": 43820
    },
    {
      "epoch": 1.85,
      "grad_norm": 9.778351563642264,
      "learning_rate": 3.819878422868667e-06,
      "loss": 0.6458,
      "step": 43825
    },
    {
      "epoch": 1.85,
      "grad_norm": 7.896191021171513,
      "learning_rate": 3.818682113444762e-06,
      "loss": 0.6694,
      "step": 43830
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.931489361361997,
      "learning_rate": 3.8174858756406805e-06,
      "loss": 0.6821,
      "step": 43835
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.585951505558606,
      "learning_rate": 3.816289709528947e-06,
      "loss": 0.6392,
      "step": 43840
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.60279178974891,
      "learning_rate": 3.815093615182083e-06,
      "loss": 0.6395,
      "step": 43845
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.579924548857887,
      "learning_rate": 3.8138975926726024e-06,
      "loss": 0.6419,
      "step": 43850
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.6139756124382805,
      "learning_rate": 3.812701642073016e-06,
      "loss": 0.6603,
      "step": 43855
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.5943033111072085,
      "learning_rate": 3.8115057634558326e-06,
      "loss": 0.6494,
      "step": 43860
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.037318693117249,
      "learning_rate": 3.810309956893553e-06,
      "loss": 0.6683,
      "step": 43865
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.522294919260756,
      "learning_rate": 3.809114222458678e-06,
      "loss": 0.6527,
      "step": 43870
    },
    {
      "epoch": 1.86,
      "grad_norm": 3.9897730305758423,
      "learning_rate": 3.8079185602236997e-06,
      "loss": 0.6681,
      "step": 43875
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.15341322089785,
      "learning_rate": 3.806722970261106e-06,
      "loss": 0.6871,
      "step": 43880
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.672122642144972,
      "learning_rate": 3.805527452643385e-06,
      "loss": 0.6676,
      "step": 43885
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.37488663199912,
      "learning_rate": 3.8043320074430163e-06,
      "loss": 0.6385,
      "step": 43890
    },
    {
      "epoch": 1.86,
      "grad_norm": 14.776323704367,
      "learning_rate": 3.803136634732477e-06,
      "loss": 0.6288,
      "step": 43895
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.85214489135842,
      "learning_rate": 3.801941334584236e-06,
      "loss": 0.6778,
      "step": 43900
    },
    {
      "epoch": 1.86,
      "grad_norm": 14.444636540141767,
      "learning_rate": 3.800746107070764e-06,
      "loss": 0.6639,
      "step": 43905
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.925559453177563,
      "learning_rate": 3.799550952264522e-06,
      "loss": 0.6648,
      "step": 43910
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.913675003822992,
      "learning_rate": 3.7983558702379702e-06,
      "loss": 0.682,
      "step": 43915
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.432791928824231,
      "learning_rate": 3.797160861063563e-06,
      "loss": 0.6377,
      "step": 43920
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.531645697526058,
      "learning_rate": 3.795965924813748e-06,
      "loss": 0.68,
      "step": 43925
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.838037462924723,
      "learning_rate": 3.7947710615609747e-06,
      "loss": 0.6799,
      "step": 43930
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.279029454599246,
      "learning_rate": 3.793576271377681e-06,
      "loss": 0.701,
      "step": 43935
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.927936563147069,
      "learning_rate": 3.792381554336303e-06,
      "loss": 0.6824,
      "step": 43940
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.805734180570522,
      "learning_rate": 3.791186910509276e-06,
      "loss": 0.6653,
      "step": 43945
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.410782797008012,
      "learning_rate": 3.7899923399690243e-06,
      "loss": 0.6663,
      "step": 43950
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.429877460034502,
      "learning_rate": 3.788797842787974e-06,
      "loss": 0.6458,
      "step": 43955
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.775474451982006,
      "learning_rate": 3.787603419038543e-06,
      "loss": 0.6828,
      "step": 43960
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.09036353715279,
      "learning_rate": 3.7864090687931444e-06,
      "loss": 0.6529,
      "step": 43965
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.821073616239193,
      "learning_rate": 3.78521479212419e-06,
      "loss": 0.6391,
      "step": 43970
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.401276820707356,
      "learning_rate": 3.784020589104083e-06,
      "loss": 0.6678,
      "step": 43975
    },
    {
      "epoch": 1.86,
      "grad_norm": 4.283738731204707,
      "learning_rate": 3.7828264598052276e-06,
      "loss": 0.679,
      "step": 43980
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.996358048887075,
      "learning_rate": 3.7816324043000184e-06,
      "loss": 0.6631,
      "step": 43985
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.193741933782854,
      "learning_rate": 3.780438422660847e-06,
      "loss": 0.6566,
      "step": 43990
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.563023484650932,
      "learning_rate": 3.779244514960103e-06,
      "loss": 0.663,
      "step": 43995
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.825769330158614,
      "learning_rate": 3.7780506812701667e-06,
      "loss": 0.6571,
      "step": 44000
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.11848957316736,
      "learning_rate": 3.7768569216634188e-06,
      "loss": 0.6297,
      "step": 44005
    },
    {
      "epoch": 1.86,
      "grad_norm": 14.974649806379034,
      "learning_rate": 3.7756632362122336e-06,
      "loss": 0.6635,
      "step": 44010
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.983737236999226,
      "learning_rate": 3.774469624988979e-06,
      "loss": 0.6724,
      "step": 44015
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.271968449915201,
      "learning_rate": 3.773276088066021e-06,
      "loss": 0.6979,
      "step": 44020
    },
    {
      "epoch": 1.86,
      "grad_norm": 6.094329929676694,
      "learning_rate": 3.7720826255157205e-06,
      "loss": 0.6494,
      "step": 44025
    },
    {
      "epoch": 1.86,
      "grad_norm": 10.57010020108791,
      "learning_rate": 3.7708892374104345e-06,
      "loss": 0.676,
      "step": 44030
    },
    {
      "epoch": 1.86,
      "grad_norm": 9.515036477726131,
      "learning_rate": 3.769695923822513e-06,
      "loss": 0.6651,
      "step": 44035
    },
    {
      "epoch": 1.86,
      "grad_norm": 12.682589075822317,
      "learning_rate": 3.768502684824303e-06,
      "loss": 0.6694,
      "step": 44040
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.129663953747231,
      "learning_rate": 3.767309520488148e-06,
      "loss": 0.6372,
      "step": 44045
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.839351985003534,
      "learning_rate": 3.7661164308863856e-06,
      "loss": 0.6973,
      "step": 44050
    },
    {
      "epoch": 1.86,
      "grad_norm": 5.305617604360949,
      "learning_rate": 3.764923416091353e-06,
      "loss": 0.6378,
      "step": 44055
    },
    {
      "epoch": 1.86,
      "grad_norm": 7.686272101175115,
      "learning_rate": 3.763730476175371e-06,
      "loss": 0.6645,
      "step": 44060
    },
    {
      "epoch": 1.86,
      "grad_norm": 8.366691750430366,
      "learning_rate": 3.7625376112107707e-06,
      "loss": 0.6407,
      "step": 44065
    },
    {
      "epoch": 1.87,
      "grad_norm": 10.505149037502722,
      "learning_rate": 3.761344821269869e-06,
      "loss": 0.6444,
      "step": 44070
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.3949564334999,
      "learning_rate": 3.7601521064249837e-06,
      "loss": 0.6406,
      "step": 44075
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.372347017816067,
      "learning_rate": 3.758959466748423e-06,
      "loss": 0.6576,
      "step": 44080
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.484014620770744,
      "learning_rate": 3.7577669023124936e-06,
      "loss": 0.6775,
      "step": 44085
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.386541888086565,
      "learning_rate": 3.7565744131894986e-06,
      "loss": 0.6256,
      "step": 44090
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.699277612708559,
      "learning_rate": 3.755381999451734e-06,
      "loss": 0.6614,
      "step": 44095
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.481357837371291,
      "learning_rate": 3.754189661171493e-06,
      "loss": 0.6694,
      "step": 44100
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.636109677296466,
      "learning_rate": 3.7529973984210635e-06,
      "loss": 0.6577,
      "step": 44105
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.70700062560069,
      "learning_rate": 3.751805211272727e-06,
      "loss": 0.6389,
      "step": 44110
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.297900817427233,
      "learning_rate": 3.7506130997987655e-06,
      "loss": 0.6212,
      "step": 44115
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.3646774737952105,
      "learning_rate": 3.74942106407145e-06,
      "loss": 0.6937,
      "step": 44120
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.1712802889275435,
      "learning_rate": 3.7482291041630535e-06,
      "loss": 0.6481,
      "step": 44125
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.744225160207648,
      "learning_rate": 3.747037220145838e-06,
      "loss": 0.6734,
      "step": 44130
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.126949492969892,
      "learning_rate": 3.745845412092065e-06,
      "loss": 0.6454,
      "step": 44135
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.753670704981321,
      "learning_rate": 3.7446536800739917e-06,
      "loss": 0.6498,
      "step": 44140
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.168160052396411,
      "learning_rate": 3.7434620241638663e-06,
      "loss": 0.6741,
      "step": 44145
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.604387165443849,
      "learning_rate": 3.7422704444339385e-06,
      "loss": 0.6712,
      "step": 44150
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.220277388809433,
      "learning_rate": 3.7410789409564496e-06,
      "loss": 0.6784,
      "step": 44155
    },
    {
      "epoch": 1.87,
      "grad_norm": 8.881509414789399,
      "learning_rate": 3.739887513803634e-06,
      "loss": 0.7064,
      "step": 44160
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.439599977866932,
      "learning_rate": 3.7386961630477282e-06,
      "loss": 0.679,
      "step": 44165
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.902589839024459,
      "learning_rate": 3.7375048887609576e-06,
      "loss": 0.6881,
      "step": 44170
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.288170191158458,
      "learning_rate": 3.736313691015548e-06,
      "loss": 0.6701,
      "step": 44175
    },
    {
      "epoch": 1.87,
      "grad_norm": 11.135060076279958,
      "learning_rate": 3.7351225698837167e-06,
      "loss": 0.6537,
      "step": 44180
    },
    {
      "epoch": 1.87,
      "grad_norm": 12.078523652516267,
      "learning_rate": 3.7339315254376774e-06,
      "loss": 0.6709,
      "step": 44185
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.331701638701411,
      "learning_rate": 3.732740557749641e-06,
      "loss": 0.652,
      "step": 44190
    },
    {
      "epoch": 1.87,
      "grad_norm": 19.129978739265965,
      "learning_rate": 3.7315496668918128e-06,
      "loss": 0.6709,
      "step": 44195
    },
    {
      "epoch": 1.87,
      "grad_norm": 6.218896040187197,
      "learning_rate": 3.73035885293639e-06,
      "loss": 0.6566,
      "step": 44200
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.959770789211603,
      "learning_rate": 3.7291681159555716e-06,
      "loss": 0.6647,
      "step": 44205
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.501266073937086,
      "learning_rate": 3.7279774560215452e-06,
      "loss": 0.7091,
      "step": 44210
    },
    {
      "epoch": 1.87,
      "grad_norm": 10.594362686735336,
      "learning_rate": 3.7267868732065005e-06,
      "loss": 0.651,
      "step": 44215
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.268700224192172,
      "learning_rate": 3.725596367582618e-06,
      "loss": 0.6616,
      "step": 44220
    },
    {
      "epoch": 1.87,
      "grad_norm": 9.253814690426344,
      "learning_rate": 3.7244059392220733e-06,
      "loss": 0.6722,
      "step": 44225
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.685008860420982,
      "learning_rate": 3.7232155881970376e-06,
      "loss": 0.6425,
      "step": 44230
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.0870372356279745,
      "learning_rate": 3.722025314579681e-06,
      "loss": 0.6456,
      "step": 44235
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.392549351976747,
      "learning_rate": 3.7208351184421653e-06,
      "loss": 0.7355,
      "step": 44240
    },
    {
      "epoch": 1.87,
      "grad_norm": 14.40627952625747,
      "learning_rate": 3.7196449998566474e-06,
      "loss": 0.7191,
      "step": 44245
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.348079146779934,
      "learning_rate": 3.7184549588952824e-06,
      "loss": 0.6772,
      "step": 44250
    },
    {
      "epoch": 1.87,
      "grad_norm": 5.05843181151354,
      "learning_rate": 3.7172649956302176e-06,
      "loss": 0.6656,
      "step": 44255
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.653324205110013,
      "learning_rate": 3.716075110133599e-06,
      "loss": 0.6825,
      "step": 44260
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.567897227527614,
      "learning_rate": 3.7148853024775644e-06,
      "loss": 0.6475,
      "step": 44265
    },
    {
      "epoch": 1.87,
      "grad_norm": 7.187016626228679,
      "learning_rate": 3.7136955727342473e-06,
      "loss": 0.6466,
      "step": 44270
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.41432540444829,
      "learning_rate": 3.7125059209757797e-06,
      "loss": 0.6375,
      "step": 44275
    },
    {
      "epoch": 1.87,
      "grad_norm": 11.999951802505393,
      "learning_rate": 3.7113163472742846e-06,
      "loss": 0.6819,
      "step": 44280
    },
    {
      "epoch": 1.87,
      "grad_norm": 10.410040534360478,
      "learning_rate": 3.7101268517018853e-06,
      "loss": 0.6387,
      "step": 44285
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.266805942584702,
      "learning_rate": 3.7089374343306946e-06,
      "loss": 0.6571,
      "step": 44290
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.219704348557099,
      "learning_rate": 3.7077480952328233e-06,
      "loss": 0.6365,
      "step": 44295
    },
    {
      "epoch": 1.87,
      "grad_norm": 4.219626629964817,
      "learning_rate": 3.7065588344803792e-06,
      "loss": 0.6455,
      "step": 44300
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.4584018551882325,
      "learning_rate": 3.705369652145463e-06,
      "loss": 0.6644,
      "step": 44305
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.352919390332786,
      "learning_rate": 3.7041805483001717e-06,
      "loss": 0.6953,
      "step": 44310
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.600554034792909,
      "learning_rate": 3.702991523016597e-06,
      "loss": 0.6399,
      "step": 44315
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.301940191767937,
      "learning_rate": 3.7018025763668253e-06,
      "loss": 0.6361,
      "step": 44320
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.489247757823377,
      "learning_rate": 3.70061370842294e-06,
      "loss": 0.6323,
      "step": 44325
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.43452871028988,
      "learning_rate": 3.6994249192570164e-06,
      "loss": 0.6595,
      "step": 44330
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.907822086285632,
      "learning_rate": 3.6982362089411308e-06,
      "loss": 0.6404,
      "step": 44335
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.0957310320126075,
      "learning_rate": 3.697047577547349e-06,
      "loss": 0.6828,
      "step": 44340
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.529642248641528,
      "learning_rate": 3.695859025147734e-06,
      "loss": 0.6853,
      "step": 44345
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.0217240022793375,
      "learning_rate": 3.6946705518143455e-06,
      "loss": 0.6815,
      "step": 44350
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.551933143512585,
      "learning_rate": 3.693482157619236e-06,
      "loss": 0.7043,
      "step": 44355
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.415377018243964,
      "learning_rate": 3.6922938426344545e-06,
      "loss": 0.6558,
      "step": 44360
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.679312374934443,
      "learning_rate": 3.6911056069320466e-06,
      "loss": 0.6371,
      "step": 44365
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.637995054690598,
      "learning_rate": 3.689917450584049e-06,
      "loss": 0.6328,
      "step": 44370
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.204780789956788,
      "learning_rate": 3.6887293736624984e-06,
      "loss": 0.6874,
      "step": 44375
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.237011429660383,
      "learning_rate": 3.687541376239424e-06,
      "loss": 0.7083,
      "step": 44380
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.297947248663635,
      "learning_rate": 3.68635345838685e-06,
      "loss": 0.6457,
      "step": 44385
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.76440493555294,
      "learning_rate": 3.6851656201767937e-06,
      "loss": 0.6782,
      "step": 44390
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.435663587975596,
      "learning_rate": 3.683977861681275e-06,
      "loss": 0.6422,
      "step": 44395
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.28543532198817,
      "learning_rate": 3.6827901829723014e-06,
      "loss": 0.6271,
      "step": 44400
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.358507988917818,
      "learning_rate": 3.6816025841218792e-06,
      "loss": 0.6746,
      "step": 44405
    },
    {
      "epoch": 1.88,
      "grad_norm": 12.462795419666628,
      "learning_rate": 3.680415065202009e-06,
      "loss": 0.6763,
      "step": 44410
    },
    {
      "epoch": 1.88,
      "grad_norm": 12.17572599566084,
      "learning_rate": 3.679227626284686e-06,
      "loss": 0.6464,
      "step": 44415
    },
    {
      "epoch": 1.88,
      "grad_norm": 10.243739229431425,
      "learning_rate": 3.678040267441902e-06,
      "loss": 0.6521,
      "step": 44420
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.01596663579331,
      "learning_rate": 3.676852988745642e-06,
      "loss": 0.6391,
      "step": 44425
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.40865246196711,
      "learning_rate": 3.675665790267888e-06,
      "loss": 0.6379,
      "step": 44430
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.074104348512498,
      "learning_rate": 3.674478672080617e-06,
      "loss": 0.6167,
      "step": 44435
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.485346762435941,
      "learning_rate": 3.673291634255799e-06,
      "loss": 0.6601,
      "step": 44440
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.615909477516975,
      "learning_rate": 3.6721046768654013e-06,
      "loss": 0.6495,
      "step": 44445
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.371002327584805,
      "learning_rate": 3.670917799981385e-06,
      "loss": 0.6379,
      "step": 44450
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.299480539305591,
      "learning_rate": 3.6697310036757084e-06,
      "loss": 0.6459,
      "step": 44455
    },
    {
      "epoch": 1.88,
      "grad_norm": 6.752574073229355,
      "learning_rate": 3.668544288020323e-06,
      "loss": 0.6721,
      "step": 44460
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.193309266251985,
      "learning_rate": 3.6673576530871748e-06,
      "loss": 0.6474,
      "step": 44465
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.954155210563334,
      "learning_rate": 3.666171098948207e-06,
      "loss": 0.6593,
      "step": 44470
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.781614179318758,
      "learning_rate": 3.6649846256753564e-06,
      "loss": 0.6596,
      "step": 44475
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.311822699128,
      "learning_rate": 3.663798233340555e-06,
      "loss": 0.6668,
      "step": 44480
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.445068603690859,
      "learning_rate": 3.6626119220157312e-06,
      "loss": 0.6201,
      "step": 44485
    },
    {
      "epoch": 1.88,
      "grad_norm": 8.241377148734646,
      "learning_rate": 3.6614256917728065e-06,
      "loss": 0.6431,
      "step": 44490
    },
    {
      "epoch": 1.88,
      "grad_norm": 13.188462070226933,
      "learning_rate": 3.6602395426837006e-06,
      "loss": 0.6656,
      "step": 44495
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.021710764148414,
      "learning_rate": 3.6590534748203246e-06,
      "loss": 0.6674,
      "step": 44500
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.053621235001356,
      "learning_rate": 3.657867488254585e-06,
      "loss": 0.6866,
      "step": 44505
    },
    {
      "epoch": 1.88,
      "grad_norm": 5.974071741242696,
      "learning_rate": 3.656681583058388e-06,
      "loss": 0.635,
      "step": 44510
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.106047868128863,
      "learning_rate": 3.6554957593036282e-06,
      "loss": 0.6644,
      "step": 44515
    },
    {
      "epoch": 1.88,
      "grad_norm": 7.690061421448045,
      "learning_rate": 3.6543100170622016e-06,
      "loss": 0.6669,
      "step": 44520
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.978681224202517,
      "learning_rate": 3.653124356405995e-06,
      "loss": 0.6626,
      "step": 44525
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.710980572047136,
      "learning_rate": 3.6519387774068893e-06,
      "loss": 0.6568,
      "step": 44530
    },
    {
      "epoch": 1.88,
      "grad_norm": 4.349572648026729,
      "learning_rate": 3.6507532801367664e-06,
      "loss": 0.6661,
      "step": 44535
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.68165840821385,
      "learning_rate": 3.6495678646674963e-06,
      "loss": 0.6664,
      "step": 44540
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.780901752604387,
      "learning_rate": 3.6483825310709525e-06,
      "loss": 0.6724,
      "step": 44545
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.402791580836213,
      "learning_rate": 3.647197279418991e-06,
      "loss": 0.6882,
      "step": 44550
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.905953345437235,
      "learning_rate": 3.6460121097834742e-06,
      "loss": 0.6225,
      "step": 44555
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.2943104417803895,
      "learning_rate": 3.6448270222362546e-06,
      "loss": 0.662,
      "step": 44560
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.5793642379574395,
      "learning_rate": 3.643642016849182e-06,
      "loss": 0.6649,
      "step": 44565
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.511713610976811,
      "learning_rate": 3.6424570936940986e-06,
      "loss": 0.6294,
      "step": 44570
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.32538589315091,
      "learning_rate": 3.6412722528428413e-06,
      "loss": 0.6609,
      "step": 44575
    },
    {
      "epoch": 1.89,
      "grad_norm": 23.305939413798264,
      "learning_rate": 3.640087494367246e-06,
      "loss": 0.6481,
      "step": 44580
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.3670008210935745,
      "learning_rate": 3.6389028183391396e-06,
      "loss": 0.6538,
      "step": 44585
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.449395938897598,
      "learning_rate": 3.637718224830347e-06,
      "loss": 0.6771,
      "step": 44590
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.522107576959655,
      "learning_rate": 3.6365337139126857e-06,
      "loss": 0.6113,
      "step": 44595
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.881631118254926,
      "learning_rate": 3.635349285657968e-06,
      "loss": 0.6519,
      "step": 44600
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.466151157650065,
      "learning_rate": 3.6341649401380043e-06,
      "loss": 0.6635,
      "step": 44605
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.51232343423555,
      "learning_rate": 3.6329806774245963e-06,
      "loss": 0.64,
      "step": 44610
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.906065495845976,
      "learning_rate": 3.6317964975895442e-06,
      "loss": 0.6541,
      "step": 44615
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.6029962836525495,
      "learning_rate": 3.63061240070464e-06,
      "loss": 0.6374,
      "step": 44620
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.02030287257848,
      "learning_rate": 3.6294283868416714e-06,
      "loss": 0.6591,
      "step": 44625
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.060862699542064,
      "learning_rate": 3.628244456072424e-06,
      "loss": 0.6484,
      "step": 44630
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.563518026792678,
      "learning_rate": 3.627060608468673e-06,
      "loss": 0.6576,
      "step": 44635
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.723319087263991,
      "learning_rate": 3.6258768441021943e-06,
      "loss": 0.655,
      "step": 44640
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.264733957462176,
      "learning_rate": 3.6246931630447548e-06,
      "loss": 0.6289,
      "step": 44645
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.200621514320712,
      "learning_rate": 3.6235095653681167e-06,
      "loss": 0.6522,
      "step": 44650
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.416073621557099,
      "learning_rate": 3.6223260511440407e-06,
      "loss": 0.6263,
      "step": 44655
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.716151681518997,
      "learning_rate": 3.6211426204442768e-06,
      "loss": 0.6523,
      "step": 44660
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.172433992365676,
      "learning_rate": 3.6199592733405746e-06,
      "loss": 0.6721,
      "step": 44665
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.273415107719127,
      "learning_rate": 3.6187760099046774e-06,
      "loss": 0.6726,
      "step": 44670
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.516759959038405,
      "learning_rate": 3.617592830208321e-06,
      "loss": 0.6572,
      "step": 44675
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.670927772111441,
      "learning_rate": 3.6164097343232402e-06,
      "loss": 0.6671,
      "step": 44680
    },
    {
      "epoch": 1.89,
      "grad_norm": 3.919116688751525,
      "learning_rate": 3.615226722321161e-06,
      "loss": 0.6244,
      "step": 44685
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.1650510615775795,
      "learning_rate": 3.614043794273807e-06,
      "loss": 0.6586,
      "step": 44690
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.816618947748804,
      "learning_rate": 3.6128609502528963e-06,
      "loss": 0.659,
      "step": 44695
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.26059501547986,
      "learning_rate": 3.611678190330138e-06,
      "loss": 0.6362,
      "step": 44700
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.788416578470977,
      "learning_rate": 3.6104955145772445e-06,
      "loss": 0.689,
      "step": 44705
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.674565215168355,
      "learning_rate": 3.6093129230659143e-06,
      "loss": 0.6971,
      "step": 44710
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.689096346233111,
      "learning_rate": 3.6081304158678443e-06,
      "loss": 0.6371,
      "step": 44715
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.223490064082183,
      "learning_rate": 3.606947993054727e-06,
      "loss": 0.6484,
      "step": 44720
    },
    {
      "epoch": 1.89,
      "grad_norm": 6.940292153112864,
      "learning_rate": 3.60576565469825e-06,
      "loss": 0.6198,
      "step": 44725
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.442786455891358,
      "learning_rate": 3.604583400870093e-06,
      "loss": 0.6822,
      "step": 44730
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.893123389925185,
      "learning_rate": 3.6034012316419358e-06,
      "loss": 0.6654,
      "step": 44735
    },
    {
      "epoch": 1.89,
      "grad_norm": 7.219036345636989,
      "learning_rate": 3.6022191470854484e-06,
      "loss": 0.7001,
      "step": 44740
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.831970149389107,
      "learning_rate": 3.601037147272295e-06,
      "loss": 0.6497,
      "step": 44745
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.62155156573502,
      "learning_rate": 3.5998552322741404e-06,
      "loss": 0.63,
      "step": 44750
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.7965635258314885,
      "learning_rate": 3.598673402162637e-06,
      "loss": 0.6726,
      "step": 44755
    },
    {
      "epoch": 1.89,
      "grad_norm": 4.276739326914388,
      "learning_rate": 3.5974916570094393e-06,
      "loss": 0.6413,
      "step": 44760
    },
    {
      "epoch": 1.89,
      "grad_norm": 12.388191735773463,
      "learning_rate": 3.5963099968861905e-06,
      "loss": 0.6734,
      "step": 44765
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.627701816732168,
      "learning_rate": 3.595128421864531e-06,
      "loss": 0.6518,
      "step": 44770
    },
    {
      "epoch": 1.89,
      "grad_norm": 5.174378873460719,
      "learning_rate": 3.5939469320160987e-06,
      "loss": 0.6575,
      "step": 44775
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.3445151023155315,
      "learning_rate": 3.5927655274125217e-06,
      "loss": 0.6134,
      "step": 44780
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.764080039728375,
      "learning_rate": 3.5915842081254248e-06,
      "loss": 0.6625,
      "step": 44785
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.834115723605342,
      "learning_rate": 3.59040297422643e-06,
      "loss": 0.6227,
      "step": 44790
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.26725284834661,
      "learning_rate": 3.589221825787149e-06,
      "loss": 0.6419,
      "step": 44795
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.352050766409466,
      "learning_rate": 3.588040762879195e-06,
      "loss": 0.6173,
      "step": 44800
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.896229313998804,
      "learning_rate": 3.58685978557417e-06,
      "loss": 0.6528,
      "step": 44805
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.051583581058532,
      "learning_rate": 3.5856788939436724e-06,
      "loss": 0.6394,
      "step": 44810
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.277739363039084,
      "learning_rate": 3.5844980880592995e-06,
      "loss": 0.676,
      "step": 44815
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.292977386185439,
      "learning_rate": 3.583317367992636e-06,
      "loss": 0.6668,
      "step": 44820
    },
    {
      "epoch": 1.9,
      "grad_norm": 9.210157182157122,
      "learning_rate": 3.582136733815269e-06,
      "loss": 0.6336,
      "step": 44825
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.057567042820143,
      "learning_rate": 3.580956185598775e-06,
      "loss": 0.6555,
      "step": 44830
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.791431581048354,
      "learning_rate": 3.579775723414727e-06,
      "loss": 0.6451,
      "step": 44835
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.286405328547201,
      "learning_rate": 3.5785953473346933e-06,
      "loss": 0.6364,
      "step": 44840
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.8697762053581055,
      "learning_rate": 3.577415057430237e-06,
      "loss": 0.6849,
      "step": 44845
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.204917099644239,
      "learning_rate": 3.576234853772916e-06,
      "loss": 0.6175,
      "step": 44850
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.042863141688427,
      "learning_rate": 3.5750547364342823e-06,
      "loss": 0.6385,
      "step": 44855
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.691394252136263,
      "learning_rate": 3.5738747054858812e-06,
      "loss": 0.6835,
      "step": 44860
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.047136578018468,
      "learning_rate": 3.5726947609992585e-06,
      "loss": 0.6381,
      "step": 44865
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.160475461695407,
      "learning_rate": 3.571514903045947e-06,
      "loss": 0.6276,
      "step": 44870
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.401212819277797,
      "learning_rate": 3.5703351316974787e-06,
      "loss": 0.6738,
      "step": 44875
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.501935376921121,
      "learning_rate": 3.5691554470253796e-06,
      "loss": 0.6776,
      "step": 44880
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.54656871780731,
      "learning_rate": 3.5679758491011717e-06,
      "loss": 0.6659,
      "step": 44885
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.2961646383851875,
      "learning_rate": 3.5667963379963687e-06,
      "loss": 0.6432,
      "step": 44890
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.6476604863741215,
      "learning_rate": 3.5656169137824836e-06,
      "loss": 0.6629,
      "step": 44895
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.170380621099829,
      "learning_rate": 3.5644375765310203e-06,
      "loss": 0.6848,
      "step": 44900
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.616275601781574,
      "learning_rate": 3.563258326313477e-06,
      "loss": 0.6457,
      "step": 44905
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.251624826277175,
      "learning_rate": 3.5620791632013495e-06,
      "loss": 0.6532,
      "step": 44910
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.958784897243633,
      "learning_rate": 3.560900087266126e-06,
      "loss": 0.6317,
      "step": 44915
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.82456531873667,
      "learning_rate": 3.559721098579293e-06,
      "loss": 0.7084,
      "step": 44920
    },
    {
      "epoch": 1.9,
      "grad_norm": 12.857695889974073,
      "learning_rate": 3.558542197212326e-06,
      "loss": 0.6209,
      "step": 44925
    },
    {
      "epoch": 1.9,
      "grad_norm": 11.113439912938025,
      "learning_rate": 3.5573633832366994e-06,
      "loss": 0.6661,
      "step": 44930
    },
    {
      "epoch": 1.9,
      "grad_norm": 14.446699449200743,
      "learning_rate": 3.556184656723883e-06,
      "loss": 0.6422,
      "step": 44935
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.59533747962382,
      "learning_rate": 3.555006017745336e-06,
      "loss": 0.6181,
      "step": 44940
    },
    {
      "epoch": 1.9,
      "grad_norm": 24.873130660378386,
      "learning_rate": 3.5538274663725192e-06,
      "loss": 0.6507,
      "step": 44945
    },
    {
      "epoch": 1.9,
      "grad_norm": 15.665247878620855,
      "learning_rate": 3.5526490026768833e-06,
      "loss": 0.6483,
      "step": 44950
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.955561939850273,
      "learning_rate": 3.5514706267298736e-06,
      "loss": 0.6783,
      "step": 44955
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.6593053473092985,
      "learning_rate": 3.5502923386029355e-06,
      "loss": 0.6421,
      "step": 44960
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.196117810003258,
      "learning_rate": 3.5491141383675e-06,
      "loss": 0.6601,
      "step": 44965
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.760959060021661,
      "learning_rate": 3.5479360260950023e-06,
      "loss": 0.6483,
      "step": 44970
    },
    {
      "epoch": 1.9,
      "grad_norm": 7.334754722085779,
      "learning_rate": 3.546758001856866e-06,
      "loss": 0.6436,
      "step": 44975
    },
    {
      "epoch": 1.9,
      "grad_norm": 8.0903609451562,
      "learning_rate": 3.54558006572451e-06,
      "loss": 0.6434,
      "step": 44980
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.09644229027741,
      "learning_rate": 3.544402217769352e-06,
      "loss": 0.636,
      "step": 44985
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.7306181457901495,
      "learning_rate": 3.543224458062799e-06,
      "loss": 0.6412,
      "step": 44990
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.692614154806212,
      "learning_rate": 3.5420467866762565e-06,
      "loss": 0.6364,
      "step": 44995
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.627130481402302,
      "learning_rate": 3.5408692036811234e-06,
      "loss": 0.6543,
      "step": 45000
    },
    {
      "epoch": 1.9,
      "grad_norm": 4.249746776839556,
      "learning_rate": 3.5396917091487907e-06,
      "loss": 0.674,
      "step": 45005
    },
    {
      "epoch": 1.9,
      "grad_norm": 6.774863664854658,
      "learning_rate": 3.53851430315065e-06,
      "loss": 0.6671,
      "step": 45010
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.859105662751808,
      "learning_rate": 3.5373369857580807e-06,
      "loss": 0.6327,
      "step": 45015
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.988088410636686,
      "learning_rate": 3.5361597570424628e-06,
      "loss": 0.631,
      "step": 45020
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.072233309313598,
      "learning_rate": 3.5349826170751665e-06,
      "loss": 0.7023,
      "step": 45025
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.913391968382866,
      "learning_rate": 3.5338055659275605e-06,
      "loss": 0.6248,
      "step": 45030
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.378224479194328,
      "learning_rate": 3.5326286036710017e-06,
      "loss": 0.7001,
      "step": 45035
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.092740629656298,
      "learning_rate": 3.531451730376849e-06,
      "loss": 0.6623,
      "step": 45040
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.636800725818945,
      "learning_rate": 3.5302749461164533e-06,
      "loss": 0.6431,
      "step": 45045
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.3715367662667175,
      "learning_rate": 3.5290982509611567e-06,
      "loss": 0.6124,
      "step": 45050
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.672022547363441,
      "learning_rate": 3.527921644982302e-06,
      "loss": 0.6249,
      "step": 45055
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.1904369484872515,
      "learning_rate": 3.52674512825122e-06,
      "loss": 0.6138,
      "step": 45060
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.19605496824398,
      "learning_rate": 3.5255687008392425e-06,
      "loss": 0.6618,
      "step": 45065
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.144042280162643,
      "learning_rate": 3.5243923628176914e-06,
      "loss": 0.6589,
      "step": 45070
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.237017772517374,
      "learning_rate": 3.5232161142578836e-06,
      "loss": 0.6536,
      "step": 45075
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.534056114779388,
      "learning_rate": 3.5220399552311346e-06,
      "loss": 0.618,
      "step": 45080
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.307479248222899,
      "learning_rate": 3.52086388580875e-06,
      "loss": 0.6429,
      "step": 45085
    },
    {
      "epoch": 1.91,
      "grad_norm": 13.560355134535,
      "learning_rate": 3.5196879060620286e-06,
      "loss": 0.6676,
      "step": 45090
    },
    {
      "epoch": 1.91,
      "grad_norm": 9.47544758678445,
      "learning_rate": 3.518512016062271e-06,
      "loss": 0.6703,
      "step": 45095
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.799166683171064,
      "learning_rate": 3.5173362158807643e-06,
      "loss": 0.6364,
      "step": 45100
    },
    {
      "epoch": 1.91,
      "grad_norm": 7.909903853368974,
      "learning_rate": 3.5161605055887973e-06,
      "loss": 0.6506,
      "step": 45105
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.45293110294423,
      "learning_rate": 3.5149848852576474e-06,
      "loss": 0.6492,
      "step": 45110
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.426069364358127,
      "learning_rate": 3.5138093549585893e-06,
      "loss": 0.689,
      "step": 45115
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.768298832492627,
      "learning_rate": 3.512633914762893e-06,
      "loss": 0.6291,
      "step": 45120
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.219609560024683,
      "learning_rate": 3.51145856474182e-06,
      "loss": 0.6611,
      "step": 45125
    },
    {
      "epoch": 1.91,
      "grad_norm": 3.9343306322600107,
      "learning_rate": 3.5102833049666307e-06,
      "loss": 0.6336,
      "step": 45130
    },
    {
      "epoch": 1.91,
      "grad_norm": 3.973331859137556,
      "learning_rate": 3.5091081355085763e-06,
      "loss": 0.6248,
      "step": 45135
    },
    {
      "epoch": 1.91,
      "grad_norm": 8.813870047716192,
      "learning_rate": 3.507933056438904e-06,
      "loss": 0.6382,
      "step": 45140
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.719698739752371,
      "learning_rate": 3.506758067828856e-06,
      "loss": 0.6486,
      "step": 45145
    },
    {
      "epoch": 1.91,
      "grad_norm": 10.309332937060988,
      "learning_rate": 3.5055831697496667e-06,
      "loss": 0.6529,
      "step": 45150
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.194784020741479,
      "learning_rate": 3.5044083622725688e-06,
      "loss": 0.6329,
      "step": 45155
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.807824937213841,
      "learning_rate": 3.5032336454687867e-06,
      "loss": 0.6419,
      "step": 45160
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.25104281768087,
      "learning_rate": 3.5020590194095383e-06,
      "loss": 0.6267,
      "step": 45165
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.501850128628913,
      "learning_rate": 3.5008844841660408e-06,
      "loss": 0.667,
      "step": 45170
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.338320866381703,
      "learning_rate": 3.4997100398095003e-06,
      "loss": 0.642,
      "step": 45175
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.480245221234186,
      "learning_rate": 3.4985356864111215e-06,
      "loss": 0.6134,
      "step": 45180
    },
    {
      "epoch": 1.91,
      "grad_norm": 6.434761186358677,
      "learning_rate": 3.4973614240421018e-06,
      "loss": 0.6542,
      "step": 45185
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.922162208983262,
      "learning_rate": 3.496187252773633e-06,
      "loss": 0.6635,
      "step": 45190
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.633068683052903,
      "learning_rate": 3.4950131726769e-06,
      "loss": 0.6623,
      "step": 45195
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.310601094445128,
      "learning_rate": 3.4938391838230855e-06,
      "loss": 0.6409,
      "step": 45200
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.279081427197244,
      "learning_rate": 3.492665286283366e-06,
      "loss": 0.6248,
      "step": 45205
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.3555152370178245,
      "learning_rate": 3.4914914801289086e-06,
      "loss": 0.6935,
      "step": 45210
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.59167451299066,
      "learning_rate": 3.49031776543088e-06,
      "loss": 0.64,
      "step": 45215
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.4084231211790605,
      "learning_rate": 3.4891441422604377e-06,
      "loss": 0.6525,
      "step": 45220
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.705064326137762,
      "learning_rate": 3.4879706106887367e-06,
      "loss": 0.671,
      "step": 45225
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.291866420477239,
      "learning_rate": 3.4867971707869236e-06,
      "loss": 0.6444,
      "step": 45230
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.3499356752475595,
      "learning_rate": 3.48562382262614e-06,
      "loss": 0.6001,
      "step": 45235
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.680347670419772,
      "learning_rate": 3.4844505662775243e-06,
      "loss": 0.6516,
      "step": 45240
    },
    {
      "epoch": 1.91,
      "grad_norm": 4.374486625621403,
      "learning_rate": 3.483277401812205e-06,
      "loss": 0.6618,
      "step": 45245
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.885142577134379,
      "learning_rate": 3.48210432930131e-06,
      "loss": 0.6421,
      "step": 45250
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.50862123955264,
      "learning_rate": 3.480931348815959e-06,
      "loss": 0.6344,
      "step": 45255
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.34030285205759,
      "learning_rate": 3.479758460427265e-06,
      "loss": 0.6208,
      "step": 45260
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.980381917399348,
      "learning_rate": 3.478585664206338e-06,
      "loss": 0.6389,
      "step": 45265
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.441674112278496,
      "learning_rate": 3.4774129602242803e-06,
      "loss": 0.6372,
      "step": 45270
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.384317833165869,
      "learning_rate": 3.47624034855219e-06,
      "loss": 0.6896,
      "step": 45275
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.996378917166695,
      "learning_rate": 3.4750678292611594e-06,
      "loss": 0.621,
      "step": 45280
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.3055117919830215,
      "learning_rate": 3.473895402422274e-06,
      "loss": 0.6587,
      "step": 45285
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.550251939016165,
      "learning_rate": 3.472723068106616e-06,
      "loss": 0.6374,
      "step": 45290
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.217032013369853,
      "learning_rate": 3.471550826385258e-06,
      "loss": 0.6458,
      "step": 45295
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.275140809942955,
      "learning_rate": 3.4703786773292725e-06,
      "loss": 0.657,
      "step": 45300
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.014511494782284,
      "learning_rate": 3.4692066210097218e-06,
      "loss": 0.6436,
      "step": 45305
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.6458925905416155,
      "learning_rate": 3.4680346574976644e-06,
      "loss": 0.6379,
      "step": 45310
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.500046154213909,
      "learning_rate": 3.4668627868641537e-06,
      "loss": 0.6312,
      "step": 45315
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.834121992633179,
      "learning_rate": 3.4656910091802346e-06,
      "loss": 0.6666,
      "step": 45320
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.752444694831217,
      "learning_rate": 3.4645193245169517e-06,
      "loss": 0.6232,
      "step": 45325
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.42214176923696,
      "learning_rate": 3.4633477329453395e-06,
      "loss": 0.6427,
      "step": 45330
    },
    {
      "epoch": 1.92,
      "grad_norm": 10.973104035260567,
      "learning_rate": 3.4621762345364264e-06,
      "loss": 0.6038,
      "step": 45335
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.567708285093062,
      "learning_rate": 3.4610048293612397e-06,
      "loss": 0.6484,
      "step": 45340
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.635953752822431,
      "learning_rate": 3.459833517490797e-06,
      "loss": 0.622,
      "step": 45345
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.468031644076369,
      "learning_rate": 3.4586622989961127e-06,
      "loss": 0.6426,
      "step": 45350
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.714235821334394,
      "learning_rate": 3.4574911739481907e-06,
      "loss": 0.6617,
      "step": 45355
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.248707188603817,
      "learning_rate": 3.456320142418036e-06,
      "loss": 0.6336,
      "step": 45360
    },
    {
      "epoch": 1.92,
      "grad_norm": 16.237906247248514,
      "learning_rate": 3.4551492044766444e-06,
      "loss": 0.6956,
      "step": 45365
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.042196472785983,
      "learning_rate": 3.453978360195005e-06,
      "loss": 0.6261,
      "step": 45370
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.417801576332172,
      "learning_rate": 3.4528076096441044e-06,
      "loss": 0.6393,
      "step": 45375
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.158103615600723,
      "learning_rate": 3.45163695289492e-06,
      "loss": 0.6699,
      "step": 45380
    },
    {
      "epoch": 1.92,
      "grad_norm": 14.036393587692878,
      "learning_rate": 3.4504663900184277e-06,
      "loss": 0.6537,
      "step": 45385
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.903788158320299,
      "learning_rate": 3.4492959210855936e-06,
      "loss": 0.6255,
      "step": 45390
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.238431591391574,
      "learning_rate": 3.448125546167379e-06,
      "loss": 0.6237,
      "step": 45395
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.449758078284561,
      "learning_rate": 3.446955265334742e-06,
      "loss": 0.6502,
      "step": 45400
    },
    {
      "epoch": 1.92,
      "grad_norm": 6.462202171715268,
      "learning_rate": 3.445785078658632e-06,
      "loss": 0.6199,
      "step": 45405
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.490308446448917,
      "learning_rate": 3.4446149862099955e-06,
      "loss": 0.6344,
      "step": 45410
    },
    {
      "epoch": 1.92,
      "grad_norm": 10.267292511315063,
      "learning_rate": 3.443444988059771e-06,
      "loss": 0.6591,
      "step": 45415
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.593758709757272,
      "learning_rate": 3.442275084278891e-06,
      "loss": 0.6395,
      "step": 45420
    },
    {
      "epoch": 1.92,
      "grad_norm": 5.51366980802667,
      "learning_rate": 3.441105274938285e-06,
      "loss": 0.6146,
      "step": 45425
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.301590409914633,
      "learning_rate": 3.4399355601088735e-06,
      "loss": 0.6242,
      "step": 45430
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.204318838542918,
      "learning_rate": 3.4387659398615746e-06,
      "loss": 0.6259,
      "step": 45435
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.503210359347232,
      "learning_rate": 3.437596414267298e-06,
      "loss": 0.6417,
      "step": 45440
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.75473444726216,
      "learning_rate": 3.436426983396947e-06,
      "loss": 0.677,
      "step": 45445
    },
    {
      "epoch": 1.92,
      "grad_norm": 7.024313056666088,
      "learning_rate": 3.4352576473214238e-06,
      "loss": 0.6319,
      "step": 45450
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.181929513580396,
      "learning_rate": 3.43408840611162e-06,
      "loss": 0.6577,
      "step": 45455
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.491345230249925,
      "learning_rate": 3.4329192598384236e-06,
      "loss": 0.6504,
      "step": 45460
    },
    {
      "epoch": 1.92,
      "grad_norm": 8.44920034533202,
      "learning_rate": 3.4317502085727172e-06,
      "loss": 0.6447,
      "step": 45465
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.877218677330802,
      "learning_rate": 3.430581252385374e-06,
      "loss": 0.6111,
      "step": 45470
    },
    {
      "epoch": 1.92,
      "grad_norm": 4.661850280878438,
      "learning_rate": 3.4294123913472694e-06,
      "loss": 0.6126,
      "step": 45475
    },
    {
      "epoch": 1.92,
      "grad_norm": 9.080635964206005,
      "learning_rate": 3.428243625529263e-06,
      "loss": 0.6175,
      "step": 45480
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.2005051644733005,
      "learning_rate": 3.4270749550022176e-06,
      "loss": 0.6249,
      "step": 45485
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.485702976987765,
      "learning_rate": 3.4259063798369842e-06,
      "loss": 0.6747,
      "step": 45490
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.399072599858354,
      "learning_rate": 3.4247379001044085e-06,
      "loss": 0.6516,
      "step": 45495
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.0871058696232145,
      "learning_rate": 3.423569515875336e-06,
      "loss": 0.6362,
      "step": 45500
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.857536226276318,
      "learning_rate": 3.4224012272205982e-06,
      "loss": 0.6536,
      "step": 45505
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.300340740463707,
      "learning_rate": 3.421233034211029e-06,
      "loss": 0.6256,
      "step": 45510
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.338711691489341,
      "learning_rate": 3.420064936917451e-06,
      "loss": 0.6338,
      "step": 45515
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.272017307438853,
      "learning_rate": 3.4188969354106805e-06,
      "loss": 0.6657,
      "step": 45520
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.450867501709167,
      "learning_rate": 3.4177290297615314e-06,
      "loss": 0.604,
      "step": 45525
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.512853510261402,
      "learning_rate": 3.416561220040811e-06,
      "loss": 0.6549,
      "step": 45530
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.36361541965981,
      "learning_rate": 3.41539350631932e-06,
      "loss": 0.6046,
      "step": 45535
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.454954899309723,
      "learning_rate": 3.4142258886678513e-06,
      "loss": 0.6334,
      "step": 45540
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.772077205290527,
      "learning_rate": 3.413058367157198e-06,
      "loss": 0.6318,
      "step": 45545
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.503724298710208,
      "learning_rate": 3.4118909418581405e-06,
      "loss": 0.597,
      "step": 45550
    },
    {
      "epoch": 1.93,
      "grad_norm": 12.623475782035229,
      "learning_rate": 3.4107236128414573e-06,
      "loss": 0.6733,
      "step": 45555
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.097557427822642,
      "learning_rate": 3.4095563801779206e-06,
      "loss": 0.6495,
      "step": 45560
    },
    {
      "epoch": 1.93,
      "grad_norm": 21.7313838759738,
      "learning_rate": 3.4083892439382947e-06,
      "loss": 0.6339,
      "step": 45565
    },
    {
      "epoch": 1.93,
      "grad_norm": 6.393503216674844,
      "learning_rate": 3.407222204193341e-06,
      "loss": 0.6165,
      "step": 45570
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.440694518791359,
      "learning_rate": 3.4060552610138135e-06,
      "loss": 0.666,
      "step": 45575
    },
    {
      "epoch": 1.93,
      "grad_norm": 10.732716293410824,
      "learning_rate": 3.4048884144704606e-06,
      "loss": 0.6778,
      "step": 45580
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.939427483936214,
      "learning_rate": 3.4037216646340253e-06,
      "loss": 0.6379,
      "step": 45585
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.9883694612238685,
      "learning_rate": 3.402555011575242e-06,
      "loss": 0.6469,
      "step": 45590
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.176765190840058,
      "learning_rate": 3.4013884553648445e-06,
      "loss": 0.6567,
      "step": 45595
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.542131464514124,
      "learning_rate": 3.4002219960735545e-06,
      "loss": 0.6235,
      "step": 45600
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.194349170981911,
      "learning_rate": 3.3990556337720935e-06,
      "loss": 0.6105,
      "step": 45605
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.404356559462518,
      "learning_rate": 3.3978893685311744e-06,
      "loss": 0.6318,
      "step": 45610
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.750007107627502,
      "learning_rate": 3.3967232004215017e-06,
      "loss": 0.6178,
      "step": 45615
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.69505036307874,
      "learning_rate": 3.39555712951378e-06,
      "loss": 0.6368,
      "step": 45620
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.117163220933203,
      "learning_rate": 3.394391155878704e-06,
      "loss": 0.6416,
      "step": 45625
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.112462350359201,
      "learning_rate": 3.3932252795869615e-06,
      "loss": 0.674,
      "step": 45630
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.132069219570511,
      "learning_rate": 3.392059500709238e-06,
      "loss": 0.6729,
      "step": 45635
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.203002296058865,
      "learning_rate": 3.39089381931621e-06,
      "loss": 0.6618,
      "step": 45640
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.579340664910821,
      "learning_rate": 3.389728235478551e-06,
      "loss": 0.6486,
      "step": 45645
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.260551560755507,
      "learning_rate": 3.388562749266926e-06,
      "loss": 0.6392,
      "step": 45650
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.766336994927626,
      "learning_rate": 3.387397360751994e-06,
      "loss": 0.6226,
      "step": 45655
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.0340342118630845,
      "learning_rate": 3.3862320700044104e-06,
      "loss": 0.6266,
      "step": 45660
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.095916294720115,
      "learning_rate": 3.3850668770948224e-06,
      "loss": 0.6492,
      "step": 45665
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.473676532956748,
      "learning_rate": 3.3839017820938735e-06,
      "loss": 0.6551,
      "step": 45670
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.131206676174092,
      "learning_rate": 3.3827367850722004e-06,
      "loss": 0.6401,
      "step": 45675
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.578780078412708,
      "learning_rate": 3.3815718861004304e-06,
      "loss": 0.673,
      "step": 45680
    },
    {
      "epoch": 1.93,
      "grad_norm": 8.040065874460522,
      "learning_rate": 3.3804070852491895e-06,
      "loss": 0.6498,
      "step": 45685
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.011626671836225,
      "learning_rate": 3.379242382589097e-06,
      "loss": 0.6073,
      "step": 45690
    },
    {
      "epoch": 1.93,
      "grad_norm": 11.913637730490201,
      "learning_rate": 3.378077778190765e-06,
      "loss": 0.6561,
      "step": 45695
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.848933547400499,
      "learning_rate": 3.376913272124799e-06,
      "loss": 0.6157,
      "step": 45700
    },
    {
      "epoch": 1.93,
      "grad_norm": 5.492708959668048,
      "learning_rate": 3.3757488644618e-06,
      "loss": 0.6128,
      "step": 45705
    },
    {
      "epoch": 1.93,
      "grad_norm": 7.8988249480678805,
      "learning_rate": 3.3745845552723633e-06,
      "loss": 0.6337,
      "step": 45710
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.567005021576079,
      "learning_rate": 3.3734203446270774e-06,
      "loss": 0.6387,
      "step": 45715
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.9329980037692085,
      "learning_rate": 3.372256232596526e-06,
      "loss": 0.6169,
      "step": 45720
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.269886785569335,
      "learning_rate": 3.3710922192512825e-06,
      "loss": 0.6546,
      "step": 45725
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.919019736793612,
      "learning_rate": 3.369928304661921e-06,
      "loss": 0.6254,
      "step": 45730
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.66442794497914,
      "learning_rate": 3.3687644888990033e-06,
      "loss": 0.638,
      "step": 45735
    },
    {
      "epoch": 1.94,
      "grad_norm": 15.493949347183731,
      "learning_rate": 3.367600772033091e-06,
      "loss": 0.6299,
      "step": 45740
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.354606585995278,
      "learning_rate": 3.366437154134736e-06,
      "loss": 0.6197,
      "step": 45745
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.921038050777262,
      "learning_rate": 3.365273635274483e-06,
      "loss": 0.6411,
      "step": 45750
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.9344316667966535,
      "learning_rate": 3.3641102155228756e-06,
      "loss": 0.6316,
      "step": 45755
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.4496752183477275,
      "learning_rate": 3.3629468949504456e-06,
      "loss": 0.6299,
      "step": 45760
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.235743501385427,
      "learning_rate": 3.3617836736277244e-06,
      "loss": 0.6454,
      "step": 45765
    },
    {
      "epoch": 1.94,
      "grad_norm": 16.034535703715164,
      "learning_rate": 3.3606205516252346e-06,
      "loss": 0.647,
      "step": 45770
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.108209852288871,
      "learning_rate": 3.359457529013489e-06,
      "loss": 0.6352,
      "step": 45775
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.967952029220819,
      "learning_rate": 3.3582946058630032e-06,
      "loss": 0.6236,
      "step": 45780
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.250529989577205,
      "learning_rate": 3.3571317822442783e-06,
      "loss": 0.6348,
      "step": 45785
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.32677807080038,
      "learning_rate": 3.3559690582278158e-06,
      "loss": 0.6225,
      "step": 45790
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.972028537489398,
      "learning_rate": 3.354806433884106e-06,
      "loss": 0.6627,
      "step": 45795
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.7083080350903685,
      "learning_rate": 3.353643909283635e-06,
      "loss": 0.6438,
      "step": 45800
    },
    {
      "epoch": 1.94,
      "grad_norm": 11.592423877122028,
      "learning_rate": 3.3524814844968855e-06,
      "loss": 0.6853,
      "step": 45805
    },
    {
      "epoch": 1.94,
      "grad_norm": 10.427975091157588,
      "learning_rate": 3.3513191595943297e-06,
      "loss": 0.6357,
      "step": 45810
    },
    {
      "epoch": 1.94,
      "grad_norm": 9.880817789950196,
      "learning_rate": 3.350156934646438e-06,
      "loss": 0.627,
      "step": 45815
    },
    {
      "epoch": 1.94,
      "grad_norm": 12.047327347134507,
      "learning_rate": 3.3489948097236712e-06,
      "loss": 0.6876,
      "step": 45820
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.44083308623586,
      "learning_rate": 3.347832784896485e-06,
      "loss": 0.6063,
      "step": 45825
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.801802607179371,
      "learning_rate": 3.3466708602353323e-06,
      "loss": 0.6282,
      "step": 45830
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.735548984809331,
      "learning_rate": 3.3455090358106536e-06,
      "loss": 0.5939,
      "step": 45835
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.962240227191571,
      "learning_rate": 3.3443473116928917e-06,
      "loss": 0.6186,
      "step": 45840
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.225505360361397,
      "learning_rate": 3.3431856879524715e-06,
      "loss": 0.6178,
      "step": 45845
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.397160821929835,
      "learning_rate": 3.342024164659825e-06,
      "loss": 0.6234,
      "step": 45850
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.032425510398778,
      "learning_rate": 3.340862741885368e-06,
      "loss": 0.6018,
      "step": 45855
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.670062510645949,
      "learning_rate": 3.339701419699517e-06,
      "loss": 0.6264,
      "step": 45860
    },
    {
      "epoch": 1.94,
      "grad_norm": 11.067726178184103,
      "learning_rate": 3.3385401981726785e-06,
      "loss": 0.6366,
      "step": 45865
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.28763534656678,
      "learning_rate": 3.3373790773752524e-06,
      "loss": 0.618,
      "step": 45870
    },
    {
      "epoch": 1.94,
      "grad_norm": 10.574772222727425,
      "learning_rate": 3.336218057377637e-06,
      "loss": 0.6456,
      "step": 45875
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.214404047278807,
      "learning_rate": 3.3350571382502184e-06,
      "loss": 0.6253,
      "step": 45880
    },
    {
      "epoch": 1.94,
      "grad_norm": 10.8430885459688,
      "learning_rate": 3.3338963200633824e-06,
      "loss": 0.646,
      "step": 45885
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.5060356945676965,
      "learning_rate": 3.332735602887505e-06,
      "loss": 0.6154,
      "step": 45890
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.5533715988187495,
      "learning_rate": 3.331574986792956e-06,
      "loss": 0.6329,
      "step": 45895
    },
    {
      "epoch": 1.94,
      "grad_norm": 6.515450463660592,
      "learning_rate": 3.3304144718501017e-06,
      "loss": 0.6429,
      "step": 45900
    },
    {
      "epoch": 1.94,
      "grad_norm": 8.21207190775567,
      "learning_rate": 3.3292540581292996e-06,
      "loss": 0.6332,
      "step": 45905
    },
    {
      "epoch": 1.94,
      "grad_norm": 5.196766114786937,
      "learning_rate": 3.3280937457009033e-06,
      "loss": 0.6482,
      "step": 45910
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.36533027224504,
      "learning_rate": 3.3269335346352584e-06,
      "loss": 0.6967,
      "step": 45915
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.236632562202655,
      "learning_rate": 3.325773425002704e-06,
      "loss": 0.6276,
      "step": 45920
    },
    {
      "epoch": 1.94,
      "grad_norm": 3.9829729003756187,
      "learning_rate": 3.3246134168735763e-06,
      "loss": 0.6111,
      "step": 45925
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.809042970021163,
      "learning_rate": 3.323453510318202e-06,
      "loss": 0.6435,
      "step": 45930
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.290893326992605,
      "learning_rate": 3.3222937054069022e-06,
      "loss": 0.6469,
      "step": 45935
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.044868727381685,
      "learning_rate": 3.321134002209994e-06,
      "loss": 0.6816,
      "step": 45940
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.363964059078036,
      "learning_rate": 3.319974400797785e-06,
      "loss": 0.6354,
      "step": 45945
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.494400399115878,
      "learning_rate": 3.3188149012405807e-06,
      "loss": 0.6524,
      "step": 45950
    },
    {
      "epoch": 1.94,
      "grad_norm": 4.215074240108514,
      "learning_rate": 3.3176555036086767e-06,
      "loss": 0.5911,
      "step": 45955
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.883991756835394,
      "learning_rate": 3.3164962079723627e-06,
      "loss": 0.6514,
      "step": 45960
    },
    {
      "epoch": 1.95,
      "grad_norm": 10.079120046684297,
      "learning_rate": 3.315337014401926e-06,
      "loss": 0.6155,
      "step": 45965
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.406521268089119,
      "learning_rate": 3.314177922967643e-06,
      "loss": 0.6617,
      "step": 45970
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.9655004200961494,
      "learning_rate": 3.3130189337397874e-06,
      "loss": 0.6425,
      "step": 45975
    },
    {
      "epoch": 1.95,
      "grad_norm": 10.389588515092377,
      "learning_rate": 3.3118600467886246e-06,
      "loss": 0.6239,
      "step": 45980
    },
    {
      "epoch": 1.95,
      "grad_norm": 21.67557341637365,
      "learning_rate": 3.3107012621844135e-06,
      "loss": 0.6373,
      "step": 45985
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.9094337528096395,
      "learning_rate": 3.3095425799974103e-06,
      "loss": 0.6548,
      "step": 45990
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.223563564680734,
      "learning_rate": 3.3083840002978595e-06,
      "loss": 0.6191,
      "step": 45995
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.430397517965408,
      "learning_rate": 3.307225523156008e-06,
      "loss": 0.642,
      "step": 46000
    },
    {
      "epoch": 1.95,
      "grad_norm": 7.396713176379317,
      "learning_rate": 3.306067148642083e-06,
      "loss": 0.641,
      "step": 46005
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.514384050287509,
      "learning_rate": 3.3049088768263182e-06,
      "loss": 0.65,
      "step": 46010
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.86124879340204,
      "learning_rate": 3.3037507077789345e-06,
      "loss": 0.6375,
      "step": 46015
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.790318871201607,
      "learning_rate": 3.302592641570149e-06,
      "loss": 0.6196,
      "step": 46020
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.821584714640911,
      "learning_rate": 3.3014346782701727e-06,
      "loss": 0.5791,
      "step": 46025
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.840158443681289,
      "learning_rate": 3.300276817949207e-06,
      "loss": 0.6014,
      "step": 46030
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.82757733318781,
      "learning_rate": 3.2991190606774527e-06,
      "loss": 0.6567,
      "step": 46035
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.672958695752343,
      "learning_rate": 3.2979614065250986e-06,
      "loss": 0.6507,
      "step": 46040
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.889145133827618,
      "learning_rate": 3.2968038555623317e-06,
      "loss": 0.6342,
      "step": 46045
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.467186230525887,
      "learning_rate": 3.29564640785933e-06,
      "loss": 0.6154,
      "step": 46050
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.500253827801541,
      "learning_rate": 3.2944890634862653e-06,
      "loss": 0.6428,
      "step": 46055
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.839399150012095,
      "learning_rate": 3.293331822513306e-06,
      "loss": 0.6356,
      "step": 46060
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.578976660232285,
      "learning_rate": 3.2921746850106095e-06,
      "loss": 0.6553,
      "step": 46065
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.862085021592242,
      "learning_rate": 3.291017651048333e-06,
      "loss": 0.6477,
      "step": 46070
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.458583766146514,
      "learning_rate": 3.2898607206966226e-06,
      "loss": 0.6231,
      "step": 46075
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.740639885305135,
      "learning_rate": 3.2887038940256176e-06,
      "loss": 0.6436,
      "step": 46080
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.760726312295329,
      "learning_rate": 3.2875471711054564e-06,
      "loss": 0.6388,
      "step": 46085
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.347443661246086,
      "learning_rate": 3.2863905520062643e-06,
      "loss": 0.6423,
      "step": 46090
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.4424496483677665,
      "learning_rate": 3.2852340367981673e-06,
      "loss": 0.6261,
      "step": 46095
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.888106935293909,
      "learning_rate": 3.284077625551279e-06,
      "loss": 0.6441,
      "step": 46100
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.888889296232943,
      "learning_rate": 3.2829213183357088e-06,
      "loss": 0.6348,
      "step": 46105
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.72029219062407,
      "learning_rate": 3.2817651152215625e-06,
      "loss": 0.6307,
      "step": 46110
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.326520750235513,
      "learning_rate": 3.2806090162789352e-06,
      "loss": 0.6721,
      "step": 46115
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.687512741651495,
      "learning_rate": 3.279453021577919e-06,
      "loss": 0.6616,
      "step": 46120
    },
    {
      "epoch": 1.95,
      "grad_norm": 8.069785976189747,
      "learning_rate": 3.2782971311885985e-06,
      "loss": 0.6561,
      "step": 46125
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.284538894788858,
      "learning_rate": 3.2771413451810506e-06,
      "loss": 0.5985,
      "step": 46130
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.755525355905471,
      "learning_rate": 3.275985663625349e-06,
      "loss": 0.5981,
      "step": 46135
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.408298168423517,
      "learning_rate": 3.2748300865915573e-06,
      "loss": 0.6329,
      "step": 46140
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.266761951064814,
      "learning_rate": 3.2736746141497365e-06,
      "loss": 0.5762,
      "step": 46145
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.114212000773811,
      "learning_rate": 3.2725192463699397e-06,
      "loss": 0.5831,
      "step": 46150
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.592012214619939,
      "learning_rate": 3.2713639833222105e-06,
      "loss": 0.6433,
      "step": 46155
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.065914483302372,
      "learning_rate": 3.270208825076594e-06,
      "loss": 0.648,
      "step": 46160
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.251480447530903,
      "learning_rate": 3.2690537717031194e-06,
      "loss": 0.6444,
      "step": 46165
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.252766212017174,
      "learning_rate": 3.267898823271817e-06,
      "loss": 0.6661,
      "step": 46170
    },
    {
      "epoch": 1.95,
      "grad_norm": 5.5036786748066415,
      "learning_rate": 3.2667439798527046e-06,
      "loss": 0.6909,
      "step": 46175
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.414793603928873,
      "learning_rate": 3.2655892415158007e-06,
      "loss": 0.6533,
      "step": 46180
    },
    {
      "epoch": 1.95,
      "grad_norm": 4.205093549194959,
      "learning_rate": 3.2644346083311113e-06,
      "loss": 0.6648,
      "step": 46185
    },
    {
      "epoch": 1.95,
      "grad_norm": 6.246429880134112,
      "learning_rate": 3.2632800803686405e-06,
      "loss": 0.6459,
      "step": 46190
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.201117143664794,
      "learning_rate": 3.2621256576983823e-06,
      "loss": 0.6116,
      "step": 46195
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.996890794660394,
      "learning_rate": 3.2609713403903255e-06,
      "loss": 0.6677,
      "step": 46200
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.136461825442796,
      "learning_rate": 3.259817128514455e-06,
      "loss": 0.6279,
      "step": 46205
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.54420595395502,
      "learning_rate": 3.2586630221407457e-06,
      "loss": 0.6056,
      "step": 46210
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.180766402213014,
      "learning_rate": 3.257509021339168e-06,
      "loss": 0.6007,
      "step": 46215
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.74045489944561,
      "learning_rate": 3.2563551261796865e-06,
      "loss": 0.6278,
      "step": 46220
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.837661691052112,
      "learning_rate": 3.255201336732256e-06,
      "loss": 0.6158,
      "step": 46225
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.538245831689887,
      "learning_rate": 3.25404765306683e-06,
      "loss": 0.6305,
      "step": 46230
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.759702231750561,
      "learning_rate": 3.2528940752533522e-06,
      "loss": 0.6457,
      "step": 46235
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.126159591537414,
      "learning_rate": 3.2517406033617595e-06,
      "loss": 0.6193,
      "step": 46240
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.2848869354476005,
      "learning_rate": 3.250587237461985e-06,
      "loss": 0.6235,
      "step": 46245
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.889217353836298,
      "learning_rate": 3.2494339776239525e-06,
      "loss": 0.6515,
      "step": 46250
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.5114275920757425,
      "learning_rate": 3.2482808239175826e-06,
      "loss": 0.6171,
      "step": 46255
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.991292830232938,
      "learning_rate": 3.247127776412786e-06,
      "loss": 0.6414,
      "step": 46260
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.977066022495548,
      "learning_rate": 3.2459748351794683e-06,
      "loss": 0.6235,
      "step": 46265
    },
    {
      "epoch": 1.96,
      "grad_norm": 10.262167364400819,
      "learning_rate": 3.2448220002875313e-06,
      "loss": 0.6562,
      "step": 46270
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.655723918808195,
      "learning_rate": 3.2436692718068654e-06,
      "loss": 0.6272,
      "step": 46275
    },
    {
      "epoch": 1.96,
      "grad_norm": 9.384576239066247,
      "learning_rate": 3.242516649807358e-06,
      "loss": 0.6388,
      "step": 46280
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.608291571213547,
      "learning_rate": 3.2413641343588897e-06,
      "loss": 0.6475,
      "step": 46285
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.7114517458120035,
      "learning_rate": 3.2402117255313327e-06,
      "loss": 0.636,
      "step": 46290
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.264695212115154,
      "learning_rate": 3.2390594233945568e-06,
      "loss": 0.6274,
      "step": 46295
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.172386351747783,
      "learning_rate": 3.2379072280184195e-06,
      "loss": 0.6317,
      "step": 46300
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.613526803223688,
      "learning_rate": 3.236755139472777e-06,
      "loss": 0.6202,
      "step": 46305
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.651683753453762,
      "learning_rate": 3.235603157827477e-06,
      "loss": 0.6415,
      "step": 46310
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.0517205975773445,
      "learning_rate": 3.23445128315236e-06,
      "loss": 0.6424,
      "step": 46315
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.649946827934103,
      "learning_rate": 3.233299515517262e-06,
      "loss": 0.6412,
      "step": 46320
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.261054805422806,
      "learning_rate": 3.2321478549920104e-06,
      "loss": 0.651,
      "step": 46325
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.7366286658296595,
      "learning_rate": 3.230996301646427e-06,
      "loss": 0.6163,
      "step": 46330
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.2007178344769285,
      "learning_rate": 3.2298448555503255e-06,
      "loss": 0.6251,
      "step": 46335
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.486332633224456,
      "learning_rate": 3.228693516773517e-06,
      "loss": 0.596,
      "step": 46340
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.055566125001305,
      "learning_rate": 3.2275422853858027e-06,
      "loss": 0.662,
      "step": 46345
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.657242884892401,
      "learning_rate": 3.226391161456979e-06,
      "loss": 0.601,
      "step": 46350
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.235320775098148,
      "learning_rate": 3.2252401450568354e-06,
      "loss": 0.6186,
      "step": 46355
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.699293184880735,
      "learning_rate": 3.224089236255153e-06,
      "loss": 0.644,
      "step": 46360
    },
    {
      "epoch": 1.96,
      "grad_norm": 7.835230626583173,
      "learning_rate": 3.2229384351217096e-06,
      "loss": 0.6231,
      "step": 46365
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.280775743768032,
      "learning_rate": 3.2217877417262745e-06,
      "loss": 0.6559,
      "step": 46370
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.055690437459158,
      "learning_rate": 3.220637156138611e-06,
      "loss": 0.6385,
      "step": 46375
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.225587105394276,
      "learning_rate": 3.219486678428475e-06,
      "loss": 0.6543,
      "step": 46380
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.991628963972974,
      "learning_rate": 3.2183363086656164e-06,
      "loss": 0.6144,
      "step": 46385
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.120547908226655,
      "learning_rate": 3.2171860469197802e-06,
      "loss": 0.6511,
      "step": 46390
    },
    {
      "epoch": 1.96,
      "grad_norm": 8.24200400520017,
      "learning_rate": 3.2160358932607018e-06,
      "loss": 0.6215,
      "step": 46395
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.511490936426109,
      "learning_rate": 3.2148858477581135e-06,
      "loss": 0.6366,
      "step": 46400
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.18077413080865,
      "learning_rate": 3.213735910481738e-06,
      "loss": 0.6431,
      "step": 46405
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.568808169848295,
      "learning_rate": 3.2125860815012912e-06,
      "loss": 0.6414,
      "step": 46410
    },
    {
      "epoch": 1.96,
      "grad_norm": 4.480480991063598,
      "learning_rate": 3.211436360886486e-06,
      "loss": 0.6051,
      "step": 46415
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.687428771859702,
      "learning_rate": 3.2102867487070254e-06,
      "loss": 0.6195,
      "step": 46420
    },
    {
      "epoch": 1.96,
      "grad_norm": 10.440458962487813,
      "learning_rate": 3.2091372450326085e-06,
      "loss": 0.625,
      "step": 46425
    },
    {
      "epoch": 1.96,
      "grad_norm": 5.138163854485248,
      "learning_rate": 3.2079878499329254e-06,
      "loss": 0.6189,
      "step": 46430
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.332813901169031,
      "learning_rate": 3.2068385634776594e-06,
      "loss": 0.6063,
      "step": 46435
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.111931666593924,
      "learning_rate": 3.20568938573649e-06,
      "loss": 0.6404,
      "step": 46440
    },
    {
      "epoch": 1.97,
      "grad_norm": 12.894188818141188,
      "learning_rate": 3.2045403167790866e-06,
      "loss": 0.6378,
      "step": 46445
    },
    {
      "epoch": 1.97,
      "grad_norm": 11.732185759794607,
      "learning_rate": 3.2033913566751175e-06,
      "loss": 0.6172,
      "step": 46450
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.723825270650447,
      "learning_rate": 3.202242505494237e-06,
      "loss": 0.6224,
      "step": 46455
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.039691356704054,
      "learning_rate": 3.2010937633060966e-06,
      "loss": 0.6371,
      "step": 46460
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.334629584991646,
      "learning_rate": 3.199945130180345e-06,
      "loss": 0.6262,
      "step": 46465
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.8948631683887385,
      "learning_rate": 3.198796606186616e-06,
      "loss": 0.6206,
      "step": 46470
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.265846080712185,
      "learning_rate": 3.197648191394544e-06,
      "loss": 0.6473,
      "step": 46475
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.16205564117438,
      "learning_rate": 3.1964998858737536e-06,
      "loss": 0.5934,
      "step": 46480
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.0936364925505995,
      "learning_rate": 3.195351689693863e-06,
      "loss": 0.6069,
      "step": 46485
    },
    {
      "epoch": 1.97,
      "grad_norm": 9.474963773750005,
      "learning_rate": 3.1942036029244827e-06,
      "loss": 0.6083,
      "step": 46490
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.922126300218031,
      "learning_rate": 3.193055625635219e-06,
      "loss": 0.6358,
      "step": 46495
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.261451480397849,
      "learning_rate": 3.1919077578956704e-06,
      "loss": 0.6325,
      "step": 46500
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.8411999635717295,
      "learning_rate": 3.190759999775428e-06,
      "loss": 0.5961,
      "step": 46505
    },
    {
      "epoch": 1.97,
      "grad_norm": 11.582305485153292,
      "learning_rate": 3.189612351344078e-06,
      "loss": 0.6252,
      "step": 46510
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.634585402936844,
      "learning_rate": 3.188464812671198e-06,
      "loss": 0.6465,
      "step": 46515
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.522395089026834,
      "learning_rate": 3.1873173838263603e-06,
      "loss": 0.6153,
      "step": 46520
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.254912837329435,
      "learning_rate": 3.186170064879131e-06,
      "loss": 0.62,
      "step": 46525
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.1235950846996134,
      "learning_rate": 3.185022855899067e-06,
      "loss": 0.6173,
      "step": 46530
    },
    {
      "epoch": 1.97,
      "grad_norm": 3.9879173494100844,
      "learning_rate": 3.183875756955721e-06,
      "loss": 0.6285,
      "step": 46535
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.057685357645161,
      "learning_rate": 3.182728768118639e-06,
      "loss": 0.6052,
      "step": 46540
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.438452022472454,
      "learning_rate": 3.181581889457358e-06,
      "loss": 0.6082,
      "step": 46545
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.211037911354311,
      "learning_rate": 3.1804351210414118e-06,
      "loss": 0.6132,
      "step": 46550
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.417088172029665,
      "learning_rate": 3.179288462940323e-06,
      "loss": 0.6437,
      "step": 46555
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.34058702002519,
      "learning_rate": 3.1781419152236135e-06,
      "loss": 0.6434,
      "step": 46560
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.1262098205843145,
      "learning_rate": 3.1769954779607927e-06,
      "loss": 0.5811,
      "step": 46565
    },
    {
      "epoch": 1.97,
      "grad_norm": 6.593190620281481,
      "learning_rate": 3.175849151221365e-06,
      "loss": 0.6192,
      "step": 46570
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.50319941357607,
      "learning_rate": 3.1747029350748316e-06,
      "loss": 0.6816,
      "step": 46575
    },
    {
      "epoch": 1.97,
      "grad_norm": 7.145322094430233,
      "learning_rate": 3.1735568295906805e-06,
      "loss": 0.6226,
      "step": 46580
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.430310278295654,
      "learning_rate": 3.172410834838401e-06,
      "loss": 0.6448,
      "step": 46585
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.479118272701115,
      "learning_rate": 3.1712649508874688e-06,
      "loss": 0.6426,
      "step": 46590
    },
    {
      "epoch": 1.97,
      "grad_norm": 5.027445150246773,
      "learning_rate": 3.170119177807355e-06,
      "loss": 0.6086,
      "step": 46595
    },
    {
      "epoch": 1.97,
      "grad_norm": 12.2505219023973,
      "learning_rate": 3.168973515667526e-06,
      "loss": 0.6211,
      "step": 46600
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.601870945224483,
      "learning_rate": 3.167827964537439e-06,
      "loss": 0.6435,
      "step": 46605
    },
    {
      "epoch": 1.97,
      "grad_norm": 8.981888257153422,
      "learning_rate": 3.166682524486546e-06,
      "loss": 0.6527,
      "step": 46610
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.741489578123719,
      "learning_rate": 3.1655371955842917e-06,
      "loss": 0.6023,
      "step": 46615
    },
    {
      "epoch": 1.97,
      "grad_norm": 16.638431468401823,
      "learning_rate": 3.1643919779001133e-06,
      "loss": 0.621,
      "step": 46620
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.896211754000867,
      "learning_rate": 3.1632468715034424e-06,
      "loss": 0.5929,
      "step": 46625
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.595735014594119,
      "learning_rate": 3.1621018764637025e-06,
      "loss": 0.6311,
      "step": 46630
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.558233563644817,
      "learning_rate": 3.1609569928503124e-06,
      "loss": 0.6301,
      "step": 46635
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.35762206616698,
      "learning_rate": 3.1598122207326844e-06,
      "loss": 0.6305,
      "step": 46640
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.125061682513744,
      "learning_rate": 3.1586675601802185e-06,
      "loss": 0.6185,
      "step": 46645
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.6341609068490595,
      "learning_rate": 3.157523011262318e-06,
      "loss": 0.5991,
      "step": 46650
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.888579897394806,
      "learning_rate": 3.1563785740483677e-06,
      "loss": 0.6515,
      "step": 46655
    },
    {
      "epoch": 1.97,
      "grad_norm": 11.847300158494878,
      "learning_rate": 3.1552342486077546e-06,
      "loss": 0.6052,
      "step": 46660
    },
    {
      "epoch": 1.97,
      "grad_norm": 4.763821469586007,
      "learning_rate": 3.154090035009854e-06,
      "loss": 0.6205,
      "step": 46665
    },
    {
      "epoch": 1.98,
      "grad_norm": 10.162321763741094,
      "learning_rate": 3.152945933324039e-06,
      "loss": 0.5883,
      "step": 46670
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.387937504583639,
      "learning_rate": 3.151801943619669e-06,
      "loss": 0.6429,
      "step": 46675
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.684205240756767,
      "learning_rate": 3.1506580659661046e-06,
      "loss": 0.6434,
      "step": 46680
    },
    {
      "epoch": 1.98,
      "grad_norm": 12.421610626633672,
      "learning_rate": 3.1495143004326944e-06,
      "loss": 0.6181,
      "step": 46685
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.429681014119205,
      "learning_rate": 3.148370647088781e-06,
      "loss": 0.616,
      "step": 46690
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.777285286288175,
      "learning_rate": 3.147227106003701e-06,
      "loss": 0.6412,
      "step": 46695
    },
    {
      "epoch": 1.98,
      "grad_norm": 12.826352553655871,
      "learning_rate": 3.146083677246784e-06,
      "loss": 0.6221,
      "step": 46700
    },
    {
      "epoch": 1.98,
      "grad_norm": 11.31903637921935,
      "learning_rate": 3.144940360887353e-06,
      "loss": 0.5993,
      "step": 46705
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.270590648377889,
      "learning_rate": 3.143797156994724e-06,
      "loss": 0.6465,
      "step": 46710
    },
    {
      "epoch": 1.98,
      "grad_norm": 9.339541665265655,
      "learning_rate": 3.1426540656382047e-06,
      "loss": 0.6034,
      "step": 46715
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.945666816275763,
      "learning_rate": 3.1415110868870994e-06,
      "loss": 0.6473,
      "step": 46720
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.212152578476717,
      "learning_rate": 3.1403682208107015e-06,
      "loss": 0.5978,
      "step": 46725
    },
    {
      "epoch": 1.98,
      "grad_norm": 9.036458940961808,
      "learning_rate": 3.139225467478302e-06,
      "loss": 0.654,
      "step": 46730
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.971596171819852,
      "learning_rate": 3.138082826959181e-06,
      "loss": 0.6805,
      "step": 46735
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.925432467243013,
      "learning_rate": 3.136940299322613e-06,
      "loss": 0.5843,
      "step": 46740
    },
    {
      "epoch": 1.98,
      "grad_norm": 10.761712516313626,
      "learning_rate": 3.135797884637867e-06,
      "loss": 0.637,
      "step": 46745
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.491754094052424,
      "learning_rate": 3.134655582974204e-06,
      "loss": 0.6178,
      "step": 46750
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.868578846769974,
      "learning_rate": 3.133513394400879e-06,
      "loss": 0.6232,
      "step": 46755
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.225187227322583,
      "learning_rate": 3.1323713189871396e-06,
      "loss": 0.6425,
      "step": 46760
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.440747971591411,
      "learning_rate": 3.131229356802224e-06,
      "loss": 0.6294,
      "step": 46765
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.634832900460518,
      "learning_rate": 3.130087507915369e-06,
      "loss": 0.6102,
      "step": 46770
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.800312612927686,
      "learning_rate": 3.1289457723958005e-06,
      "loss": 0.6147,
      "step": 46775
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.16424808361495,
      "learning_rate": 3.1278041503127376e-06,
      "loss": 0.6233,
      "step": 46780
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.263129373261356,
      "learning_rate": 3.1266626417353957e-06,
      "loss": 0.6241,
      "step": 46785
    },
    {
      "epoch": 1.98,
      "grad_norm": 12.457306859331743,
      "learning_rate": 3.125521246732978e-06,
      "loss": 0.6176,
      "step": 46790
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.4080910944801275,
      "learning_rate": 3.1243799653746877e-06,
      "loss": 0.6343,
      "step": 46795
    },
    {
      "epoch": 1.98,
      "grad_norm": 8.777589661068728,
      "learning_rate": 3.1232387977297143e-06,
      "loss": 0.6109,
      "step": 46800
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.32213921654426,
      "learning_rate": 3.1220977438672433e-06,
      "loss": 0.6375,
      "step": 46805
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.580621833082404,
      "learning_rate": 3.1209568038564574e-06,
      "loss": 0.6875,
      "step": 46810
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.604946762870719,
      "learning_rate": 3.1198159777665237e-06,
      "loss": 0.6364,
      "step": 46815
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.283053467752125,
      "learning_rate": 3.1186752656666097e-06,
      "loss": 0.6329,
      "step": 46820
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.094788332827578,
      "learning_rate": 3.1175346676258712e-06,
      "loss": 0.6542,
      "step": 46825
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.40663322536381,
      "learning_rate": 3.116394183713462e-06,
      "loss": 0.6215,
      "step": 46830
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.025858793450196,
      "learning_rate": 3.1152538139985245e-06,
      "loss": 0.6319,
      "step": 46835
    },
    {
      "epoch": 1.98,
      "grad_norm": 9.977742969996442,
      "learning_rate": 3.114113558550197e-06,
      "loss": 0.6087,
      "step": 46840
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.727026109270043,
      "learning_rate": 3.112973417437609e-06,
      "loss": 0.5802,
      "step": 46845
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.11985387639585,
      "learning_rate": 3.1118333907298843e-06,
      "loss": 0.6445,
      "step": 46850
    },
    {
      "epoch": 1.98,
      "grad_norm": 13.697498840480531,
      "learning_rate": 3.1106934784961394e-06,
      "loss": 0.599,
      "step": 46855
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.421122232118819,
      "learning_rate": 3.109553680805484e-06,
      "loss": 0.6134,
      "step": 46860
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.2473322106608675,
      "learning_rate": 3.1084139977270207e-06,
      "loss": 0.6099,
      "step": 46865
    },
    {
      "epoch": 1.98,
      "grad_norm": 6.245696152850468,
      "learning_rate": 3.1072744293298446e-06,
      "loss": 0.6463,
      "step": 46870
    },
    {
      "epoch": 1.98,
      "grad_norm": 12.45783626369971,
      "learning_rate": 3.1061349756830437e-06,
      "loss": 0.597,
      "step": 46875
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.461279453089836,
      "learning_rate": 3.104995636855702e-06,
      "loss": 0.6219,
      "step": 46880
    },
    {
      "epoch": 1.98,
      "grad_norm": 4.676007299294277,
      "learning_rate": 3.103856412916892e-06,
      "loss": 0.6447,
      "step": 46885
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.2950696905504415,
      "learning_rate": 3.1027173039356827e-06,
      "loss": 0.6514,
      "step": 46890
    },
    {
      "epoch": 1.98,
      "grad_norm": 7.398479795081675,
      "learning_rate": 3.1015783099811357e-06,
      "loss": 0.6138,
      "step": 46895
    },
    {
      "epoch": 1.98,
      "grad_norm": 5.044479169464052,
      "learning_rate": 3.100439431122302e-06,
      "loss": 0.6054,
      "step": 46900
    },
    {
      "epoch": 1.99,
      "grad_norm": 11.804052362360174,
      "learning_rate": 3.0993006674282316e-06,
      "loss": 0.6449,
      "step": 46905
    },
    {
      "epoch": 1.99,
      "grad_norm": 10.975069219231203,
      "learning_rate": 3.0981620189679616e-06,
      "loss": 0.6155,
      "step": 46910
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.229553714779008,
      "learning_rate": 3.0970234858105276e-06,
      "loss": 0.6426,
      "step": 46915
    },
    {
      "epoch": 1.99,
      "grad_norm": 12.128456260333069,
      "learning_rate": 3.0958850680249548e-06,
      "loss": 0.6324,
      "step": 46920
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.458938698591031,
      "learning_rate": 3.0947467656802597e-06,
      "loss": 0.6175,
      "step": 46925
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.15558306401921,
      "learning_rate": 3.0936085788454574e-06,
      "loss": 0.6006,
      "step": 46930
    },
    {
      "epoch": 1.99,
      "grad_norm": 9.598198601089736,
      "learning_rate": 3.0924705075895513e-06,
      "loss": 0.6499,
      "step": 46935
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.767678372819063,
      "learning_rate": 3.0913325519815396e-06,
      "loss": 0.6308,
      "step": 46940
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.900721759057035,
      "learning_rate": 3.090194712090413e-06,
      "loss": 0.6418,
      "step": 46945
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.82450697601192,
      "learning_rate": 3.089056987985155e-06,
      "loss": 0.6155,
      "step": 46950
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.728015496275644,
      "learning_rate": 3.0879193797347433e-06,
      "loss": 0.6167,
      "step": 46955
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.742949349784359,
      "learning_rate": 3.086781887408147e-06,
      "loss": 0.6079,
      "step": 46960
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.5879537730708915,
      "learning_rate": 3.0856445110743305e-06,
      "loss": 0.6121,
      "step": 46965
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.3228519702352,
      "learning_rate": 3.0845072508022487e-06,
      "loss": 0.6126,
      "step": 46970
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.6610079079695,
      "learning_rate": 3.083370106660849e-06,
      "loss": 0.6267,
      "step": 46975
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.529583406288785,
      "learning_rate": 3.0822330787190736e-06,
      "loss": 0.6532,
      "step": 46980
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.686243257580459,
      "learning_rate": 3.081096167045858e-06,
      "loss": 0.618,
      "step": 46985
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.865513355628833,
      "learning_rate": 3.0799593717101307e-06,
      "loss": 0.6138,
      "step": 46990
    },
    {
      "epoch": 1.99,
      "grad_norm": 10.96422710705633,
      "learning_rate": 3.078822692780809e-06,
      "loss": 0.611,
      "step": 46995
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.794855510431477,
      "learning_rate": 3.0776861303268104e-06,
      "loss": 0.6108,
      "step": 47000
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.947990803639829,
      "learning_rate": 3.0765496844170374e-06,
      "loss": 0.6406,
      "step": 47005
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.211926810267644,
      "learning_rate": 3.0754133551203923e-06,
      "loss": 0.649,
      "step": 47010
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.8510144909303285,
      "learning_rate": 3.074277142505767e-06,
      "loss": 0.6171,
      "step": 47015
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.4852303674809,
      "learning_rate": 3.073141046642045e-06,
      "loss": 0.6159,
      "step": 47020
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.810037219446173,
      "learning_rate": 3.0720050675981063e-06,
      "loss": 0.6219,
      "step": 47025
    },
    {
      "epoch": 1.99,
      "grad_norm": 12.041243225205607,
      "learning_rate": 3.07086920544282e-06,
      "loss": 0.6427,
      "step": 47030
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.929232061116423,
      "learning_rate": 3.069733460245053e-06,
      "loss": 0.6298,
      "step": 47035
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.549853995408155,
      "learning_rate": 3.0685978320736604e-06,
      "loss": 0.6161,
      "step": 47040
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.801933256377474,
      "learning_rate": 3.067462320997491e-06,
      "loss": 0.6165,
      "step": 47045
    },
    {
      "epoch": 1.99,
      "grad_norm": 11.999573797118748,
      "learning_rate": 3.066326927085389e-06,
      "loss": 0.6374,
      "step": 47050
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.627220885243319,
      "learning_rate": 3.0651916504061895e-06,
      "loss": 0.6213,
      "step": 47055
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.775374496663006,
      "learning_rate": 3.064056491028722e-06,
      "loss": 0.6486,
      "step": 47060
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.979715372933483,
      "learning_rate": 3.062921449021806e-06,
      "loss": 0.6071,
      "step": 47065
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.297321311676706,
      "learning_rate": 3.0617865244542565e-06,
      "loss": 0.6069,
      "step": 47070
    },
    {
      "epoch": 1.99,
      "grad_norm": 11.398372254633383,
      "learning_rate": 3.0606517173948824e-06,
      "loss": 0.5869,
      "step": 47075
    },
    {
      "epoch": 1.99,
      "grad_norm": 8.213627464121108,
      "learning_rate": 3.0595170279124808e-06,
      "loss": 0.5774,
      "step": 47080
    },
    {
      "epoch": 1.99,
      "grad_norm": 7.847612780284359,
      "learning_rate": 3.0583824560758456e-06,
      "loss": 0.6361,
      "step": 47085
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.230833664308463,
      "learning_rate": 3.0572480019537643e-06,
      "loss": 0.6126,
      "step": 47090
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.87824339017857,
      "learning_rate": 3.0561136656150126e-06,
      "loss": 0.6206,
      "step": 47095
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.034740510501605,
      "learning_rate": 3.0549794471283645e-06,
      "loss": 0.6209,
      "step": 47100
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.2982236032587195,
      "learning_rate": 3.0538453465625826e-06,
      "loss": 0.6171,
      "step": 47105
    },
    {
      "epoch": 1.99,
      "grad_norm": 6.429817191328955,
      "learning_rate": 3.0527113639864248e-06,
      "loss": 0.6423,
      "step": 47110
    },
    {
      "epoch": 1.99,
      "grad_norm": 5.750921463542665,
      "learning_rate": 3.051577499468642e-06,
      "loss": 0.6386,
      "step": 47115
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.561544068931166,
      "learning_rate": 3.0504437530779747e-06,
      "loss": 0.6201,
      "step": 47120
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.249087963466114,
      "learning_rate": 3.0493101248831607e-06,
      "loss": 0.6402,
      "step": 47125
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.099603054673989,
      "learning_rate": 3.0481766149529286e-06,
      "loss": 0.6406,
      "step": 47130
    },
    {
      "epoch": 1.99,
      "grad_norm": 4.376779125790554,
      "learning_rate": 3.047043223356e-06,
      "loss": 0.6189,
      "step": 47135
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.488704818790552,
      "learning_rate": 3.0459099501610856e-06,
      "loss": 0.5909,
      "step": 47140
    },
    {
      "epoch": 2.0,
      "grad_norm": 10.966580928447296,
      "learning_rate": 3.044776795436897e-06,
      "loss": 0.5981,
      "step": 47145
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.322300020749453,
      "learning_rate": 3.0436437592521308e-06,
      "loss": 0.5779,
      "step": 47150
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.84558432774676,
      "learning_rate": 3.04251084167548e-06,
      "loss": 0.5873,
      "step": 47155
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.497170623213144,
      "learning_rate": 3.0413780427756323e-06,
      "loss": 0.6044,
      "step": 47160
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.704093094519641,
      "learning_rate": 3.0402453626212637e-06,
      "loss": 0.6012,
      "step": 47165
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.978748573025106,
      "learning_rate": 3.039112801281047e-06,
      "loss": 0.5737,
      "step": 47170
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.53526559971311,
      "learning_rate": 3.0379803588236443e-06,
      "loss": 0.6037,
      "step": 47175
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.954032451193737,
      "learning_rate": 3.0368480353177133e-06,
      "loss": 0.5919,
      "step": 47180
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.161137178889893,
      "learning_rate": 3.035715830831904e-06,
      "loss": 0.6486,
      "step": 47185
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.82406453908264,
      "learning_rate": 3.034583745434857e-06,
      "loss": 0.6545,
      "step": 47190
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.232416342902947,
      "learning_rate": 3.033451779195209e-06,
      "loss": 0.6171,
      "step": 47195
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.53915098172456,
      "learning_rate": 3.0323199321815866e-06,
      "loss": 0.6451,
      "step": 47200
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.242573088464062,
      "learning_rate": 3.0311882044626094e-06,
      "loss": 0.6336,
      "step": 47205
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.681813637211751,
      "learning_rate": 3.030056596106894e-06,
      "loss": 0.6199,
      "step": 47210
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.287132884725201,
      "learning_rate": 3.0289251071830434e-06,
      "loss": 0.6114,
      "step": 47215
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.937589762642504,
      "learning_rate": 3.027793737759658e-06,
      "loss": 0.6609,
      "step": 47220
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.1908081433656745,
      "learning_rate": 3.0266624879053297e-06,
      "loss": 0.6242,
      "step": 47225
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.440861334593197,
      "learning_rate": 3.02553135768864e-06,
      "loss": 0.6008,
      "step": 47230
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.804001209664816,
      "learning_rate": 3.0244003471781703e-06,
      "loss": 0.651,
      "step": 47235
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.856443484067078,
      "learning_rate": 3.023269456442487e-06,
      "loss": 0.5801,
      "step": 47240
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.520627380087969,
      "learning_rate": 3.0221386855501545e-06,
      "loss": 0.6238,
      "step": 47245
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.76037385146742,
      "learning_rate": 3.021008034569727e-06,
      "loss": 0.6219,
      "step": 47250
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.448442979736266,
      "learning_rate": 3.0198775035697532e-06,
      "loss": 0.6403,
      "step": 47255
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.8787976503372192,
      "eval_runtime": 36.9564,
      "eval_samples_per_second": 31.632,
      "eval_steps_per_second": 3.978,
      "step": 47257
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.406558505351557,
      "learning_rate": 3.0187470926187747e-06,
      "loss": 0.4693,
      "step": 47260
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.7125083707601454,
      "learning_rate": 3.017616801785322e-06,
      "loss": 0.3843,
      "step": 47265
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.479499005769949,
      "learning_rate": 3.0164866311379246e-06,
      "loss": 0.428,
      "step": 47270
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.055686527703872,
      "learning_rate": 3.0153565807451e-06,
      "loss": 0.3967,
      "step": 47275
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.282920902201872,
      "learning_rate": 3.01422665067536e-06,
      "loss": 0.4125,
      "step": 47280
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.34626229991146,
      "learning_rate": 3.013096840997209e-06,
      "loss": 0.3679,
      "step": 47285
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.8383342170736694,
      "learning_rate": 3.0119671517791425e-06,
      "loss": 0.4228,
      "step": 47290
    },
    {
      "epoch": 2.0,
      "grad_norm": 8.587227508857335,
      "learning_rate": 3.010837583089656e-06,
      "loss": 0.4017,
      "step": 47295
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.801434011877231,
      "learning_rate": 3.0097081349972236e-06,
      "loss": 0.3895,
      "step": 47300
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.063233546911788,
      "learning_rate": 3.008578807570326e-06,
      "loss": 0.4089,
      "step": 47305
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.4089489198017775,
      "learning_rate": 3.007449600877428e-06,
      "loss": 0.4324,
      "step": 47310
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.267716439156547,
      "learning_rate": 3.0063205149869924e-06,
      "loss": 0.4019,
      "step": 47315
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.694874467645835,
      "learning_rate": 3.005191549967471e-06,
      "loss": 0.3914,
      "step": 47320
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.624501837834187,
      "learning_rate": 3.0040627058873097e-06,
      "loss": 0.4038,
      "step": 47325
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.608713061113824,
      "learning_rate": 3.0029339828149477e-06,
      "loss": 0.3948,
      "step": 47330
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.1808167089937465,
      "learning_rate": 3.001805380818815e-06,
      "loss": 0.4122,
      "step": 47335
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.349164693119417,
      "learning_rate": 3.0006768999673375e-06,
      "loss": 0.3734,
      "step": 47340
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.148108784119843,
      "learning_rate": 2.9995485403289306e-06,
      "loss": 0.3989,
      "step": 47345
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.757899254225337,
      "learning_rate": 2.9984203019720014e-06,
      "loss": 0.3911,
      "step": 47350
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.967542362222487,
      "learning_rate": 2.997292184964955e-06,
      "loss": 0.3816,
      "step": 47355
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.9765544164382205,
      "learning_rate": 2.996164189376183e-06,
      "loss": 0.4181,
      "step": 47360
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.575063373682107,
      "learning_rate": 2.9950363152740747e-06,
      "loss": 0.4045,
      "step": 47365
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.8164892321274517,
      "learning_rate": 2.993908562727008e-06,
      "loss": 0.3978,
      "step": 47370
    },
    {
      "epoch": 2.0,
      "grad_norm": 5.123280446546897,
      "learning_rate": 2.9927809318033565e-06,
      "loss": 0.4334,
      "step": 47375
    },
    {
      "epoch": 2.01,
      "grad_norm": 14.684753100301338,
      "learning_rate": 2.9916534225714846e-06,
      "loss": 0.3957,
      "step": 47380
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.653272048105198,
      "learning_rate": 2.9905260350997505e-06,
      "loss": 0.3888,
      "step": 47385
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.271621887287846,
      "learning_rate": 2.9893987694565023e-06,
      "loss": 0.3797,
      "step": 47390
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.443935735667931,
      "learning_rate": 2.9882716257100856e-06,
      "loss": 0.4037,
      "step": 47395
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.038228259834561,
      "learning_rate": 2.9871446039288337e-06,
      "loss": 0.4026,
      "step": 47400
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.022961434538157,
      "learning_rate": 2.9860177041810763e-06,
      "loss": 0.4348,
      "step": 47405
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.382533210689673,
      "learning_rate": 2.984890926535133e-06,
      "loss": 0.3571,
      "step": 47410
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.715919155041624,
      "learning_rate": 2.9837642710593162e-06,
      "loss": 0.4028,
      "step": 47415
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.976456413190658,
      "learning_rate": 2.982637737821933e-06,
      "loss": 0.4097,
      "step": 47420
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.544803547755578,
      "learning_rate": 2.981511326891281e-06,
      "loss": 0.4091,
      "step": 47425
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.1863930014457384,
      "learning_rate": 2.9803850383356525e-06,
      "loss": 0.3616,
      "step": 47430
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.813989691485184,
      "learning_rate": 2.9792588722233307e-06,
      "loss": 0.4137,
      "step": 47435
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.411035134292161,
      "learning_rate": 2.9781328286225895e-06,
      "loss": 0.4104,
      "step": 47440
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.286110766397366,
      "learning_rate": 2.9770069076017007e-06,
      "loss": 0.3881,
      "step": 47445
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.1930344244648055,
      "learning_rate": 2.975881109228923e-06,
      "loss": 0.392,
      "step": 47450
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.048276347620639,
      "learning_rate": 2.974755433572514e-06,
      "loss": 0.3951,
      "step": 47455
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.498821357179391,
      "learning_rate": 2.9736298807007145e-06,
      "loss": 0.3726,
      "step": 47460
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.4904755563052365,
      "learning_rate": 2.9725044506817685e-06,
      "loss": 0.4013,
      "step": 47465
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.966527817370967,
      "learning_rate": 2.9713791435839033e-06,
      "loss": 0.413,
      "step": 47470
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.941165482959042,
      "learning_rate": 2.970253959475347e-06,
      "loss": 0.396,
      "step": 47475
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.236806687808002,
      "learning_rate": 2.969128898424314e-06,
      "loss": 0.3768,
      "step": 47480
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.643563266388056,
      "learning_rate": 2.968003960499013e-06,
      "loss": 0.4003,
      "step": 47485
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.31465131212337,
      "learning_rate": 2.9668791457676473e-06,
      "loss": 0.4034,
      "step": 47490
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.784586985494391,
      "learning_rate": 2.9657544542984086e-06,
      "loss": 0.3918,
      "step": 47495
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.647443403147733,
      "learning_rate": 2.9646298861594867e-06,
      "loss": 0.3783,
      "step": 47500
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.830395460228566,
      "learning_rate": 2.96350544141906e-06,
      "loss": 0.4024,
      "step": 47505
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.073805944001501,
      "learning_rate": 2.962381120145298e-06,
      "loss": 0.39,
      "step": 47510
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.097842659651942,
      "learning_rate": 2.961256922406368e-06,
      "loss": 0.428,
      "step": 47515
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.2687924730478075,
      "learning_rate": 2.9601328482704237e-06,
      "loss": 0.3887,
      "step": 47520
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.8454766960321742,
      "learning_rate": 2.9590088978056185e-06,
      "loss": 0.3798,
      "step": 47525
    },
    {
      "epoch": 2.01,
      "grad_norm": 7.77419988240331,
      "learning_rate": 2.957885071080091e-06,
      "loss": 0.3894,
      "step": 47530
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.34451387447682,
      "learning_rate": 2.9567613681619754e-06,
      "loss": 0.3796,
      "step": 47535
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.5956675373415585,
      "learning_rate": 2.955637789119401e-06,
      "loss": 0.3945,
      "step": 47540
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.617278596110945,
      "learning_rate": 2.954514334020484e-06,
      "loss": 0.3968,
      "step": 47545
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.067198771902128,
      "learning_rate": 2.953391002933339e-06,
      "loss": 0.3977,
      "step": 47550
    },
    {
      "epoch": 2.01,
      "grad_norm": 8.695716488099286,
      "learning_rate": 2.9522677959260694e-06,
      "loss": 0.4003,
      "step": 47555
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.982660863579801,
      "learning_rate": 2.9511447130667702e-06,
      "loss": 0.3887,
      "step": 47560
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.674794757926067,
      "learning_rate": 2.9500217544235326e-06,
      "loss": 0.3941,
      "step": 47565
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.66708020258648,
      "learning_rate": 2.948898920064437e-06,
      "loss": 0.3925,
      "step": 47570
    },
    {
      "epoch": 2.01,
      "grad_norm": 7.6983697439890335,
      "learning_rate": 2.947776210057559e-06,
      "loss": 0.3829,
      "step": 47575
    },
    {
      "epoch": 2.01,
      "grad_norm": 3.9040654156844323,
      "learning_rate": 2.9466536244709647e-06,
      "loss": 0.4048,
      "step": 47580
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.6199754344666895,
      "learning_rate": 2.945531163372711e-06,
      "loss": 0.3724,
      "step": 47585
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.420611029965048,
      "learning_rate": 2.9444088268308535e-06,
      "loss": 0.3792,
      "step": 47590
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.570304254007392,
      "learning_rate": 2.943286614913432e-06,
      "loss": 0.3974,
      "step": 47595
    },
    {
      "epoch": 2.01,
      "grad_norm": 5.712081437431294,
      "learning_rate": 2.942164527688486e-06,
      "loss": 0.3844,
      "step": 47600
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.715592466413024,
      "learning_rate": 2.941042565224043e-06,
      "loss": 0.4118,
      "step": 47605
    },
    {
      "epoch": 2.01,
      "grad_norm": 6.168183477980882,
      "learning_rate": 2.9399207275881237e-06,
      "loss": 0.3862,
      "step": 47610
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.071783064772214,
      "learning_rate": 2.9387990148487433e-06,
      "loss": 0.41,
      "step": 47615
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.90123964207469,
      "learning_rate": 2.9376774270739085e-06,
      "loss": 0.3961,
      "step": 47620
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.348298075396367,
      "learning_rate": 2.9365559643316154e-06,
      "loss": 0.3712,
      "step": 47625
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.296818125288639,
      "learning_rate": 2.935434626689855e-06,
      "loss": 0.4092,
      "step": 47630
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.977071053657004,
      "learning_rate": 2.9343134142166123e-06,
      "loss": 0.3871,
      "step": 47635
    },
    {
      "epoch": 2.02,
      "grad_norm": 8.71759936216889,
      "learning_rate": 2.933192326979862e-06,
      "loss": 0.3992,
      "step": 47640
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.842694774223993,
      "learning_rate": 2.932071365047574e-06,
      "loss": 0.3774,
      "step": 47645
    },
    {
      "epoch": 2.02,
      "grad_norm": 5.259355052271828,
      "learning_rate": 2.930950528487708e-06,
      "loss": 0.3621,
      "step": 47650
    },
    {
      "epoch": 2.02,
      "grad_norm": 5.18225565917172,
      "learning_rate": 2.929829817368216e-06,
      "loss": 0.3843,
      "step": 47655
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.925843051594825,
      "learning_rate": 2.928709231757045e-06,
      "loss": 0.3704,
      "step": 47660
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.997059658212725,
      "learning_rate": 2.9275887717221318e-06,
      "loss": 0.3616,
      "step": 47665
    },
    {
      "epoch": 2.02,
      "grad_norm": 6.273733940560973,
      "learning_rate": 2.9264684373314067e-06,
      "loss": 0.3791,
      "step": 47670
    },
    {
      "epoch": 2.02,
      "grad_norm": 10.010193607783602,
      "learning_rate": 2.925348228652793e-06,
      "loss": 0.3829,
      "step": 47675
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.502786673034213,
      "learning_rate": 2.924228145754204e-06,
      "loss": 0.3866,
      "step": 47680
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.064099294271605,
      "learning_rate": 2.9231081887035494e-06,
      "loss": 0.4221,
      "step": 47685
    },
    {
      "epoch": 2.02,
      "grad_norm": 6.656586130220156,
      "learning_rate": 2.921988357568727e-06,
      "loss": 0.3767,
      "step": 47690
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.545874618160335,
      "learning_rate": 2.920868652417629e-06,
      "loss": 0.3984,
      "step": 47695
    },
    {
      "epoch": 2.02,
      "grad_norm": 7.857123149454434,
      "learning_rate": 2.919749073318141e-06,
      "loss": 0.4016,
      "step": 47700
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.342110253228481,
      "learning_rate": 2.91862962033814e-06,
      "loss": 0.397,
      "step": 47705
    },
    {
      "epoch": 2.02,
      "grad_norm": 8.202751623725467,
      "learning_rate": 2.9175102935454926e-06,
      "loss": 0.4101,
      "step": 47710
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.432775513127382,
      "learning_rate": 2.916391093008062e-06,
      "loss": 0.386,
      "step": 47715
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.7077636641253324,
      "learning_rate": 2.9152720187937035e-06,
      "loss": 0.3842,
      "step": 47720
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.9534942903313044,
      "learning_rate": 2.9141530709702604e-06,
      "loss": 0.4235,
      "step": 47725
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.2377277916748,
      "learning_rate": 2.9130342496055717e-06,
      "loss": 0.4148,
      "step": 47730
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.215758128269374,
      "learning_rate": 2.9119155547674698e-06,
      "loss": 0.39,
      "step": 47735
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.0780200293333735,
      "learning_rate": 2.9107969865237794e-06,
      "loss": 0.4015,
      "step": 47740
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.17709725900854,
      "learning_rate": 2.9096785449423114e-06,
      "loss": 0.3853,
      "step": 47745
    },
    {
      "epoch": 2.02,
      "grad_norm": 6.800077962782613,
      "learning_rate": 2.9085602300908756e-06,
      "loss": 0.42,
      "step": 47750
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.8035581374777836,
      "learning_rate": 2.907442042037275e-06,
      "loss": 0.3845,
      "step": 47755
    },
    {
      "epoch": 2.02,
      "grad_norm": 5.46898508667038,
      "learning_rate": 2.9063239808492973e-06,
      "loss": 0.4063,
      "step": 47760
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.927107264142804,
      "learning_rate": 2.9052060465947296e-06,
      "loss": 0.3929,
      "step": 47765
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.466200293620653,
      "learning_rate": 2.904088239341351e-06,
      "loss": 0.3707,
      "step": 47770
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.110460981541154,
      "learning_rate": 2.902970559156927e-06,
      "loss": 0.42,
      "step": 47775
    },
    {
      "epoch": 2.02,
      "grad_norm": 5.286904665721644,
      "learning_rate": 2.9018530061092225e-06,
      "loss": 0.3996,
      "step": 47780
    },
    {
      "epoch": 2.02,
      "grad_norm": 5.816565170761535,
      "learning_rate": 2.900735580265988e-06,
      "loss": 0.4084,
      "step": 47785
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.30919406035898,
      "learning_rate": 2.8996182816949735e-06,
      "loss": 0.4203,
      "step": 47790
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.464092131797559,
      "learning_rate": 2.8985011104639143e-06,
      "loss": 0.3971,
      "step": 47795
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.8381139630388628,
      "learning_rate": 2.8973840666405424e-06,
      "loss": 0.3726,
      "step": 47800
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.819212224980174,
      "learning_rate": 2.896267150292583e-06,
      "loss": 0.4025,
      "step": 47805
    },
    {
      "epoch": 2.02,
      "grad_norm": 9.015863328472255,
      "learning_rate": 2.8951503614877475e-06,
      "loss": 0.4059,
      "step": 47810
    },
    {
      "epoch": 2.02,
      "grad_norm": 24.369607105062965,
      "learning_rate": 2.894033700293746e-06,
      "loss": 0.4389,
      "step": 47815
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.289849586678744,
      "learning_rate": 2.892917166778277e-06,
      "loss": 0.3554,
      "step": 47820
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.31593109655073,
      "learning_rate": 2.891800761009036e-06,
      "loss": 0.3947,
      "step": 47825
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.640556423552882,
      "learning_rate": 2.890684483053703e-06,
      "loss": 0.3721,
      "step": 47830
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.729831882273159,
      "learning_rate": 2.8895683329799563e-06,
      "loss": 0.3697,
      "step": 47835
    },
    {
      "epoch": 2.02,
      "grad_norm": 4.028984164154781,
      "learning_rate": 2.8884523108554672e-06,
      "loss": 0.4006,
      "step": 47840
    },
    {
      "epoch": 2.02,
      "grad_norm": 3.892084887615599,
      "learning_rate": 2.8873364167478935e-06,
      "loss": 0.3685,
      "step": 47845
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.9450597584169,
      "learning_rate": 2.8862206507248893e-06,
      "loss": 0.387,
      "step": 47850
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.97601429158614,
      "learning_rate": 2.8851050128541024e-06,
      "loss": 0.3925,
      "step": 47855
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.661187136926373,
      "learning_rate": 2.883989503203167e-06,
      "loss": 0.3878,
      "step": 47860
    },
    {
      "epoch": 2.03,
      "grad_norm": 12.622871963056166,
      "learning_rate": 2.882874121839715e-06,
      "loss": 0.4059,
      "step": 47865
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.401097743443097,
      "learning_rate": 2.881758868831369e-06,
      "loss": 0.358,
      "step": 47870
    },
    {
      "epoch": 2.03,
      "grad_norm": 6.711589675533919,
      "learning_rate": 2.8806437442457457e-06,
      "loss": 0.4128,
      "step": 47875
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.2245819868186905,
      "learning_rate": 2.879528748150447e-06,
      "loss": 0.3654,
      "step": 47880
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.060696159350697,
      "learning_rate": 2.8784138806130746e-06,
      "loss": 0.4016,
      "step": 47885
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.8312903214940777,
      "learning_rate": 2.877299141701221e-06,
      "loss": 0.3616,
      "step": 47890
    },
    {
      "epoch": 2.03,
      "grad_norm": 6.765022755362178,
      "learning_rate": 2.8761845314824653e-06,
      "loss": 0.4007,
      "step": 47895
    },
    {
      "epoch": 2.03,
      "grad_norm": 6.807441965558301,
      "learning_rate": 2.8750700500243867e-06,
      "loss": 0.398,
      "step": 47900
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.180072075119044,
      "learning_rate": 2.8739556973945527e-06,
      "loss": 0.4146,
      "step": 47905
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.082194429377859,
      "learning_rate": 2.8728414736605208e-06,
      "loss": 0.3929,
      "step": 47910
    },
    {
      "epoch": 2.03,
      "grad_norm": 6.099911697422608,
      "learning_rate": 2.871727378889845e-06,
      "loss": 0.3859,
      "step": 47915
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.542975038218331,
      "learning_rate": 2.8706134131500683e-06,
      "loss": 0.396,
      "step": 47920
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.0134557152627215,
      "learning_rate": 2.869499576508731e-06,
      "loss": 0.383,
      "step": 47925
    },
    {
      "epoch": 2.03,
      "grad_norm": 8.532318568264783,
      "learning_rate": 2.8683858690333555e-06,
      "loss": 0.3989,
      "step": 47930
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.299556875424699,
      "learning_rate": 2.867272290791467e-06,
      "loss": 0.3632,
      "step": 47935
    },
    {
      "epoch": 2.03,
      "grad_norm": 13.551811332320883,
      "learning_rate": 2.8661588418505793e-06,
      "loss": 0.3953,
      "step": 47940
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.49397292238517,
      "learning_rate": 2.8650455222781947e-06,
      "loss": 0.3662,
      "step": 47945
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.16883034192617,
      "learning_rate": 2.86393233214181e-06,
      "loss": 0.3831,
      "step": 47950
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.9712516729283847,
      "learning_rate": 2.8628192715089165e-06,
      "loss": 0.3712,
      "step": 47955
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.117031164377064,
      "learning_rate": 2.8617063404469964e-06,
      "loss": 0.3997,
      "step": 47960
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.252525765986604,
      "learning_rate": 2.8605935390235206e-06,
      "loss": 0.3911,
      "step": 47965
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.875140145132414,
      "learning_rate": 2.8594808673059572e-06,
      "loss": 0.4097,
      "step": 47970
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.485034875311755,
      "learning_rate": 2.858368325361765e-06,
      "loss": 0.4076,
      "step": 47975
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.501660928661121,
      "learning_rate": 2.857255913258391e-06,
      "loss": 0.3784,
      "step": 47980
    },
    {
      "epoch": 2.03,
      "grad_norm": 6.4661822513601805,
      "learning_rate": 2.8561436310632793e-06,
      "loss": 0.3823,
      "step": 47985
    },
    {
      "epoch": 2.03,
      "grad_norm": 9.04588973113391,
      "learning_rate": 2.855031478843865e-06,
      "loss": 0.382,
      "step": 47990
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.195991934620597,
      "learning_rate": 2.8539194566675752e-06,
      "loss": 0.4198,
      "step": 47995
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.840824558497126,
      "learning_rate": 2.852807564601825e-06,
      "loss": 0.4005,
      "step": 48000
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.170442593148155,
      "learning_rate": 2.8516958027140284e-06,
      "loss": 0.4006,
      "step": 48005
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.359858673157745,
      "learning_rate": 2.850584171071589e-06,
      "loss": 0.4029,
      "step": 48010
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.277054442590462,
      "learning_rate": 2.8494726697418974e-06,
      "loss": 0.3779,
      "step": 48015
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.403124936217612,
      "learning_rate": 2.8483612987923435e-06,
      "loss": 0.3809,
      "step": 48020
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.9906868772275375,
      "learning_rate": 2.8472500582903084e-06,
      "loss": 0.3762,
      "step": 48025
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.614173722470552,
      "learning_rate": 2.8461389483031586e-06,
      "loss": 0.3806,
      "step": 48030
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.640790744750531,
      "learning_rate": 2.8450279688982603e-06,
      "loss": 0.3797,
      "step": 48035
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.1857230043837115,
      "learning_rate": 2.8439171201429684e-06,
      "loss": 0.4142,
      "step": 48040
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.231838896996696,
      "learning_rate": 2.842806402104633e-06,
      "loss": 0.372,
      "step": 48045
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.126745866215367,
      "learning_rate": 2.841695814850589e-06,
      "loss": 0.3977,
      "step": 48050
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.054999098469555,
      "learning_rate": 2.84058535844817e-06,
      "loss": 0.3869,
      "step": 48055
    },
    {
      "epoch": 2.03,
      "grad_norm": 4.070293201148474,
      "learning_rate": 2.8394750329647025e-06,
      "loss": 0.3824,
      "step": 48060
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.99780252063637,
      "learning_rate": 2.8383648384674973e-06,
      "loss": 0.393,
      "step": 48065
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.9781577660559595,
      "learning_rate": 2.837254775023864e-06,
      "loss": 0.3669,
      "step": 48070
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.236269854909774,
      "learning_rate": 2.836144842701106e-06,
      "loss": 0.4292,
      "step": 48075
    },
    {
      "epoch": 2.03,
      "grad_norm": 5.110770039245152,
      "learning_rate": 2.835035041566511e-06,
      "loss": 0.3921,
      "step": 48080
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.773325234350164,
      "learning_rate": 2.833925371687363e-06,
      "loss": 0.3582,
      "step": 48085
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.288653456996363,
      "learning_rate": 2.832815833130942e-06,
      "loss": 0.4091,
      "step": 48090
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.924388218237496,
      "learning_rate": 2.831706425964511e-06,
      "loss": 0.3531,
      "step": 48095
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.042470814114546,
      "learning_rate": 2.830597150255333e-06,
      "loss": 0.4236,
      "step": 48100
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.7041070601447075,
      "learning_rate": 2.829488006070661e-06,
      "loss": 0.3924,
      "step": 48105
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.3330071802938175,
      "learning_rate": 2.8283789934777383e-06,
      "loss": 0.392,
      "step": 48110
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.175850130255339,
      "learning_rate": 2.8272701125437986e-06,
      "loss": 0.4072,
      "step": 48115
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.984236367039781,
      "learning_rate": 2.8261613633360708e-06,
      "loss": 0.383,
      "step": 48120
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.155621785129845,
      "learning_rate": 2.8250527459217775e-06,
      "loss": 0.3698,
      "step": 48125
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.017320981212969,
      "learning_rate": 2.823944260368131e-06,
      "loss": 0.3719,
      "step": 48130
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.088609117737862,
      "learning_rate": 2.822835906742333e-06,
      "loss": 0.3972,
      "step": 48135
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.473783393776434,
      "learning_rate": 2.8217276851115805e-06,
      "loss": 0.3805,
      "step": 48140
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.028859481683899,
      "learning_rate": 2.820619595543064e-06,
      "loss": 0.3896,
      "step": 48145
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.23202135574207,
      "learning_rate": 2.81951163810396e-06,
      "loss": 0.3766,
      "step": 48150
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.1775451629764415,
      "learning_rate": 2.818403812861442e-06,
      "loss": 0.3866,
      "step": 48155
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.7352576058573073,
      "learning_rate": 2.8172961198826775e-06,
      "loss": 0.3768,
      "step": 48160
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.651892688530061,
      "learning_rate": 2.816188559234818e-06,
      "loss": 0.3727,
      "step": 48165
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.331925586133346,
      "learning_rate": 2.8150811309850124e-06,
      "loss": 0.3939,
      "step": 48170
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.165468639382936,
      "learning_rate": 2.813973835200403e-06,
      "loss": 0.3807,
      "step": 48175
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.434481766797425,
      "learning_rate": 2.812866671948122e-06,
      "loss": 0.3909,
      "step": 48180
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.294882379298864,
      "learning_rate": 2.811759641295291e-06,
      "loss": 0.3986,
      "step": 48185
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.400347963344931,
      "learning_rate": 2.810652743309027e-06,
      "loss": 0.4312,
      "step": 48190
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.7200872540214545,
      "learning_rate": 2.8095459780564404e-06,
      "loss": 0.3877,
      "step": 48195
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.5877589160667362,
      "learning_rate": 2.8084393456046265e-06,
      "loss": 0.3736,
      "step": 48200
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.699831064953315,
      "learning_rate": 2.8073328460206804e-06,
      "loss": 0.4073,
      "step": 48205
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.434159599386055,
      "learning_rate": 2.8062264793716865e-06,
      "loss": 0.4155,
      "step": 48210
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.0739280884905735,
      "learning_rate": 2.8051202457247173e-06,
      "loss": 0.369,
      "step": 48215
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.944469726262241,
      "learning_rate": 2.8040141451468426e-06,
      "loss": 0.3648,
      "step": 48220
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.767980413361959,
      "learning_rate": 2.8029081777051213e-06,
      "loss": 0.3872,
      "step": 48225
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.217062336095852,
      "learning_rate": 2.8018023434666073e-06,
      "loss": 0.3766,
      "step": 48230
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.801639687007457,
      "learning_rate": 2.8006966424983413e-06,
      "loss": 0.3897,
      "step": 48235
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.292093732864242,
      "learning_rate": 2.7995910748673586e-06,
      "loss": 0.3776,
      "step": 48240
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.035631103481674,
      "learning_rate": 2.79848564064069e-06,
      "loss": 0.3778,
      "step": 48245
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.05459240264595,
      "learning_rate": 2.7973803398853506e-06,
      "loss": 0.3736,
      "step": 48250
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.231650065518774,
      "learning_rate": 2.796275172668352e-06,
      "loss": 0.3745,
      "step": 48255
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.223637167027155,
      "learning_rate": 2.7951701390567014e-06,
      "loss": 0.3812,
      "step": 48260
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.853259572625339,
      "learning_rate": 2.7940652391173906e-06,
      "loss": 0.4185,
      "step": 48265
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.108730016375332,
      "learning_rate": 2.7929604729174044e-06,
      "loss": 0.3968,
      "step": 48270
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.195627575098824,
      "learning_rate": 2.7918558405237232e-06,
      "loss": 0.3659,
      "step": 48275
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.99909738886082,
      "learning_rate": 2.7907513420033207e-06,
      "loss": 0.3948,
      "step": 48280
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.04313488262699,
      "learning_rate": 2.7896469774231548e-06,
      "loss": 0.4016,
      "step": 48285
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.94339342572917,
      "learning_rate": 2.7885427468501813e-06,
      "loss": 0.3799,
      "step": 48290
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.6890326192917984,
      "learning_rate": 2.787438650351347e-06,
      "loss": 0.3845,
      "step": 48295
    },
    {
      "epoch": 2.04,
      "grad_norm": 6.355299529250054,
      "learning_rate": 2.7863346879935926e-06,
      "loss": 0.4236,
      "step": 48300
    },
    {
      "epoch": 2.04,
      "grad_norm": 5.87706970564047,
      "learning_rate": 2.7852308598438437e-06,
      "loss": 0.4043,
      "step": 48305
    },
    {
      "epoch": 2.04,
      "grad_norm": 7.240536451779711,
      "learning_rate": 2.784127165969024e-06,
      "loss": 0.3466,
      "step": 48310
    },
    {
      "epoch": 2.04,
      "grad_norm": 6.003576672084176,
      "learning_rate": 2.7830236064360485e-06,
      "loss": 0.4018,
      "step": 48315
    },
    {
      "epoch": 2.04,
      "grad_norm": 3.870110644213413,
      "learning_rate": 2.781920181311819e-06,
      "loss": 0.402,
      "step": 48320
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.426806177657531,
      "learning_rate": 2.780816890663236e-06,
      "loss": 0.374,
      "step": 48325
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.344707046671713,
      "learning_rate": 2.77971373455719e-06,
      "loss": 0.4182,
      "step": 48330
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.514210849108016,
      "learning_rate": 2.778610713060559e-06,
      "loss": 0.394,
      "step": 48335
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.592807150078073,
      "learning_rate": 2.777507826240216e-06,
      "loss": 0.374,
      "step": 48340
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.569838100045995,
      "learning_rate": 2.7764050741630278e-06,
      "loss": 0.4111,
      "step": 48345
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.264369401870004,
      "learning_rate": 2.7753024568958508e-06,
      "loss": 0.393,
      "step": 48350
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.69277871852729,
      "learning_rate": 2.774199974505532e-06,
      "loss": 0.4024,
      "step": 48355
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.9145106005925556,
      "learning_rate": 2.773097627058912e-06,
      "loss": 0.3963,
      "step": 48360
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.656961185937571,
      "learning_rate": 2.7719954146228254e-06,
      "loss": 0.3946,
      "step": 48365
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.321301624594661,
      "learning_rate": 2.7708933372640925e-06,
      "loss": 0.3788,
      "step": 48370
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.02214173024217,
      "learning_rate": 2.7697913950495298e-06,
      "loss": 0.3725,
      "step": 48375
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.9796984419359673,
      "learning_rate": 2.7686895880459475e-06,
      "loss": 0.3787,
      "step": 48380
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.258481266938924,
      "learning_rate": 2.7675879163201414e-06,
      "loss": 0.4112,
      "step": 48385
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.901298453053275,
      "learning_rate": 2.766486379938904e-06,
      "loss": 0.3796,
      "step": 48390
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.312260190227359,
      "learning_rate": 2.7653849789690203e-06,
      "loss": 0.3675,
      "step": 48395
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.291482365128579,
      "learning_rate": 2.764283713477261e-06,
      "loss": 0.4042,
      "step": 48400
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.42267969844581,
      "learning_rate": 2.7631825835303948e-06,
      "loss": 0.3847,
      "step": 48405
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.403607079143448,
      "learning_rate": 2.762081589195179e-06,
      "loss": 0.4052,
      "step": 48410
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.185346648236187,
      "learning_rate": 2.7609807305383675e-06,
      "loss": 0.3743,
      "step": 48415
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.34310104095834,
      "learning_rate": 2.759880007626697e-06,
      "loss": 0.3935,
      "step": 48420
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.7844747328018946,
      "learning_rate": 2.7587794205269024e-06,
      "loss": 0.3797,
      "step": 48425
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.299545195137125,
      "learning_rate": 2.7576789693057125e-06,
      "loss": 0.3884,
      "step": 48430
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.963007533460066,
      "learning_rate": 2.7565786540298413e-06,
      "loss": 0.4162,
      "step": 48435
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.617904577702721,
      "learning_rate": 2.7554784747659962e-06,
      "loss": 0.3868,
      "step": 48440
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.551583945865222,
      "learning_rate": 2.7543784315808796e-06,
      "loss": 0.38,
      "step": 48445
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.624987964056512,
      "learning_rate": 2.7532785245411863e-06,
      "loss": 0.3734,
      "step": 48450
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.329332866545534,
      "learning_rate": 2.7521787537135956e-06,
      "loss": 0.4195,
      "step": 48455
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.507290462139366,
      "learning_rate": 2.751079119164787e-06,
      "loss": 0.3637,
      "step": 48460
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.053213369785701,
      "learning_rate": 2.7499796209614283e-06,
      "loss": 0.3878,
      "step": 48465
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.040708612867266,
      "learning_rate": 2.7488802591701754e-06,
      "loss": 0.426,
      "step": 48470
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.169649148186469,
      "learning_rate": 2.7477810338576817e-06,
      "loss": 0.405,
      "step": 48475
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.9586866184857987,
      "learning_rate": 2.7466819450905893e-06,
      "loss": 0.3906,
      "step": 48480
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.167320348648449,
      "learning_rate": 2.7455829929355353e-06,
      "loss": 0.374,
      "step": 48485
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.058896872929868,
      "learning_rate": 2.7444841774591425e-06,
      "loss": 0.3902,
      "step": 48490
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.612775650778324,
      "learning_rate": 2.7433854987280295e-06,
      "loss": 0.3887,
      "step": 48495
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.373570165706974,
      "learning_rate": 2.742286956808809e-06,
      "loss": 0.3901,
      "step": 48500
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.4190735712246,
      "learning_rate": 2.7411885517680782e-06,
      "loss": 0.3956,
      "step": 48505
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.301074962439415,
      "learning_rate": 2.7400902836724313e-06,
      "loss": 0.3909,
      "step": 48510
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.001962786369771,
      "learning_rate": 2.738992152588456e-06,
      "loss": 0.3926,
      "step": 48515
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.789725858477086,
      "learning_rate": 2.7378941585827246e-06,
      "loss": 0.4055,
      "step": 48520
    },
    {
      "epoch": 2.05,
      "grad_norm": 3.5982589723521654,
      "learning_rate": 2.736796301721807e-06,
      "loss": 0.3787,
      "step": 48525
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.404394226673754,
      "learning_rate": 2.735698582072264e-06,
      "loss": 0.4263,
      "step": 48530
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.32801112748244,
      "learning_rate": 2.7346009997006474e-06,
      "loss": 0.3711,
      "step": 48535
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.021115381284519,
      "learning_rate": 2.733503554673497e-06,
      "loss": 0.392,
      "step": 48540
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.054467299470257,
      "learning_rate": 2.73240624705735e-06,
      "loss": 0.3923,
      "step": 48545
    },
    {
      "epoch": 2.05,
      "grad_norm": 4.642101356379261,
      "learning_rate": 2.731309076918735e-06,
      "loss": 0.3778,
      "step": 48550
    },
    {
      "epoch": 2.05,
      "grad_norm": 5.90088598323868,
      "learning_rate": 2.7302120443241665e-06,
      "loss": 0.3997,
      "step": 48555
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.950109874480378,
      "learning_rate": 2.7291151493401545e-06,
      "loss": 0.3865,
      "step": 48560
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.5499005039360556,
      "learning_rate": 2.7280183920332056e-06,
      "loss": 0.3666,
      "step": 48565
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.956143559045745,
      "learning_rate": 2.726921772469806e-06,
      "loss": 0.4051,
      "step": 48570
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.6392363595031822,
      "learning_rate": 2.725825290716444e-06,
      "loss": 0.3691,
      "step": 48575
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.893745130583176,
      "learning_rate": 2.7247289468395965e-06,
      "loss": 0.3983,
      "step": 48580
    },
    {
      "epoch": 2.06,
      "grad_norm": 6.5570696978935015,
      "learning_rate": 2.723632740905732e-06,
      "loss": 0.4088,
      "step": 48585
    },
    {
      "epoch": 2.06,
      "grad_norm": 9.698394165038511,
      "learning_rate": 2.7225366729813095e-06,
      "loss": 0.3877,
      "step": 48590
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.450383148938726,
      "learning_rate": 2.7214407431327783e-06,
      "loss": 0.3782,
      "step": 48595
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.944339661267222,
      "learning_rate": 2.7203449514265833e-06,
      "loss": 0.3916,
      "step": 48600
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.775718101785444,
      "learning_rate": 2.7192492979291606e-06,
      "loss": 0.3764,
      "step": 48605
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.800730545085146,
      "learning_rate": 2.7181537827069328e-06,
      "loss": 0.3919,
      "step": 48610
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.966451241805173,
      "learning_rate": 2.7170584058263196e-06,
      "loss": 0.3913,
      "step": 48615
    },
    {
      "epoch": 2.06,
      "grad_norm": 6.93660263204623,
      "learning_rate": 2.715963167353733e-06,
      "loss": 0.4016,
      "step": 48620
    },
    {
      "epoch": 2.06,
      "grad_norm": 7.698625819136889,
      "learning_rate": 2.7148680673555695e-06,
      "loss": 0.3889,
      "step": 48625
    },
    {
      "epoch": 2.06,
      "grad_norm": 6.232238118570845,
      "learning_rate": 2.713773105898224e-06,
      "loss": 0.3714,
      "step": 48630
    },
    {
      "epoch": 2.06,
      "grad_norm": 8.823150222903811,
      "learning_rate": 2.7126782830480827e-06,
      "loss": 0.3978,
      "step": 48635
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.042317024595186,
      "learning_rate": 2.7115835988715177e-06,
      "loss": 0.3769,
      "step": 48640
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.003916360256172,
      "learning_rate": 2.710489053434898e-06,
      "loss": 0.3805,
      "step": 48645
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.520191329108922,
      "learning_rate": 2.709394646804584e-06,
      "loss": 0.3653,
      "step": 48650
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.049544885302632,
      "learning_rate": 2.7083003790469262e-06,
      "loss": 0.3794,
      "step": 48655
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.964407085722664,
      "learning_rate": 2.707206250228265e-06,
      "loss": 0.3751,
      "step": 48660
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.330884655345749,
      "learning_rate": 2.706112260414935e-06,
      "loss": 0.3828,
      "step": 48665
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.1978876822049935,
      "learning_rate": 2.705018409673263e-06,
      "loss": 0.4039,
      "step": 48670
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.8487417803422064,
      "learning_rate": 2.703924698069563e-06,
      "loss": 0.3695,
      "step": 48675
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.4957794396976087,
      "learning_rate": 2.7028311256701457e-06,
      "loss": 0.3739,
      "step": 48680
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.901395195338321,
      "learning_rate": 2.7017376925413123e-06,
      "loss": 0.3523,
      "step": 48685
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.272880650340702,
      "learning_rate": 2.7006443987493513e-06,
      "loss": 0.3918,
      "step": 48690
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.552460420620594,
      "learning_rate": 2.699551244360547e-06,
      "loss": 0.3615,
      "step": 48695
    },
    {
      "epoch": 2.06,
      "grad_norm": 6.09029001793615,
      "learning_rate": 2.6984582294411763e-06,
      "loss": 0.3888,
      "step": 48700
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.178009153647491,
      "learning_rate": 2.697365354057503e-06,
      "loss": 0.3941,
      "step": 48705
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.30048215653837,
      "learning_rate": 2.696272618275785e-06,
      "loss": 0.3769,
      "step": 48710
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.094727503777293,
      "learning_rate": 2.6951800221622714e-06,
      "loss": 0.4215,
      "step": 48715
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.8872646963910995,
      "learning_rate": 2.6940875657832068e-06,
      "loss": 0.3962,
      "step": 48720
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.159269793740176,
      "learning_rate": 2.6929952492048184e-06,
      "loss": 0.3791,
      "step": 48725
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.013258462664164,
      "learning_rate": 2.6919030724933333e-06,
      "loss": 0.408,
      "step": 48730
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.282336799805424,
      "learning_rate": 2.6908110357149674e-06,
      "loss": 0.425,
      "step": 48735
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.50757335706558,
      "learning_rate": 2.6897191389359253e-06,
      "loss": 0.4026,
      "step": 48740
    },
    {
      "epoch": 2.06,
      "grad_norm": 6.971045764146771,
      "learning_rate": 2.688627382222406e-06,
      "loss": 0.3961,
      "step": 48745
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.661024779287403,
      "learning_rate": 2.687535765640602e-06,
      "loss": 0.3937,
      "step": 48750
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.04324946428947,
      "learning_rate": 2.686444289256693e-06,
      "loss": 0.3865,
      "step": 48755
    },
    {
      "epoch": 2.06,
      "grad_norm": 7.579909932334052,
      "learning_rate": 2.6853529531368505e-06,
      "loss": 0.3963,
      "step": 48760
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.347597687549111,
      "learning_rate": 2.6842617573472403e-06,
      "loss": 0.3717,
      "step": 48765
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.8794700100244914,
      "learning_rate": 2.68317070195402e-06,
      "loss": 0.3488,
      "step": 48770
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.178796846730626,
      "learning_rate": 2.682079787023334e-06,
      "loss": 0.3866,
      "step": 48775
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.16766330119346,
      "learning_rate": 2.680989012621323e-06,
      "loss": 0.3815,
      "step": 48780
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.375902402039784,
      "learning_rate": 2.679898378814117e-06,
      "loss": 0.3849,
      "step": 48785
    },
    {
      "epoch": 2.06,
      "grad_norm": 5.36328391655569,
      "learning_rate": 2.67880788566784e-06,
      "loss": 0.3894,
      "step": 48790
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.1341531474382105,
      "learning_rate": 2.677717533248602e-06,
      "loss": 0.4083,
      "step": 48795
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.597446799046262,
      "learning_rate": 2.6766273216225093e-06,
      "loss": 0.3923,
      "step": 48800
    },
    {
      "epoch": 2.07,
      "grad_norm": 3.7483899072627427,
      "learning_rate": 2.6755372508556603e-06,
      "loss": 0.3751,
      "step": 48805
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.689037156362096,
      "learning_rate": 2.6744473210141385e-06,
      "loss": 0.3792,
      "step": 48810
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.405152396891129,
      "learning_rate": 2.6733575321640264e-06,
      "loss": 0.3683,
      "step": 48815
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.601083531395233,
      "learning_rate": 2.6722678843713958e-06,
      "loss": 0.3676,
      "step": 48820
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.163212067135756,
      "learning_rate": 2.6711783777023036e-06,
      "loss": 0.3734,
      "step": 48825
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.3582318833968365,
      "learning_rate": 2.6700890122228074e-06,
      "loss": 0.3766,
      "step": 48830
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.660450556949528,
      "learning_rate": 2.6689997879989516e-06,
      "loss": 0.3734,
      "step": 48835
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.161179407442511,
      "learning_rate": 2.667910705096774e-06,
      "loss": 0.402,
      "step": 48840
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.371897780123862,
      "learning_rate": 2.666821763582299e-06,
      "loss": 0.3697,
      "step": 48845
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.496835219583497,
      "learning_rate": 2.665732963521548e-06,
      "loss": 0.3996,
      "step": 48850
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.528942984797106,
      "learning_rate": 2.664644304980534e-06,
      "loss": 0.3725,
      "step": 48855
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.577663673743399,
      "learning_rate": 2.663555788025254e-06,
      "loss": 0.3653,
      "step": 48860
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.1040313893328255,
      "learning_rate": 2.6624674127217054e-06,
      "loss": 0.3763,
      "step": 48865
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.294053441864696,
      "learning_rate": 2.6613791791358734e-06,
      "loss": 0.3891,
      "step": 48870
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.480008742807418,
      "learning_rate": 2.6602910873337307e-06,
      "loss": 0.3832,
      "step": 48875
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.0473120786164465,
      "learning_rate": 2.6592031373812484e-06,
      "loss": 0.382,
      "step": 48880
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.450227310997142,
      "learning_rate": 2.658115329344384e-06,
      "loss": 0.3829,
      "step": 48885
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.135922977015602,
      "learning_rate": 2.6570276632890917e-06,
      "loss": 0.3888,
      "step": 48890
    },
    {
      "epoch": 2.07,
      "grad_norm": 6.515891820904272,
      "learning_rate": 2.655940139281309e-06,
      "loss": 0.3785,
      "step": 48895
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.344462821168202,
      "learning_rate": 2.6548527573869698e-06,
      "loss": 0.3946,
      "step": 48900
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.109356002022949,
      "learning_rate": 2.653765517672003e-06,
      "loss": 0.378,
      "step": 48905
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.2609856172907765,
      "learning_rate": 2.6526784202023194e-06,
      "loss": 0.3644,
      "step": 48910
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.614365807863728,
      "learning_rate": 2.6515914650438314e-06,
      "loss": 0.4074,
      "step": 48915
    },
    {
      "epoch": 2.07,
      "grad_norm": 6.06455249508406,
      "learning_rate": 2.6505046522624326e-06,
      "loss": 0.3869,
      "step": 48920
    },
    {
      "epoch": 2.07,
      "grad_norm": 6.006835085881209,
      "learning_rate": 2.6494179819240184e-06,
      "loss": 0.3879,
      "step": 48925
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.090195172031044,
      "learning_rate": 2.648331454094466e-06,
      "loss": 0.3663,
      "step": 48930
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.140840771231388,
      "learning_rate": 2.6472450688396513e-06,
      "loss": 0.4142,
      "step": 48935
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.115114507990791,
      "learning_rate": 2.6461588262254388e-06,
      "loss": 0.3821,
      "step": 48940
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.722847534864015,
      "learning_rate": 2.6450727263176817e-06,
      "loss": 0.3778,
      "step": 48945
    },
    {
      "epoch": 2.07,
      "grad_norm": 3.827257159887921,
      "learning_rate": 2.6439867691822283e-06,
      "loss": 0.3782,
      "step": 48950
    },
    {
      "epoch": 2.07,
      "grad_norm": 7.277556924058113,
      "learning_rate": 2.6429009548849173e-06,
      "loss": 0.4098,
      "step": 48955
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.058902488040372,
      "learning_rate": 2.6418152834915806e-06,
      "loss": 0.3627,
      "step": 48960
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.048965176677439,
      "learning_rate": 2.6407297550680346e-06,
      "loss": 0.3868,
      "step": 48965
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.762467868739015,
      "learning_rate": 2.639644369680094e-06,
      "loss": 0.3825,
      "step": 48970
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.645496806393459,
      "learning_rate": 2.638559127393565e-06,
      "loss": 0.3908,
      "step": 48975
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.2444726141926274,
      "learning_rate": 2.6374740282742383e-06,
      "loss": 0.3985,
      "step": 48980
    },
    {
      "epoch": 2.07,
      "grad_norm": 10.766393375915584,
      "learning_rate": 2.6363890723879026e-06,
      "loss": 0.3849,
      "step": 48985
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.040058211840637,
      "learning_rate": 2.6353042598003363e-06,
      "loss": 0.3969,
      "step": 48990
    },
    {
      "epoch": 2.07,
      "grad_norm": 6.57164803620502,
      "learning_rate": 2.634219590577306e-06,
      "loss": 0.3958,
      "step": 48995
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.143096539548772,
      "learning_rate": 2.6331350647845727e-06,
      "loss": 0.3926,
      "step": 49000
    },
    {
      "epoch": 2.07,
      "grad_norm": 5.546735037849133,
      "learning_rate": 2.632050682487891e-06,
      "loss": 0.3752,
      "step": 49005
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.44884087597974,
      "learning_rate": 2.630966443752999e-06,
      "loss": 0.3693,
      "step": 49010
    },
    {
      "epoch": 2.07,
      "grad_norm": 4.970880980102137,
      "learning_rate": 2.629882348645634e-06,
      "loss": 0.4049,
      "step": 49015
    },
    {
      "epoch": 2.07,
      "grad_norm": 3.8810543456140656,
      "learning_rate": 2.62879839723152e-06,
      "loss": 0.3579,
      "step": 49020
    },
    {
      "epoch": 2.07,
      "grad_norm": 8.451817284269396,
      "learning_rate": 2.627714589576377e-06,
      "loss": 0.3783,
      "step": 49025
    },
    {
      "epoch": 2.08,
      "grad_norm": 6.580980237386432,
      "learning_rate": 2.6266309257459085e-06,
      "loss": 0.3731,
      "step": 49030
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.348971145075655,
      "learning_rate": 2.6255474058058163e-06,
      "loss": 0.3921,
      "step": 49035
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.203429737016677,
      "learning_rate": 2.6244640298217917e-06,
      "loss": 0.3601,
      "step": 49040
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.588063239393658,
      "learning_rate": 2.6233807978595143e-06,
      "loss": 0.3976,
      "step": 49045
    },
    {
      "epoch": 2.08,
      "grad_norm": 12.491402518908926,
      "learning_rate": 2.622297709984659e-06,
      "loss": 0.404,
      "step": 49050
    },
    {
      "epoch": 2.08,
      "grad_norm": 13.691072458845982,
      "learning_rate": 2.621214766262891e-06,
      "loss": 0.3907,
      "step": 49055
    },
    {
      "epoch": 2.08,
      "grad_norm": 7.581929565525865,
      "learning_rate": 2.6201319667598633e-06,
      "loss": 0.378,
      "step": 49060
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.747502036354879,
      "learning_rate": 2.619049311541224e-06,
      "loss": 0.3916,
      "step": 49065
    },
    {
      "epoch": 2.08,
      "grad_norm": 11.947588096786621,
      "learning_rate": 2.6179668006726124e-06,
      "loss": 0.376,
      "step": 49070
    },
    {
      "epoch": 2.08,
      "grad_norm": 6.828127093266616,
      "learning_rate": 2.6168844342196597e-06,
      "loss": 0.3843,
      "step": 49075
    },
    {
      "epoch": 2.08,
      "grad_norm": 7.896040292874367,
      "learning_rate": 2.6158022122479807e-06,
      "loss": 0.4004,
      "step": 49080
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.567006714401726,
      "learning_rate": 2.6147201348231916e-06,
      "loss": 0.3537,
      "step": 49085
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.807868812847661,
      "learning_rate": 2.6136382020108943e-06,
      "loss": 0.4177,
      "step": 49090
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.981771887222753,
      "learning_rate": 2.6125564138766846e-06,
      "loss": 0.4008,
      "step": 49095
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.236785748551839,
      "learning_rate": 2.611474770486146e-06,
      "loss": 0.3975,
      "step": 49100
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.061875568407436,
      "learning_rate": 2.6103932719048564e-06,
      "loss": 0.4048,
      "step": 49105
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.095892521061463,
      "learning_rate": 2.609311918198385e-06,
      "loss": 0.3802,
      "step": 49110
    },
    {
      "epoch": 2.08,
      "grad_norm": 6.063882065852188,
      "learning_rate": 2.608230709432289e-06,
      "loss": 0.4112,
      "step": 49115
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.0820637444667724,
      "learning_rate": 2.607149645672119e-06,
      "loss": 0.3867,
      "step": 49120
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.114988874060213,
      "learning_rate": 2.6060687269834207e-06,
      "loss": 0.3664,
      "step": 49125
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.041203185459699,
      "learning_rate": 2.604987953431721e-06,
      "loss": 0.3946,
      "step": 49130
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.998551344833177,
      "learning_rate": 2.6039073250825476e-06,
      "loss": 0.377,
      "step": 49135
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.120079708441207,
      "learning_rate": 2.6028268420014146e-06,
      "loss": 0.3781,
      "step": 49140
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.839990058698662,
      "learning_rate": 2.6017465042538314e-06,
      "loss": 0.3723,
      "step": 49145
    },
    {
      "epoch": 2.08,
      "grad_norm": 6.046238072447489,
      "learning_rate": 2.6006663119052913e-06,
      "loss": 0.4018,
      "step": 49150
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.934221379241571,
      "learning_rate": 2.5995862650212856e-06,
      "loss": 0.3632,
      "step": 49155
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.565988227474438,
      "learning_rate": 2.598506363667295e-06,
      "loss": 0.3711,
      "step": 49160
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.125449991708251,
      "learning_rate": 2.5974266079087884e-06,
      "loss": 0.3767,
      "step": 49165
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.8424417805420745,
      "learning_rate": 2.59634699781123e-06,
      "loss": 0.4194,
      "step": 49170
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.678483842287691,
      "learning_rate": 2.5952675334400745e-06,
      "loss": 0.3714,
      "step": 49175
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.9783348503739715,
      "learning_rate": 2.594188214860763e-06,
      "loss": 0.3954,
      "step": 49180
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.362954186255089,
      "learning_rate": 2.593109042138733e-06,
      "loss": 0.3896,
      "step": 49185
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.691186314651187,
      "learning_rate": 2.592030015339413e-06,
      "loss": 0.3806,
      "step": 49190
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.683857388953216,
      "learning_rate": 2.590951134528222e-06,
      "loss": 0.391,
      "step": 49195
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.0321210169962605,
      "learning_rate": 2.5898723997705657e-06,
      "loss": 0.3884,
      "step": 49200
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.2841414946300995,
      "learning_rate": 2.5887938111318455e-06,
      "loss": 0.3768,
      "step": 49205
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.9920102444496095,
      "learning_rate": 2.587715368677457e-06,
      "loss": 0.3981,
      "step": 49210
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.7900466935067825,
      "learning_rate": 2.5866370724727784e-06,
      "loss": 0.3809,
      "step": 49215
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.8137774098630923,
      "learning_rate": 2.5855589225831844e-06,
      "loss": 0.3873,
      "step": 49220
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.4578510942446075,
      "learning_rate": 2.5844809190740437e-06,
      "loss": 0.3991,
      "step": 49225
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.194832317260989,
      "learning_rate": 2.5834030620107078e-06,
      "loss": 0.3843,
      "step": 49230
    },
    {
      "epoch": 2.08,
      "grad_norm": 5.1262913405352775,
      "learning_rate": 2.582325351458525e-06,
      "loss": 0.3571,
      "step": 49235
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.7949040845686275,
      "learning_rate": 2.5812477874828373e-06,
      "loss": 0.3943,
      "step": 49240
    },
    {
      "epoch": 2.08,
      "grad_norm": 7.0427948804486435,
      "learning_rate": 2.5801703701489713e-06,
      "loss": 0.3844,
      "step": 49245
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.711364121142358,
      "learning_rate": 2.5790930995222463e-06,
      "loss": 0.3736,
      "step": 49250
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.9603745763761222,
      "learning_rate": 2.578015975667976e-06,
      "loss": 0.377,
      "step": 49255
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.119100866087835,
      "learning_rate": 2.5769389986514637e-06,
      "loss": 0.3782,
      "step": 49260
    },
    {
      "epoch": 2.08,
      "grad_norm": 4.739832962603566,
      "learning_rate": 2.5758621685380016e-06,
      "loss": 0.4062,
      "step": 49265
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.53863607520641,
      "learning_rate": 2.574785485392875e-06,
      "loss": 0.3619,
      "step": 49270
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.3492833110385085,
      "learning_rate": 2.5737089492813612e-06,
      "loss": 0.3997,
      "step": 49275
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.946245638803383,
      "learning_rate": 2.5726325602687285e-06,
      "loss": 0.3302,
      "step": 49280
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.617055321970091,
      "learning_rate": 2.571556318420232e-06,
      "loss": 0.3671,
      "step": 49285
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.870217948071273,
      "learning_rate": 2.5704802238011226e-06,
      "loss": 0.3877,
      "step": 49290
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.817464712010223,
      "learning_rate": 2.569404276476643e-06,
      "loss": 0.4373,
      "step": 49295
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.80271860260757,
      "learning_rate": 2.5683284765120204e-06,
      "loss": 0.3877,
      "step": 49300
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.8786578448434077,
      "learning_rate": 2.5672528239724794e-06,
      "loss": 0.3765,
      "step": 49305
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.840724826597111,
      "learning_rate": 2.5661773189232365e-06,
      "loss": 0.3767,
      "step": 49310
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.9607700841651874,
      "learning_rate": 2.56510196142949e-06,
      "loss": 0.3947,
      "step": 49315
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.392638736642099,
      "learning_rate": 2.564026751556441e-06,
      "loss": 0.3965,
      "step": 49320
    },
    {
      "epoch": 2.09,
      "grad_norm": 6.844300602395313,
      "learning_rate": 2.562951689369274e-06,
      "loss": 0.4104,
      "step": 49325
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.5335264704752785,
      "learning_rate": 2.5618767749331696e-06,
      "loss": 0.3938,
      "step": 49330
    },
    {
      "epoch": 2.09,
      "grad_norm": 6.3056748648663214,
      "learning_rate": 2.5608020083132935e-06,
      "loss": 0.4149,
      "step": 49335
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.602749764289394,
      "learning_rate": 2.5597273895748054e-06,
      "loss": 0.3902,
      "step": 49340
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.602597555011603,
      "learning_rate": 2.5586529187828603e-06,
      "loss": 0.3698,
      "step": 49345
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.4850721070595,
      "learning_rate": 2.557578596002595e-06,
      "loss": 0.3848,
      "step": 49350
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.9814981383252155,
      "learning_rate": 2.556504421299145e-06,
      "loss": 0.3772,
      "step": 49355
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.164175277475107,
      "learning_rate": 2.555430394737637e-06,
      "loss": 0.3438,
      "step": 49360
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.72081236370228,
      "learning_rate": 2.554356516383182e-06,
      "loss": 0.4124,
      "step": 49365
    },
    {
      "epoch": 2.09,
      "grad_norm": 6.148949610442225,
      "learning_rate": 2.5532827863008862e-06,
      "loss": 0.4052,
      "step": 49370
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.784977006109457,
      "learning_rate": 2.5522092045558487e-06,
      "loss": 0.3867,
      "step": 49375
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.812493989430617,
      "learning_rate": 2.5511357712131587e-06,
      "loss": 0.37,
      "step": 49380
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.110668952923217,
      "learning_rate": 2.5500624863378916e-06,
      "loss": 0.3836,
      "step": 49385
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.351742140320441,
      "learning_rate": 2.548989349995119e-06,
      "loss": 0.3675,
      "step": 49390
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.306976478065809,
      "learning_rate": 2.547916362249905e-06,
      "loss": 0.4087,
      "step": 49395
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.863035877359422,
      "learning_rate": 2.5468435231672995e-06,
      "loss": 0.3838,
      "step": 49400
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.050191686008576,
      "learning_rate": 2.545770832812342e-06,
      "loss": 0.3926,
      "step": 49405
    },
    {
      "epoch": 2.09,
      "grad_norm": 9.555796270167855,
      "learning_rate": 2.5446982912500706e-06,
      "loss": 0.3983,
      "step": 49410
    },
    {
      "epoch": 2.09,
      "grad_norm": 6.1891855453256985,
      "learning_rate": 2.543625898545511e-06,
      "loss": 0.3832,
      "step": 49415
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.2132842778285164,
      "learning_rate": 2.5425536547636763e-06,
      "loss": 0.3744,
      "step": 49420
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.603209848717021,
      "learning_rate": 2.5414815599695743e-06,
      "loss": 0.3965,
      "step": 49425
    },
    {
      "epoch": 2.09,
      "grad_norm": 3.995517650234465,
      "learning_rate": 2.5404096142282057e-06,
      "loss": 0.4101,
      "step": 49430
    },
    {
      "epoch": 2.09,
      "grad_norm": 6.309811905580369,
      "learning_rate": 2.5393378176045556e-06,
      "loss": 0.4008,
      "step": 49435
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.606835131566553,
      "learning_rate": 2.5382661701636046e-06,
      "loss": 0.369,
      "step": 49440
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.509785120845635,
      "learning_rate": 2.5371946719703257e-06,
      "loss": 0.3814,
      "step": 49445
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.3606537445326925,
      "learning_rate": 2.53612332308968e-06,
      "loss": 0.3469,
      "step": 49450
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.143428631512332,
      "learning_rate": 2.5350521235866188e-06,
      "loss": 0.3928,
      "step": 49455
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.397818090890851,
      "learning_rate": 2.5339810735260863e-06,
      "loss": 0.3969,
      "step": 49460
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.570348234496646,
      "learning_rate": 2.532910172973019e-06,
      "loss": 0.3865,
      "step": 49465
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.299430828387545,
      "learning_rate": 2.531839421992339e-06,
      "loss": 0.377,
      "step": 49470
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.637197922302898,
      "learning_rate": 2.5307688206489644e-06,
      "loss": 0.3854,
      "step": 49475
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.952569899156878,
      "learning_rate": 2.5296983690078054e-06,
      "loss": 0.3943,
      "step": 49480
    },
    {
      "epoch": 2.09,
      "grad_norm": 5.744601122212498,
      "learning_rate": 2.528628067133756e-06,
      "loss": 0.3721,
      "step": 49485
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.205104201256576,
      "learning_rate": 2.527557915091706e-06,
      "loss": 0.3624,
      "step": 49490
    },
    {
      "epoch": 2.09,
      "grad_norm": 6.051912784040933,
      "learning_rate": 2.526487912946538e-06,
      "loss": 0.3788,
      "step": 49495
    },
    {
      "epoch": 2.09,
      "grad_norm": 4.115852812523828,
      "learning_rate": 2.5254180607631236e-06,
      "loss": 0.3864,
      "step": 49500
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.7616855038523815,
      "learning_rate": 2.5243483586063206e-06,
      "loss": 0.3686,
      "step": 49505
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.280639664026874,
      "learning_rate": 2.5232788065409847e-06,
      "loss": 0.3966,
      "step": 49510
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.297314428993249,
      "learning_rate": 2.522209404631961e-06,
      "loss": 0.4087,
      "step": 49515
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.9481993530155446,
      "learning_rate": 2.521140152944081e-06,
      "loss": 0.3625,
      "step": 49520
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.786545880772812,
      "learning_rate": 2.5200710515421717e-06,
      "loss": 0.393,
      "step": 49525
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.251369715162034,
      "learning_rate": 2.5190021004910516e-06,
      "loss": 0.3987,
      "step": 49530
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.957075682599359,
      "learning_rate": 2.517933299855524e-06,
      "loss": 0.3978,
      "step": 49535
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.9582525929647363,
      "learning_rate": 2.5168646497003895e-06,
      "loss": 0.3675,
      "step": 49540
    },
    {
      "epoch": 2.1,
      "grad_norm": 6.983028463144516,
      "learning_rate": 2.515796150090439e-06,
      "loss": 0.3666,
      "step": 49545
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.6195673020099015,
      "learning_rate": 2.514727801090448e-06,
      "loss": 0.3957,
      "step": 49550
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.6001685427389,
      "learning_rate": 2.5136596027651905e-06,
      "loss": 0.3647,
      "step": 49555
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.721425414917315,
      "learning_rate": 2.512591555179429e-06,
      "loss": 0.3775,
      "step": 49560
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.670864627243677,
      "learning_rate": 2.5115236583979145e-06,
      "loss": 0.4264,
      "step": 49565
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.676794173034663,
      "learning_rate": 2.510455912485389e-06,
      "loss": 0.371,
      "step": 49570
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.43474250570863,
      "learning_rate": 2.509388317506588e-06,
      "loss": 0.4064,
      "step": 49575
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.224012140003875,
      "learning_rate": 2.508320873526237e-06,
      "loss": 0.3953,
      "step": 49580
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.32758956943303,
      "learning_rate": 2.5072535806090542e-06,
      "loss": 0.4153,
      "step": 49585
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.560613241540756,
      "learning_rate": 2.506186438819742e-06,
      "loss": 0.3663,
      "step": 49590
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.9372993658854702,
      "learning_rate": 2.5051194482230003e-06,
      "loss": 0.3548,
      "step": 49595
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.158865974125944,
      "learning_rate": 2.5040526088835192e-06,
      "loss": 0.3692,
      "step": 49600
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.8617992983720666,
      "learning_rate": 2.5029859208659753e-06,
      "loss": 0.3936,
      "step": 49605
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.270589598654198,
      "learning_rate": 2.5019193842350387e-06,
      "loss": 0.394,
      "step": 49610
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.225232404923594,
      "learning_rate": 2.5008529990553743e-06,
      "loss": 0.3847,
      "step": 49615
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.218055627216,
      "learning_rate": 2.4997867653916286e-06,
      "loss": 0.3978,
      "step": 49620
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.2932021414246915,
      "learning_rate": 2.498720683308447e-06,
      "loss": 0.3676,
      "step": 49625
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.9182320791886864,
      "learning_rate": 2.4976547528704627e-06,
      "loss": 0.3799,
      "step": 49630
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.4105785979391126,
      "learning_rate": 2.4965889741423015e-06,
      "loss": 0.3755,
      "step": 49635
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.8279685439912345,
      "learning_rate": 2.4955233471885747e-06,
      "loss": 0.4094,
      "step": 49640
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.303876463827186,
      "learning_rate": 2.4944578720738906e-06,
      "loss": 0.3712,
      "step": 49645
    },
    {
      "epoch": 2.1,
      "grad_norm": 7.364460746140534,
      "learning_rate": 2.4933925488628473e-06,
      "loss": 0.3781,
      "step": 49650
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.492039106257755,
      "learning_rate": 2.492327377620028e-06,
      "loss": 0.3781,
      "step": 49655
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.224287065740411,
      "learning_rate": 2.4912623584100137e-06,
      "loss": 0.3828,
      "step": 49660
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.627652283633303,
      "learning_rate": 2.490197491297374e-06,
      "loss": 0.3707,
      "step": 49665
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.796643170941925,
      "learning_rate": 2.4891327763466668e-06,
      "loss": 0.3893,
      "step": 49670
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.38920195627582,
      "learning_rate": 2.4880682136224427e-06,
      "loss": 0.4081,
      "step": 49675
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.559810741139777,
      "learning_rate": 2.4870038031892437e-06,
      "loss": 0.3735,
      "step": 49680
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.470549831519377,
      "learning_rate": 2.4859395451116038e-06,
      "loss": 0.3817,
      "step": 49685
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.0109098537896255,
      "learning_rate": 2.4848754394540415e-06,
      "loss": 0.3935,
      "step": 49690
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.763284059541512,
      "learning_rate": 2.483811486281073e-06,
      "loss": 0.3634,
      "step": 49695
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.136611052150635,
      "learning_rate": 2.482747685657205e-06,
      "loss": 0.3724,
      "step": 49700
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.444272159245875,
      "learning_rate": 2.481684037646927e-06,
      "loss": 0.3839,
      "step": 49705
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.449399016862127,
      "learning_rate": 2.4806205423147284e-06,
      "loss": 0.368,
      "step": 49710
    },
    {
      "epoch": 2.1,
      "grad_norm": 5.908726942079969,
      "learning_rate": 2.479557199725087e-06,
      "loss": 0.393,
      "step": 49715
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.240950711045852,
      "learning_rate": 2.478494009942467e-06,
      "loss": 0.3554,
      "step": 49720
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.0333119426723885,
      "learning_rate": 2.4774309730313294e-06,
      "loss": 0.3765,
      "step": 49725
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.146407327590279,
      "learning_rate": 2.4763680890561196e-06,
      "loss": 0.4146,
      "step": 49730
    },
    {
      "epoch": 2.1,
      "grad_norm": 3.8480734753855494,
      "learning_rate": 2.475305358081281e-06,
      "loss": 0.3645,
      "step": 49735
    },
    {
      "epoch": 2.11,
      "grad_norm": 6.657829288946752,
      "learning_rate": 2.4742427801712393e-06,
      "loss": 0.3883,
      "step": 49740
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.131902751353832,
      "learning_rate": 2.473180355390418e-06,
      "loss": 0.4095,
      "step": 49745
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.147268937645226,
      "learning_rate": 2.4721180838032296e-06,
      "loss": 0.3763,
      "step": 49750
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.084015032056912,
      "learning_rate": 2.4710559654740773e-06,
      "loss": 0.3596,
      "step": 49755
    },
    {
      "epoch": 2.11,
      "grad_norm": 6.080179734460404,
      "learning_rate": 2.46999400046735e-06,
      "loss": 0.3781,
      "step": 49760
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.020642133077635,
      "learning_rate": 2.468932188847435e-06,
      "loss": 0.4006,
      "step": 49765
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.7836456282610076,
      "learning_rate": 2.467870530678707e-06,
      "loss": 0.3662,
      "step": 49770
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.271451236352591,
      "learning_rate": 2.4668090260255285e-06,
      "loss": 0.3862,
      "step": 49775
    },
    {
      "epoch": 2.11,
      "grad_norm": 6.404612159231934,
      "learning_rate": 2.4657476749522568e-06,
      "loss": 0.3779,
      "step": 49780
    },
    {
      "epoch": 2.11,
      "grad_norm": 7.126695836322258,
      "learning_rate": 2.464686477523241e-06,
      "loss": 0.4025,
      "step": 49785
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.2224258109326485,
      "learning_rate": 2.463625433802814e-06,
      "loss": 0.3666,
      "step": 49790
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.71235947766091,
      "learning_rate": 2.4625645438553057e-06,
      "loss": 0.3843,
      "step": 49795
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.7269993203341003,
      "learning_rate": 2.461503807745035e-06,
      "loss": 0.3565,
      "step": 49800
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.586395355051144,
      "learning_rate": 2.4604432255363126e-06,
      "loss": 0.3468,
      "step": 49805
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.763742919086334,
      "learning_rate": 2.4593827972934354e-06,
      "loss": 0.3985,
      "step": 49810
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.631216144703485,
      "learning_rate": 2.4583225230806955e-06,
      "loss": 0.3819,
      "step": 49815
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.316063192548243,
      "learning_rate": 2.457262402962376e-06,
      "loss": 0.3811,
      "step": 49820
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.080869557647586,
      "learning_rate": 2.4562024370027454e-06,
      "loss": 0.3716,
      "step": 49825
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.8456964305603085,
      "learning_rate": 2.4551426252660682e-06,
      "loss": 0.3696,
      "step": 49830
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.268892212304196,
      "learning_rate": 2.454082967816599e-06,
      "loss": 0.3847,
      "step": 49835
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.8704386788553897,
      "learning_rate": 2.4530234647185788e-06,
      "loss": 0.3505,
      "step": 49840
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.32426826315786,
      "learning_rate": 2.4519641160362427e-06,
      "loss": 0.3984,
      "step": 49845
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.027185671168084,
      "learning_rate": 2.4509049218338193e-06,
      "loss": 0.3804,
      "step": 49850
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.217048525038172,
      "learning_rate": 2.44984588217552e-06,
      "loss": 0.3965,
      "step": 49855
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.13185634949303,
      "learning_rate": 2.448786997125553e-06,
      "loss": 0.3649,
      "step": 49860
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.9757673531192483,
      "learning_rate": 2.4477282667481157e-06,
      "loss": 0.3882,
      "step": 49865
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.46467420358684,
      "learning_rate": 2.446669691107397e-06,
      "loss": 0.3924,
      "step": 49870
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.2558098145137375,
      "learning_rate": 2.4456112702675726e-06,
      "loss": 0.4079,
      "step": 49875
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.675542984130239,
      "learning_rate": 2.444553004292812e-06,
      "loss": 0.3686,
      "step": 49880
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.974364702495152,
      "learning_rate": 2.443494893247278e-06,
      "loss": 0.3768,
      "step": 49885
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.569564416705734,
      "learning_rate": 2.442436937195118e-06,
      "loss": 0.402,
      "step": 49890
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.9497043939560132,
      "learning_rate": 2.4413791362004713e-06,
      "loss": 0.3743,
      "step": 49895
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.249458144394923,
      "learning_rate": 2.44032149032747e-06,
      "loss": 0.3724,
      "step": 49900
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.162657587320998,
      "learning_rate": 2.43926399964024e-06,
      "loss": 0.3865,
      "step": 49905
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.017711903993835,
      "learning_rate": 2.4382066642028884e-06,
      "loss": 0.3867,
      "step": 49910
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.97519195441277,
      "learning_rate": 2.4371494840795206e-06,
      "loss": 0.3525,
      "step": 49915
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.391077735217333,
      "learning_rate": 2.436092459334232e-06,
      "loss": 0.3822,
      "step": 49920
    },
    {
      "epoch": 2.11,
      "grad_norm": 6.260823648024386,
      "learning_rate": 2.4350355900311044e-06,
      "loss": 0.3691,
      "step": 49925
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.771614016617556,
      "learning_rate": 2.433978876234213e-06,
      "loss": 0.3525,
      "step": 49930
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.895896132051187,
      "learning_rate": 2.432922318007624e-06,
      "loss": 0.378,
      "step": 49935
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.771203482913908,
      "learning_rate": 2.431865915415395e-06,
      "loss": 0.3774,
      "step": 49940
    },
    {
      "epoch": 2.11,
      "grad_norm": 3.532206117323641,
      "learning_rate": 2.4308096685215694e-06,
      "loss": 0.3684,
      "step": 49945
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.318721255751485,
      "learning_rate": 2.4297535773901854e-06,
      "loss": 0.3712,
      "step": 49950
    },
    {
      "epoch": 2.11,
      "grad_norm": 8.96153361146327,
      "learning_rate": 2.428697642085273e-06,
      "loss": 0.3716,
      "step": 49955
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.444379912833598,
      "learning_rate": 2.427641862670847e-06,
      "loss": 0.3775,
      "step": 49960
    },
    {
      "epoch": 2.11,
      "grad_norm": 5.428909623541166,
      "learning_rate": 2.4265862392109175e-06,
      "loss": 0.3653,
      "step": 49965
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.3561449934485825,
      "learning_rate": 2.425530771769486e-06,
      "loss": 0.3805,
      "step": 49970
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.883735944730505,
      "learning_rate": 2.4244754604105384e-06,
      "loss": 0.3868,
      "step": 49975
    },
    {
      "epoch": 2.12,
      "grad_norm": 6.384482524497086,
      "learning_rate": 2.4234203051980577e-06,
      "loss": 0.3926,
      "step": 49980
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.225179001660537,
      "learning_rate": 2.422365306196014e-06,
      "loss": 0.3626,
      "step": 49985
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.162008239120738,
      "learning_rate": 2.421310463468371e-06,
      "loss": 0.3775,
      "step": 49990
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.16745171513817,
      "learning_rate": 2.420255777079077e-06,
      "loss": 0.3818,
      "step": 49995
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.503633308041489,
      "learning_rate": 2.4192012470920763e-06,
      "loss": 0.3794,
      "step": 50000
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.046081256510451,
      "learning_rate": 2.4181468735713037e-06,
      "loss": 0.3868,
      "step": 50005
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.6688619739946144,
      "learning_rate": 2.417092656580679e-06,
      "loss": 0.3847,
      "step": 50010
    },
    {
      "epoch": 2.12,
      "grad_norm": 7.7663393019947735,
      "learning_rate": 2.4160385961841188e-06,
      "loss": 0.3732,
      "step": 50015
    },
    {
      "epoch": 2.12,
      "grad_norm": 7.877827143175252,
      "learning_rate": 2.414984692445529e-06,
      "loss": 0.3826,
      "step": 50020
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.211286575101403,
      "learning_rate": 2.4139309454288013e-06,
      "loss": 0.3855,
      "step": 50025
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.9243580697673135,
      "learning_rate": 2.4128773551978224e-06,
      "loss": 0.3993,
      "step": 50030
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.9718384765724535,
      "learning_rate": 2.4118239218164684e-06,
      "loss": 0.3924,
      "step": 50035
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.039118369219979,
      "learning_rate": 2.410770645348609e-06,
      "loss": 0.3518,
      "step": 50040
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.082315817751445,
      "learning_rate": 2.4097175258580964e-06,
      "loss": 0.3622,
      "step": 50045
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.146929517895016,
      "learning_rate": 2.4086645634087826e-06,
      "loss": 0.4051,
      "step": 50050
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.845863747071786,
      "learning_rate": 2.4076117580645004e-06,
      "loss": 0.3896,
      "step": 50055
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.354437884313686,
      "learning_rate": 2.406559109889084e-06,
      "loss": 0.3772,
      "step": 50060
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.94783565526839,
      "learning_rate": 2.4055066189463473e-06,
      "loss": 0.376,
      "step": 50065
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.8024253464910562,
      "learning_rate": 2.4044542853001018e-06,
      "loss": 0.3586,
      "step": 50070
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.892252759936716,
      "learning_rate": 2.4034021090141498e-06,
      "loss": 0.3747,
      "step": 50075
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.301806944445829,
      "learning_rate": 2.4023500901522767e-06,
      "loss": 0.3639,
      "step": 50080
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.2826572367742495,
      "learning_rate": 2.4012982287782667e-06,
      "loss": 0.3558,
      "step": 50085
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.297069211754986,
      "learning_rate": 2.400246524955892e-06,
      "loss": 0.408,
      "step": 50090
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.075429312802378,
      "learning_rate": 2.399194978748911e-06,
      "loss": 0.3618,
      "step": 50095
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.141510757619812,
      "learning_rate": 2.3981435902210776e-06,
      "loss": 0.404,
      "step": 50100
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.4452110645637335,
      "learning_rate": 2.397092359436134e-06,
      "loss": 0.3802,
      "step": 50105
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.219697315086441,
      "learning_rate": 2.396041286457816e-06,
      "loss": 0.3735,
      "step": 50110
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.9661014527545277,
      "learning_rate": 2.394990371349842e-06,
      "loss": 0.3714,
      "step": 50115
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.6277987578383777,
      "learning_rate": 2.3939396141759287e-06,
      "loss": 0.3893,
      "step": 50120
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.167748381548549,
      "learning_rate": 2.392889014999782e-06,
      "loss": 0.3742,
      "step": 50125
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.597812997401016,
      "learning_rate": 2.391838573885093e-06,
      "loss": 0.3837,
      "step": 50130
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.461404742437002,
      "learning_rate": 2.3907882908955486e-06,
      "loss": 0.3763,
      "step": 50135
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.692538058787547,
      "learning_rate": 2.3897381660948265e-06,
      "loss": 0.3839,
      "step": 50140
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.369217026815814,
      "learning_rate": 2.388688199546589e-06,
      "loss": 0.3962,
      "step": 50145
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.089548058217305,
      "learning_rate": 2.3876383913144934e-06,
      "loss": 0.3924,
      "step": 50150
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.172585048318543,
      "learning_rate": 2.3865887414621897e-06,
      "loss": 0.407,
      "step": 50155
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.110677182727945,
      "learning_rate": 2.3855392500533103e-06,
      "loss": 0.3696,
      "step": 50160
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.427670251872314,
      "learning_rate": 2.384489917151485e-06,
      "loss": 0.3987,
      "step": 50165
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.506669351712094,
      "learning_rate": 2.383440742820332e-06,
      "loss": 0.4063,
      "step": 50170
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.569927951381089,
      "learning_rate": 2.3823917271234614e-06,
      "loss": 0.3862,
      "step": 50175
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.400852669837989,
      "learning_rate": 2.381342870124468e-06,
      "loss": 0.3584,
      "step": 50180
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.193714869002925,
      "learning_rate": 2.3802941718869426e-06,
      "loss": 0.3441,
      "step": 50185
    },
    {
      "epoch": 2.12,
      "grad_norm": 6.03284537146811,
      "learning_rate": 2.3792456324744667e-06,
      "loss": 0.375,
      "step": 50190
    },
    {
      "epoch": 2.12,
      "grad_norm": 6.061948575366682,
      "learning_rate": 2.378197251950607e-06,
      "loss": 0.3722,
      "step": 50195
    },
    {
      "epoch": 2.12,
      "grad_norm": 5.704703005457088,
      "learning_rate": 2.377149030378925e-06,
      "loss": 0.354,
      "step": 50200
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.715195587032702,
      "learning_rate": 2.3761009678229736e-06,
      "loss": 0.3666,
      "step": 50205
    },
    {
      "epoch": 2.12,
      "grad_norm": 4.235779860225067,
      "learning_rate": 2.375053064346291e-06,
      "loss": 0.3873,
      "step": 50210
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.2137413878824574,
      "learning_rate": 2.3740053200124076e-06,
      "loss": 0.3672,
      "step": 50215
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.5808093551978804,
      "learning_rate": 2.3729577348848464e-06,
      "loss": 0.3783,
      "step": 50220
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.452500082790417,
      "learning_rate": 2.3719103090271213e-06,
      "loss": 0.3502,
      "step": 50225
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.1752993322630045,
      "learning_rate": 2.3708630425027314e-06,
      "loss": 0.3686,
      "step": 50230
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.861356559290214,
      "learning_rate": 2.3698159353751704e-06,
      "loss": 0.3771,
      "step": 50235
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.556076181390272,
      "learning_rate": 2.3687689877079217e-06,
      "loss": 0.3767,
      "step": 50240
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.476052008178158,
      "learning_rate": 2.367722199564461e-06,
      "loss": 0.366,
      "step": 50245
    },
    {
      "epoch": 2.13,
      "grad_norm": 7.236177133365301,
      "learning_rate": 2.3666755710082473e-06,
      "loss": 0.3688,
      "step": 50250
    },
    {
      "epoch": 2.13,
      "grad_norm": 8.376589189582377,
      "learning_rate": 2.3656291021027376e-06,
      "loss": 0.4004,
      "step": 50255
    },
    {
      "epoch": 2.13,
      "grad_norm": 6.20965374969848,
      "learning_rate": 2.3645827929113774e-06,
      "loss": 0.3895,
      "step": 50260
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.001237307254817,
      "learning_rate": 2.3635366434975975e-06,
      "loss": 0.3271,
      "step": 50265
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.562536140586596,
      "learning_rate": 2.362490653924825e-06,
      "loss": 0.3966,
      "step": 50270
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.593750425558845,
      "learning_rate": 2.361444824256477e-06,
      "loss": 0.3866,
      "step": 50275
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.133215604047289,
      "learning_rate": 2.3603991545559556e-06,
      "loss": 0.3795,
      "step": 50280
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.653949473834183,
      "learning_rate": 2.3593536448866578e-06,
      "loss": 0.3706,
      "step": 50285
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.157975445181617,
      "learning_rate": 2.3583082953119708e-06,
      "loss": 0.3726,
      "step": 50290
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.042881334685599,
      "learning_rate": 2.3572631058952716e-06,
      "loss": 0.3481,
      "step": 50295
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.167037811031089,
      "learning_rate": 2.356218076699925e-06,
      "loss": 0.3775,
      "step": 50300
    },
    {
      "epoch": 2.13,
      "grad_norm": 7.668433576834251,
      "learning_rate": 2.355173207789288e-06,
      "loss": 0.3797,
      "step": 50305
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.896701667204212,
      "learning_rate": 2.354128499226711e-06,
      "loss": 0.351,
      "step": 50310
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.668711073387136,
      "learning_rate": 2.3530839510755277e-06,
      "loss": 0.35,
      "step": 50315
    },
    {
      "epoch": 2.13,
      "grad_norm": 6.853118180994254,
      "learning_rate": 2.3520395633990674e-06,
      "loss": 0.3861,
      "step": 50320
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.320294182639153,
      "learning_rate": 2.3509953362606503e-06,
      "loss": 0.3755,
      "step": 50325
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.573031600142931,
      "learning_rate": 2.3499512697235814e-06,
      "loss": 0.3953,
      "step": 50330
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.212673635948971,
      "learning_rate": 2.348907363851161e-06,
      "loss": 0.3511,
      "step": 50335
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.067905488706705,
      "learning_rate": 2.3478636187066783e-06,
      "loss": 0.3819,
      "step": 50340
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.609118523630749,
      "learning_rate": 2.346820034353414e-06,
      "loss": 0.3591,
      "step": 50345
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.779343776010606,
      "learning_rate": 2.3457766108546343e-06,
      "loss": 0.4008,
      "step": 50350
    },
    {
      "epoch": 2.13,
      "grad_norm": 6.120065677225579,
      "learning_rate": 2.3447333482736e-06,
      "loss": 0.3653,
      "step": 50355
    },
    {
      "epoch": 2.13,
      "grad_norm": 12.66347312711895,
      "learning_rate": 2.3436902466735632e-06,
      "loss": 0.3893,
      "step": 50360
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.9585936718256085,
      "learning_rate": 2.3426473061177606e-06,
      "loss": 0.3819,
      "step": 50365
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.278607432923751,
      "learning_rate": 2.3416045266694265e-06,
      "loss": 0.4004,
      "step": 50370
    },
    {
      "epoch": 2.13,
      "grad_norm": 5.233860298690419,
      "learning_rate": 2.3405619083917773e-06,
      "loss": 0.4086,
      "step": 50375
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.044596122486876,
      "learning_rate": 2.3395194513480278e-06,
      "loss": 0.3893,
      "step": 50380
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.3305520126800126,
      "learning_rate": 2.338477155601375e-06,
      "loss": 0.4049,
      "step": 50385
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.292684482287587,
      "learning_rate": 2.337435021215013e-06,
      "loss": 0.4026,
      "step": 50390
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.9413607865978926,
      "learning_rate": 2.3363930482521253e-06,
      "loss": 0.3502,
      "step": 50395
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.657990302429634,
      "learning_rate": 2.3353512367758786e-06,
      "loss": 0.366,
      "step": 50400
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.1175086497455995,
      "learning_rate": 2.334309586849437e-06,
      "loss": 0.3646,
      "step": 50405
    },
    {
      "epoch": 2.13,
      "grad_norm": 7.08414716030964,
      "learning_rate": 2.3332680985359557e-06,
      "loss": 0.3891,
      "step": 50410
    },
    {
      "epoch": 2.13,
      "grad_norm": 3.901863339868657,
      "learning_rate": 2.332226771898572e-06,
      "loss": 0.3711,
      "step": 50415
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.163917329710865,
      "learning_rate": 2.331185607000421e-06,
      "loss": 0.3752,
      "step": 50420
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.435059414610699,
      "learning_rate": 2.330144603904626e-06,
      "loss": 0.3592,
      "step": 50425
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.042912333737766,
      "learning_rate": 2.3291037626743006e-06,
      "loss": 0.4071,
      "step": 50430
    },
    {
      "epoch": 2.13,
      "grad_norm": 6.695729774111038,
      "learning_rate": 2.3280630833725452e-06,
      "loss": 0.3786,
      "step": 50435
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.693785131950566,
      "learning_rate": 2.3270225660624545e-06,
      "loss": 0.3792,
      "step": 50440
    },
    {
      "epoch": 2.13,
      "grad_norm": 4.591554681300314,
      "learning_rate": 2.325982210807114e-06,
      "loss": 0.3509,
      "step": 50445
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.269179696655977,
      "learning_rate": 2.3249420176695938e-06,
      "loss": 0.3836,
      "step": 50450
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.908920454745094,
      "learning_rate": 2.3239019867129588e-06,
      "loss": 0.4186,
      "step": 50455
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.286860635930135,
      "learning_rate": 2.322862118000266e-06,
      "loss": 0.4171,
      "step": 50460
    },
    {
      "epoch": 2.14,
      "grad_norm": 5.299826387909762,
      "learning_rate": 2.3218224115945553e-06,
      "loss": 0.3799,
      "step": 50465
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.282772459292189,
      "learning_rate": 2.3207828675588627e-06,
      "loss": 0.3711,
      "step": 50470
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.9776379064244893,
      "learning_rate": 2.3197434859562135e-06,
      "loss": 0.366,
      "step": 50475
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.846682034736626,
      "learning_rate": 2.318704266849623e-06,
      "loss": 0.3854,
      "step": 50480
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.700855002782642,
      "learning_rate": 2.317665210302093e-06,
      "loss": 0.3739,
      "step": 50485
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.3009172930782515,
      "learning_rate": 2.3166263163766206e-06,
      "loss": 0.3961,
      "step": 50490
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.079679362055096,
      "learning_rate": 2.3155875851361924e-06,
      "loss": 0.3788,
      "step": 50495
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.275290053587404,
      "learning_rate": 2.3145490166437793e-06,
      "loss": 0.3711,
      "step": 50500
    },
    {
      "epoch": 2.14,
      "grad_norm": 5.038353232740384,
      "learning_rate": 2.3135106109623494e-06,
      "loss": 0.3698,
      "step": 50505
    },
    {
      "epoch": 2.14,
      "grad_norm": 7.3784742231605875,
      "learning_rate": 2.31247236815486e-06,
      "loss": 0.38,
      "step": 50510
    },
    {
      "epoch": 2.14,
      "grad_norm": 8.677001944116634,
      "learning_rate": 2.3114342882842523e-06,
      "loss": 0.4145,
      "step": 50515
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.423570557876477,
      "learning_rate": 2.3103963714134647e-06,
      "loss": 0.3844,
      "step": 50520
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.983621066241579,
      "learning_rate": 2.3093586176054223e-06,
      "loss": 0.3792,
      "step": 50525
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.064275571815297,
      "learning_rate": 2.3083210269230434e-06,
      "loss": 0.3717,
      "step": 50530
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.699126525000278,
      "learning_rate": 2.3072835994292312e-06,
      "loss": 0.3931,
      "step": 50535
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.006200984998097,
      "learning_rate": 2.306246335186882e-06,
      "loss": 0.383,
      "step": 50540
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.774542194307398,
      "learning_rate": 2.3052092342588823e-06,
      "loss": 0.3783,
      "step": 50545
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.5839880311907875,
      "learning_rate": 2.304172296708111e-06,
      "loss": 0.3876,
      "step": 50550
    },
    {
      "epoch": 2.14,
      "grad_norm": 6.259050357289544,
      "learning_rate": 2.30313552259743e-06,
      "loss": 0.3884,
      "step": 50555
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.976538532929298,
      "learning_rate": 2.302098911989699e-06,
      "loss": 0.3681,
      "step": 50560
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.345438457859317,
      "learning_rate": 2.301062464947766e-06,
      "loss": 0.3793,
      "step": 50565
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.9192401196155164,
      "learning_rate": 2.300026181534464e-06,
      "loss": 0.3724,
      "step": 50570
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.092022787303198,
      "learning_rate": 2.298990061812621e-06,
      "loss": 0.3669,
      "step": 50575
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.9344946740917663,
      "learning_rate": 2.2979541058450565e-06,
      "loss": 0.38,
      "step": 50580
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.9973193778708347,
      "learning_rate": 2.296918313694574e-06,
      "loss": 0.3936,
      "step": 50585
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.208261080433807,
      "learning_rate": 2.2958826854239714e-06,
      "loss": 0.3653,
      "step": 50590
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.9184189305912267,
      "learning_rate": 2.2948472210960367e-06,
      "loss": 0.3411,
      "step": 50595
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.52101815736275,
      "learning_rate": 2.293811920773548e-06,
      "loss": 0.3773,
      "step": 50600
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.0849819119092645,
      "learning_rate": 2.29277678451927e-06,
      "loss": 0.3729,
      "step": 50605
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.506825155390957,
      "learning_rate": 2.2917418123959604e-06,
      "loss": 0.3888,
      "step": 50610
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.504393941969743,
      "learning_rate": 2.29070700446637e-06,
      "loss": 0.4016,
      "step": 50615
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.0558706310688555,
      "learning_rate": 2.2896723607932315e-06,
      "loss": 0.3742,
      "step": 50620
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.465244562518348,
      "learning_rate": 2.288637881439274e-06,
      "loss": 0.3708,
      "step": 50625
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.634756276872223,
      "learning_rate": 2.287603566467217e-06,
      "loss": 0.3962,
      "step": 50630
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.1301317656661025,
      "learning_rate": 2.286569415939764e-06,
      "loss": 0.371,
      "step": 50635
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.582733968358318,
      "learning_rate": 2.285535429919615e-06,
      "loss": 0.3978,
      "step": 50640
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.6912932808825785,
      "learning_rate": 2.2845016084694572e-06,
      "loss": 0.3909,
      "step": 50645
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.3816800423906015,
      "learning_rate": 2.28346795165197e-06,
      "loss": 0.4069,
      "step": 50650
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.113223028788747,
      "learning_rate": 2.282434459529817e-06,
      "loss": 0.3459,
      "step": 50655
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.434120460227427,
      "learning_rate": 2.281401132165658e-06,
      "loss": 0.3679,
      "step": 50660
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.066238234409624,
      "learning_rate": 2.280367969622142e-06,
      "loss": 0.3603,
      "step": 50665
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.028461449172315,
      "learning_rate": 2.2793349719619036e-06,
      "loss": 0.3916,
      "step": 50670
    },
    {
      "epoch": 2.14,
      "grad_norm": 3.757003421187071,
      "learning_rate": 2.2783021392475717e-06,
      "loss": 0.3617,
      "step": 50675
    },
    {
      "epoch": 2.14,
      "grad_norm": 4.070517854659836,
      "learning_rate": 2.2772694715417658e-06,
      "loss": 0.4051,
      "step": 50680
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.311564882763757,
      "learning_rate": 2.2762369689070903e-06,
      "loss": 0.367,
      "step": 50685
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.3411079329766435,
      "learning_rate": 2.2752046314061436e-06,
      "loss": 0.3862,
      "step": 50690
    },
    {
      "epoch": 2.15,
      "grad_norm": 3.6180860993451187,
      "learning_rate": 2.274172459101517e-06,
      "loss": 0.4064,
      "step": 50695
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.726395907339224,
      "learning_rate": 2.2731404520557836e-06,
      "loss": 0.3684,
      "step": 50700
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.465517734535609,
      "learning_rate": 2.2721086103315117e-06,
      "loss": 0.4041,
      "step": 50705
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.149715113016545,
      "learning_rate": 2.2710769339912588e-06,
      "loss": 0.353,
      "step": 50710
    },
    {
      "epoch": 2.15,
      "grad_norm": 3.901392965811322,
      "learning_rate": 2.270045423097575e-06,
      "loss": 0.3764,
      "step": 50715
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.447604627125826,
      "learning_rate": 2.269014077712994e-06,
      "loss": 0.4183,
      "step": 50720
    },
    {
      "epoch": 2.15,
      "grad_norm": 3.8174424379871974,
      "learning_rate": 2.2679828979000456e-06,
      "loss": 0.3595,
      "step": 50725
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.462026469959688,
      "learning_rate": 2.266951883721246e-06,
      "loss": 0.385,
      "step": 50730
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.180264338912058,
      "learning_rate": 2.2659210352391055e-06,
      "loss": 0.378,
      "step": 50735
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.881279386411326,
      "learning_rate": 2.264890352516117e-06,
      "loss": 0.372,
      "step": 50740
    },
    {
      "epoch": 2.15,
      "grad_norm": 3.8155366830853468,
      "learning_rate": 2.26385983561477e-06,
      "loss": 0.3555,
      "step": 50745
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.734346289277913,
      "learning_rate": 2.262829484597544e-06,
      "loss": 0.3884,
      "step": 50750
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.043472371146545,
      "learning_rate": 2.2617992995269013e-06,
      "loss": 0.3557,
      "step": 50755
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.119800570632844,
      "learning_rate": 2.2607692804653014e-06,
      "loss": 0.3768,
      "step": 50760
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.601288320000802,
      "learning_rate": 2.2597394274751932e-06,
      "loss": 0.3474,
      "step": 50765
    },
    {
      "epoch": 2.15,
      "grad_norm": 7.678334891168863,
      "learning_rate": 2.2587097406190102e-06,
      "loss": 0.3673,
      "step": 50770
    },
    {
      "epoch": 2.15,
      "grad_norm": 7.976398426379458,
      "learning_rate": 2.257680219959181e-06,
      "loss": 0.3924,
      "step": 50775
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.53633594368111,
      "learning_rate": 2.256650865558122e-06,
      "loss": 0.3677,
      "step": 50780
    },
    {
      "epoch": 2.15,
      "grad_norm": 3.757943587039481,
      "learning_rate": 2.2556216774782427e-06,
      "loss": 0.3646,
      "step": 50785
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.809714445900774,
      "learning_rate": 2.2545926557819344e-06,
      "loss": 0.3599,
      "step": 50790
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.583126285015245,
      "learning_rate": 2.2535638005315867e-06,
      "loss": 0.3843,
      "step": 50795
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.017375823145854,
      "learning_rate": 2.2525351117895783e-06,
      "loss": 0.3919,
      "step": 50800
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.046980258206454,
      "learning_rate": 2.251506589618271e-06,
      "loss": 0.3607,
      "step": 50805
    },
    {
      "epoch": 2.15,
      "grad_norm": 7.1304231819622155,
      "learning_rate": 2.2504782340800234e-06,
      "loss": 0.3802,
      "step": 50810
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.543585660689358,
      "learning_rate": 2.249450045237183e-06,
      "loss": 0.3627,
      "step": 50815
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.948765733882174,
      "learning_rate": 2.2484220231520825e-06,
      "loss": 0.3711,
      "step": 50820
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.788088857808289,
      "learning_rate": 2.24739416788705e-06,
      "loss": 0.3746,
      "step": 50825
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.339496384859222,
      "learning_rate": 2.246366479504401e-06,
      "loss": 0.3833,
      "step": 50830
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.135418584638771,
      "learning_rate": 2.245338958066443e-06,
      "loss": 0.3586,
      "step": 50835
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.430434323466136,
      "learning_rate": 2.244311603635468e-06,
      "loss": 0.3588,
      "step": 50840
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.216925922270115,
      "learning_rate": 2.243284416273763e-06,
      "loss": 0.376,
      "step": 50845
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.076710326214254,
      "learning_rate": 2.2422573960436063e-06,
      "loss": 0.3867,
      "step": 50850
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.498541750925796,
      "learning_rate": 2.24123054300726e-06,
      "loss": 0.3587,
      "step": 50855
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.3990030999973095,
      "learning_rate": 2.240203857226978e-06,
      "loss": 0.3843,
      "step": 50860
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.20784503874404,
      "learning_rate": 2.239177338765007e-06,
      "loss": 0.3928,
      "step": 50865
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.9201851433335735,
      "learning_rate": 2.238150987683583e-06,
      "loss": 0.4107,
      "step": 50870
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.360094697411207,
      "learning_rate": 2.237124804044928e-06,
      "loss": 0.3517,
      "step": 50875
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.044257098536264,
      "learning_rate": 2.236098787911257e-06,
      "loss": 0.3773,
      "step": 50880
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.034921453634541,
      "learning_rate": 2.2350729393447778e-06,
      "loss": 0.349,
      "step": 50885
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.6596562738585545,
      "learning_rate": 2.23404725840768e-06,
      "loss": 0.3859,
      "step": 50890
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.3763265822479225,
      "learning_rate": 2.233021745162149e-06,
      "loss": 0.4168,
      "step": 50895
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.018084908790607,
      "learning_rate": 2.23199639967036e-06,
      "loss": 0.357,
      "step": 50900
    },
    {
      "epoch": 2.15,
      "grad_norm": 5.034819798944364,
      "learning_rate": 2.230971221994477e-06,
      "loss": 0.3469,
      "step": 50905
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.575473937712187,
      "learning_rate": 2.229946212196651e-06,
      "loss": 0.362,
      "step": 50910
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.347200211362484,
      "learning_rate": 2.228921370339026e-06,
      "loss": 0.364,
      "step": 50915
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.121996669574978,
      "learning_rate": 2.2278966964837386e-06,
      "loss": 0.3688,
      "step": 50920
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.0957359069168495,
      "learning_rate": 2.2268721906929063e-06,
      "loss": 0.38,
      "step": 50925
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.775776662087176,
      "learning_rate": 2.225847853028645e-06,
      "loss": 0.358,
      "step": 50930
    },
    {
      "epoch": 2.16,
      "grad_norm": 6.002876685241673,
      "learning_rate": 2.224823683553059e-06,
      "loss": 0.3885,
      "step": 50935
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.82532733256446,
      "learning_rate": 2.2237996823282364e-06,
      "loss": 0.4085,
      "step": 50940
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.061653621125829,
      "learning_rate": 2.2227758494162617e-06,
      "loss": 0.3669,
      "step": 50945
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.324036788469857,
      "learning_rate": 2.221752184879206e-06,
      "loss": 0.3667,
      "step": 50950
    },
    {
      "epoch": 2.16,
      "grad_norm": 6.86716899218463,
      "learning_rate": 2.220728688779134e-06,
      "loss": 0.3756,
      "step": 50955
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.573622293763051,
      "learning_rate": 2.2197053611780933e-06,
      "loss": 0.3425,
      "step": 50960
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.3927309024147405,
      "learning_rate": 2.2186822021381266e-06,
      "loss": 0.3745,
      "step": 50965
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.738554777109576,
      "learning_rate": 2.2176592117212675e-06,
      "loss": 0.3652,
      "step": 50970
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.131962176264198,
      "learning_rate": 2.2166363899895325e-06,
      "loss": 0.3824,
      "step": 50975
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.606439973467614,
      "learning_rate": 2.2156137370049346e-06,
      "loss": 0.3589,
      "step": 50980
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.69896633364378,
      "learning_rate": 2.214591252829476e-06,
      "loss": 0.3777,
      "step": 50985
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.235984561991574,
      "learning_rate": 2.2135689375251436e-06,
      "loss": 0.369,
      "step": 50990
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.812624575996112,
      "learning_rate": 2.212546791153918e-06,
      "loss": 0.3656,
      "step": 50995
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.88585961869959,
      "learning_rate": 2.211524813777773e-06,
      "loss": 0.3738,
      "step": 51000
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.921244715278705,
      "learning_rate": 2.2105030054586616e-06,
      "loss": 0.3916,
      "step": 51005
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.2388940789529626,
      "learning_rate": 2.2094813662585366e-06,
      "loss": 0.3536,
      "step": 51010
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.4243669129879954,
      "learning_rate": 2.208459896239337e-06,
      "loss": 0.3792,
      "step": 51015
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.655373553583169,
      "learning_rate": 2.2074385954629947e-06,
      "loss": 0.355,
      "step": 51020
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.8428372723257587,
      "learning_rate": 2.206417463991421e-06,
      "loss": 0.3686,
      "step": 51025
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.845743450724909,
      "learning_rate": 2.2053965018865276e-06,
      "loss": 0.3832,
      "step": 51030
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.0654617722646,
      "learning_rate": 2.2043757092102123e-06,
      "loss": 0.385,
      "step": 51035
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.408820900111711,
      "learning_rate": 2.2033550860243648e-06,
      "loss": 0.3547,
      "step": 51040
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.8739037114327792,
      "learning_rate": 2.202334632390859e-06,
      "loss": 0.3475,
      "step": 51045
    },
    {
      "epoch": 2.16,
      "grad_norm": 7.678318964065676,
      "learning_rate": 2.2013143483715637e-06,
      "loss": 0.3682,
      "step": 51050
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.594879435767553,
      "learning_rate": 2.2002942340283375e-06,
      "loss": 0.3845,
      "step": 51055
    },
    {
      "epoch": 2.16,
      "grad_norm": 6.884627620035937,
      "learning_rate": 2.1992742894230227e-06,
      "loss": 0.3879,
      "step": 51060
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.202578743256359,
      "learning_rate": 2.1982545146174584e-06,
      "loss": 0.3713,
      "step": 51065
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.664514550360664,
      "learning_rate": 2.197234909673472e-06,
      "loss": 0.3505,
      "step": 51070
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.122267390009706,
      "learning_rate": 2.196215474652875e-06,
      "loss": 0.3818,
      "step": 51075
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.992298514568472,
      "learning_rate": 2.1951962096174755e-06,
      "loss": 0.3745,
      "step": 51080
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.122598217497837,
      "learning_rate": 2.1941771146290675e-06,
      "loss": 0.3715,
      "step": 51085
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.4436339058779866,
      "learning_rate": 2.1931581897494385e-06,
      "loss": 0.3582,
      "step": 51090
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.039835660364556,
      "learning_rate": 2.1921394350403587e-06,
      "loss": 0.3436,
      "step": 51095
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.402455971415569,
      "learning_rate": 2.191120850563594e-06,
      "loss": 0.3904,
      "step": 51100
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.0670076649141595,
      "learning_rate": 2.1901024363808997e-06,
      "loss": 0.3579,
      "step": 51105
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.9168415397135177,
      "learning_rate": 2.1890841925540165e-06,
      "loss": 0.3836,
      "step": 51110
    },
    {
      "epoch": 2.16,
      "grad_norm": 5.204152111030827,
      "learning_rate": 2.188066119144679e-06,
      "loss": 0.3855,
      "step": 51115
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.369177487586138,
      "learning_rate": 2.187048216214611e-06,
      "loss": 0.3774,
      "step": 51120
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.9763553558216485,
      "learning_rate": 2.186030483825523e-06,
      "loss": 0.3592,
      "step": 51125
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.507869513798007,
      "learning_rate": 2.1850129220391176e-06,
      "loss": 0.3745,
      "step": 51130
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.088523272742548,
      "learning_rate": 2.1839955309170862e-06,
      "loss": 0.3691,
      "step": 51135
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.197804250986726,
      "learning_rate": 2.1829783105211137e-06,
      "loss": 0.3611,
      "step": 51140
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.114975302622314,
      "learning_rate": 2.181961260912866e-06,
      "loss": 0.3696,
      "step": 51145
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.20858623205499,
      "learning_rate": 2.1809443821540067e-06,
      "loss": 0.3618,
      "step": 51150
    },
    {
      "epoch": 2.16,
      "grad_norm": 4.78329479625587,
      "learning_rate": 2.179927674306187e-06,
      "loss": 0.3681,
      "step": 51155
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.032861455970911,
      "learning_rate": 2.1789111374310445e-06,
      "loss": 0.3831,
      "step": 51160
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.835971674965703,
      "learning_rate": 2.1778947715902093e-06,
      "loss": 0.3808,
      "step": 51165
    },
    {
      "epoch": 2.17,
      "grad_norm": 10.612648986202283,
      "learning_rate": 2.176878576845304e-06,
      "loss": 0.3745,
      "step": 51170
    },
    {
      "epoch": 2.17,
      "grad_norm": 8.823951336420272,
      "learning_rate": 2.175862553257932e-06,
      "loss": 0.3443,
      "step": 51175
    },
    {
      "epoch": 2.17,
      "grad_norm": 6.5224714912929596,
      "learning_rate": 2.1748467008896966e-06,
      "loss": 0.3724,
      "step": 51180
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.434440456786207,
      "learning_rate": 2.1738310198021824e-06,
      "loss": 0.3877,
      "step": 51185
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.130317387170686,
      "learning_rate": 2.172815510056971e-06,
      "loss": 0.3638,
      "step": 51190
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.090067295015865,
      "learning_rate": 2.171800171715625e-06,
      "loss": 0.4122,
      "step": 51195
    },
    {
      "epoch": 2.17,
      "grad_norm": 6.258493522486268,
      "learning_rate": 2.170785004839704e-06,
      "loss": 0.3623,
      "step": 51200
    },
    {
      "epoch": 2.17,
      "grad_norm": 14.981679850780715,
      "learning_rate": 2.1697700094907543e-06,
      "loss": 0.3704,
      "step": 51205
    },
    {
      "epoch": 2.17,
      "grad_norm": 9.409403803041403,
      "learning_rate": 2.168755185730314e-06,
      "loss": 0.3731,
      "step": 51210
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.4132859461817295,
      "learning_rate": 2.1677405336199053e-06,
      "loss": 0.3483,
      "step": 51215
    },
    {
      "epoch": 2.17,
      "grad_norm": 8.94575902133439,
      "learning_rate": 2.166726053221046e-06,
      "loss": 0.3691,
      "step": 51220
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.8243649025030644,
      "learning_rate": 2.1657117445952415e-06,
      "loss": 0.3857,
      "step": 51225
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.105115141071039,
      "learning_rate": 2.164697607803984e-06,
      "loss": 0.3868,
      "step": 51230
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.023860583391261,
      "learning_rate": 2.1636836429087582e-06,
      "loss": 0.354,
      "step": 51235
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.114160904750753,
      "learning_rate": 2.162669849971041e-06,
      "loss": 0.3569,
      "step": 51240
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.134121297964744,
      "learning_rate": 2.1616562290522907e-06,
      "loss": 0.3624,
      "step": 51245
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.93559317758632,
      "learning_rate": 2.160642780213963e-06,
      "loss": 0.3807,
      "step": 51250
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.320209098793835,
      "learning_rate": 2.1596295035175e-06,
      "loss": 0.381,
      "step": 51255
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.430699333227567,
      "learning_rate": 2.158616399024335e-06,
      "loss": 0.374,
      "step": 51260
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.9457623870660394,
      "learning_rate": 2.1576034667958865e-06,
      "loss": 0.3673,
      "step": 51265
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.8085622861495594,
      "learning_rate": 2.1565907068935674e-06,
      "loss": 0.3806,
      "step": 51270
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.803521610420573,
      "learning_rate": 2.15557811937878e-06,
      "loss": 0.3944,
      "step": 51275
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.124632698326712,
      "learning_rate": 2.154565704312911e-06,
      "loss": 0.3551,
      "step": 51280
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.00429236004617,
      "learning_rate": 2.1535534617573415e-06,
      "loss": 0.3573,
      "step": 51285
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.457261112581591,
      "learning_rate": 2.152541391773443e-06,
      "loss": 0.3747,
      "step": 51290
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.899646483049162,
      "learning_rate": 2.1515294944225713e-06,
      "loss": 0.3738,
      "step": 51295
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.814709750417389,
      "learning_rate": 2.1505177697660756e-06,
      "loss": 0.383,
      "step": 51300
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.956381493431635,
      "learning_rate": 2.1495062178652953e-06,
      "loss": 0.3913,
      "step": 51305
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.2358200718095045,
      "learning_rate": 2.148494838781555e-06,
      "loss": 0.384,
      "step": 51310
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.021572536853781,
      "learning_rate": 2.147483632576174e-06,
      "loss": 0.3854,
      "step": 51315
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.145993945556708,
      "learning_rate": 2.1464725993104574e-06,
      "loss": 0.3481,
      "step": 51320
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.253283009564978,
      "learning_rate": 2.145461739045704e-06,
      "loss": 0.346,
      "step": 51325
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.0624195530810505,
      "learning_rate": 2.144451051843194e-06,
      "loss": 0.3512,
      "step": 51330
    },
    {
      "epoch": 2.17,
      "grad_norm": 6.586653923156549,
      "learning_rate": 2.143440537764207e-06,
      "loss": 0.3729,
      "step": 51335
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.7158730649238367,
      "learning_rate": 2.142430196870007e-06,
      "loss": 0.3714,
      "step": 51340
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.472858670509909,
      "learning_rate": 2.141420029221847e-06,
      "loss": 0.3669,
      "step": 51345
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.750748239236262,
      "learning_rate": 2.140410034880968e-06,
      "loss": 0.3583,
      "step": 51350
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.497897233037699,
      "learning_rate": 2.1394002139086064e-06,
      "loss": 0.3833,
      "step": 51355
    },
    {
      "epoch": 2.17,
      "grad_norm": 3.7936110680022295,
      "learning_rate": 2.1383905663659847e-06,
      "loss": 0.3954,
      "step": 51360
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.101175928254679,
      "learning_rate": 2.1373810923143124e-06,
      "loss": 0.3698,
      "step": 51365
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.003769428948448,
      "learning_rate": 2.1363717918147926e-06,
      "loss": 0.373,
      "step": 51370
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.394080987851491,
      "learning_rate": 2.135362664928617e-06,
      "loss": 0.3977,
      "step": 51375
    },
    {
      "epoch": 2.17,
      "grad_norm": 5.823549119322144,
      "learning_rate": 2.134353711716964e-06,
      "loss": 0.3531,
      "step": 51380
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.054676946497687,
      "learning_rate": 2.1333449322410037e-06,
      "loss": 0.3506,
      "step": 51385
    },
    {
      "epoch": 2.17,
      "grad_norm": 4.88462917993755,
      "learning_rate": 2.1323363265618964e-06,
      "loss": 0.3877,
      "step": 51390
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.266221650146213,
      "learning_rate": 2.131327894740792e-06,
      "loss": 0.3616,
      "step": 51395
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.206376712803528,
      "learning_rate": 2.1303196368388267e-06,
      "loss": 0.3667,
      "step": 51400
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.255717954921978,
      "learning_rate": 2.1293115529171284e-06,
      "loss": 0.3778,
      "step": 51405
    },
    {
      "epoch": 2.18,
      "grad_norm": 7.041399534268732,
      "learning_rate": 2.1283036430368165e-06,
      "loss": 0.3717,
      "step": 51410
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.8543742200587174,
      "learning_rate": 2.127295907258994e-06,
      "loss": 0.3398,
      "step": 51415
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.5582613575096165,
      "learning_rate": 2.1262883456447588e-06,
      "loss": 0.3634,
      "step": 51420
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.1305461447214205,
      "learning_rate": 2.1252809582551983e-06,
      "loss": 0.3972,
      "step": 51425
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.746268164953898,
      "learning_rate": 2.1242737451513835e-06,
      "loss": 0.3691,
      "step": 51430
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.924606015904837,
      "learning_rate": 2.123266706394381e-06,
      "loss": 0.3735,
      "step": 51435
    },
    {
      "epoch": 2.18,
      "grad_norm": 5.236784904148432,
      "learning_rate": 2.122259842045244e-06,
      "loss": 0.374,
      "step": 51440
    },
    {
      "epoch": 2.18,
      "grad_norm": 6.921410838405361,
      "learning_rate": 2.1212531521650188e-06,
      "loss": 0.3513,
      "step": 51445
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.51288889518615,
      "learning_rate": 2.120246636814733e-06,
      "loss": 0.3709,
      "step": 51450
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.794559015941084,
      "learning_rate": 2.119240296055411e-06,
      "loss": 0.3642,
      "step": 51455
    },
    {
      "epoch": 2.18,
      "grad_norm": 5.17300180789268,
      "learning_rate": 2.1182341299480657e-06,
      "loss": 0.3804,
      "step": 51460
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.499227865092043,
      "learning_rate": 2.1172281385536953e-06,
      "loss": 0.373,
      "step": 51465
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.605381574162916,
      "learning_rate": 2.116222321933291e-06,
      "loss": 0.3648,
      "step": 51470
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.078348104362696,
      "learning_rate": 2.1152166801478352e-06,
      "loss": 0.3582,
      "step": 51475
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.276052317327856,
      "learning_rate": 2.1142112132582925e-06,
      "loss": 0.3834,
      "step": 51480
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.8700669775199654,
      "learning_rate": 2.113205921325624e-06,
      "loss": 0.3629,
      "step": 51485
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.145527827304661,
      "learning_rate": 2.1122008044107765e-06,
      "loss": 0.3657,
      "step": 51490
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.200229295045798,
      "learning_rate": 2.1111958625746905e-06,
      "loss": 0.3587,
      "step": 51495
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.109583220687445,
      "learning_rate": 2.1101910958782877e-06,
      "loss": 0.3503,
      "step": 51500
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.0092540391647296,
      "learning_rate": 2.109186504382489e-06,
      "loss": 0.3936,
      "step": 51505
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.978861233971614,
      "learning_rate": 2.108182088148195e-06,
      "loss": 0.3537,
      "step": 51510
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.261005355654024,
      "learning_rate": 2.1071778472363045e-06,
      "loss": 0.3833,
      "step": 51515
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.613900951337735,
      "learning_rate": 2.106173781707699e-06,
      "loss": 0.3589,
      "step": 51520
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.4904199785574037,
      "learning_rate": 2.1051698916232527e-06,
      "loss": 0.3376,
      "step": 51525
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.105349777872789,
      "learning_rate": 2.1041661770438303e-06,
      "loss": 0.3703,
      "step": 51530
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.11274391101578,
      "learning_rate": 2.1031626380302813e-06,
      "loss": 0.3728,
      "step": 51535
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.8544893034824677,
      "learning_rate": 2.1021592746434484e-06,
      "loss": 0.3814,
      "step": 51540
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.195565490792251,
      "learning_rate": 2.1011560869441644e-06,
      "loss": 0.3633,
      "step": 51545
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.700153869312193,
      "learning_rate": 2.1001530749932466e-06,
      "loss": 0.3774,
      "step": 51550
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.054351867797535,
      "learning_rate": 2.099150238851506e-06,
      "loss": 0.3513,
      "step": 51555
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.962867752971523,
      "learning_rate": 2.0981475785797435e-06,
      "loss": 0.3855,
      "step": 51560
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.7072800348990236,
      "learning_rate": 2.097145094238744e-06,
      "loss": 0.3415,
      "step": 51565
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.105364614842066,
      "learning_rate": 2.096142785889286e-06,
      "loss": 0.3705,
      "step": 51570
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.118636932486681,
      "learning_rate": 2.0951406535921383e-06,
      "loss": 0.3635,
      "step": 51575
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.043424260924906,
      "learning_rate": 2.0941386974080573e-06,
      "loss": 0.3726,
      "step": 51580
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.815175529899832,
      "learning_rate": 2.093136917397786e-06,
      "loss": 0.3739,
      "step": 51585
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.888037287389602,
      "learning_rate": 2.0921353136220614e-06,
      "loss": 0.3719,
      "step": 51590
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.183301608276359,
      "learning_rate": 2.0911338861416087e-06,
      "loss": 0.3587,
      "step": 51595
    },
    {
      "epoch": 2.18,
      "grad_norm": 3.8824926396972694,
      "learning_rate": 2.090132635017138e-06,
      "loss": 0.3829,
      "step": 51600
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.962896790676248,
      "learning_rate": 2.089131560309355e-06,
      "loss": 0.3563,
      "step": 51605
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.083195457470332,
      "learning_rate": 2.088130662078953e-06,
      "loss": 0.3662,
      "step": 51610
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.370295758380475,
      "learning_rate": 2.0871299403866096e-06,
      "loss": 0.4075,
      "step": 51615
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.536411881897502,
      "learning_rate": 2.086129395292998e-06,
      "loss": 0.367,
      "step": 51620
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.081032954210912,
      "learning_rate": 2.085129026858778e-06,
      "loss": 0.3824,
      "step": 51625
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.9554990247378115,
      "learning_rate": 2.084128835144601e-06,
      "loss": 0.3952,
      "step": 51630
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.9446208025323477,
      "learning_rate": 2.0831288202111017e-06,
      "loss": 0.3861,
      "step": 51635
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.083792622722573,
      "learning_rate": 2.08212898211891e-06,
      "loss": 0.362,
      "step": 51640
    },
    {
      "epoch": 2.19,
      "grad_norm": 6.14534859097579,
      "learning_rate": 2.0811293209286453e-06,
      "loss": 0.3723,
      "step": 51645
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.983637357190407,
      "learning_rate": 2.0801298367009105e-06,
      "loss": 0.3804,
      "step": 51650
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.212778105253001,
      "learning_rate": 2.0791305294963027e-06,
      "loss": 0.4083,
      "step": 51655
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.239878358727457,
      "learning_rate": 2.078131399375409e-06,
      "loss": 0.3703,
      "step": 51660
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.5231199356800325,
      "learning_rate": 2.077132446398802e-06,
      "loss": 0.3615,
      "step": 51665
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.995279890712481,
      "learning_rate": 2.076133670627043e-06,
      "loss": 0.3972,
      "step": 51670
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.156443819449974,
      "learning_rate": 2.0751350721206874e-06,
      "loss": 0.3746,
      "step": 51675
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.785049446921461,
      "learning_rate": 2.074136650940278e-06,
      "loss": 0.3823,
      "step": 51680
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.21327953631408,
      "learning_rate": 2.0731384071463446e-06,
      "loss": 0.3769,
      "step": 51685
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.185043372646848,
      "learning_rate": 2.0721403407994075e-06,
      "loss": 0.3533,
      "step": 51690
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.084063701857908,
      "learning_rate": 2.071142451959977e-06,
      "loss": 0.3874,
      "step": 51695
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.756882068372619,
      "learning_rate": 2.0701447406885543e-06,
      "loss": 0.3908,
      "step": 51700
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.97596081544535,
      "learning_rate": 2.0691472070456242e-06,
      "loss": 0.3525,
      "step": 51705
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.981541293731341,
      "learning_rate": 2.0681498510916653e-06,
      "loss": 0.3682,
      "step": 51710
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.050526559263437,
      "learning_rate": 2.0671526728871473e-06,
      "loss": 0.3304,
      "step": 51715
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.222969853161382,
      "learning_rate": 2.066155672492522e-06,
      "loss": 0.3582,
      "step": 51720
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.7534704966933,
      "learning_rate": 2.0651588499682364e-06,
      "loss": 0.3639,
      "step": 51725
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.187612091421166,
      "learning_rate": 2.064162205374726e-06,
      "loss": 0.3812,
      "step": 51730
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.227192675561793,
      "learning_rate": 2.063165738772412e-06,
      "loss": 0.3715,
      "step": 51735
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.7752378694010176,
      "learning_rate": 2.062169450221708e-06,
      "loss": 0.3637,
      "step": 51740
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.725752789852882,
      "learning_rate": 2.0611733397830174e-06,
      "loss": 0.3698,
      "step": 51745
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.4193213644975735,
      "learning_rate": 2.0601774075167317e-06,
      "loss": 0.3795,
      "step": 51750
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.990852042380881,
      "learning_rate": 2.059181653483229e-06,
      "loss": 0.3555,
      "step": 51755
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.869377026977483,
      "learning_rate": 2.05818607774288e-06,
      "loss": 0.3612,
      "step": 51760
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.236327299799827,
      "learning_rate": 2.0571906803560453e-06,
      "loss": 0.3494,
      "step": 51765
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.026272768522512,
      "learning_rate": 2.0561954613830696e-06,
      "loss": 0.359,
      "step": 51770
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.9085209536453,
      "learning_rate": 2.055200420884292e-06,
      "loss": 0.3829,
      "step": 51775
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.9944548526479524,
      "learning_rate": 2.0542055589200406e-06,
      "loss": 0.3728,
      "step": 51780
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.447119160316467,
      "learning_rate": 2.0532108755506263e-06,
      "loss": 0.3623,
      "step": 51785
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.659615341526252,
      "learning_rate": 2.0522163708363573e-06,
      "loss": 0.3822,
      "step": 51790
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.1250720666244,
      "learning_rate": 2.051222044837527e-06,
      "loss": 0.364,
      "step": 51795
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.188667157657558,
      "learning_rate": 2.0502278976144197e-06,
      "loss": 0.3905,
      "step": 51800
    },
    {
      "epoch": 2.19,
      "grad_norm": 5.084043544631529,
      "learning_rate": 2.0492339292273046e-06,
      "loss": 0.4074,
      "step": 51805
    },
    {
      "epoch": 2.19,
      "grad_norm": 8.513666690445548,
      "learning_rate": 2.0482401397364445e-06,
      "loss": 0.3927,
      "step": 51810
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.088418700873688,
      "learning_rate": 2.0472465292020917e-06,
      "loss": 0.3549,
      "step": 51815
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.95418219143137,
      "learning_rate": 2.046253097684482e-06,
      "loss": 0.3718,
      "step": 51820
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.243837819608366,
      "learning_rate": 2.0452598452438473e-06,
      "loss": 0.3715,
      "step": 51825
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.110171573977456,
      "learning_rate": 2.044266771940406e-06,
      "loss": 0.3808,
      "step": 51830
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.120271922361783,
      "learning_rate": 2.0432738778343634e-06,
      "loss": 0.3843,
      "step": 51835
    },
    {
      "epoch": 2.19,
      "grad_norm": 7.992287059383662,
      "learning_rate": 2.042281162985915e-06,
      "loss": 0.365,
      "step": 51840
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.114072674781159,
      "learning_rate": 2.041288627455248e-06,
      "loss": 0.3639,
      "step": 51845
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.8843517415091005,
      "learning_rate": 2.0402962713025376e-06,
      "loss": 0.3706,
      "step": 51850
    },
    {
      "epoch": 2.19,
      "grad_norm": 4.735312003394535,
      "learning_rate": 2.0393040945879446e-06,
      "loss": 0.3848,
      "step": 51855
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.826196607110661,
      "learning_rate": 2.038312097371623e-06,
      "loss": 0.3639,
      "step": 51860
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.204349597294005,
      "learning_rate": 2.0373202797137167e-06,
      "loss": 0.3785,
      "step": 51865
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.8561300614769127,
      "learning_rate": 2.0363286416743533e-06,
      "loss": 0.3535,
      "step": 51870
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.165358976097341,
      "learning_rate": 2.0353371833136547e-06,
      "loss": 0.3773,
      "step": 51875
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.0839575592203285,
      "learning_rate": 2.03434590469173e-06,
      "loss": 0.4007,
      "step": 51880
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.935437244851414,
      "learning_rate": 2.033354805868679e-06,
      "loss": 0.3479,
      "step": 51885
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.756361962332113,
      "learning_rate": 2.0323638869045866e-06,
      "loss": 0.3827,
      "step": 51890
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.4512864180963705,
      "learning_rate": 2.0313731478595295e-06,
      "loss": 0.3841,
      "step": 51895
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.270440806479402,
      "learning_rate": 2.0303825887935763e-06,
      "loss": 0.3587,
      "step": 51900
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.721564408715685,
      "learning_rate": 2.0293922097667774e-06,
      "loss": 0.3632,
      "step": 51905
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.474115041100587,
      "learning_rate": 2.0284020108391782e-06,
      "loss": 0.3851,
      "step": 51910
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.0845033704801565,
      "learning_rate": 2.0274119920708146e-06,
      "loss": 0.3777,
      "step": 51915
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.7185346015757257,
      "learning_rate": 2.0264221535217032e-06,
      "loss": 0.3654,
      "step": 51920
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.454504208450664,
      "learning_rate": 2.0254324952518584e-06,
      "loss": 0.3672,
      "step": 51925
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.7530132158441925,
      "learning_rate": 2.0244430173212793e-06,
      "loss": 0.3425,
      "step": 51930
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.063953293364503,
      "learning_rate": 2.0234537197899575e-06,
      "loss": 0.3606,
      "step": 51935
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.1000237722405,
      "learning_rate": 2.022464602717866e-06,
      "loss": 0.3486,
      "step": 51940
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.412934234675003,
      "learning_rate": 2.021475666164976e-06,
      "loss": 0.3774,
      "step": 51945
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.134408646621687,
      "learning_rate": 2.0204869101912445e-06,
      "loss": 0.341,
      "step": 51950
    },
    {
      "epoch": 2.2,
      "grad_norm": 6.4598563774912305,
      "learning_rate": 2.0194983348566133e-06,
      "loss": 0.3673,
      "step": 51955
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.9002987481566875,
      "learning_rate": 2.0185099402210183e-06,
      "loss": 0.3573,
      "step": 51960
    },
    {
      "epoch": 2.2,
      "grad_norm": 6.175541955717893,
      "learning_rate": 2.017521726344386e-06,
      "loss": 0.3904,
      "step": 51965
    },
    {
      "epoch": 2.2,
      "grad_norm": 6.6204911772146,
      "learning_rate": 2.0165336932866235e-06,
      "loss": 0.3711,
      "step": 51970
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.579069146498737,
      "learning_rate": 2.0155458411076356e-06,
      "loss": 0.3619,
      "step": 51975
    },
    {
      "epoch": 2.2,
      "grad_norm": 6.7153989030271255,
      "learning_rate": 2.014558169867312e-06,
      "loss": 0.3554,
      "step": 51980
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.013134493082191,
      "learning_rate": 2.013570679625535e-06,
      "loss": 0.3647,
      "step": 51985
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.352710991603538,
      "learning_rate": 2.0125833704421703e-06,
      "loss": 0.385,
      "step": 51990
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.5270311850192515,
      "learning_rate": 2.011596242377075e-06,
      "loss": 0.3601,
      "step": 51995
    },
    {
      "epoch": 2.2,
      "grad_norm": 6.137427258235416,
      "learning_rate": 2.0106092954900965e-06,
      "loss": 0.3773,
      "step": 52000
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.676219143585936,
      "learning_rate": 2.0096225298410725e-06,
      "loss": 0.3668,
      "step": 52005
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.1243149611227405,
      "learning_rate": 2.0086359454898247e-06,
      "loss": 0.3847,
      "step": 52010
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.6235355890623606,
      "learning_rate": 2.0076495424961677e-06,
      "loss": 0.3746,
      "step": 52015
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.012469569119401,
      "learning_rate": 2.006663320919907e-06,
      "loss": 0.3769,
      "step": 52020
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.044297928019615,
      "learning_rate": 2.0056772808208303e-06,
      "loss": 0.3825,
      "step": 52025
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.015911371663097,
      "learning_rate": 2.0046914222587195e-06,
      "loss": 0.3829,
      "step": 52030
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.991488613149469,
      "learning_rate": 2.0037057452933465e-06,
      "loss": 0.3796,
      "step": 52035
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.6813101717331835,
      "learning_rate": 2.0027202499844666e-06,
      "loss": 0.3514,
      "step": 52040
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.919994615143753,
      "learning_rate": 2.0017349363918297e-06,
      "loss": 0.3446,
      "step": 52045
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.889472870652354,
      "learning_rate": 2.000749804575171e-06,
      "loss": 0.347,
      "step": 52050
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.071881291099214,
      "learning_rate": 1.99976485459422e-06,
      "loss": 0.3852,
      "step": 52055
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.013320510136518,
      "learning_rate": 1.998780086508686e-06,
      "loss": 0.3703,
      "step": 52060
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.030597314352833,
      "learning_rate": 1.9977955003782747e-06,
      "loss": 0.3894,
      "step": 52065
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.834619976275307,
      "learning_rate": 1.9968110962626807e-06,
      "loss": 0.3567,
      "step": 52070
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.362719555751729,
      "learning_rate": 1.9958268742215824e-06,
      "loss": 0.3876,
      "step": 52075
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.44066083565376,
      "learning_rate": 1.9948428343146512e-06,
      "loss": 0.3733,
      "step": 52080
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.995028036302467,
      "learning_rate": 1.993858976601549e-06,
      "loss": 0.369,
      "step": 52085
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.906914702349831,
      "learning_rate": 1.9928753011419205e-06,
      "loss": 0.3622,
      "step": 52090
    },
    {
      "epoch": 2.2,
      "grad_norm": 5.093732474817955,
      "learning_rate": 1.9918918079954047e-06,
      "loss": 0.3404,
      "step": 52095
    },
    {
      "epoch": 2.2,
      "grad_norm": 4.741861705574664,
      "learning_rate": 1.990908497221628e-06,
      "loss": 0.3388,
      "step": 52100
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.782002767313741,
      "learning_rate": 1.9899253688802073e-06,
      "loss": 0.3953,
      "step": 52105
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.10076632090282,
      "learning_rate": 1.988942423030743e-06,
      "loss": 0.3632,
      "step": 52110
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.474426153669389,
      "learning_rate": 1.9879596597328306e-06,
      "loss": 0.3667,
      "step": 52115
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.359230629121613,
      "learning_rate": 1.986977079046053e-06,
      "loss": 0.3622,
      "step": 52120
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.275009392004105,
      "learning_rate": 1.9859946810299784e-06,
      "loss": 0.3973,
      "step": 52125
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.749293205796794,
      "learning_rate": 1.9850124657441683e-06,
      "loss": 0.3339,
      "step": 52130
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.9336687103217254,
      "learning_rate": 1.984030433248173e-06,
      "loss": 0.3668,
      "step": 52135
    },
    {
      "epoch": 2.21,
      "grad_norm": 6.19449399804296,
      "learning_rate": 1.9830485836015273e-06,
      "loss": 0.3712,
      "step": 52140
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.617219363891496,
      "learning_rate": 1.982066916863759e-06,
      "loss": 0.3695,
      "step": 52145
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.0899390001602205,
      "learning_rate": 1.9810854330943857e-06,
      "loss": 0.3569,
      "step": 52150
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.7095346688128936,
      "learning_rate": 1.9801041323529104e-06,
      "loss": 0.3636,
      "step": 52155
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.9853679060097043,
      "learning_rate": 1.9791230146988243e-06,
      "loss": 0.3669,
      "step": 52160
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.96138389585488,
      "learning_rate": 1.978142080191611e-06,
      "loss": 0.3826,
      "step": 52165
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.747042031105779,
      "learning_rate": 1.9771613288907447e-06,
      "loss": 0.3809,
      "step": 52170
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.046840603569988,
      "learning_rate": 1.976180760855681e-06,
      "loss": 0.3893,
      "step": 52175
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.8691562524940775,
      "learning_rate": 1.9752003761458703e-06,
      "loss": 0.3622,
      "step": 52180
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.062471620183667,
      "learning_rate": 1.974220174820752e-06,
      "loss": 0.3564,
      "step": 52185
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.025426498823294,
      "learning_rate": 1.9732401569397526e-06,
      "loss": 0.3632,
      "step": 52190
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.9743812865491153,
      "learning_rate": 1.9722603225622855e-06,
      "loss": 0.3693,
      "step": 52195
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.068860289679473,
      "learning_rate": 1.971280671747756e-06,
      "loss": 0.3713,
      "step": 52200
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.679569382675868,
      "learning_rate": 1.97030120455556e-06,
      "loss": 0.3657,
      "step": 52205
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.949474841144683,
      "learning_rate": 1.9693219210450763e-06,
      "loss": 0.3496,
      "step": 52210
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.2565806285270265,
      "learning_rate": 1.968342821275677e-06,
      "loss": 0.3591,
      "step": 52215
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.446382583126833,
      "learning_rate": 1.967363905306725e-06,
      "loss": 0.359,
      "step": 52220
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.9600743225860504,
      "learning_rate": 1.9663851731975635e-06,
      "loss": 0.3688,
      "step": 52225
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.3558852888730355,
      "learning_rate": 1.9654066250075344e-06,
      "loss": 0.3674,
      "step": 52230
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.248378836482522,
      "learning_rate": 1.9644282607959617e-06,
      "loss": 0.3731,
      "step": 52235
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.339411432024546,
      "learning_rate": 1.9634500806221644e-06,
      "loss": 0.3641,
      "step": 52240
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.419979276318935,
      "learning_rate": 1.9624720845454424e-06,
      "loss": 0.3585,
      "step": 52245
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.087273234751124,
      "learning_rate": 1.9614942726250906e-06,
      "loss": 0.3808,
      "step": 52250
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.006498356702361,
      "learning_rate": 1.9605166449203927e-06,
      "loss": 0.3615,
      "step": 52255
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.899700520635498,
      "learning_rate": 1.9595392014906157e-06,
      "loss": 0.3629,
      "step": 52260
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.0693599621914895,
      "learning_rate": 1.958561942395021e-06,
      "loss": 0.3893,
      "step": 52265
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.824579001211704,
      "learning_rate": 1.957584867692859e-06,
      "loss": 0.3862,
      "step": 52270
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.718757857068727,
      "learning_rate": 1.956607977443362e-06,
      "loss": 0.3705,
      "step": 52275
    },
    {
      "epoch": 2.21,
      "grad_norm": 7.00674977584657,
      "learning_rate": 1.9556312717057597e-06,
      "loss": 0.3662,
      "step": 52280
    },
    {
      "epoch": 2.21,
      "grad_norm": 7.961240252252757,
      "learning_rate": 1.9546547505392653e-06,
      "loss": 0.3709,
      "step": 52285
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.3828703940668365,
      "learning_rate": 1.953678414003085e-06,
      "loss": 0.3584,
      "step": 52290
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.086542332194679,
      "learning_rate": 1.9527022621564075e-06,
      "loss": 0.3574,
      "step": 52295
    },
    {
      "epoch": 2.21,
      "grad_norm": 6.943109211727939,
      "learning_rate": 1.9517262950584153e-06,
      "loss": 0.3669,
      "step": 52300
    },
    {
      "epoch": 2.21,
      "grad_norm": 5.714653688908005,
      "learning_rate": 1.9507505127682814e-06,
      "loss": 0.3804,
      "step": 52305
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.7338210356256787,
      "learning_rate": 1.9497749153451597e-06,
      "loss": 0.3616,
      "step": 52310
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.276891112505889,
      "learning_rate": 1.9487995028482023e-06,
      "loss": 0.3457,
      "step": 52315
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.179078575684295,
      "learning_rate": 1.9478242753365413e-06,
      "loss": 0.3973,
      "step": 52320
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.823555683691143,
      "learning_rate": 1.9468492328693057e-06,
      "loss": 0.3654,
      "step": 52325
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.886830111675473,
      "learning_rate": 1.945874375505606e-06,
      "loss": 0.3463,
      "step": 52330
    },
    {
      "epoch": 2.21,
      "grad_norm": 4.678463769405813,
      "learning_rate": 1.944899703304547e-06,
      "loss": 0.3605,
      "step": 52335
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.314466737027587,
      "learning_rate": 1.9439252163252214e-06,
      "loss": 0.3504,
      "step": 52340
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.158457508563852,
      "learning_rate": 1.942950914626707e-06,
      "loss": 0.387,
      "step": 52345
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.052021047208834,
      "learning_rate": 1.9419767982680725e-06,
      "loss": 0.3497,
      "step": 52350
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.191669331438492,
      "learning_rate": 1.941002867308378e-06,
      "loss": 0.3601,
      "step": 52355
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.289202422339016,
      "learning_rate": 1.9400291218066703e-06,
      "loss": 0.3631,
      "step": 52360
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.163567958282461,
      "learning_rate": 1.9390555618219818e-06,
      "loss": 0.3589,
      "step": 52365
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.764923268161796,
      "learning_rate": 1.938082187413339e-06,
      "loss": 0.3346,
      "step": 52370
    },
    {
      "epoch": 2.22,
      "grad_norm": 6.141513764941518,
      "learning_rate": 1.937108998639755e-06,
      "loss": 0.3725,
      "step": 52375
    },
    {
      "epoch": 2.22,
      "grad_norm": 8.46875793632904,
      "learning_rate": 1.9361359955602287e-06,
      "loss": 0.3733,
      "step": 52380
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.904148435574873,
      "learning_rate": 1.935163178233752e-06,
      "loss": 0.3613,
      "step": 52385
    },
    {
      "epoch": 2.22,
      "grad_norm": 7.285022073967542,
      "learning_rate": 1.934190546719306e-06,
      "loss": 0.3497,
      "step": 52390
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.109518472245119,
      "learning_rate": 1.933218101075855e-06,
      "loss": 0.3472,
      "step": 52395
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.113303233106136,
      "learning_rate": 1.932245841362357e-06,
      "loss": 0.3752,
      "step": 52400
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.8783738037222895,
      "learning_rate": 1.931273767637757e-06,
      "loss": 0.3413,
      "step": 52405
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.477465744155155,
      "learning_rate": 1.930301879960991e-06,
      "loss": 0.3575,
      "step": 52410
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.476051325252788,
      "learning_rate": 1.9293301783909784e-06,
      "loss": 0.3685,
      "step": 52415
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.391132174108271,
      "learning_rate": 1.928358662986632e-06,
      "loss": 0.3618,
      "step": 52420
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.545256176396989,
      "learning_rate": 1.927387333806854e-06,
      "loss": 0.3623,
      "step": 52425
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.175017610564468,
      "learning_rate": 1.926416190910529e-06,
      "loss": 0.3542,
      "step": 52430
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.123149654600157,
      "learning_rate": 1.925445234356537e-06,
      "loss": 0.3397,
      "step": 52435
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.025658209469571,
      "learning_rate": 1.9244744642037462e-06,
      "loss": 0.3607,
      "step": 52440
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.551473052274857,
      "learning_rate": 1.923503880511007e-06,
      "loss": 0.4065,
      "step": 52445
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.9309635901021744,
      "learning_rate": 1.9225334833371655e-06,
      "loss": 0.3594,
      "step": 52450
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.325724890301639,
      "learning_rate": 1.921563272741056e-06,
      "loss": 0.3896,
      "step": 52455
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.818908116741205,
      "learning_rate": 1.9205932487814954e-06,
      "loss": 0.3511,
      "step": 52460
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.583659448863868,
      "learning_rate": 1.919623411517296e-06,
      "loss": 0.3845,
      "step": 52465
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.349346092745814,
      "learning_rate": 1.9186537610072555e-06,
      "loss": 0.362,
      "step": 52470
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.080288114784697,
      "learning_rate": 1.917684297310164e-06,
      "loss": 0.3517,
      "step": 52475
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.598254865537337,
      "learning_rate": 1.9167150204847913e-06,
      "loss": 0.355,
      "step": 52480
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.15440874555474,
      "learning_rate": 1.915745930589904e-06,
      "loss": 0.3446,
      "step": 52485
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.948977984371894,
      "learning_rate": 1.914777027684257e-06,
      "loss": 0.3546,
      "step": 52490
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.7714640450693815,
      "learning_rate": 1.913808311826592e-06,
      "loss": 0.3823,
      "step": 52495
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.237498032240751,
      "learning_rate": 1.9128397830756368e-06,
      "loss": 0.3623,
      "step": 52500
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.289082304931002,
      "learning_rate": 1.9118714414901122e-06,
      "loss": 0.3836,
      "step": 52505
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.078741156773504,
      "learning_rate": 1.9109032871287274e-06,
      "loss": 0.342,
      "step": 52510
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.552989142599369,
      "learning_rate": 1.909935320050175e-06,
      "loss": 0.3742,
      "step": 52515
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.98392139774444,
      "learning_rate": 1.908967540313143e-06,
      "loss": 0.371,
      "step": 52520
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.86080492200221,
      "learning_rate": 1.907999947976305e-06,
      "loss": 0.3549,
      "step": 52525
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.007939545921851,
      "learning_rate": 1.9070325430983212e-06,
      "loss": 0.3861,
      "step": 52530
    },
    {
      "epoch": 2.22,
      "grad_norm": 6.767143685456613,
      "learning_rate": 1.9060653257378432e-06,
      "loss": 0.3664,
      "step": 52535
    },
    {
      "epoch": 2.22,
      "grad_norm": 9.70787830221513,
      "learning_rate": 1.9050982959535114e-06,
      "loss": 0.3548,
      "step": 52540
    },
    {
      "epoch": 2.22,
      "grad_norm": 5.398269057966513,
      "learning_rate": 1.9041314538039545e-06,
      "loss": 0.3629,
      "step": 52545
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.047948213816526,
      "learning_rate": 1.9031647993477874e-06,
      "loss": 0.375,
      "step": 52550
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.492654768160743,
      "learning_rate": 1.9021983326436166e-06,
      "loss": 0.372,
      "step": 52555
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.096637076860877,
      "learning_rate": 1.9012320537500373e-06,
      "loss": 0.3885,
      "step": 52560
    },
    {
      "epoch": 2.22,
      "grad_norm": 4.863173821493066,
      "learning_rate": 1.9002659627256292e-06,
      "loss": 0.3618,
      "step": 52565
    },
    {
      "epoch": 2.22,
      "grad_norm": 3.8221833842937722,
      "learning_rate": 1.899300059628965e-06,
      "loss": 0.3713,
      "step": 52570
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.6380844127172853,
      "learning_rate": 1.898334344518607e-06,
      "loss": 0.3418,
      "step": 52575
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.377294278632141,
      "learning_rate": 1.8973688174530996e-06,
      "loss": 0.3504,
      "step": 52580
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.8239141479937873,
      "learning_rate": 1.8964034784909819e-06,
      "loss": 0.3501,
      "step": 52585
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.877907379924693,
      "learning_rate": 1.895438327690779e-06,
      "loss": 0.371,
      "step": 52590
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.803271010009943,
      "learning_rate": 1.894473365111007e-06,
      "loss": 0.3533,
      "step": 52595
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.866671104008564,
      "learning_rate": 1.8935085908101658e-06,
      "loss": 0.3488,
      "step": 52600
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.021100002240936,
      "learning_rate": 1.8925440048467486e-06,
      "loss": 0.3724,
      "step": 52605
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.167187498662469,
      "learning_rate": 1.8915796072792363e-06,
      "loss": 0.3599,
      "step": 52610
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.562233497884677,
      "learning_rate": 1.8906153981660946e-06,
      "loss": 0.376,
      "step": 52615
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.947660863818793,
      "learning_rate": 1.8896513775657826e-06,
      "loss": 0.3504,
      "step": 52620
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.128357593299075,
      "learning_rate": 1.888687545536747e-06,
      "loss": 0.3467,
      "step": 52625
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.928489975796033,
      "learning_rate": 1.8877239021374195e-06,
      "loss": 0.3698,
      "step": 52630
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.7889570286838437,
      "learning_rate": 1.8867604474262247e-06,
      "loss": 0.3717,
      "step": 52635
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.822087568848517,
      "learning_rate": 1.8857971814615744e-06,
      "loss": 0.3613,
      "step": 52640
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.288559002450357,
      "learning_rate": 1.8848341043018685e-06,
      "loss": 0.3748,
      "step": 52645
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.43144202621119,
      "learning_rate": 1.8838712160054929e-06,
      "loss": 0.3811,
      "step": 52650
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.536662797303458,
      "learning_rate": 1.8829085166308265e-06,
      "loss": 0.3725,
      "step": 52655
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.048189720593763,
      "learning_rate": 1.881946006236235e-06,
      "loss": 0.3602,
      "step": 52660
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.693163203436942,
      "learning_rate": 1.8809836848800744e-06,
      "loss": 0.3797,
      "step": 52665
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.59335958281409,
      "learning_rate": 1.8800215526206844e-06,
      "loss": 0.3682,
      "step": 52670
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.5897458105273437,
      "learning_rate": 1.879059609516397e-06,
      "loss": 0.3652,
      "step": 52675
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.279809588318355,
      "learning_rate": 1.878097855625534e-06,
      "loss": 0.3526,
      "step": 52680
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.855734402805208,
      "learning_rate": 1.8771362910064006e-06,
      "loss": 0.3832,
      "step": 52685
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.907301637185064,
      "learning_rate": 1.8761749157172953e-06,
      "loss": 0.362,
      "step": 52690
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.213336188221828,
      "learning_rate": 1.8752137298165052e-06,
      "loss": 0.3691,
      "step": 52695
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.269830686902421,
      "learning_rate": 1.8742527333623e-06,
      "loss": 0.3613,
      "step": 52700
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.033303111413465,
      "learning_rate": 1.8732919264129451e-06,
      "loss": 0.3572,
      "step": 52705
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.1794721454088535,
      "learning_rate": 1.872331309026692e-06,
      "loss": 0.3565,
      "step": 52710
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.933123449945412,
      "learning_rate": 1.871370881261777e-06,
      "loss": 0.3746,
      "step": 52715
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.8003110831257105,
      "learning_rate": 1.8704106431764296e-06,
      "loss": 0.3603,
      "step": 52720
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.212211974765563,
      "learning_rate": 1.8694505948288673e-06,
      "loss": 0.3612,
      "step": 52725
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.298843631015071,
      "learning_rate": 1.8684907362772947e-06,
      "loss": 0.3493,
      "step": 52730
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.191727846833544,
      "learning_rate": 1.8675310675799036e-06,
      "loss": 0.3454,
      "step": 52735
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.9403184134836104,
      "learning_rate": 1.8665715887948766e-06,
      "loss": 0.3759,
      "step": 52740
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.090940384661897,
      "learning_rate": 1.865612299980386e-06,
      "loss": 0.3797,
      "step": 52745
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.1649462186129025,
      "learning_rate": 1.864653201194588e-06,
      "loss": 0.3755,
      "step": 52750
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.703855388222147,
      "learning_rate": 1.8636942924956298e-06,
      "loss": 0.3573,
      "step": 52755
    },
    {
      "epoch": 2.23,
      "grad_norm": 6.791083060422642,
      "learning_rate": 1.862735573941651e-06,
      "loss": 0.3492,
      "step": 52760
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.237379850746144,
      "learning_rate": 1.8617770455907707e-06,
      "loss": 0.3456,
      "step": 52765
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.025506823848462,
      "learning_rate": 1.8608187075011047e-06,
      "loss": 0.3641,
      "step": 52770
    },
    {
      "epoch": 2.23,
      "grad_norm": 3.662202097187108,
      "learning_rate": 1.859860559730754e-06,
      "loss": 0.3625,
      "step": 52775
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.390596568070467,
      "learning_rate": 1.8589026023378093e-06,
      "loss": 0.3587,
      "step": 52780
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.23930610963032,
      "learning_rate": 1.8579448353803459e-06,
      "loss": 0.3738,
      "step": 52785
    },
    {
      "epoch": 2.23,
      "grad_norm": 7.842964638300032,
      "learning_rate": 1.8569872589164317e-06,
      "loss": 0.3529,
      "step": 52790
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.009353901865072,
      "learning_rate": 1.8560298730041237e-06,
      "loss": 0.3383,
      "step": 52795
    },
    {
      "epoch": 2.23,
      "grad_norm": 4.3152747534048865,
      "learning_rate": 1.8550726777014633e-06,
      "loss": 0.3524,
      "step": 52800
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.217652354038226,
      "learning_rate": 1.8541156730664817e-06,
      "loss": 0.3632,
      "step": 52805
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.181123017580436,
      "learning_rate": 1.8531588591571998e-06,
      "loss": 0.3787,
      "step": 52810
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.834009309587241,
      "learning_rate": 1.852202236031629e-06,
      "loss": 0.392,
      "step": 52815
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.431219782249487,
      "learning_rate": 1.8512458037477626e-06,
      "loss": 0.3963,
      "step": 52820
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8475311036454407,
      "learning_rate": 1.850289562363588e-06,
      "loss": 0.3562,
      "step": 52825
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.9014367275676376,
      "learning_rate": 1.8493335119370814e-06,
      "loss": 0.3483,
      "step": 52830
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.718757910136919,
      "learning_rate": 1.848377652526202e-06,
      "loss": 0.393,
      "step": 52835
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8967628671227765,
      "learning_rate": 1.847421984188902e-06,
      "loss": 0.3461,
      "step": 52840
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.0272393495300856,
      "learning_rate": 1.8464665069831205e-06,
      "loss": 0.3456,
      "step": 52845
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.147982336466948,
      "learning_rate": 1.8455112209667875e-06,
      "loss": 0.3658,
      "step": 52850
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.015118316576194,
      "learning_rate": 1.8445561261978156e-06,
      "loss": 0.3353,
      "step": 52855
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.796401477746836,
      "learning_rate": 1.843601222734111e-06,
      "loss": 0.383,
      "step": 52860
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.188307939391202,
      "learning_rate": 1.8426465106335685e-06,
      "loss": 0.3621,
      "step": 52865
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.5333895220065905,
      "learning_rate": 1.8416919899540663e-06,
      "loss": 0.3486,
      "step": 52870
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.9254561126277734,
      "learning_rate": 1.8407376607534754e-06,
      "loss": 0.3623,
      "step": 52875
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8879028391273223,
      "learning_rate": 1.839783523089656e-06,
      "loss": 0.345,
      "step": 52880
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.235122960089528,
      "learning_rate": 1.838829577020451e-06,
      "loss": 0.3504,
      "step": 52885
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.021847320569711,
      "learning_rate": 1.837875822603698e-06,
      "loss": 0.3809,
      "step": 52890
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.389994328576631,
      "learning_rate": 1.8369222598972187e-06,
      "loss": 0.364,
      "step": 52895
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.414931627394944,
      "learning_rate": 1.8359688889588278e-06,
      "loss": 0.3325,
      "step": 52900
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.1916310647214745,
      "learning_rate": 1.8350157098463212e-06,
      "loss": 0.3668,
      "step": 52905
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.161141304947531,
      "learning_rate": 1.8340627226174896e-06,
      "loss": 0.392,
      "step": 52910
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8422776779919365,
      "learning_rate": 1.8331099273301112e-06,
      "loss": 0.3777,
      "step": 52915
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.486709164494434,
      "learning_rate": 1.8321573240419476e-06,
      "loss": 0.3639,
      "step": 52920
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.87158017382092,
      "learning_rate": 1.8312049128107545e-06,
      "loss": 0.3641,
      "step": 52925
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.217925218782692,
      "learning_rate": 1.8302526936942749e-06,
      "loss": 0.3785,
      "step": 52930
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.355501497210609,
      "learning_rate": 1.829300666750236e-06,
      "loss": 0.3365,
      "step": 52935
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.063589917072802,
      "learning_rate": 1.8283488320363585e-06,
      "loss": 0.3552,
      "step": 52940
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.122445580458788,
      "learning_rate": 1.8273971896103488e-06,
      "loss": 0.3463,
      "step": 52945
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.936382037431601,
      "learning_rate": 1.826445739529904e-06,
      "loss": 0.378,
      "step": 52950
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.101368925141458,
      "learning_rate": 1.8254944818527038e-06,
      "loss": 0.3489,
      "step": 52955
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.304381004406447,
      "learning_rate": 1.8245434166364246e-06,
      "loss": 0.3741,
      "step": 52960
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.911301933481199,
      "learning_rate": 1.8235925439387225e-06,
      "loss": 0.3409,
      "step": 52965
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.0932602824018005,
      "learning_rate": 1.8226418638172498e-06,
      "loss": 0.3627,
      "step": 52970
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.853821227355291,
      "learning_rate": 1.8216913763296401e-06,
      "loss": 0.3594,
      "step": 52975
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.39686085366871,
      "learning_rate": 1.8207410815335203e-06,
      "loss": 0.3309,
      "step": 52980
    },
    {
      "epoch": 2.24,
      "grad_norm": 6.102895459780268,
      "learning_rate": 1.8197909794865054e-06,
      "loss": 0.3576,
      "step": 52985
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8446937586439263,
      "learning_rate": 1.818841070246194e-06,
      "loss": 0.3737,
      "step": 52990
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.616285824894671,
      "learning_rate": 1.8178913538701782e-06,
      "loss": 0.3572,
      "step": 52995
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.119704979213325,
      "learning_rate": 1.816941830416038e-06,
      "loss": 0.378,
      "step": 53000
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.173626965155049,
      "learning_rate": 1.815992499941337e-06,
      "loss": 0.3658,
      "step": 53005
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.924961337143304,
      "learning_rate": 1.8150433625036324e-06,
      "loss": 0.3634,
      "step": 53010
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.119049885767718,
      "learning_rate": 1.814094418160468e-06,
      "loss": 0.3996,
      "step": 53015
    },
    {
      "epoch": 2.24,
      "grad_norm": 5.358812973288078,
      "learning_rate": 1.8131456669693736e-06,
      "loss": 0.3529,
      "step": 53020
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.580032077984705,
      "learning_rate": 1.8121971089878698e-06,
      "loss": 0.3714,
      "step": 53025
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.948483304481978,
      "learning_rate": 1.8112487442734656e-06,
      "loss": 0.3765,
      "step": 53030
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8361594368064384,
      "learning_rate": 1.8103005728836587e-06,
      "loss": 0.3595,
      "step": 53035
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.116040656169031,
      "learning_rate": 1.8093525948759311e-06,
      "loss": 0.3774,
      "step": 53040
    },
    {
      "epoch": 2.24,
      "grad_norm": 4.421928918688815,
      "learning_rate": 1.8084048103077573e-06,
      "loss": 0.3633,
      "step": 53045
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.128638318632504,
      "learning_rate": 1.8074572192366002e-06,
      "loss": 0.3545,
      "step": 53050
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.958646606347638,
      "learning_rate": 1.8065098217199061e-06,
      "loss": 0.3324,
      "step": 53055
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.815161411482528,
      "learning_rate": 1.8055626178151154e-06,
      "loss": 0.3731,
      "step": 53060
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.8586910485184274,
      "learning_rate": 1.8046156075796545e-06,
      "loss": 0.3431,
      "step": 53065
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.020434190253334,
      "learning_rate": 1.803668791070936e-06,
      "loss": 0.3692,
      "step": 53070
    },
    {
      "epoch": 2.25,
      "grad_norm": 5.139718093080879,
      "learning_rate": 1.802722168346363e-06,
      "loss": 0.367,
      "step": 53075
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.5604930248365037,
      "learning_rate": 1.801775739463328e-06,
      "loss": 0.3629,
      "step": 53080
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.296991113426738,
      "learning_rate": 1.8008295044792102e-06,
      "loss": 0.3661,
      "step": 53085
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.323002687686565,
      "learning_rate": 1.7998834634513745e-06,
      "loss": 0.3569,
      "step": 53090
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.6347778054673068,
      "learning_rate": 1.798937616437178e-06,
      "loss": 0.3454,
      "step": 53095
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.013264099885689,
      "learning_rate": 1.7979919634939663e-06,
      "loss": 0.3725,
      "step": 53100
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.597132063916742,
      "learning_rate": 1.797046504679069e-06,
      "loss": 0.337,
      "step": 53105
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.0651751369632505,
      "learning_rate": 1.7961012400498073e-06,
      "loss": 0.3589,
      "step": 53110
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.8606636170674413,
      "learning_rate": 1.7951561696634912e-06,
      "loss": 0.3635,
      "step": 53115
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.9580716053831773,
      "learning_rate": 1.7942112935774153e-06,
      "loss": 0.3663,
      "step": 53120
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.088798366856965,
      "learning_rate": 1.7932666118488674e-06,
      "loss": 0.3581,
      "step": 53125
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.343914207042321,
      "learning_rate": 1.7923221245351168e-06,
      "loss": 0.3606,
      "step": 53130
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.813265053430516,
      "learning_rate": 1.7913778316934293e-06,
      "loss": 0.3442,
      "step": 53135
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.884118493906515,
      "learning_rate": 1.7904337333810512e-06,
      "loss": 0.3262,
      "step": 53140
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.248908872095434,
      "learning_rate": 1.789489829655221e-06,
      "loss": 0.3877,
      "step": 53145
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.39241482898308,
      "learning_rate": 1.7885461205731664e-06,
      "loss": 0.3351,
      "step": 53150
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.386977583539001,
      "learning_rate": 1.787602606192102e-06,
      "loss": 0.3398,
      "step": 53155
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.181755995726041,
      "learning_rate": 1.7866592865692272e-06,
      "loss": 0.3642,
      "step": 53160
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.402468719281442,
      "learning_rate": 1.7857161617617352e-06,
      "loss": 0.3671,
      "step": 53165
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.515984720811214,
      "learning_rate": 1.7847732318268063e-06,
      "loss": 0.3666,
      "step": 53170
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.184225551302388,
      "learning_rate": 1.783830496821603e-06,
      "loss": 0.3667,
      "step": 53175
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.137313444840027,
      "learning_rate": 1.7828879568032835e-06,
      "loss": 0.3715,
      "step": 53180
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.241981363542599,
      "learning_rate": 1.7819456118289928e-06,
      "loss": 0.3709,
      "step": 53185
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.437289420148114,
      "learning_rate": 1.7810034619558592e-06,
      "loss": 0.3566,
      "step": 53190
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.647701738805234,
      "learning_rate": 1.7800615072410032e-06,
      "loss": 0.3565,
      "step": 53195
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.720576199353198,
      "learning_rate": 1.7791197477415333e-06,
      "loss": 0.3913,
      "step": 53200
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.453086032747109,
      "learning_rate": 1.7781781835145479e-06,
      "loss": 0.3895,
      "step": 53205
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.116628271761173,
      "learning_rate": 1.777236814617127e-06,
      "loss": 0.3938,
      "step": 53210
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.885827225269734,
      "learning_rate": 1.7762956411063454e-06,
      "loss": 0.3683,
      "step": 53215
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.0550655185288145,
      "learning_rate": 1.7753546630392644e-06,
      "loss": 0.3717,
      "step": 53220
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.56891051689047,
      "learning_rate": 1.7744138804729305e-06,
      "loss": 0.3645,
      "step": 53225
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.9699939071558257,
      "learning_rate": 1.7734732934643816e-06,
      "loss": 0.3348,
      "step": 53230
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.097134850863129,
      "learning_rate": 1.7725329020706443e-06,
      "loss": 0.3631,
      "step": 53235
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.038300638656339,
      "learning_rate": 1.7715927063487293e-06,
      "loss": 0.3229,
      "step": 53240
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.633620662524091,
      "learning_rate": 1.7706527063556384e-06,
      "loss": 0.3886,
      "step": 53245
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.097900884047093,
      "learning_rate": 1.7697129021483616e-06,
      "loss": 0.3472,
      "step": 53250
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.195587040045473,
      "learning_rate": 1.7687732937838787e-06,
      "loss": 0.3725,
      "step": 53255
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.299414692616899,
      "learning_rate": 1.7678338813191508e-06,
      "loss": 0.3914,
      "step": 53260
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.7194205865236696,
      "learning_rate": 1.766894664811134e-06,
      "loss": 0.3432,
      "step": 53265
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.278337404970941,
      "learning_rate": 1.7659556443167725e-06,
      "loss": 0.3571,
      "step": 53270
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.933569454614378,
      "learning_rate": 1.7650168198929919e-06,
      "loss": 0.3578,
      "step": 53275
    },
    {
      "epoch": 2.25,
      "grad_norm": 4.194446581644329,
      "learning_rate": 1.7640781915967126e-06,
      "loss": 0.3421,
      "step": 53280
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.33961023645128,
      "learning_rate": 1.7631397594848425e-06,
      "loss": 0.3732,
      "step": 53285
    },
    {
      "epoch": 2.26,
      "grad_norm": 6.134170479546423,
      "learning_rate": 1.7622015236142742e-06,
      "loss": 0.3758,
      "step": 53290
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.3237712335239005,
      "learning_rate": 1.761263484041889e-06,
      "loss": 0.3495,
      "step": 53295
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.7699836841176357,
      "learning_rate": 1.7603256408245584e-06,
      "loss": 0.3116,
      "step": 53300
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.0464377712993755,
      "learning_rate": 1.7593879940191433e-06,
      "loss": 0.3389,
      "step": 53305
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.3543430064408915,
      "learning_rate": 1.7584505436824867e-06,
      "loss": 0.3798,
      "step": 53310
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.242050900284283,
      "learning_rate": 1.7575132898714253e-06,
      "loss": 0.3731,
      "step": 53315
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.507600767474456,
      "learning_rate": 1.7565762326427843e-06,
      "loss": 0.3511,
      "step": 53320
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.376689386220052,
      "learning_rate": 1.7556393720533703e-06,
      "loss": 0.3666,
      "step": 53325
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.063002731053875,
      "learning_rate": 1.7547027081599848e-06,
      "loss": 0.3545,
      "step": 53330
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.9913007451936484,
      "learning_rate": 1.7537662410194145e-06,
      "loss": 0.3344,
      "step": 53335
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.9030341251991154,
      "learning_rate": 1.7528299706884367e-06,
      "loss": 0.3767,
      "step": 53340
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.0020783429591,
      "learning_rate": 1.7518938972238114e-06,
      "loss": 0.3446,
      "step": 53345
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.2720730395942725,
      "learning_rate": 1.7509580206822912e-06,
      "loss": 0.3699,
      "step": 53350
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.695300767476802,
      "learning_rate": 1.750022341120618e-06,
      "loss": 0.3562,
      "step": 53355
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.920804392162004,
      "learning_rate": 1.749086858595515e-06,
      "loss": 0.349,
      "step": 53360
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.9897560495512487,
      "learning_rate": 1.7481515731637006e-06,
      "loss": 0.3567,
      "step": 53365
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.600349154542182,
      "learning_rate": 1.7472164848818785e-06,
      "loss": 0.3642,
      "step": 53370
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.027130459451585,
      "learning_rate": 1.7462815938067385e-06,
      "loss": 0.3519,
      "step": 53375
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.1685124606383415,
      "learning_rate": 1.7453468999949607e-06,
      "loss": 0.3565,
      "step": 53380
    },
    {
      "epoch": 2.26,
      "grad_norm": 6.5330676899466695,
      "learning_rate": 1.744412403503214e-06,
      "loss": 0.3754,
      "step": 53385
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.22537501307672,
      "learning_rate": 1.7434781043881549e-06,
      "loss": 0.368,
      "step": 53390
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.25519038338382,
      "learning_rate": 1.7425440027064244e-06,
      "loss": 0.3507,
      "step": 53395
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.196358007450328,
      "learning_rate": 1.7416100985146561e-06,
      "loss": 0.3561,
      "step": 53400
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.330358392893522,
      "learning_rate": 1.7406763918694714e-06,
      "loss": 0.3702,
      "step": 53405
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.7685175680526726,
      "learning_rate": 1.7397428828274743e-06,
      "loss": 0.3679,
      "step": 53410
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.337856249867497,
      "learning_rate": 1.738809571445263e-06,
      "loss": 0.3404,
      "step": 53415
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.43433399364858,
      "learning_rate": 1.7378764577794227e-06,
      "loss": 0.3406,
      "step": 53420
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.479431797562968,
      "learning_rate": 1.7369435418865226e-06,
      "loss": 0.3673,
      "step": 53425
    },
    {
      "epoch": 2.26,
      "grad_norm": 6.1348623849809885,
      "learning_rate": 1.7360108238231233e-06,
      "loss": 0.3483,
      "step": 53430
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.241709787499552,
      "learning_rate": 1.7350783036457742e-06,
      "loss": 0.3483,
      "step": 53435
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.9106016096357057,
      "learning_rate": 1.7341459814110112e-06,
      "loss": 0.358,
      "step": 53440
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.09202162984694,
      "learning_rate": 1.7332138571753565e-06,
      "loss": 0.3698,
      "step": 53445
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.026678978131323,
      "learning_rate": 1.7322819309953243e-06,
      "loss": 0.3405,
      "step": 53450
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.720973626143717,
      "learning_rate": 1.7313502029274116e-06,
      "loss": 0.3489,
      "step": 53455
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.028549137275551,
      "learning_rate": 1.7304186730281097e-06,
      "loss": 0.3794,
      "step": 53460
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.154253964684513,
      "learning_rate": 1.729487341353891e-06,
      "loss": 0.3486,
      "step": 53465
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.297762908427482,
      "learning_rate": 1.7285562079612212e-06,
      "loss": 0.3736,
      "step": 53470
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.529379423898424,
      "learning_rate": 1.727625272906554e-06,
      "loss": 0.3435,
      "step": 53475
    },
    {
      "epoch": 2.26,
      "grad_norm": 6.0056473662193595,
      "learning_rate": 1.7266945362463257e-06,
      "loss": 0.3606,
      "step": 53480
    },
    {
      "epoch": 2.26,
      "grad_norm": 6.851484476646318,
      "learning_rate": 1.7257639980369657e-06,
      "loss": 0.3703,
      "step": 53485
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.737436852150534,
      "learning_rate": 1.7248336583348917e-06,
      "loss": 0.3383,
      "step": 53490
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.116955231639267,
      "learning_rate": 1.723903517196504e-06,
      "loss": 0.3493,
      "step": 53495
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.947191587972658,
      "learning_rate": 1.7229735746781961e-06,
      "loss": 0.3716,
      "step": 53500
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.9423658542145104,
      "learning_rate": 1.7220438308363474e-06,
      "loss": 0.3646,
      "step": 53505
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.534652848145454,
      "learning_rate": 1.7211142857273273e-06,
      "loss": 0.3395,
      "step": 53510
    },
    {
      "epoch": 2.26,
      "grad_norm": 5.417820561507296,
      "learning_rate": 1.7201849394074882e-06,
      "loss": 0.3439,
      "step": 53515
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.6940709494381223,
      "learning_rate": 1.7192557919331754e-06,
      "loss": 0.349,
      "step": 53520
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.517813302195003,
      "learning_rate": 1.7183268433607215e-06,
      "loss": 0.371,
      "step": 53525
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.756244759348609,
      "learning_rate": 1.717398093746443e-06,
      "loss": 0.346,
      "step": 53530
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.666612902245065,
      "learning_rate": 1.7164695431466493e-06,
      "loss": 0.357,
      "step": 53535
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.142042160920247,
      "learning_rate": 1.7155411916176363e-06,
      "loss": 0.3703,
      "step": 53540
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.771070128081749,
      "learning_rate": 1.7146130392156851e-06,
      "loss": 0.3289,
      "step": 53545
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.952037272240819,
      "learning_rate": 1.7136850859970673e-06,
      "loss": 0.3618,
      "step": 53550
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.9875735536578905,
      "learning_rate": 1.712757332018043e-06,
      "loss": 0.3435,
      "step": 53555
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.248130150239056,
      "learning_rate": 1.7118297773348603e-06,
      "loss": 0.3745,
      "step": 53560
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.693447740966035,
      "learning_rate": 1.7109024220037508e-06,
      "loss": 0.3574,
      "step": 53565
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.001178525810628,
      "learning_rate": 1.7099752660809393e-06,
      "loss": 0.3604,
      "step": 53570
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.15567987386065,
      "learning_rate": 1.709048309622638e-06,
      "loss": 0.3608,
      "step": 53575
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.458772579463961,
      "learning_rate": 1.708121552685042e-06,
      "loss": 0.3659,
      "step": 53580
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.6441195864043547,
      "learning_rate": 1.70719499532434e-06,
      "loss": 0.364,
      "step": 53585
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.60218399901867,
      "learning_rate": 1.7062686375967075e-06,
      "loss": 0.3646,
      "step": 53590
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.474015576600486,
      "learning_rate": 1.705342479558304e-06,
      "loss": 0.3669,
      "step": 53595
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.960015999850594,
      "learning_rate": 1.7044165212652814e-06,
      "loss": 0.3586,
      "step": 53600
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.120008156315747,
      "learning_rate": 1.703490762773779e-06,
      "loss": 0.3485,
      "step": 53605
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.671993652432476,
      "learning_rate": 1.7025652041399216e-06,
      "loss": 0.3586,
      "step": 53610
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.8846380423301134,
      "learning_rate": 1.7016398454198212e-06,
      "loss": 0.367,
      "step": 53615
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.633219516723773,
      "learning_rate": 1.7007146866695817e-06,
      "loss": 0.3408,
      "step": 53620
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.177864678116925,
      "learning_rate": 1.6997897279452936e-06,
      "loss": 0.3755,
      "step": 53625
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.7608927920769686,
      "learning_rate": 1.6988649693030318e-06,
      "loss": 0.3813,
      "step": 53630
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.253026317541189,
      "learning_rate": 1.697940410798863e-06,
      "loss": 0.362,
      "step": 53635
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.555835182519368,
      "learning_rate": 1.6970160524888407e-06,
      "loss": 0.3417,
      "step": 53640
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.9950158001306777,
      "learning_rate": 1.6960918944290071e-06,
      "loss": 0.3737,
      "step": 53645
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.436047335095203,
      "learning_rate": 1.6951679366753887e-06,
      "loss": 0.3515,
      "step": 53650
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.410036756666341,
      "learning_rate": 1.6942441792840037e-06,
      "loss": 0.3437,
      "step": 53655
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.7153311080252975,
      "learning_rate": 1.6933206223108577e-06,
      "loss": 0.4044,
      "step": 53660
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.24268534774769,
      "learning_rate": 1.6923972658119414e-06,
      "loss": 0.3631,
      "step": 53665
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.904438638145542,
      "learning_rate": 1.6914741098432357e-06,
      "loss": 0.3463,
      "step": 53670
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.955096463219155,
      "learning_rate": 1.6905511544607105e-06,
      "loss": 0.3788,
      "step": 53675
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.4841771134210315,
      "learning_rate": 1.6896283997203195e-06,
      "loss": 0.3582,
      "step": 53680
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.439059606654289,
      "learning_rate": 1.688705845678007e-06,
      "loss": 0.3702,
      "step": 53685
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.117388715887991,
      "learning_rate": 1.6877834923897057e-06,
      "loss": 0.3728,
      "step": 53690
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.202590326056345,
      "learning_rate": 1.686861339911336e-06,
      "loss": 0.365,
      "step": 53695
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.358550415251707,
      "learning_rate": 1.685939388298803e-06,
      "loss": 0.3402,
      "step": 53700
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.000296034883441,
      "learning_rate": 1.6850176376080025e-06,
      "loss": 0.3448,
      "step": 53705
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.139867775382686,
      "learning_rate": 1.6840960878948198e-06,
      "loss": 0.3394,
      "step": 53710
    },
    {
      "epoch": 2.27,
      "grad_norm": 6.22044320979231,
      "learning_rate": 1.6831747392151226e-06,
      "loss": 0.3514,
      "step": 53715
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.450863336664558,
      "learning_rate": 1.6822535916247707e-06,
      "loss": 0.3701,
      "step": 53720
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.044017424452533,
      "learning_rate": 1.6813326451796125e-06,
      "loss": 0.3698,
      "step": 53725
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.088624488258786,
      "learning_rate": 1.680411899935479e-06,
      "loss": 0.358,
      "step": 53730
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.889949864823278,
      "learning_rate": 1.679491355948194e-06,
      "loss": 0.3497,
      "step": 53735
    },
    {
      "epoch": 2.27,
      "grad_norm": 5.89628743852143,
      "learning_rate": 1.6785710132735665e-06,
      "loss": 0.3449,
      "step": 53740
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.331833851205384,
      "learning_rate": 1.6776508719673972e-06,
      "loss": 0.3403,
      "step": 53745
    },
    {
      "epoch": 2.27,
      "grad_norm": 4.432662568383738,
      "learning_rate": 1.6767309320854675e-06,
      "loss": 0.3597,
      "step": 53750
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.005956186508215,
      "learning_rate": 1.675811193683552e-06,
      "loss": 0.3582,
      "step": 53755
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.165452087193429,
      "learning_rate": 1.674891656817414e-06,
      "loss": 0.3514,
      "step": 53760
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.541740515105331,
      "learning_rate": 1.6739723215427989e-06,
      "loss": 0.3393,
      "step": 53765
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.393627239289327,
      "learning_rate": 1.6730531879154466e-06,
      "loss": 0.3717,
      "step": 53770
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.177844615958607,
      "learning_rate": 1.6721342559910776e-06,
      "loss": 0.3664,
      "step": 53775
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.8877402778645456,
      "learning_rate": 1.671215525825408e-06,
      "loss": 0.3792,
      "step": 53780
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.082938698169374,
      "learning_rate": 1.670296997474134e-06,
      "loss": 0.3672,
      "step": 53785
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.761175571373643,
      "learning_rate": 1.6693786709929456e-06,
      "loss": 0.3511,
      "step": 53790
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.957586432011262,
      "learning_rate": 1.6684605464375186e-06,
      "loss": 0.353,
      "step": 53795
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.862241561782925,
      "learning_rate": 1.667542623863514e-06,
      "loss": 0.3819,
      "step": 53800
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.7832819532305297,
      "learning_rate": 1.666624903326584e-06,
      "loss": 0.3572,
      "step": 53805
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.809256140248226,
      "learning_rate": 1.6657073848823674e-06,
      "loss": 0.3318,
      "step": 53810
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.678544404311857,
      "learning_rate": 1.664790068586492e-06,
      "loss": 0.3448,
      "step": 53815
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.968567111848225,
      "learning_rate": 1.6638729544945692e-06,
      "loss": 0.3687,
      "step": 53820
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.1825118829528165,
      "learning_rate": 1.662956042662202e-06,
      "loss": 0.3569,
      "step": 53825
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.373804070411063,
      "learning_rate": 1.6620393331449824e-06,
      "loss": 0.356,
      "step": 53830
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.730073924046188,
      "learning_rate": 1.661122825998484e-06,
      "loss": 0.3503,
      "step": 53835
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.324773873405594,
      "learning_rate": 1.6602065212782737e-06,
      "loss": 0.3527,
      "step": 53840
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.9049361054855707,
      "learning_rate": 1.6592904190399056e-06,
      "loss": 0.3586,
      "step": 53845
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.529696152015119,
      "learning_rate": 1.6583745193389172e-06,
      "loss": 0.3777,
      "step": 53850
    },
    {
      "epoch": 2.28,
      "grad_norm": 6.295943552280052,
      "learning_rate": 1.657458822230839e-06,
      "loss": 0.3517,
      "step": 53855
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.8195071766793327,
      "learning_rate": 1.6565433277711884e-06,
      "loss": 0.3741,
      "step": 53860
    },
    {
      "epoch": 2.28,
      "grad_norm": 5.976281948370772,
      "learning_rate": 1.655628036015466e-06,
      "loss": 0.3712,
      "step": 53865
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.848172801210396,
      "learning_rate": 1.6547129470191642e-06,
      "loss": 0.3372,
      "step": 53870
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.9888022135396084,
      "learning_rate": 1.6537980608377625e-06,
      "loss": 0.34,
      "step": 53875
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.884714561540338,
      "learning_rate": 1.6528833775267294e-06,
      "loss": 0.3566,
      "step": 53880
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.861421660821195,
      "learning_rate": 1.651968897141516e-06,
      "loss": 0.3564,
      "step": 53885
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.029863730122377,
      "learning_rate": 1.651054619737567e-06,
      "loss": 0.3521,
      "step": 53890
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.0401136785269784,
      "learning_rate": 1.650140545370313e-06,
      "loss": 0.3689,
      "step": 53895
    },
    {
      "epoch": 2.28,
      "grad_norm": 5.992637725002927,
      "learning_rate": 1.649226674095169e-06,
      "loss": 0.3786,
      "step": 53900
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.603948815781125,
      "learning_rate": 1.6483130059675418e-06,
      "loss": 0.3807,
      "step": 53905
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.314905509092832,
      "learning_rate": 1.6473995410428256e-06,
      "loss": 0.3813,
      "step": 53910
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.742272897624811,
      "learning_rate": 1.6464862793763987e-06,
      "loss": 0.3423,
      "step": 53915
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.140519683103931,
      "learning_rate": 1.6455732210236302e-06,
      "loss": 0.354,
      "step": 53920
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.9568868787192897,
      "learning_rate": 1.6446603660398763e-06,
      "loss": 0.3618,
      "step": 53925
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.632563943908224,
      "learning_rate": 1.6437477144804831e-06,
      "loss": 0.3693,
      "step": 53930
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.511488474185682,
      "learning_rate": 1.6428352664007796e-06,
      "loss": 0.3894,
      "step": 53935
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.9434512065821616,
      "learning_rate": 1.6419230218560834e-06,
      "loss": 0.3592,
      "step": 53940
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.8746032384581377,
      "learning_rate": 1.6410109809017028e-06,
      "loss": 0.3417,
      "step": 53945
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.348400461265312,
      "learning_rate": 1.6400991435929336e-06,
      "loss": 0.3358,
      "step": 53950
    },
    {
      "epoch": 2.28,
      "grad_norm": 6.545580126116169,
      "learning_rate": 1.639187509985055e-06,
      "loss": 0.3457,
      "step": 53955
    },
    {
      "epoch": 2.28,
      "grad_norm": 6.090312521635988,
      "learning_rate": 1.6382760801333381e-06,
      "loss": 0.3776,
      "step": 53960
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.490438032829979,
      "learning_rate": 1.637364854093042e-06,
      "loss": 0.3778,
      "step": 53965
    },
    {
      "epoch": 2.28,
      "grad_norm": 6.061629522302544,
      "learning_rate": 1.636453831919408e-06,
      "loss": 0.3633,
      "step": 53970
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.938051934754005,
      "learning_rate": 1.6355430136676702e-06,
      "loss": 0.3592,
      "step": 53975
    },
    {
      "epoch": 2.28,
      "grad_norm": 6.940753640065751,
      "learning_rate": 1.6346323993930513e-06,
      "loss": 0.3467,
      "step": 53980
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.1890036538453215,
      "learning_rate": 1.6337219891507549e-06,
      "loss": 0.3504,
      "step": 53985
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.9636722645967186,
      "learning_rate": 1.6328117829959795e-06,
      "loss": 0.3724,
      "step": 53990
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.440576461666571,
      "learning_rate": 1.6319017809839066e-06,
      "loss": 0.3636,
      "step": 53995
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.980594556174794,
      "learning_rate": 1.6309919831697092e-06,
      "loss": 0.353,
      "step": 54000
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.38553438441278,
      "learning_rate": 1.6300823896085428e-06,
      "loss": 0.3745,
      "step": 54005
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.89829446284253,
      "learning_rate": 1.6291730003555545e-06,
      "loss": 0.3358,
      "step": 54010
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.086969881960062,
      "learning_rate": 1.6282638154658798e-06,
      "loss": 0.3262,
      "step": 54015
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.195947151810937,
      "learning_rate": 1.6273548349946366e-06,
      "loss": 0.3579,
      "step": 54020
    },
    {
      "epoch": 2.29,
      "grad_norm": 5.698918289129321,
      "learning_rate": 1.626446058996935e-06,
      "loss": 0.3538,
      "step": 54025
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.583908630926258,
      "learning_rate": 1.625537487527874e-06,
      "loss": 0.3641,
      "step": 54030
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.0732035707624,
      "learning_rate": 1.624629120642533e-06,
      "loss": 0.3575,
      "step": 54035
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.12736534965869,
      "learning_rate": 1.6237209583959862e-06,
      "loss": 0.3829,
      "step": 54040
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.341394177900301,
      "learning_rate": 1.622813000843293e-06,
      "loss": 0.3453,
      "step": 54045
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.4463287160370895,
      "learning_rate": 1.6219052480395008e-06,
      "loss": 0.3461,
      "step": 54050
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.736034861548534,
      "learning_rate": 1.6209977000396415e-06,
      "loss": 0.3475,
      "step": 54055
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.941732501968647,
      "learning_rate": 1.620090356898738e-06,
      "loss": 0.3634,
      "step": 54060
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.113572760744246,
      "learning_rate": 1.6191832186718026e-06,
      "loss": 0.363,
      "step": 54065
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.316465409997266,
      "learning_rate": 1.6182762854138284e-06,
      "loss": 0.3669,
      "step": 54070
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.902919805053425,
      "learning_rate": 1.6173695571798016e-06,
      "loss": 0.3724,
      "step": 54075
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.5987229966565293,
      "learning_rate": 1.6164630340246962e-06,
      "loss": 0.3544,
      "step": 54080
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.638539649125324,
      "learning_rate": 1.6155567160034692e-06,
      "loss": 0.3779,
      "step": 54085
    },
    {
      "epoch": 2.29,
      "grad_norm": 5.080423649524689,
      "learning_rate": 1.614650603171069e-06,
      "loss": 0.3598,
      "step": 54090
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.096004492737438,
      "learning_rate": 1.6137446955824327e-06,
      "loss": 0.3701,
      "step": 54095
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.9395172309290376,
      "learning_rate": 1.612838993292481e-06,
      "loss": 0.3298,
      "step": 54100
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.510374501224488,
      "learning_rate": 1.6119334963561218e-06,
      "loss": 0.375,
      "step": 54105
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.7503732736481483,
      "learning_rate": 1.6110282048282554e-06,
      "loss": 0.3841,
      "step": 54110
    },
    {
      "epoch": 2.29,
      "grad_norm": 6.05969682723419,
      "learning_rate": 1.610123118763766e-06,
      "loss": 0.3658,
      "step": 54115
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.92240969821514,
      "learning_rate": 1.6092182382175287e-06,
      "loss": 0.3448,
      "step": 54120
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.969431447082445,
      "learning_rate": 1.6083135632444002e-06,
      "loss": 0.3707,
      "step": 54125
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.865708207745226,
      "learning_rate": 1.6074090938992294e-06,
      "loss": 0.3731,
      "step": 54130
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.678214546814403,
      "learning_rate": 1.606504830236854e-06,
      "loss": 0.3281,
      "step": 54135
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.147521254906464,
      "learning_rate": 1.605600772312093e-06,
      "loss": 0.3434,
      "step": 54140
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.892202008027043,
      "learning_rate": 1.6046969201797585e-06,
      "loss": 0.3651,
      "step": 54145
    },
    {
      "epoch": 2.29,
      "grad_norm": 5.662867757518447,
      "learning_rate": 1.6037932738946504e-06,
      "loss": 0.3849,
      "step": 54150
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.533982260557702,
      "learning_rate": 1.6028898335115512e-06,
      "loss": 0.3697,
      "step": 54155
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.9329376652717856,
      "learning_rate": 1.601986599085234e-06,
      "loss": 0.3381,
      "step": 54160
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.928528047824139,
      "learning_rate": 1.6010835706704619e-06,
      "loss": 0.3533,
      "step": 54165
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.274240210819913,
      "learning_rate": 1.60018074832198e-06,
      "loss": 0.321,
      "step": 54170
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.707991744920607,
      "learning_rate": 1.5992781320945244e-06,
      "loss": 0.3396,
      "step": 54175
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.676013712742363,
      "learning_rate": 1.5983757220428182e-06,
      "loss": 0.3376,
      "step": 54180
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.939713354876155,
      "learning_rate": 1.5974735182215745e-06,
      "loss": 0.3757,
      "step": 54185
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.242562341679641,
      "learning_rate": 1.5965715206854865e-06,
      "loss": 0.3601,
      "step": 54190
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.066942725453382,
      "learning_rate": 1.5956697294892426e-06,
      "loss": 0.3715,
      "step": 54195
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.109300405520384,
      "learning_rate": 1.5947681446875168e-06,
      "loss": 0.3469,
      "step": 54200
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.252417062689107,
      "learning_rate": 1.593866766334966e-06,
      "loss": 0.3737,
      "step": 54205
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.2258718928596775,
      "learning_rate": 1.5929655944862405e-06,
      "loss": 0.3621,
      "step": 54210
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.036467177114945,
      "learning_rate": 1.5920646291959768e-06,
      "loss": 0.3386,
      "step": 54215
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.9361625880776874,
      "learning_rate": 1.5911638705187943e-06,
      "loss": 0.352,
      "step": 54220
    },
    {
      "epoch": 2.29,
      "grad_norm": 3.953098875507561,
      "learning_rate": 1.5902633185093058e-06,
      "loss": 0.3347,
      "step": 54225
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.964339110373232,
      "learning_rate": 1.5893629732221084e-06,
      "loss": 0.3668,
      "step": 54230
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.647972259228924,
      "learning_rate": 1.5884628347117887e-06,
      "loss": 0.367,
      "step": 54235
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.7886093819231967,
      "learning_rate": 1.587562903032917e-06,
      "loss": 0.3539,
      "step": 54240
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.171963771955341,
      "learning_rate": 1.5866631782400549e-06,
      "loss": 0.3662,
      "step": 54245
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.8748684689839,
      "learning_rate": 1.5857636603877518e-06,
      "loss": 0.3691,
      "step": 54250
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.9899482962934925,
      "learning_rate": 1.5848643495305389e-06,
      "loss": 0.3676,
      "step": 54255
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.861892248603292,
      "learning_rate": 1.5839652457229433e-06,
      "loss": 0.3909,
      "step": 54260
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.832836269589828,
      "learning_rate": 1.5830663490194703e-06,
      "loss": 0.3186,
      "step": 54265
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.576846228157823,
      "learning_rate": 1.5821676594746211e-06,
      "loss": 0.3345,
      "step": 54270
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.243372339194075,
      "learning_rate": 1.5812691771428778e-06,
      "loss": 0.3501,
      "step": 54275
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.799899103952548,
      "learning_rate": 1.5803709020787144e-06,
      "loss": 0.3661,
      "step": 54280
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.9924482757247732,
      "learning_rate": 1.5794728343365916e-06,
      "loss": 0.3308,
      "step": 54285
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.9579642517354436,
      "learning_rate": 1.5785749739709538e-06,
      "loss": 0.3516,
      "step": 54290
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.6627928235415723,
      "learning_rate": 1.5776773210362373e-06,
      "loss": 0.3574,
      "step": 54295
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.621492241620413,
      "learning_rate": 1.576779875586864e-06,
      "loss": 0.3765,
      "step": 54300
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.273438159161337,
      "learning_rate": 1.5758826376772452e-06,
      "loss": 0.3492,
      "step": 54305
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.888323589021636,
      "learning_rate": 1.5749856073617743e-06,
      "loss": 0.3863,
      "step": 54310
    },
    {
      "epoch": 2.3,
      "grad_norm": 6.8475785737430135,
      "learning_rate": 1.574088784694837e-06,
      "loss": 0.3925,
      "step": 54315
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.2768154266216465,
      "learning_rate": 1.5731921697308072e-06,
      "loss": 0.3478,
      "step": 54320
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.644933878248935,
      "learning_rate": 1.5722957625240403e-06,
      "loss": 0.3456,
      "step": 54325
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.9362747564959126,
      "learning_rate": 1.5713995631288853e-06,
      "loss": 0.3455,
      "step": 54330
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.198280074306882,
      "learning_rate": 1.5705035715996764e-06,
      "loss": 0.3642,
      "step": 54335
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.5689660725406824,
      "learning_rate": 1.5696077879907329e-06,
      "loss": 0.3526,
      "step": 54340
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.383629727664212,
      "learning_rate": 1.5687122123563647e-06,
      "loss": 0.3638,
      "step": 54345
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.111614594638951,
      "learning_rate": 1.5678168447508673e-06,
      "loss": 0.3539,
      "step": 54350
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.748408307940443,
      "learning_rate": 1.566921685228527e-06,
      "loss": 0.3457,
      "step": 54355
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.9945781072614888,
      "learning_rate": 1.5660267338436107e-06,
      "loss": 0.3314,
      "step": 54360
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.040786609556089,
      "learning_rate": 1.5651319906503782e-06,
      "loss": 0.3486,
      "step": 54365
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.258217828415274,
      "learning_rate": 1.564237455703077e-06,
      "loss": 0.3925,
      "step": 54370
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.030547960070974,
      "learning_rate": 1.563343129055937e-06,
      "loss": 0.3454,
      "step": 54375
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.711019105295505,
      "learning_rate": 1.56244901076318e-06,
      "loss": 0.3578,
      "step": 54380
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.293216098227058,
      "learning_rate": 1.5615551008790158e-06,
      "loss": 0.3568,
      "step": 54385
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.345646803166752,
      "learning_rate": 1.5606613994576358e-06,
      "loss": 0.3434,
      "step": 54390
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.870418942169404,
      "learning_rate": 1.559767906553224e-06,
      "loss": 0.3383,
      "step": 54395
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.917107018287111,
      "learning_rate": 1.5588746222199508e-06,
      "loss": 0.3568,
      "step": 54400
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.608625122301406,
      "learning_rate": 1.5579815465119746e-06,
      "loss": 0.3297,
      "step": 54405
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.237061293918931,
      "learning_rate": 1.5570886794834366e-06,
      "loss": 0.3553,
      "step": 54410
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.890301608714642,
      "learning_rate": 1.5561960211884709e-06,
      "loss": 0.3636,
      "step": 54415
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.672545805304668,
      "learning_rate": 1.5553035716811975e-06,
      "loss": 0.3545,
      "step": 54420
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.285513502410438,
      "learning_rate": 1.5544113310157222e-06,
      "loss": 0.3621,
      "step": 54425
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.195289254714602,
      "learning_rate": 1.5535192992461372e-06,
      "loss": 0.3431,
      "step": 54430
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.975971656082313,
      "learning_rate": 1.5526274764265248e-06,
      "loss": 0.3564,
      "step": 54435
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.6232849208679685,
      "learning_rate": 1.5517358626109557e-06,
      "loss": 0.3481,
      "step": 54440
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.8645433259674706,
      "learning_rate": 1.550844457853482e-06,
      "loss": 0.3708,
      "step": 54445
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.030236020665636,
      "learning_rate": 1.5499532622081497e-06,
      "loss": 0.3477,
      "step": 54450
    },
    {
      "epoch": 2.3,
      "grad_norm": 3.8824498299346897,
      "learning_rate": 1.5490622757289898e-06,
      "loss": 0.3585,
      "step": 54455
    },
    {
      "epoch": 2.3,
      "grad_norm": 4.55262600883059,
      "learning_rate": 1.5481714984700175e-06,
      "loss": 0.335,
      "step": 54460
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.217657656721012,
      "learning_rate": 1.5472809304852398e-06,
      "loss": 0.3586,
      "step": 54465
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.9833721421229535,
      "learning_rate": 1.5463905718286503e-06,
      "loss": 0.3441,
      "step": 54470
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.768387392257179,
      "learning_rate": 1.5455004225542258e-06,
      "loss": 0.3505,
      "step": 54475
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.126358559302379,
      "learning_rate": 1.5446104827159352e-06,
      "loss": 0.3399,
      "step": 54480
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.1187375614476265,
      "learning_rate": 1.5437207523677332e-06,
      "loss": 0.3527,
      "step": 54485
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.905465351404854,
      "learning_rate": 1.5428312315635623e-06,
      "loss": 0.3341,
      "step": 54490
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.426213750926166,
      "learning_rate": 1.5419419203573493e-06,
      "loss": 0.3896,
      "step": 54495
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.271299589331983,
      "learning_rate": 1.5410528188030117e-06,
      "loss": 0.3403,
      "step": 54500
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.613707296371552,
      "learning_rate": 1.5401639269544538e-06,
      "loss": 0.3567,
      "step": 54505
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.09305132112167,
      "learning_rate": 1.5392752448655652e-06,
      "loss": 0.3676,
      "step": 54510
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.488485063581074,
      "learning_rate": 1.5383867725902236e-06,
      "loss": 0.3362,
      "step": 54515
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.8484101083788125,
      "learning_rate": 1.5374985101822976e-06,
      "loss": 0.3652,
      "step": 54520
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.8554772214689526,
      "learning_rate": 1.536610457695636e-06,
      "loss": 0.3479,
      "step": 54525
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.712726932185907,
      "learning_rate": 1.535722615184081e-06,
      "loss": 0.3596,
      "step": 54530
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.846021487830665,
      "learning_rate": 1.534834982701459e-06,
      "loss": 0.3692,
      "step": 54535
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.242382626604983,
      "learning_rate": 1.533947560301587e-06,
      "loss": 0.3297,
      "step": 54540
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.68054097154349,
      "learning_rate": 1.5330603480382627e-06,
      "loss": 0.3594,
      "step": 54545
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.1638448492734765,
      "learning_rate": 1.5321733459652776e-06,
      "loss": 0.3747,
      "step": 54550
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.8901708019677828,
      "learning_rate": 1.5312865541364091e-06,
      "loss": 0.3532,
      "step": 54555
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.184278415049343,
      "learning_rate": 1.530399972605418e-06,
      "loss": 0.3414,
      "step": 54560
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.194221233850725,
      "learning_rate": 1.529513601426057e-06,
      "loss": 0.3571,
      "step": 54565
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.057713950237074,
      "learning_rate": 1.5286274406520647e-06,
      "loss": 0.3403,
      "step": 54570
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.290901052679981,
      "learning_rate": 1.527741490337164e-06,
      "loss": 0.3651,
      "step": 54575
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.8386264888140786,
      "learning_rate": 1.52685575053507e-06,
      "loss": 0.3461,
      "step": 54580
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.030908843819734,
      "learning_rate": 1.5259702212994803e-06,
      "loss": 0.3616,
      "step": 54585
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.58900924208917,
      "learning_rate": 1.5250849026840847e-06,
      "loss": 0.3449,
      "step": 54590
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.165105511076327,
      "learning_rate": 1.524199794742554e-06,
      "loss": 0.3585,
      "step": 54595
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.949481408332809,
      "learning_rate": 1.5233148975285511e-06,
      "loss": 0.3779,
      "step": 54600
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.8051735822646995,
      "learning_rate": 1.5224302110957256e-06,
      "loss": 0.3275,
      "step": 54605
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.234490750159869,
      "learning_rate": 1.5215457354977142e-06,
      "loss": 0.3627,
      "step": 54610
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.696538255496593,
      "learning_rate": 1.5206614707881367e-06,
      "loss": 0.3456,
      "step": 54615
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.142745589436863,
      "learning_rate": 1.519777417020606e-06,
      "loss": 0.3626,
      "step": 54620
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.455452016624465,
      "learning_rate": 1.518893574248721e-06,
      "loss": 0.3481,
      "step": 54625
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.756356387535101,
      "learning_rate": 1.5180099425260625e-06,
      "loss": 0.3673,
      "step": 54630
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.132544145561781,
      "learning_rate": 1.5171265219062048e-06,
      "loss": 0.3539,
      "step": 54635
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.224546957526565,
      "learning_rate": 1.516243312442709e-06,
      "loss": 0.3725,
      "step": 54640
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.417291184044669,
      "learning_rate": 1.515360314189117e-06,
      "loss": 0.3601,
      "step": 54645
    },
    {
      "epoch": 2.31,
      "grad_norm": 6.015782224684628,
      "learning_rate": 1.5144775271989653e-06,
      "loss": 0.3319,
      "step": 54650
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.818772574066071,
      "learning_rate": 1.513594951525774e-06,
      "loss": 0.3899,
      "step": 54655
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.155469987480383,
      "learning_rate": 1.5127125872230536e-06,
      "loss": 0.3472,
      "step": 54660
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.129356873051151,
      "learning_rate": 1.5118304343442947e-06,
      "loss": 0.3573,
      "step": 54665
    },
    {
      "epoch": 2.31,
      "grad_norm": 5.24383810235975,
      "learning_rate": 1.5109484929429818e-06,
      "loss": 0.3569,
      "step": 54670
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.729161233369576,
      "learning_rate": 1.5100667630725856e-06,
      "loss": 0.3514,
      "step": 54675
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.661766609408095,
      "learning_rate": 1.5091852447865607e-06,
      "loss": 0.3387,
      "step": 54680
    },
    {
      "epoch": 2.31,
      "grad_norm": 7.3151684908094445,
      "learning_rate": 1.508303938138352e-06,
      "loss": 0.3551,
      "step": 54685
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.029977165359439,
      "learning_rate": 1.5074228431813914e-06,
      "loss": 0.3425,
      "step": 54690
    },
    {
      "epoch": 2.31,
      "grad_norm": 4.311875253260442,
      "learning_rate": 1.5065419599690944e-06,
      "loss": 0.3704,
      "step": 54695
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.870749532162497,
      "learning_rate": 1.5056612885548688e-06,
      "loss": 0.387,
      "step": 54700
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.842078782439271,
      "learning_rate": 1.5047808289921056e-06,
      "loss": 0.3701,
      "step": 54705
    },
    {
      "epoch": 2.32,
      "grad_norm": 6.040159327264844,
      "learning_rate": 1.5039005813341873e-06,
      "loss": 0.3776,
      "step": 54710
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.793420440112233,
      "learning_rate": 1.5030205456344772e-06,
      "loss": 0.3645,
      "step": 54715
    },
    {
      "epoch": 2.32,
      "grad_norm": 5.0452165662670065,
      "learning_rate": 1.5021407219463302e-06,
      "loss": 0.374,
      "step": 54720
    },
    {
      "epoch": 2.32,
      "grad_norm": 5.5105079348272366,
      "learning_rate": 1.50126111032309e-06,
      "loss": 0.3564,
      "step": 54725
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.900807702448875,
      "learning_rate": 1.5003817108180807e-06,
      "loss": 0.3663,
      "step": 54730
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.933343415261964,
      "learning_rate": 1.4995025234846205e-06,
      "loss": 0.3652,
      "step": 54735
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.457793374976396,
      "learning_rate": 1.4986235483760126e-06,
      "loss": 0.3318,
      "step": 54740
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.7322497060818405,
      "learning_rate": 1.4977447855455451e-06,
      "loss": 0.3618,
      "step": 54745
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.280188274763539,
      "learning_rate": 1.4968662350464935e-06,
      "loss": 0.3522,
      "step": 54750
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.7128033553288846,
      "learning_rate": 1.4959878969321228e-06,
      "loss": 0.3415,
      "step": 54755
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.490439323355305,
      "learning_rate": 1.4951097712556867e-06,
      "loss": 0.3298,
      "step": 54760
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.189145788249495,
      "learning_rate": 1.4942318580704195e-06,
      "loss": 0.3571,
      "step": 54765
    },
    {
      "epoch": 2.32,
      "grad_norm": 5.145431798540984,
      "learning_rate": 1.4933541574295484e-06,
      "loss": 0.3365,
      "step": 54770
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.5612836306860545,
      "learning_rate": 1.492476669386287e-06,
      "loss": 0.3381,
      "step": 54775
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.323649033873212,
      "learning_rate": 1.4915993939938317e-06,
      "loss": 0.3385,
      "step": 54780
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.740094399127605,
      "learning_rate": 1.4907223313053704e-06,
      "loss": 0.3505,
      "step": 54785
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.251059386330171,
      "learning_rate": 1.4898454813740775e-06,
      "loss": 0.333,
      "step": 54790
    },
    {
      "epoch": 2.32,
      "grad_norm": 5.115704031736812,
      "learning_rate": 1.4889688442531154e-06,
      "loss": 0.3244,
      "step": 54795
    },
    {
      "epoch": 2.32,
      "grad_norm": 6.894299328325798,
      "learning_rate": 1.4880924199956287e-06,
      "loss": 0.3859,
      "step": 54800
    },
    {
      "epoch": 2.32,
      "grad_norm": 5.623584526767408,
      "learning_rate": 1.4872162086547538e-06,
      "loss": 0.3363,
      "step": 54805
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.340472513680994,
      "learning_rate": 1.4863402102836144e-06,
      "loss": 0.3438,
      "step": 54810
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.562430085590519,
      "learning_rate": 1.4854644249353161e-06,
      "loss": 0.3354,
      "step": 54815
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.675050383311123,
      "learning_rate": 1.4845888526629582e-06,
      "loss": 0.3632,
      "step": 54820
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.310334601145407,
      "learning_rate": 1.4837134935196247e-06,
      "loss": 0.3641,
      "step": 54825
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.7025482435075645,
      "learning_rate": 1.4828383475583824e-06,
      "loss": 0.3679,
      "step": 54830
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.293721810385579,
      "learning_rate": 1.481963414832291e-06,
      "loss": 0.3686,
      "step": 54835
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.9702048547296207,
      "learning_rate": 1.4810886953943953e-06,
      "loss": 0.3234,
      "step": 54840
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.8790463707830636,
      "learning_rate": 1.4802141892977284e-06,
      "loss": 0.3703,
      "step": 54845
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.8246093269370887,
      "learning_rate": 1.4793398965953054e-06,
      "loss": 0.3558,
      "step": 54850
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.9693095388012964,
      "learning_rate": 1.4784658173401344e-06,
      "loss": 0.3722,
      "step": 54855
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.559594673145694,
      "learning_rate": 1.4775919515852093e-06,
      "loss": 0.358,
      "step": 54860
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.5047566984407292,
      "learning_rate": 1.4767182993835073e-06,
      "loss": 0.3264,
      "step": 54865
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.015975326022307,
      "learning_rate": 1.475844860787996e-06,
      "loss": 0.3599,
      "step": 54870
    },
    {
      "epoch": 2.32,
      "grad_norm": 6.612837539950845,
      "learning_rate": 1.4749716358516319e-06,
      "loss": 0.3724,
      "step": 54875
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.5090104021344315,
      "learning_rate": 1.4740986246273525e-06,
      "loss": 0.3566,
      "step": 54880
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.848175949299008,
      "learning_rate": 1.4732258271680877e-06,
      "loss": 0.3396,
      "step": 54885
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.145539148320909,
      "learning_rate": 1.4723532435267528e-06,
      "loss": 0.3664,
      "step": 54890
    },
    {
      "epoch": 2.32,
      "grad_norm": 6.2972418476206355,
      "learning_rate": 1.4714808737562513e-06,
      "loss": 0.3325,
      "step": 54895
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.965644271616134,
      "learning_rate": 1.4706087179094691e-06,
      "loss": 0.3274,
      "step": 54900
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.040050757815286,
      "learning_rate": 1.4697367760392855e-06,
      "loss": 0.3478,
      "step": 54905
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.7491681622397164,
      "learning_rate": 1.4688650481985611e-06,
      "loss": 0.3473,
      "step": 54910
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.9314535362925818,
      "learning_rate": 1.467993534440149e-06,
      "loss": 0.3621,
      "step": 54915
    },
    {
      "epoch": 2.32,
      "grad_norm": 4.161599878075956,
      "learning_rate": 1.4671222348168835e-06,
      "loss": 0.3239,
      "step": 54920
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.9503418436398356,
      "learning_rate": 1.4662511493815906e-06,
      "loss": 0.3513,
      "step": 54925
    },
    {
      "epoch": 2.32,
      "grad_norm": 5.302651321008852,
      "learning_rate": 1.465380278187083e-06,
      "loss": 0.334,
      "step": 54930
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.749175738056873,
      "learning_rate": 1.4645096212861559e-06,
      "loss": 0.334,
      "step": 54935
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.9269051051274264,
      "learning_rate": 1.4636391787315968e-06,
      "loss": 0.367,
      "step": 54940
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.499298834293157,
      "learning_rate": 1.462768950576179e-06,
      "loss": 0.3291,
      "step": 54945
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.058165514177957,
      "learning_rate": 1.4618989368726583e-06,
      "loss": 0.3569,
      "step": 54950
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.229824167909258,
      "learning_rate": 1.461029137673784e-06,
      "loss": 0.3643,
      "step": 54955
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.964847259972748,
      "learning_rate": 1.4601595530322882e-06,
      "loss": 0.3739,
      "step": 54960
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.812806702163245,
      "learning_rate": 1.4592901830008932e-06,
      "loss": 0.3725,
      "step": 54965
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.903635657175875,
      "learning_rate": 1.4584210276323036e-06,
      "loss": 0.3636,
      "step": 54970
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.013918381004673,
      "learning_rate": 1.4575520869792153e-06,
      "loss": 0.3402,
      "step": 54975
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.981856069566361,
      "learning_rate": 1.4566833610943104e-06,
      "loss": 0.3348,
      "step": 54980
    },
    {
      "epoch": 2.33,
      "grad_norm": 5.0819493093990955,
      "learning_rate": 1.455814850030255e-06,
      "loss": 0.3851,
      "step": 54985
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.9158626436302235,
      "learning_rate": 1.4549465538397057e-06,
      "loss": 0.343,
      "step": 54990
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.1561099321644015,
      "learning_rate": 1.4540784725753054e-06,
      "loss": 0.3607,
      "step": 54995
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.330581672876071,
      "learning_rate": 1.453210606289681e-06,
      "loss": 0.3564,
      "step": 55000
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.7598439176710032,
      "learning_rate": 1.4523429550354506e-06,
      "loss": 0.3496,
      "step": 55005
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.110514276993395,
      "learning_rate": 1.451475518865218e-06,
      "loss": 0.3598,
      "step": 55010
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.425936554800969,
      "learning_rate": 1.450608297831571e-06,
      "loss": 0.3385,
      "step": 55015
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.409645223191734,
      "learning_rate": 1.4497412919870873e-06,
      "loss": 0.3342,
      "step": 55020
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.7035321145067583,
      "learning_rate": 1.448874501384332e-06,
      "loss": 0.3493,
      "step": 55025
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.729207651077628,
      "learning_rate": 1.4480079260758568e-06,
      "loss": 0.3634,
      "step": 55030
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.041527196712236,
      "learning_rate": 1.4471415661141969e-06,
      "loss": 0.3683,
      "step": 55035
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.231534648775785,
      "learning_rate": 1.446275421551878e-06,
      "loss": 0.3411,
      "step": 55040
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.770910874341322,
      "learning_rate": 1.4454094924414148e-06,
      "loss": 0.3637,
      "step": 55045
    },
    {
      "epoch": 2.33,
      "grad_norm": 5.266759415682012,
      "learning_rate": 1.4445437788353016e-06,
      "loss": 0.374,
      "step": 55050
    },
    {
      "epoch": 2.33,
      "grad_norm": 6.3309693671857845,
      "learning_rate": 1.4436782807860267e-06,
      "loss": 0.3582,
      "step": 55055
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.347056760015068,
      "learning_rate": 1.442812998346063e-06,
      "loss": 0.3393,
      "step": 55060
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.9543437865546567,
      "learning_rate": 1.4419479315678698e-06,
      "loss": 0.3544,
      "step": 55065
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.370479074613875,
      "learning_rate": 1.4410830805038905e-06,
      "loss": 0.3627,
      "step": 55070
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.7904289230250656,
      "learning_rate": 1.4402184452065616e-06,
      "loss": 0.3285,
      "step": 55075
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.2537987230363425,
      "learning_rate": 1.439354025728304e-06,
      "loss": 0.3517,
      "step": 55080
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.0295388926119005,
      "learning_rate": 1.4384898221215221e-06,
      "loss": 0.3308,
      "step": 55085
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.903153173011288,
      "learning_rate": 1.4376258344386118e-06,
      "loss": 0.3471,
      "step": 55090
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.50862363989649,
      "learning_rate": 1.4367620627319533e-06,
      "loss": 0.331,
      "step": 55095
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.079267157926809,
      "learning_rate": 1.4358985070539171e-06,
      "loss": 0.337,
      "step": 55100
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.9588658514421446,
      "learning_rate": 1.4350351674568542e-06,
      "loss": 0.3365,
      "step": 55105
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.6230614158507906,
      "learning_rate": 1.4341720439931083e-06,
      "loss": 0.3528,
      "step": 55110
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.7213048380296483,
      "learning_rate": 1.4333091367150093e-06,
      "loss": 0.3281,
      "step": 55115
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.8826770753095894,
      "learning_rate": 1.43244644567487e-06,
      "loss": 0.3671,
      "step": 55120
    },
    {
      "epoch": 2.33,
      "grad_norm": 7.502305015558864,
      "learning_rate": 1.431583970924994e-06,
      "loss": 0.3777,
      "step": 55125
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.6638167148785827,
      "learning_rate": 1.430721712517672e-06,
      "loss": 0.3219,
      "step": 55130
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.2067003247315675,
      "learning_rate": 1.4298596705051775e-06,
      "loss": 0.3492,
      "step": 55135
    },
    {
      "epoch": 2.33,
      "grad_norm": 3.898883888290832,
      "learning_rate": 1.428997844939775e-06,
      "loss": 0.3371,
      "step": 55140
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.840320847532583,
      "learning_rate": 1.4281362358737145e-06,
      "loss": 0.3557,
      "step": 55145
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.1757951472849815,
      "learning_rate": 1.4272748433592343e-06,
      "loss": 0.3533,
      "step": 55150
    },
    {
      "epoch": 2.33,
      "grad_norm": 5.265764544492705,
      "learning_rate": 1.4264136674485551e-06,
      "loss": 0.3611,
      "step": 55155
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.538951808797188,
      "learning_rate": 1.4255527081938885e-06,
      "loss": 0.3682,
      "step": 55160
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.2395222661775165,
      "learning_rate": 1.4246919656474346e-06,
      "loss": 0.3556,
      "step": 55165
    },
    {
      "epoch": 2.33,
      "grad_norm": 4.303842668172034,
      "learning_rate": 1.4238314398613735e-06,
      "loss": 0.3612,
      "step": 55170
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.7933471389283695,
      "learning_rate": 1.422971130887878e-06,
      "loss": 0.3445,
      "step": 55175
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.579030712320996,
      "learning_rate": 1.4221110387791082e-06,
      "loss": 0.3525,
      "step": 55180
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.6831803856637624,
      "learning_rate": 1.421251163587205e-06,
      "loss": 0.3682,
      "step": 55185
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.1399608102221945,
      "learning_rate": 1.4203915053643031e-06,
      "loss": 0.3834,
      "step": 55190
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.4401983960482445,
      "learning_rate": 1.4195320641625193e-06,
      "loss": 0.3522,
      "step": 55195
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.845814885900577,
      "learning_rate": 1.4186728400339616e-06,
      "loss": 0.3396,
      "step": 55200
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.975743911665738,
      "learning_rate": 1.4178138330307194e-06,
      "loss": 0.3515,
      "step": 55205
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.9030223366957415,
      "learning_rate": 1.4169550432048723e-06,
      "loss": 0.3529,
      "step": 55210
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.7824323296751756,
      "learning_rate": 1.4160964706084883e-06,
      "loss": 0.3579,
      "step": 55215
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.8097082542178446,
      "learning_rate": 1.4152381152936173e-06,
      "loss": 0.3266,
      "step": 55220
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.933591243195999,
      "learning_rate": 1.4143799773123e-06,
      "loss": 0.3575,
      "step": 55225
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.146356553603192,
      "learning_rate": 1.4135220567165641e-06,
      "loss": 0.3375,
      "step": 55230
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.9988722874556673,
      "learning_rate": 1.4126643535584222e-06,
      "loss": 0.3352,
      "step": 55235
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.448258150133473,
      "learning_rate": 1.4118068678898716e-06,
      "loss": 0.3545,
      "step": 55240
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.409290774615804,
      "learning_rate": 1.410949599762902e-06,
      "loss": 0.3135,
      "step": 55245
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.287618183791643,
      "learning_rate": 1.410092549229487e-06,
      "loss": 0.3642,
      "step": 55250
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.102636634431103,
      "learning_rate": 1.4092357163415848e-06,
      "loss": 0.3587,
      "step": 55255
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.830941851288035,
      "learning_rate": 1.408379101151145e-06,
      "loss": 0.3533,
      "step": 55260
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.494206083549135,
      "learning_rate": 1.4075227037101003e-06,
      "loss": 0.3411,
      "step": 55265
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.115211419964185,
      "learning_rate": 1.4066665240703731e-06,
      "loss": 0.3359,
      "step": 55270
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.8486596680118907,
      "learning_rate": 1.405810562283869e-06,
      "loss": 0.3437,
      "step": 55275
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.71487928916392,
      "learning_rate": 1.4049548184024837e-06,
      "loss": 0.3198,
      "step": 55280
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.8722078192122,
      "learning_rate": 1.4040992924780995e-06,
      "loss": 0.3162,
      "step": 55285
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.194733275513015,
      "learning_rate": 1.4032439845625818e-06,
      "loss": 0.3285,
      "step": 55290
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.255646733236831,
      "learning_rate": 1.4023888947077863e-06,
      "loss": 0.3656,
      "step": 55295
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.9646102654367352,
      "learning_rate": 1.4015340229655566e-06,
      "loss": 0.3501,
      "step": 55300
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.827318706598373,
      "learning_rate": 1.400679369387718e-06,
      "loss": 0.3277,
      "step": 55305
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.181024183654609,
      "learning_rate": 1.3998249340260873e-06,
      "loss": 0.3528,
      "step": 55310
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.62937205311957,
      "learning_rate": 1.3989707169324678e-06,
      "loss": 0.3559,
      "step": 55315
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.076289702041772,
      "learning_rate": 1.3981167181586447e-06,
      "loss": 0.3045,
      "step": 55320
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.390021454149438,
      "learning_rate": 1.3972629377563956e-06,
      "loss": 0.3547,
      "step": 55325
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.207551287030898,
      "learning_rate": 1.3964093757774821e-06,
      "loss": 0.3782,
      "step": 55330
    },
    {
      "epoch": 2.34,
      "grad_norm": 5.0926156674937415,
      "learning_rate": 1.395556032273655e-06,
      "loss": 0.3542,
      "step": 55335
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.293898468006087,
      "learning_rate": 1.3947029072966466e-06,
      "loss": 0.3447,
      "step": 55340
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.603413405753714,
      "learning_rate": 1.3938500008981815e-06,
      "loss": 0.3292,
      "step": 55345
    },
    {
      "epoch": 2.34,
      "grad_norm": 6.188241163373703,
      "learning_rate": 1.3929973131299702e-06,
      "loss": 0.3717,
      "step": 55350
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.106368986001545,
      "learning_rate": 1.3921448440437059e-06,
      "loss": 0.3455,
      "step": 55355
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.487163671429729,
      "learning_rate": 1.3912925936910716e-06,
      "loss": 0.3369,
      "step": 55360
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.0320571019137175,
      "learning_rate": 1.3904405621237393e-06,
      "loss": 0.362,
      "step": 55365
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.756532346662424,
      "learning_rate": 1.3895887493933618e-06,
      "loss": 0.3423,
      "step": 55370
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.008639388891662,
      "learning_rate": 1.3887371555515838e-06,
      "loss": 0.3584,
      "step": 55375
    },
    {
      "epoch": 2.34,
      "grad_norm": 6.150479953739124,
      "learning_rate": 1.387885780650035e-06,
      "loss": 0.3468,
      "step": 55380
    },
    {
      "epoch": 2.34,
      "grad_norm": 6.8201034489847805,
      "learning_rate": 1.3870346247403327e-06,
      "loss": 0.3613,
      "step": 55385
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.9871027200222926,
      "learning_rate": 1.3861836878740787e-06,
      "loss": 0.3293,
      "step": 55390
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.263023714690012,
      "learning_rate": 1.3853329701028616e-06,
      "loss": 0.333,
      "step": 55395
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.351444783155533,
      "learning_rate": 1.384482471478259e-06,
      "loss": 0.3608,
      "step": 55400
    },
    {
      "epoch": 2.34,
      "grad_norm": 3.944162345614734,
      "learning_rate": 1.383632192051836e-06,
      "loss": 0.3649,
      "step": 55405
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.7974114056083184,
      "learning_rate": 1.382782131875139e-06,
      "loss": 0.3388,
      "step": 55410
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.735181743867907,
      "learning_rate": 1.381932290999707e-06,
      "loss": 0.3941,
      "step": 55415
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.8026001460057346,
      "learning_rate": 1.3810826694770635e-06,
      "loss": 0.3227,
      "step": 55420
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.212846050476077,
      "learning_rate": 1.3802332673587172e-06,
      "loss": 0.3846,
      "step": 55425
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.566705643787886,
      "learning_rate": 1.3793840846961649e-06,
      "loss": 0.3623,
      "step": 55430
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.8995268184608,
      "learning_rate": 1.3785351215408926e-06,
      "loss": 0.3407,
      "step": 55435
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.6167877910947106,
      "learning_rate": 1.3776863779443668e-06,
      "loss": 0.3639,
      "step": 55440
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.785471208077401,
      "learning_rate": 1.3768378539580456e-06,
      "loss": 0.3403,
      "step": 55445
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.939465010723387,
      "learning_rate": 1.3759895496333737e-06,
      "loss": 0.3681,
      "step": 55450
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.730612907304052,
      "learning_rate": 1.375141465021782e-06,
      "loss": 0.3424,
      "step": 55455
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.87422294839232,
      "learning_rate": 1.3742936001746842e-06,
      "loss": 0.3845,
      "step": 55460
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.110447528294041,
      "learning_rate": 1.3734459551434853e-06,
      "loss": 0.336,
      "step": 55465
    },
    {
      "epoch": 2.35,
      "grad_norm": 5.129127653410787,
      "learning_rate": 1.3725985299795774e-06,
      "loss": 0.3473,
      "step": 55470
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.5655910557715105,
      "learning_rate": 1.3717513247343345e-06,
      "loss": 0.3434,
      "step": 55475
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.05958024109856,
      "learning_rate": 1.3709043394591215e-06,
      "loss": 0.3436,
      "step": 55480
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.957447208747008,
      "learning_rate": 1.3700575742052897e-06,
      "loss": 0.3729,
      "step": 55485
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.010910135203399,
      "learning_rate": 1.3692110290241733e-06,
      "loss": 0.3346,
      "step": 55490
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.9818323022813673,
      "learning_rate": 1.3683647039670978e-06,
      "loss": 0.3143,
      "step": 55495
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.8843343325320614,
      "learning_rate": 1.3675185990853723e-06,
      "loss": 0.3683,
      "step": 55500
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.9518874183869803,
      "learning_rate": 1.366672714430296e-06,
      "loss": 0.3643,
      "step": 55505
    },
    {
      "epoch": 2.35,
      "grad_norm": 5.388592137358785,
      "learning_rate": 1.3658270500531496e-06,
      "loss": 0.3639,
      "step": 55510
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.087657438327674,
      "learning_rate": 1.364981606005204e-06,
      "loss": 0.353,
      "step": 55515
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.395500165851745,
      "learning_rate": 1.3641363823377184e-06,
      "loss": 0.3432,
      "step": 55520
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.020965168443106,
      "learning_rate": 1.363291379101932e-06,
      "loss": 0.355,
      "step": 55525
    },
    {
      "epoch": 2.35,
      "grad_norm": 5.031467185067379,
      "learning_rate": 1.3624465963490774e-06,
      "loss": 0.3268,
      "step": 55530
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.180448758465576,
      "learning_rate": 1.3616020341303721e-06,
      "loss": 0.3456,
      "step": 55535
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.081239417991336,
      "learning_rate": 1.3607576924970167e-06,
      "loss": 0.3433,
      "step": 55540
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.377178175391782,
      "learning_rate": 1.3599135715002031e-06,
      "loss": 0.3518,
      "step": 55545
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.9698833285195647,
      "learning_rate": 1.3590696711911072e-06,
      "loss": 0.3378,
      "step": 55550
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.092839442359719,
      "learning_rate": 1.3582259916208952e-06,
      "loss": 0.344,
      "step": 55555
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.296058623961259,
      "learning_rate": 1.3573825328407114e-06,
      "loss": 0.353,
      "step": 55560
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.9908689704850016,
      "learning_rate": 1.3565392949016948e-06,
      "loss": 0.3323,
      "step": 55565
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.145262859862436,
      "learning_rate": 1.3556962778549687e-06,
      "loss": 0.3625,
      "step": 55570
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.779388984786787,
      "learning_rate": 1.354853481751644e-06,
      "loss": 0.3272,
      "step": 55575
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.329013288163415,
      "learning_rate": 1.3540109066428142e-06,
      "loss": 0.3505,
      "step": 55580
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.0504946029171665,
      "learning_rate": 1.3531685525795635e-06,
      "loss": 0.3115,
      "step": 55585
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.7605621964633253,
      "learning_rate": 1.3523264196129626e-06,
      "loss": 0.3076,
      "step": 55590
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.8509597011257197,
      "learning_rate": 1.3514845077940653e-06,
      "loss": 0.3619,
      "step": 55595
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.5116914437079565,
      "learning_rate": 1.350642817173915e-06,
      "loss": 0.3848,
      "step": 55600
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.303479192892619,
      "learning_rate": 1.3498013478035426e-06,
      "loss": 0.3346,
      "step": 55605
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.267675582147344,
      "learning_rate": 1.3489600997339607e-06,
      "loss": 0.3668,
      "step": 55610
    },
    {
      "epoch": 2.35,
      "grad_norm": 5.172539554338578,
      "learning_rate": 1.3481190730161742e-06,
      "loss": 0.3695,
      "step": 55615
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.483649949038961,
      "learning_rate": 1.347278267701172e-06,
      "loss": 0.3254,
      "step": 55620
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.997130439257425,
      "learning_rate": 1.346437683839928e-06,
      "loss": 0.3705,
      "step": 55625
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.921495149472675,
      "learning_rate": 1.345597321483405e-06,
      "loss": 0.3438,
      "step": 55630
    },
    {
      "epoch": 2.35,
      "grad_norm": 4.066636128029731,
      "learning_rate": 1.3447571806825526e-06,
      "loss": 0.3367,
      "step": 55635
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.8966105533045448,
      "learning_rate": 1.3439172614883067e-06,
      "loss": 0.3496,
      "step": 55640
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.7508693037144214,
      "learning_rate": 1.3430775639515858e-06,
      "loss": 0.3449,
      "step": 55645
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.33855871275355,
      "learning_rate": 1.342238088123301e-06,
      "loss": 0.3559,
      "step": 55650
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.0971856127482456,
      "learning_rate": 1.3413988340543477e-06,
      "loss": 0.3479,
      "step": 55655
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.46747640623213,
      "learning_rate": 1.340559801795605e-06,
      "loss": 0.3593,
      "step": 55660
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.831264526982916,
      "learning_rate": 1.3397209913979425e-06,
      "loss": 0.3475,
      "step": 55665
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.162810804648072,
      "learning_rate": 1.338882402912216e-06,
      "loss": 0.3626,
      "step": 55670
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.739472971462204,
      "learning_rate": 1.3380440363892637e-06,
      "loss": 0.36,
      "step": 55675
    },
    {
      "epoch": 2.36,
      "grad_norm": 6.142191384800508,
      "learning_rate": 1.3372058918799147e-06,
      "loss": 0.3675,
      "step": 55680
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.112393245707808,
      "learning_rate": 1.3363679694349834e-06,
      "loss": 0.3438,
      "step": 55685
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.7682394535471095,
      "learning_rate": 1.3355302691052724e-06,
      "loss": 0.3467,
      "step": 55690
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.102171217715294,
      "learning_rate": 1.3346927909415652e-06,
      "loss": 0.3492,
      "step": 55695
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.7037749956670356,
      "learning_rate": 1.333855534994638e-06,
      "loss": 0.362,
      "step": 55700
    },
    {
      "epoch": 2.36,
      "grad_norm": 7.785678508909988,
      "learning_rate": 1.333018501315252e-06,
      "loss": 0.3692,
      "step": 55705
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.245461029426809,
      "learning_rate": 1.3321816899541512e-06,
      "loss": 0.3528,
      "step": 55710
    },
    {
      "epoch": 2.36,
      "grad_norm": 5.164128537186574,
      "learning_rate": 1.331345100962072e-06,
      "loss": 0.3468,
      "step": 55715
    },
    {
      "epoch": 2.36,
      "grad_norm": 6.9848189691815366,
      "learning_rate": 1.3305087343897316e-06,
      "loss": 0.3772,
      "step": 55720
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.523771692658554,
      "learning_rate": 1.3296725902878394e-06,
      "loss": 0.3539,
      "step": 55725
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.019773474493285,
      "learning_rate": 1.3288366687070852e-06,
      "loss": 0.3397,
      "step": 55730
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.270038306108495,
      "learning_rate": 1.32800096969815e-06,
      "loss": 0.3803,
      "step": 55735
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.225382394729628,
      "learning_rate": 1.3271654933117012e-06,
      "loss": 0.3192,
      "step": 55740
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.593876420552302,
      "learning_rate": 1.3263302395983886e-06,
      "loss": 0.3548,
      "step": 55745
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.390086747140461,
      "learning_rate": 1.3254952086088518e-06,
      "loss": 0.3474,
      "step": 55750
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.453808160952599,
      "learning_rate": 1.3246604003937176e-06,
      "loss": 0.3428,
      "step": 55755
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.921844285521916,
      "learning_rate": 1.3238258150035983e-06,
      "loss": 0.3603,
      "step": 55760
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.365298250537008,
      "learning_rate": 1.3229914524890902e-06,
      "loss": 0.3534,
      "step": 55765
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.817963196687594,
      "learning_rate": 1.3221573129007793e-06,
      "loss": 0.3447,
      "step": 55770
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.229219626446148,
      "learning_rate": 1.321323396289238e-06,
      "loss": 0.3524,
      "step": 55775
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.422963194825038,
      "learning_rate": 1.3204897027050224e-06,
      "loss": 0.3764,
      "step": 55780
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.489940431164984,
      "learning_rate": 1.3196562321986778e-06,
      "loss": 0.3433,
      "step": 55785
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.410909456564164,
      "learning_rate": 1.3188229848207363e-06,
      "loss": 0.3363,
      "step": 55790
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.6294574542768974,
      "learning_rate": 1.3179899606217122e-06,
      "loss": 0.3291,
      "step": 55795
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.112139769441018,
      "learning_rate": 1.3171571596521116e-06,
      "loss": 0.3564,
      "step": 55800
    },
    {
      "epoch": 2.36,
      "grad_norm": 5.732627052020113,
      "learning_rate": 1.3163245819624237e-06,
      "loss": 0.3487,
      "step": 55805
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.45956805501037,
      "learning_rate": 1.315492227603128e-06,
      "loss": 0.3471,
      "step": 55810
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.177961859652178,
      "learning_rate": 1.3146600966246832e-06,
      "loss": 0.3994,
      "step": 55815
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.049830124000791,
      "learning_rate": 1.3138281890775417e-06,
      "loss": 0.3345,
      "step": 55820
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.9092876170129225,
      "learning_rate": 1.3129965050121402e-06,
      "loss": 0.3518,
      "step": 55825
    },
    {
      "epoch": 2.36,
      "grad_norm": 6.0184514845303365,
      "learning_rate": 1.3121650444788992e-06,
      "loss": 0.3433,
      "step": 55830
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.684731485243696,
      "learning_rate": 1.3113338075282284e-06,
      "loss": 0.3503,
      "step": 55835
    },
    {
      "epoch": 2.36,
      "grad_norm": 5.227962320291398,
      "learning_rate": 1.3105027942105247e-06,
      "loss": 0.3487,
      "step": 55840
    },
    {
      "epoch": 2.36,
      "grad_norm": 3.9294827476025485,
      "learning_rate": 1.309672004576168e-06,
      "loss": 0.3484,
      "step": 55845
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.401911422461429,
      "learning_rate": 1.3088414386755272e-06,
      "loss": 0.3219,
      "step": 55850
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.111762800385693,
      "learning_rate": 1.308011096558957e-06,
      "loss": 0.3649,
      "step": 55855
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.467962077468306,
      "learning_rate": 1.3071809782768008e-06,
      "loss": 0.3347,
      "step": 55860
    },
    {
      "epoch": 2.36,
      "grad_norm": 5.588413416092084,
      "learning_rate": 1.3063510838793824e-06,
      "loss": 0.3681,
      "step": 55865
    },
    {
      "epoch": 2.36,
      "grad_norm": 5.021150635520996,
      "learning_rate": 1.305521413417018e-06,
      "loss": 0.3125,
      "step": 55870
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.391011618769206,
      "learning_rate": 1.3046919669400093e-06,
      "loss": 0.3583,
      "step": 55875
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.64237703987566,
      "learning_rate": 1.3038627444986425e-06,
      "loss": 0.3331,
      "step": 55880
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.7865438235615665,
      "learning_rate": 1.3030337461431874e-06,
      "loss": 0.3329,
      "step": 55885
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.186188349926038,
      "learning_rate": 1.3022049719239073e-06,
      "loss": 0.3562,
      "step": 55890
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.78490752133425,
      "learning_rate": 1.301376421891049e-06,
      "loss": 0.3578,
      "step": 55895
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.918538994360132,
      "learning_rate": 1.300548096094842e-06,
      "loss": 0.3183,
      "step": 55900
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.9138612279561666,
      "learning_rate": 1.2997199945855066e-06,
      "loss": 0.3275,
      "step": 55905
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.923359895247955,
      "learning_rate": 1.29889211741325e-06,
      "loss": 0.3566,
      "step": 55910
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.032891921983772,
      "learning_rate": 1.2980644646282614e-06,
      "loss": 0.3785,
      "step": 55915
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.853533660744576,
      "learning_rate": 1.2972370362807191e-06,
      "loss": 0.3153,
      "step": 55920
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.745807634438993,
      "learning_rate": 1.2964098324207902e-06,
      "loss": 0.3432,
      "step": 55925
    },
    {
      "epoch": 2.37,
      "grad_norm": 6.417016321105122,
      "learning_rate": 1.2955828530986225e-06,
      "loss": 0.3715,
      "step": 55930
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.764829857995478,
      "learning_rate": 1.2947560983643549e-06,
      "loss": 0.3669,
      "step": 55935
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.782572627433298,
      "learning_rate": 1.2939295682681108e-06,
      "loss": 0.3393,
      "step": 55940
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.9253579246168524,
      "learning_rate": 1.2931032628600016e-06,
      "loss": 0.3351,
      "step": 55945
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.218643236891965,
      "learning_rate": 1.2922771821901213e-06,
      "loss": 0.3566,
      "step": 55950
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.86567523556772,
      "learning_rate": 1.2914513263085536e-06,
      "loss": 0.3456,
      "step": 55955
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.550604741656876,
      "learning_rate": 1.29062569526537e-06,
      "loss": 0.3247,
      "step": 55960
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.13616965405224,
      "learning_rate": 1.2898002891106225e-06,
      "loss": 0.3499,
      "step": 55965
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.791048857521803,
      "learning_rate": 1.2889751078943553e-06,
      "loss": 0.3529,
      "step": 55970
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.850706860275869,
      "learning_rate": 1.2881501516665966e-06,
      "loss": 0.3288,
      "step": 55975
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.25230556934984,
      "learning_rate": 1.28732542047736e-06,
      "loss": 0.3476,
      "step": 55980
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.208737602593975,
      "learning_rate": 1.2865009143766466e-06,
      "loss": 0.3352,
      "step": 55985
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.021362939462503,
      "learning_rate": 1.2856766334144444e-06,
      "loss": 0.3421,
      "step": 55990
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.00083711490453,
      "learning_rate": 1.2848525776407283e-06,
      "loss": 0.3668,
      "step": 55995
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.676075614243951,
      "learning_rate": 1.2840287471054557e-06,
      "loss": 0.3721,
      "step": 56000
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.82609149913964,
      "learning_rate": 1.283205141858575e-06,
      "loss": 0.38,
      "step": 56005
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.81295511228173,
      "learning_rate": 1.2823817619500195e-06,
      "loss": 0.365,
      "step": 56010
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.9665553329486913,
      "learning_rate": 1.2815586074297053e-06,
      "loss": 0.3077,
      "step": 56015
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.052871564458969,
      "learning_rate": 1.28073567834754e-06,
      "loss": 0.3665,
      "step": 56020
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.8782914665998804,
      "learning_rate": 1.2799129747534166e-06,
      "loss": 0.3661,
      "step": 56025
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.938404687773065,
      "learning_rate": 1.2790904966972102e-06,
      "loss": 0.3485,
      "step": 56030
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.014476909953917,
      "learning_rate": 1.2782682442287865e-06,
      "loss": 0.3525,
      "step": 56035
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.737326330683595,
      "learning_rate": 1.277446217397998e-06,
      "loss": 0.3539,
      "step": 56040
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.278939190998384,
      "learning_rate": 1.2766244162546804e-06,
      "loss": 0.3429,
      "step": 56045
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.531774304321063,
      "learning_rate": 1.275802840848655e-06,
      "loss": 0.3494,
      "step": 56050
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.298112082051563,
      "learning_rate": 1.2749814912297337e-06,
      "loss": 0.3615,
      "step": 56055
    },
    {
      "epoch": 2.37,
      "grad_norm": 5.205917929398763,
      "learning_rate": 1.274160367447712e-06,
      "loss": 0.3476,
      "step": 56060
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.3174209372800485,
      "learning_rate": 1.2733394695523738e-06,
      "loss": 0.3408,
      "step": 56065
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.556242124937779,
      "learning_rate": 1.2725187975934849e-06,
      "loss": 0.3255,
      "step": 56070
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.336639195586417,
      "learning_rate": 1.271698351620802e-06,
      "loss": 0.3662,
      "step": 56075
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.180398745519801,
      "learning_rate": 1.2708781316840674e-06,
      "loss": 0.3499,
      "step": 56080
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.082706949604593,
      "learning_rate": 1.2700581378330052e-06,
      "loss": 0.345,
      "step": 56085
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.454175050134992,
      "learning_rate": 1.2692383701173316e-06,
      "loss": 0.3257,
      "step": 56090
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.8894647293107556,
      "learning_rate": 1.2684188285867477e-06,
      "loss": 0.3632,
      "step": 56095
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.0870018333469025,
      "learning_rate": 1.267599513290937e-06,
      "loss": 0.3608,
      "step": 56100
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.949174027432825,
      "learning_rate": 1.2667804242795734e-06,
      "loss": 0.3415,
      "step": 56105
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.10937404443264,
      "learning_rate": 1.265961561602317e-06,
      "loss": 0.3446,
      "step": 56110
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.057028937983714,
      "learning_rate": 1.2651429253088127e-06,
      "loss": 0.3552,
      "step": 56115
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.061570314602765,
      "learning_rate": 1.2643245154486899e-06,
      "loss": 0.3477,
      "step": 56120
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.82220118583394,
      "learning_rate": 1.2635063320715685e-06,
      "loss": 0.3545,
      "step": 56125
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.768856531618555,
      "learning_rate": 1.2626883752270536e-06,
      "loss": 0.3607,
      "step": 56130
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.117609707521339,
      "learning_rate": 1.2618706449647317e-06,
      "loss": 0.3363,
      "step": 56135
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.519244719455889,
      "learning_rate": 1.261053141334182e-06,
      "loss": 0.3462,
      "step": 56140
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.763057681384074,
      "learning_rate": 1.2602358643849683e-06,
      "loss": 0.3537,
      "step": 56145
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.8336662345472563,
      "learning_rate": 1.2594188141666368e-06,
      "loss": 0.3431,
      "step": 56150
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.149541099161768,
      "learning_rate": 1.2586019907287244e-06,
      "loss": 0.3275,
      "step": 56155
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.316995669464308,
      "learning_rate": 1.2577853941207547e-06,
      "loss": 0.3666,
      "step": 56160
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.66280877743433,
      "learning_rate": 1.2569690243922312e-06,
      "loss": 0.3264,
      "step": 56165
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.282029869352085,
      "learning_rate": 1.2561528815926505e-06,
      "loss": 0.3387,
      "step": 56170
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.954316961126753,
      "learning_rate": 1.255336965771493e-06,
      "loss": 0.3419,
      "step": 56175
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.9916603811696194,
      "learning_rate": 1.2545212769782262e-06,
      "loss": 0.3437,
      "step": 56180
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.294647069778002,
      "learning_rate": 1.2537058152623005e-06,
      "loss": 0.3579,
      "step": 56185
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.664136498279568,
      "learning_rate": 1.2528905806731555e-06,
      "loss": 0.3308,
      "step": 56190
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.073167455128477,
      "learning_rate": 1.2520755732602196e-06,
      "loss": 0.3278,
      "step": 56195
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.272783277771394,
      "learning_rate": 1.251260793072901e-06,
      "loss": 0.3382,
      "step": 56200
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.363907824497373,
      "learning_rate": 1.2504462401605972e-06,
      "loss": 0.3269,
      "step": 56205
    },
    {
      "epoch": 2.38,
      "grad_norm": 5.312327297114382,
      "learning_rate": 1.2496319145726932e-06,
      "loss": 0.3322,
      "step": 56210
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.569899863667851,
      "learning_rate": 1.2488178163585601e-06,
      "loss": 0.3572,
      "step": 56215
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.047102384079045,
      "learning_rate": 1.2480039455675525e-06,
      "loss": 0.3433,
      "step": 56220
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.84011461636553,
      "learning_rate": 1.2471903022490134e-06,
      "loss": 0.3188,
      "step": 56225
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.5966779973840906,
      "learning_rate": 1.2463768864522735e-06,
      "loss": 0.3444,
      "step": 56230
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.0845417939107875,
      "learning_rate": 1.245563698226645e-06,
      "loss": 0.3368,
      "step": 56235
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.050081355770733,
      "learning_rate": 1.2447507376214302e-06,
      "loss": 0.3512,
      "step": 56240
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.214356694193335,
      "learning_rate": 1.2439380046859174e-06,
      "loss": 0.3599,
      "step": 56245
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.8158448569828205,
      "learning_rate": 1.2431254994693804e-06,
      "loss": 0.3275,
      "step": 56250
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.9101584457008234,
      "learning_rate": 1.2423132220210766e-06,
      "loss": 0.3328,
      "step": 56255
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.163220685716269,
      "learning_rate": 1.2415011723902536e-06,
      "loss": 0.3479,
      "step": 56260
    },
    {
      "epoch": 2.38,
      "grad_norm": 5.317419706805609,
      "learning_rate": 1.2406893506261453e-06,
      "loss": 0.3417,
      "step": 56265
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.400425551107368,
      "learning_rate": 1.2398777567779663e-06,
      "loss": 0.3362,
      "step": 56270
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.556054707705937,
      "learning_rate": 1.2390663908949235e-06,
      "loss": 0.3435,
      "step": 56275
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.4800779373837205,
      "learning_rate": 1.2382552530262081e-06,
      "loss": 0.3359,
      "step": 56280
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.028650576657266,
      "learning_rate": 1.2374443432209954e-06,
      "loss": 0.3427,
      "step": 56285
    },
    {
      "epoch": 2.38,
      "grad_norm": 6.1591874444541075,
      "learning_rate": 1.236633661528449e-06,
      "loss": 0.3519,
      "step": 56290
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.923034852127373,
      "learning_rate": 1.2358232079977179e-06,
      "loss": 0.3326,
      "step": 56295
    },
    {
      "epoch": 2.38,
      "grad_norm": 8.29779257531179,
      "learning_rate": 1.23501298267794e-06,
      "loss": 0.3533,
      "step": 56300
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.818434293798176,
      "learning_rate": 1.234202985618233e-06,
      "loss": 0.3429,
      "step": 56305
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.2591512145228565,
      "learning_rate": 1.2333932168677066e-06,
      "loss": 0.3652,
      "step": 56310
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.6263671184116357,
      "learning_rate": 1.2325836764754563e-06,
      "loss": 0.3539,
      "step": 56315
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.070523711018789,
      "learning_rate": 1.2317743644905583e-06,
      "loss": 0.3439,
      "step": 56320
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.7271284536070377,
      "learning_rate": 1.2309652809620814e-06,
      "loss": 0.335,
      "step": 56325
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.272686173320991,
      "learning_rate": 1.2301564259390786e-06,
      "loss": 0.3516,
      "step": 56330
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.9364050290880144,
      "learning_rate": 1.229347799470586e-06,
      "loss": 0.319,
      "step": 56335
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.460413442485901,
      "learning_rate": 1.2285394016056295e-06,
      "loss": 0.3512,
      "step": 56340
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.17413430368539,
      "learning_rate": 1.22773123239322e-06,
      "loss": 0.3603,
      "step": 56345
    },
    {
      "epoch": 2.38,
      "grad_norm": 4.236662996057952,
      "learning_rate": 1.2269232918823554e-06,
      "loss": 0.3965,
      "step": 56350
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.7770242104213736,
      "learning_rate": 1.2261155801220166e-06,
      "loss": 0.3292,
      "step": 56355
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.663359595479212,
      "learning_rate": 1.2253080971611736e-06,
      "loss": 0.3214,
      "step": 56360
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.299667672535685,
      "learning_rate": 1.224500843048783e-06,
      "loss": 0.3423,
      "step": 56365
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.8169810599133935,
      "learning_rate": 1.223693817833786e-06,
      "loss": 0.3537,
      "step": 56370
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.279985295007791,
      "learning_rate": 1.2228870215651074e-06,
      "loss": 0.3581,
      "step": 56375
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.8059500991932085,
      "learning_rate": 1.222080454291663e-06,
      "loss": 0.3096,
      "step": 56380
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.183620710461634,
      "learning_rate": 1.2212741160623537e-06,
      "loss": 0.3648,
      "step": 56385
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.460469143650533,
      "learning_rate": 1.2204680069260628e-06,
      "loss": 0.3401,
      "step": 56390
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.747481355269418,
      "learning_rate": 1.2196621269316638e-06,
      "loss": 0.3394,
      "step": 56395
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.454252294094227,
      "learning_rate": 1.2188564761280159e-06,
      "loss": 0.3656,
      "step": 56400
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.5021637823003036,
      "learning_rate": 1.2180510545639612e-06,
      "loss": 0.3592,
      "step": 56405
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.8405870111259075,
      "learning_rate": 1.2172458622883303e-06,
      "loss": 0.3625,
      "step": 56410
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.367901583823025,
      "learning_rate": 1.2164408993499404e-06,
      "loss": 0.3456,
      "step": 56415
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.068501791686017,
      "learning_rate": 1.2156361657975957e-06,
      "loss": 0.3619,
      "step": 56420
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.456982934607577,
      "learning_rate": 1.2148316616800815e-06,
      "loss": 0.3716,
      "step": 56425
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.01996956647707,
      "learning_rate": 1.2140273870461737e-06,
      "loss": 0.3459,
      "step": 56430
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.100836662742025,
      "learning_rate": 1.2132233419446354e-06,
      "loss": 0.348,
      "step": 56435
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.216336199715205,
      "learning_rate": 1.2124195264242101e-06,
      "loss": 0.3527,
      "step": 56440
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.7200618430058805,
      "learning_rate": 1.211615940533632e-06,
      "loss": 0.3659,
      "step": 56445
    },
    {
      "epoch": 2.39,
      "grad_norm": 7.882473699789652,
      "learning_rate": 1.2108125843216223e-06,
      "loss": 0.3448,
      "step": 56450
    },
    {
      "epoch": 2.39,
      "grad_norm": 6.606248741664309,
      "learning_rate": 1.210009457836882e-06,
      "loss": 0.3697,
      "step": 56455
    },
    {
      "epoch": 2.39,
      "grad_norm": 5.876830757682971,
      "learning_rate": 1.209206561128105e-06,
      "loss": 0.3246,
      "step": 56460
    },
    {
      "epoch": 2.39,
      "grad_norm": 6.807438903928233,
      "learning_rate": 1.2084038942439697e-06,
      "loss": 0.342,
      "step": 56465
    },
    {
      "epoch": 2.39,
      "grad_norm": 5.5609946872091225,
      "learning_rate": 1.2076014572331356e-06,
      "loss": 0.333,
      "step": 56470
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.731673141284543,
      "learning_rate": 1.2067992501442549e-06,
      "loss": 0.3551,
      "step": 56475
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.184057591087504,
      "learning_rate": 1.2059972730259623e-06,
      "loss": 0.3445,
      "step": 56480
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.4390993490406245,
      "learning_rate": 1.2051955259268805e-06,
      "loss": 0.3263,
      "step": 56485
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.918815012539708,
      "learning_rate": 1.2043940088956147e-06,
      "loss": 0.3436,
      "step": 56490
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.7354162698167492,
      "learning_rate": 1.2035927219807604e-06,
      "loss": 0.3631,
      "step": 56495
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.913744920111048,
      "learning_rate": 1.2027916652308975e-06,
      "loss": 0.3288,
      "step": 56500
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.308056665449875,
      "learning_rate": 1.2019908386945894e-06,
      "loss": 0.3466,
      "step": 56505
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.473366687726201,
      "learning_rate": 1.2011902424203897e-06,
      "loss": 0.3456,
      "step": 56510
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.75336022461932,
      "learning_rate": 1.2003898764568372e-06,
      "loss": 0.3521,
      "step": 56515
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.141934935456006,
      "learning_rate": 1.1995897408524527e-06,
      "loss": 0.3312,
      "step": 56520
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.137933199754448,
      "learning_rate": 1.1987898356557493e-06,
      "loss": 0.3309,
      "step": 56525
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.926767726938109,
      "learning_rate": 1.19799016091522e-06,
      "loss": 0.356,
      "step": 56530
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.743462704212318,
      "learning_rate": 1.1971907166793496e-06,
      "loss": 0.328,
      "step": 56535
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.662777440762261,
      "learning_rate": 1.1963915029966028e-06,
      "loss": 0.3705,
      "step": 56540
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.254489488529074,
      "learning_rate": 1.1955925199154356e-06,
      "loss": 0.3516,
      "step": 56545
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.103481329540277,
      "learning_rate": 1.1947937674842874e-06,
      "loss": 0.368,
      "step": 56550
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.785153847180542,
      "learning_rate": 1.1939952457515858e-06,
      "loss": 0.3235,
      "step": 56555
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.8150842846257937,
      "learning_rate": 1.19319695476574e-06,
      "loss": 0.3236,
      "step": 56560
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.7591610191463336,
      "learning_rate": 1.1923988945751502e-06,
      "loss": 0.3154,
      "step": 56565
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.473851978568733,
      "learning_rate": 1.1916010652282005e-06,
      "loss": 0.3331,
      "step": 56570
    },
    {
      "epoch": 2.39,
      "grad_norm": 5.5208392503754045,
      "learning_rate": 1.1908034667732593e-06,
      "loss": 0.3529,
      "step": 56575
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.8793528002780695,
      "learning_rate": 1.190006099258683e-06,
      "loss": 0.3544,
      "step": 56580
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.265155831274544,
      "learning_rate": 1.1892089627328157e-06,
      "loss": 0.3466,
      "step": 56585
    },
    {
      "epoch": 2.39,
      "grad_norm": 4.9854106775049,
      "learning_rate": 1.1884120572439827e-06,
      "loss": 0.3462,
      "step": 56590
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.8964229123107885,
      "learning_rate": 1.1876153828404995e-06,
      "loss": 0.3599,
      "step": 56595
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.374570532710603,
      "learning_rate": 1.1868189395706658e-06,
      "loss": 0.3576,
      "step": 56600
    },
    {
      "epoch": 2.4,
      "grad_norm": 7.799192059631832,
      "learning_rate": 1.186022727482769e-06,
      "loss": 0.3649,
      "step": 56605
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.223432559321248,
      "learning_rate": 1.1852267466250783e-06,
      "loss": 0.3044,
      "step": 56610
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.4106607448899435,
      "learning_rate": 1.1844309970458534e-06,
      "loss": 0.3278,
      "step": 56615
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.595113762495233,
      "learning_rate": 1.183635478793339e-06,
      "loss": 0.3299,
      "step": 56620
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.604172959403088,
      "learning_rate": 1.1828401919157628e-06,
      "loss": 0.3157,
      "step": 56625
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.7892305894727145,
      "learning_rate": 1.1820451364613423e-06,
      "loss": 0.3393,
      "step": 56630
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.9769497888625605,
      "learning_rate": 1.1812503124782803e-06,
      "loss": 0.3272,
      "step": 56635
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.880632401665747,
      "learning_rate": 1.1804557200147615e-06,
      "loss": 0.3051,
      "step": 56640
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.017454781649578,
      "learning_rate": 1.179661359118962e-06,
      "loss": 0.3418,
      "step": 56645
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.63058356214561,
      "learning_rate": 1.1788672298390413e-06,
      "loss": 0.3342,
      "step": 56650
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.154782479950823,
      "learning_rate": 1.1780733322231458e-06,
      "loss": 0.3748,
      "step": 56655
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.0865863330178716,
      "learning_rate": 1.1772796663194052e-06,
      "loss": 0.3528,
      "step": 56660
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.491277325552022,
      "learning_rate": 1.1764862321759384e-06,
      "loss": 0.3639,
      "step": 56665
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.155548428228232,
      "learning_rate": 1.1756930298408499e-06,
      "loss": 0.3213,
      "step": 56670
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.882242553454759,
      "learning_rate": 1.1749000593622267e-06,
      "loss": 0.3585,
      "step": 56675
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.260700333042391,
      "learning_rate": 1.174107320788146e-06,
      "loss": 0.3658,
      "step": 56680
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.078747061421399,
      "learning_rate": 1.17331481416667e-06,
      "loss": 0.356,
      "step": 56685
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.9670077186311135,
      "learning_rate": 1.172522539545845e-06,
      "loss": 0.3635,
      "step": 56690
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.023438981837194,
      "learning_rate": 1.1717304969737031e-06,
      "loss": 0.3508,
      "step": 56695
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.823090998694218,
      "learning_rate": 1.1709386864982646e-06,
      "loss": 0.3637,
      "step": 56700
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.854170596248708,
      "learning_rate": 1.170147108167536e-06,
      "loss": 0.3736,
      "step": 56705
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.654391944424784,
      "learning_rate": 1.1693557620295053e-06,
      "loss": 0.345,
      "step": 56710
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.238560403086151,
      "learning_rate": 1.1685646481321515e-06,
      "loss": 0.3376,
      "step": 56715
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.158361449649931,
      "learning_rate": 1.1677737665234373e-06,
      "loss": 0.3488,
      "step": 56720
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.544770183822335,
      "learning_rate": 1.166983117251313e-06,
      "loss": 0.3585,
      "step": 56725
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.6140877954686355,
      "learning_rate": 1.1661927003637097e-06,
      "loss": 0.3253,
      "step": 56730
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.6973941741159484,
      "learning_rate": 1.1654025159085508e-06,
      "loss": 0.3446,
      "step": 56735
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.437665175818403,
      "learning_rate": 1.1646125639337435e-06,
      "loss": 0.3398,
      "step": 56740
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.2111065634403655,
      "learning_rate": 1.1638228444871774e-06,
      "loss": 0.3508,
      "step": 56745
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.589798529588758,
      "learning_rate": 1.1630333576167324e-06,
      "loss": 0.3353,
      "step": 56750
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.503344616811239,
      "learning_rate": 1.1622441033702742e-06,
      "loss": 0.3392,
      "step": 56755
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.7342863826040205,
      "learning_rate": 1.1614550817956498e-06,
      "loss": 0.3568,
      "step": 56760
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.4872082940221,
      "learning_rate": 1.1606662929406975e-06,
      "loss": 0.3531,
      "step": 56765
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.354552228237745,
      "learning_rate": 1.1598777368532398e-06,
      "loss": 0.3226,
      "step": 56770
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.7670790249664683,
      "learning_rate": 1.159089413581082e-06,
      "loss": 0.3539,
      "step": 56775
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.022191338765902,
      "learning_rate": 1.1583013231720197e-06,
      "loss": 0.3609,
      "step": 56780
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.467486150618449,
      "learning_rate": 1.1575134656738318e-06,
      "loss": 0.3562,
      "step": 56785
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.626950937858999,
      "learning_rate": 1.1567258411342851e-06,
      "loss": 0.3273,
      "step": 56790
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.840100248745007,
      "learning_rate": 1.1559384496011294e-06,
      "loss": 0.313,
      "step": 56795
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.612093788714251,
      "learning_rate": 1.155151291122102e-06,
      "loss": 0.3399,
      "step": 56800
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.945557251560856,
      "learning_rate": 1.1543643657449277e-06,
      "loss": 0.3526,
      "step": 56805
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.4700650585877755,
      "learning_rate": 1.1535776735173127e-06,
      "loss": 0.3461,
      "step": 56810
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.9393605197990986,
      "learning_rate": 1.1527912144869536e-06,
      "loss": 0.3218,
      "step": 56815
    },
    {
      "epoch": 2.4,
      "grad_norm": 5.3111464394638555,
      "learning_rate": 1.1520049887015322e-06,
      "loss": 0.353,
      "step": 56820
    },
    {
      "epoch": 2.4,
      "grad_norm": 4.342555915373743,
      "learning_rate": 1.1512189962087123e-06,
      "loss": 0.3217,
      "step": 56825
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.864373750013248,
      "learning_rate": 1.1504332370561477e-06,
      "loss": 0.3392,
      "step": 56830
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.806553216063425,
      "learning_rate": 1.1496477112914767e-06,
      "loss": 0.3553,
      "step": 56835
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.854469413987072,
      "learning_rate": 1.1488624189623249e-06,
      "loss": 0.3476,
      "step": 56840
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.275119593561502,
      "learning_rate": 1.1480773601162992e-06,
      "loss": 0.3511,
      "step": 56845
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.970161339987659,
      "learning_rate": 1.1472925348009984e-06,
      "loss": 0.3305,
      "step": 56850
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.142120702832089,
      "learning_rate": 1.1465079430640008e-06,
      "loss": 0.3397,
      "step": 56855
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.017284552602058,
      "learning_rate": 1.145723584952878e-06,
      "loss": 0.3456,
      "step": 56860
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.6748169488463605,
      "learning_rate": 1.1449394605151792e-06,
      "loss": 0.3224,
      "step": 56865
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.542752211364065,
      "learning_rate": 1.1441555697984453e-06,
      "loss": 0.3047,
      "step": 56870
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.369654380259122,
      "learning_rate": 1.143371912850203e-06,
      "loss": 0.3082,
      "step": 56875
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.897751019700783,
      "learning_rate": 1.1425884897179607e-06,
      "loss": 0.3325,
      "step": 56880
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.472942981051483,
      "learning_rate": 1.1418053004492152e-06,
      "loss": 0.3615,
      "step": 56885
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.615938007710609,
      "learning_rate": 1.1410223450914514e-06,
      "loss": 0.3681,
      "step": 56890
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.307134665168528,
      "learning_rate": 1.1402396236921343e-06,
      "loss": 0.3337,
      "step": 56895
    },
    {
      "epoch": 2.41,
      "grad_norm": 5.7385537536736795,
      "learning_rate": 1.1394571362987195e-06,
      "loss": 0.3574,
      "step": 56900
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.496709311583137,
      "learning_rate": 1.1386748829586468e-06,
      "loss": 0.3265,
      "step": 56905
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.9961132678714324,
      "learning_rate": 1.1378928637193438e-06,
      "loss": 0.3483,
      "step": 56910
    },
    {
      "epoch": 2.41,
      "grad_norm": 5.167455808327841,
      "learning_rate": 1.1371110786282185e-06,
      "loss": 0.3259,
      "step": 56915
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.6139097669997216,
      "learning_rate": 1.1363295277326703e-06,
      "loss": 0.324,
      "step": 56920
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.175104134970415,
      "learning_rate": 1.135548211080083e-06,
      "loss": 0.3471,
      "step": 56925
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.055510692413433,
      "learning_rate": 1.1347671287178235e-06,
      "loss": 0.3219,
      "step": 56930
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.069856583483291,
      "learning_rate": 1.1339862806932478e-06,
      "loss": 0.3349,
      "step": 56935
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.8233800481673206,
      "learning_rate": 1.1332056670536974e-06,
      "loss": 0.339,
      "step": 56940
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.9253397380128967,
      "learning_rate": 1.1324252878464958e-06,
      "loss": 0.329,
      "step": 56945
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.634972787058376,
      "learning_rate": 1.1316451431189567e-06,
      "loss": 0.3455,
      "step": 56950
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.200900944922554,
      "learning_rate": 1.1308652329183778e-06,
      "loss": 0.3607,
      "step": 56955
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.918487083661899,
      "learning_rate": 1.1300855572920445e-06,
      "loss": 0.3495,
      "step": 56960
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.113599656671094,
      "learning_rate": 1.1293061162872232e-06,
      "loss": 0.3224,
      "step": 56965
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.549136236937687,
      "learning_rate": 1.128526909951171e-06,
      "loss": 0.3433,
      "step": 56970
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.150902529307994,
      "learning_rate": 1.127747938331129e-06,
      "loss": 0.3144,
      "step": 56975
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.039845476152868,
      "learning_rate": 1.1269692014743227e-06,
      "loss": 0.3368,
      "step": 56980
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.6695552906140594,
      "learning_rate": 1.1261906994279652e-06,
      "loss": 0.3142,
      "step": 56985
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.881461509640357,
      "learning_rate": 1.1254124322392562e-06,
      "loss": 0.378,
      "step": 56990
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.350278752790158,
      "learning_rate": 1.124634399955377e-06,
      "loss": 0.3489,
      "step": 56995
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.156181304261934,
      "learning_rate": 1.1238566026234993e-06,
      "loss": 0.3305,
      "step": 57000
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.019306395727137,
      "learning_rate": 1.1230790402907776e-06,
      "loss": 0.3508,
      "step": 57005
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.168698725812384,
      "learning_rate": 1.1223017130043574e-06,
      "loss": 0.3589,
      "step": 57010
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.554145242819516,
      "learning_rate": 1.1215246208113584e-06,
      "loss": 0.3187,
      "step": 57015
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.451892932916253,
      "learning_rate": 1.1207477637588982e-06,
      "loss": 0.3697,
      "step": 57020
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.9727925653567415,
      "learning_rate": 1.1199711418940739e-06,
      "loss": 0.3418,
      "step": 57025
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.6742176836623375,
      "learning_rate": 1.1191947552639714e-06,
      "loss": 0.3522,
      "step": 57030
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.9566178267034964,
      "learning_rate": 1.1184186039156585e-06,
      "loss": 0.3449,
      "step": 57035
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.6715870793680265,
      "learning_rate": 1.117642687896192e-06,
      "loss": 0.3304,
      "step": 57040
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.299037989487199,
      "learning_rate": 1.116867007252615e-06,
      "loss": 0.3465,
      "step": 57045
    },
    {
      "epoch": 2.41,
      "grad_norm": 5.842260670438582,
      "learning_rate": 1.1160915620319519e-06,
      "loss": 0.3149,
      "step": 57050
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.327095322186797,
      "learning_rate": 1.1153163522812166e-06,
      "loss": 0.3645,
      "step": 57055
    },
    {
      "epoch": 2.41,
      "grad_norm": 4.009791496917735,
      "learning_rate": 1.1145413780474102e-06,
      "loss": 0.3038,
      "step": 57060
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.219170096938655,
      "learning_rate": 1.113766639377513e-06,
      "loss": 0.3463,
      "step": 57065
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.17106153268378,
      "learning_rate": 1.1129921363184976e-06,
      "loss": 0.3369,
      "step": 57070
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.176857000090576,
      "learning_rate": 1.11221786891732e-06,
      "loss": 0.3354,
      "step": 57075
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.39145399245273,
      "learning_rate": 1.1114438372209208e-06,
      "loss": 0.346,
      "step": 57080
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.798230384768873,
      "learning_rate": 1.1106700412762273e-06,
      "loss": 0.3407,
      "step": 57085
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.676647171519071,
      "learning_rate": 1.1098964811301528e-06,
      "loss": 0.3621,
      "step": 57090
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.095255275659979,
      "learning_rate": 1.1091231568295974e-06,
      "loss": 0.3391,
      "step": 57095
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.207240671239678,
      "learning_rate": 1.108350068421442e-06,
      "loss": 0.341,
      "step": 57100
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.349139686535518,
      "learning_rate": 1.1075772159525593e-06,
      "loss": 0.3391,
      "step": 57105
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.674538039444013,
      "learning_rate": 1.1068045994698062e-06,
      "loss": 0.3356,
      "step": 57110
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.82336304763771,
      "learning_rate": 1.1060322190200206e-06,
      "loss": 0.3316,
      "step": 57115
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.14540455272532,
      "learning_rate": 1.105260074650032e-06,
      "loss": 0.3588,
      "step": 57120
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.9799566895007277,
      "learning_rate": 1.1044881664066538e-06,
      "loss": 0.3468,
      "step": 57125
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.323147081653709,
      "learning_rate": 1.1037164943366819e-06,
      "loss": 0.3781,
      "step": 57130
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.177913963804417,
      "learning_rate": 1.1029450584869017e-06,
      "loss": 0.3465,
      "step": 57135
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.943590613600536,
      "learning_rate": 1.1021738589040843e-06,
      "loss": 0.3086,
      "step": 57140
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.5844919474233095,
      "learning_rate": 1.1014028956349854e-06,
      "loss": 0.3309,
      "step": 57145
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.505840638835024,
      "learning_rate": 1.1006321687263438e-06,
      "loss": 0.3451,
      "step": 57150
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.920962168656925,
      "learning_rate": 1.0998616782248878e-06,
      "loss": 0.3523,
      "step": 57155
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.278549877096805,
      "learning_rate": 1.0990914241773314e-06,
      "loss": 0.3304,
      "step": 57160
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.279815723608131,
      "learning_rate": 1.0983214066303705e-06,
      "loss": 0.3192,
      "step": 57165
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.332700203661917,
      "learning_rate": 1.0975516256306913e-06,
      "loss": 0.3359,
      "step": 57170
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.058974227370347,
      "learning_rate": 1.0967820812249602e-06,
      "loss": 0.3475,
      "step": 57175
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.920267009674776,
      "learning_rate": 1.0960127734598358e-06,
      "loss": 0.3496,
      "step": 57180
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.950268933823067,
      "learning_rate": 1.0952437023819556e-06,
      "loss": 0.3415,
      "step": 57185
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.303089897269296,
      "learning_rate": 1.0944748680379485e-06,
      "loss": 0.3223,
      "step": 57190
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.522371916364081,
      "learning_rate": 1.0937062704744273e-06,
      "loss": 0.3361,
      "step": 57195
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.1440315944873385,
      "learning_rate": 1.0929379097379867e-06,
      "loss": 0.3302,
      "step": 57200
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.4479956726598364,
      "learning_rate": 1.0921697858752123e-06,
      "loss": 0.3173,
      "step": 57205
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.9852513369908555,
      "learning_rate": 1.091401898932673e-06,
      "loss": 0.3142,
      "step": 57210
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.6148319317635105,
      "learning_rate": 1.090634248956925e-06,
      "loss": 0.3568,
      "step": 57215
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.153989987777079,
      "learning_rate": 1.0898668359945052e-06,
      "loss": 0.3581,
      "step": 57220
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.043643162056907,
      "learning_rate": 1.0890996600919417e-06,
      "loss": 0.3175,
      "step": 57225
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.769632091804468,
      "learning_rate": 1.0883327212957473e-06,
      "loss": 0.3437,
      "step": 57230
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.1185154700071855,
      "learning_rate": 1.0875660196524162e-06,
      "loss": 0.3705,
      "step": 57235
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.8860089998557124,
      "learning_rate": 1.086799555208433e-06,
      "loss": 0.343,
      "step": 57240
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.422458694447115,
      "learning_rate": 1.0860333280102676e-06,
      "loss": 0.337,
      "step": 57245
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.081317646390591,
      "learning_rate": 1.085267338104371e-06,
      "loss": 0.3443,
      "step": 57250
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.629507468223407,
      "learning_rate": 1.0845015855371844e-06,
      "loss": 0.366,
      "step": 57255
    },
    {
      "epoch": 2.42,
      "grad_norm": 7.457854345957066,
      "learning_rate": 1.083736070355133e-06,
      "loss": 0.3448,
      "step": 57260
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.645145445076018,
      "learning_rate": 1.08297079260463e-06,
      "loss": 0.3301,
      "step": 57265
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.4120901532230996,
      "learning_rate": 1.0822057523320677e-06,
      "loss": 0.3042,
      "step": 57270
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.314199996264397,
      "learning_rate": 1.0814409495838314e-06,
      "loss": 0.3383,
      "step": 57275
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.847019796796359,
      "learning_rate": 1.0806763844062884e-06,
      "loss": 0.3256,
      "step": 57280
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.168166630142644,
      "learning_rate": 1.0799120568457899e-06,
      "loss": 0.3228,
      "step": 57285
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.122705712166449,
      "learning_rate": 1.079147966948677e-06,
      "loss": 0.3518,
      "step": 57290
    },
    {
      "epoch": 2.42,
      "grad_norm": 4.3011995455462815,
      "learning_rate": 1.078384114761275e-06,
      "loss": 0.3421,
      "step": 57295
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.8623773092730533,
      "learning_rate": 1.0776205003298906e-06,
      "loss": 0.3476,
      "step": 57300
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.079811980004402,
      "learning_rate": 1.076857123700822e-06,
      "loss": 0.3434,
      "step": 57305
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.671007921855317,
      "learning_rate": 1.0760939849203512e-06,
      "loss": 0.366,
      "step": 57310
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.971629722778755,
      "learning_rate": 1.0753310840347425e-06,
      "loss": 0.2955,
      "step": 57315
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.056073541879393,
      "learning_rate": 1.0745684210902503e-06,
      "loss": 0.327,
      "step": 57320
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.010705608054268,
      "learning_rate": 1.073805996133111e-06,
      "loss": 0.3525,
      "step": 57325
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.006245451207343,
      "learning_rate": 1.0730438092095512e-06,
      "loss": 0.3459,
      "step": 57330
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.833995367390576,
      "learning_rate": 1.0722818603657776e-06,
      "loss": 0.3537,
      "step": 57335
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.18623663348344,
      "learning_rate": 1.0715201496479844e-06,
      "loss": 0.3384,
      "step": 57340
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.031412104001203,
      "learning_rate": 1.0707586771023526e-06,
      "loss": 0.3422,
      "step": 57345
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.10377783078606,
      "learning_rate": 1.0699974427750493e-06,
      "loss": 0.3578,
      "step": 57350
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.8387256636012395,
      "learning_rate": 1.069236446712224e-06,
      "loss": 0.3487,
      "step": 57355
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.8569554352473374,
      "learning_rate": 1.068475688960014e-06,
      "loss": 0.312,
      "step": 57360
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.9106425919117127,
      "learning_rate": 1.0677151695645443e-06,
      "loss": 0.3204,
      "step": 57365
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.623182944435181,
      "learning_rate": 1.066954888571919e-06,
      "loss": 0.3383,
      "step": 57370
    },
    {
      "epoch": 2.43,
      "grad_norm": 6.396277386889992,
      "learning_rate": 1.066194846028234e-06,
      "loss": 0.3266,
      "step": 57375
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.708517370015805,
      "learning_rate": 1.065435041979569e-06,
      "loss": 0.307,
      "step": 57380
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.776082362017705,
      "learning_rate": 1.0646754764719868e-06,
      "loss": 0.3255,
      "step": 57385
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.651993225058832,
      "learning_rate": 1.0639161495515382e-06,
      "loss": 0.3476,
      "step": 57390
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.041922501312812,
      "learning_rate": 1.06315706126426e-06,
      "loss": 0.3297,
      "step": 57395
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.05497325401343,
      "learning_rate": 1.0623982116561737e-06,
      "loss": 0.3708,
      "step": 57400
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.1239151536547904,
      "learning_rate": 1.0616396007732837e-06,
      "loss": 0.3367,
      "step": 57405
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.176437728071655,
      "learning_rate": 1.0608812286615844e-06,
      "loss": 0.3412,
      "step": 57410
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.8292579727100775,
      "learning_rate": 1.0601230953670544e-06,
      "loss": 0.3287,
      "step": 57415
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.018676745502739,
      "learning_rate": 1.0593652009356541e-06,
      "loss": 0.3396,
      "step": 57420
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.757641193806856,
      "learning_rate": 1.0586075454133344e-06,
      "loss": 0.3205,
      "step": 57425
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.854202971176708,
      "learning_rate": 1.0578501288460307e-06,
      "loss": 0.327,
      "step": 57430
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.048296795232814,
      "learning_rate": 1.0570929512796608e-06,
      "loss": 0.3136,
      "step": 57435
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.4253108170231386,
      "learning_rate": 1.0563360127601307e-06,
      "loss": 0.3087,
      "step": 57440
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.344310223586198,
      "learning_rate": 1.0555793133333313e-06,
      "loss": 0.3226,
      "step": 57445
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.147417267918798,
      "learning_rate": 1.0548228530451416e-06,
      "loss": 0.3477,
      "step": 57450
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.442360431712468,
      "learning_rate": 1.0540666319414195e-06,
      "loss": 0.3122,
      "step": 57455
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.676397447636436,
      "learning_rate": 1.0533106500680147e-06,
      "loss": 0.3635,
      "step": 57460
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.754240180536058,
      "learning_rate": 1.0525549074707613e-06,
      "loss": 0.3475,
      "step": 57465
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.002888257892067,
      "learning_rate": 1.0517994041954742e-06,
      "loss": 0.3425,
      "step": 57470
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.301489663317671,
      "learning_rate": 1.05104414028796e-06,
      "loss": 0.3327,
      "step": 57475
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.011149077297497,
      "learning_rate": 1.0502891157940092e-06,
      "loss": 0.3588,
      "step": 57480
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.696351906337958,
      "learning_rate": 1.049534330759393e-06,
      "loss": 0.3518,
      "step": 57485
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.04155955519014,
      "learning_rate": 1.0487797852298742e-06,
      "loss": 0.3415,
      "step": 57490
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.975640996381281,
      "learning_rate": 1.0480254792511996e-06,
      "loss": 0.3081,
      "step": 57495
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.7245848930658427,
      "learning_rate": 1.0472714128690992e-06,
      "loss": 0.3249,
      "step": 57500
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.7477953543032463,
      "learning_rate": 1.0465175861292887e-06,
      "loss": 0.3507,
      "step": 57505
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.0224121271880255,
      "learning_rate": 1.0457639990774715e-06,
      "loss": 0.3422,
      "step": 57510
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.721288576188212,
      "learning_rate": 1.0450106517593355e-06,
      "loss": 0.343,
      "step": 57515
    },
    {
      "epoch": 2.43,
      "grad_norm": 4.158340769514438,
      "learning_rate": 1.0442575442205556e-06,
      "loss": 0.32,
      "step": 57520
    },
    {
      "epoch": 2.43,
      "grad_norm": 10.097333264311956,
      "learning_rate": 1.0435046765067869e-06,
      "loss": 0.3354,
      "step": 57525
    },
    {
      "epoch": 2.43,
      "grad_norm": 3.88933683718144,
      "learning_rate": 1.0427520486636754e-06,
      "loss": 0.3374,
      "step": 57530
    },
    {
      "epoch": 2.43,
      "grad_norm": 5.088273658952878,
      "learning_rate": 1.0419996607368521e-06,
      "loss": 0.3453,
      "step": 57535
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.142753100821465,
      "learning_rate": 1.0412475127719296e-06,
      "loss": 0.288,
      "step": 57540
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.08556700288829,
      "learning_rate": 1.0404956048145094e-06,
      "loss": 0.3279,
      "step": 57545
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.286816117138864,
      "learning_rate": 1.0397439369101791e-06,
      "loss": 0.3048,
      "step": 57550
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.164524582619652,
      "learning_rate": 1.0389925091045073e-06,
      "loss": 0.3646,
      "step": 57555
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.794924632062213,
      "learning_rate": 1.0382413214430515e-06,
      "loss": 0.338,
      "step": 57560
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.899638909029547,
      "learning_rate": 1.0374903739713549e-06,
      "loss": 0.324,
      "step": 57565
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.109486744421262,
      "learning_rate": 1.0367396667349466e-06,
      "loss": 0.356,
      "step": 57570
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.160693008928198,
      "learning_rate": 1.035989199779337e-06,
      "loss": 0.3433,
      "step": 57575
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.24856131113691,
      "learning_rate": 1.035238973150025e-06,
      "loss": 0.3618,
      "step": 57580
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.078613817089558,
      "learning_rate": 1.0344889868924973e-06,
      "loss": 0.3472,
      "step": 57585
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.726913556723607,
      "learning_rate": 1.0337392410522207e-06,
      "loss": 0.3559,
      "step": 57590
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.598044912764058,
      "learning_rate": 1.03298973567465e-06,
      "loss": 0.3411,
      "step": 57595
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.728965349914194,
      "learning_rate": 1.032240470805228e-06,
      "loss": 0.3453,
      "step": 57600
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.7912803911807837,
      "learning_rate": 1.0314914464893776e-06,
      "loss": 0.359,
      "step": 57605
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.9649811004033784,
      "learning_rate": 1.0307426627725115e-06,
      "loss": 0.3065,
      "step": 57610
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.998160224698893,
      "learning_rate": 1.0299941197000268e-06,
      "loss": 0.3439,
      "step": 57615
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.157528765418892,
      "learning_rate": 1.0292458173173037e-06,
      "loss": 0.3434,
      "step": 57620
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.126924245463731,
      "learning_rate": 1.0284977556697102e-06,
      "loss": 0.3657,
      "step": 57625
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.3799710129198606,
      "learning_rate": 1.0277499348025993e-06,
      "loss": 0.3178,
      "step": 57630
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.198746943937231,
      "learning_rate": 1.0270023547613106e-06,
      "loss": 0.3293,
      "step": 57635
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.9167831767502874,
      "learning_rate": 1.0262550155911648e-06,
      "loss": 0.3106,
      "step": 57640
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.114667366858954,
      "learning_rate": 1.0255079173374727e-06,
      "loss": 0.3523,
      "step": 57645
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.158913337625498,
      "learning_rate": 1.0247610600455294e-06,
      "loss": 0.3596,
      "step": 57650
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.176295388025411,
      "learning_rate": 1.024014443760612e-06,
      "loss": 0.3534,
      "step": 57655
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.934322932853985,
      "learning_rate": 1.0232680685279889e-06,
      "loss": 0.3221,
      "step": 57660
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.956755526343446,
      "learning_rate": 1.0225219343929078e-06,
      "loss": 0.3128,
      "step": 57665
    },
    {
      "epoch": 2.44,
      "grad_norm": 7.917637726100076,
      "learning_rate": 1.0217760414006067e-06,
      "loss": 0.3568,
      "step": 57670
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.308814510640868,
      "learning_rate": 1.0210303895963047e-06,
      "loss": 0.3322,
      "step": 57675
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.201381113507919,
      "learning_rate": 1.02028497902521e-06,
      "loss": 0.3514,
      "step": 57680
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.788767981748645,
      "learning_rate": 1.0195398097325154e-06,
      "loss": 0.3315,
      "step": 57685
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.390397389469206,
      "learning_rate": 1.0187948817633963e-06,
      "loss": 0.3234,
      "step": 57690
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.030540324968917,
      "learning_rate": 1.0180501951630163e-06,
      "loss": 0.3272,
      "step": 57695
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.186967476322787,
      "learning_rate": 1.0173057499765237e-06,
      "loss": 0.313,
      "step": 57700
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.7722700856943385,
      "learning_rate": 1.0165615462490535e-06,
      "loss": 0.3264,
      "step": 57705
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.712247172111406,
      "learning_rate": 1.0158175840257218e-06,
      "loss": 0.3365,
      "step": 57710
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.216025199584915,
      "learning_rate": 1.0150738633516345e-06,
      "loss": 0.3357,
      "step": 57715
    },
    {
      "epoch": 2.44,
      "grad_norm": 5.912224589842646,
      "learning_rate": 1.0143303842718816e-06,
      "loss": 0.3367,
      "step": 57720
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.683108902872549,
      "learning_rate": 1.0135871468315361e-06,
      "loss": 0.328,
      "step": 57725
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.9188250624071963,
      "learning_rate": 1.0128441510756598e-06,
      "loss": 0.3188,
      "step": 57730
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.380345490609457,
      "learning_rate": 1.0121013970492993e-06,
      "loss": 0.3269,
      "step": 57735
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.2589191332529905,
      "learning_rate": 1.0113588847974831e-06,
      "loss": 0.3494,
      "step": 57740
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.363706637948588,
      "learning_rate": 1.010616614365229e-06,
      "loss": 0.3336,
      "step": 57745
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.8149506810636016,
      "learning_rate": 1.0098745857975377e-06,
      "loss": 0.3549,
      "step": 57750
    },
    {
      "epoch": 2.44,
      "grad_norm": 3.5876270672835022,
      "learning_rate": 1.009132799139399e-06,
      "loss": 0.3352,
      "step": 57755
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.863643630747743,
      "learning_rate": 1.0083912544357815e-06,
      "loss": 0.3339,
      "step": 57760
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.1185300834446155,
      "learning_rate": 1.0076499517316452e-06,
      "loss": 0.3359,
      "step": 57765
    },
    {
      "epoch": 2.44,
      "grad_norm": 4.494898866158834,
      "learning_rate": 1.0069088910719332e-06,
      "loss": 0.3536,
      "step": 57770
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.81819663192064,
      "learning_rate": 1.0061680725015716e-06,
      "loss": 0.329,
      "step": 57775
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.106686848080117,
      "learning_rate": 1.0054274960654758e-06,
      "loss": 0.353,
      "step": 57780
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.540619469094639,
      "learning_rate": 1.0046871618085463e-06,
      "loss": 0.3633,
      "step": 57785
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.170156793140971,
      "learning_rate": 1.0039470697756636e-06,
      "loss": 0.337,
      "step": 57790
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.783237611291634,
      "learning_rate": 1.0032072200116993e-06,
      "loss": 0.338,
      "step": 57795
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.841812060669276,
      "learning_rate": 1.0024676125615091e-06,
      "loss": 0.3297,
      "step": 57800
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.307104071591766,
      "learning_rate": 1.001728247469933e-06,
      "loss": 0.3347,
      "step": 57805
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.672964346078582,
      "learning_rate": 1.0009891247817949e-06,
      "loss": 0.3353,
      "step": 57810
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.529553607824022,
      "learning_rate": 1.0002502445419066e-06,
      "loss": 0.3676,
      "step": 57815
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.589727959893476,
      "learning_rate": 9.995116067950654e-07,
      "loss": 0.3516,
      "step": 57820
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.051004716091924,
      "learning_rate": 9.987732115860521e-07,
      "loss": 0.3179,
      "step": 57825
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.015515864329743,
      "learning_rate": 9.980350589596315e-07,
      "loss": 0.3207,
      "step": 57830
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.111480734761745,
      "learning_rate": 9.972971489605572e-07,
      "loss": 0.3558,
      "step": 57835
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.34980478291329,
      "learning_rate": 9.965594816335677e-07,
      "loss": 0.3442,
      "step": 57840
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.030695302400033,
      "learning_rate": 9.95822057023383e-07,
      "loss": 0.3362,
      "step": 57845
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.0218469246048825,
      "learning_rate": 9.950848751747123e-07,
      "loss": 0.3321,
      "step": 57850
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.084027852516597,
      "learning_rate": 9.9434793613225e-07,
      "loss": 0.3481,
      "step": 57855
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.918143296888962,
      "learning_rate": 9.936112399406717e-07,
      "loss": 0.3273,
      "step": 57860
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.8605849080771644,
      "learning_rate": 9.928747866446432e-07,
      "loss": 0.3569,
      "step": 57865
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.562713887654944,
      "learning_rate": 9.921385762888126e-07,
      "loss": 0.3239,
      "step": 57870
    },
    {
      "epoch": 2.45,
      "grad_norm": 6.868384103138348,
      "learning_rate": 9.914026089178158e-07,
      "loss": 0.3381,
      "step": 57875
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.955612731139122,
      "learning_rate": 9.906668845762702e-07,
      "loss": 0.3303,
      "step": 57880
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.144102524507659,
      "learning_rate": 9.899314033087816e-07,
      "loss": 0.3496,
      "step": 57885
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.104408580151624,
      "learning_rate": 9.891961651599407e-07,
      "loss": 0.3294,
      "step": 57890
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.114335058040739,
      "learning_rate": 9.884611701743214e-07,
      "loss": 0.3657,
      "step": 57895
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.884242465242713,
      "learning_rate": 9.877264183964847e-07,
      "loss": 0.332,
      "step": 57900
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.638497784630729,
      "learning_rate": 9.869919098709786e-07,
      "loss": 0.326,
      "step": 57905
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.9003398205687696,
      "learning_rate": 9.8625764464233e-07,
      "loss": 0.3542,
      "step": 57910
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.503952794337806,
      "learning_rate": 9.85523622755058e-07,
      "loss": 0.3478,
      "step": 57915
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.990727130641454,
      "learning_rate": 9.847898442536653e-07,
      "loss": 0.3239,
      "step": 57920
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.119265451764556,
      "learning_rate": 9.840563091826356e-07,
      "loss": 0.3126,
      "step": 57925
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.10291623220106,
      "learning_rate": 9.833230175864421e-07,
      "loss": 0.3438,
      "step": 57930
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.011869773449787,
      "learning_rate": 9.82589969509543e-07,
      "loss": 0.3451,
      "step": 57935
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.997151637525354,
      "learning_rate": 9.818571649963815e-07,
      "loss": 0.3231,
      "step": 57940
    },
    {
      "epoch": 2.45,
      "grad_norm": 9.557629859936185,
      "learning_rate": 9.81124604091383e-07,
      "loss": 0.3356,
      "step": 57945
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.921715430224236,
      "learning_rate": 9.803922868389621e-07,
      "loss": 0.319,
      "step": 57950
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.2303711254001435,
      "learning_rate": 9.796602132835175e-07,
      "loss": 0.3398,
      "step": 57955
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.6323853353094595,
      "learning_rate": 9.789283834694314e-07,
      "loss": 0.2968,
      "step": 57960
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.311430514944957,
      "learning_rate": 9.781967974410722e-07,
      "loss": 0.302,
      "step": 57965
    },
    {
      "epoch": 2.45,
      "grad_norm": 5.043047224285803,
      "learning_rate": 9.77465455242796e-07,
      "loss": 0.3449,
      "step": 57970
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.9690928373540038,
      "learning_rate": 9.76734356918939e-07,
      "loss": 0.3133,
      "step": 57975
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.048784510867277,
      "learning_rate": 9.760035025138287e-07,
      "loss": 0.3408,
      "step": 57980
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.421545900895566,
      "learning_rate": 9.752728920717714e-07,
      "loss": 0.3474,
      "step": 57985
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.014967560969097,
      "learning_rate": 9.74542525637065e-07,
      "loss": 0.3394,
      "step": 57990
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.091823324036924,
      "learning_rate": 9.738124032539863e-07,
      "loss": 0.3306,
      "step": 57995
    },
    {
      "epoch": 2.45,
      "grad_norm": 3.8277986429628736,
      "learning_rate": 9.730825249668025e-07,
      "loss": 0.3106,
      "step": 58000
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.211996454142268,
      "learning_rate": 9.723528908197627e-07,
      "loss": 0.3227,
      "step": 58005
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.7358848196516883,
      "learning_rate": 9.716235008571057e-07,
      "loss": 0.3329,
      "step": 58010
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.022325646444267,
      "learning_rate": 9.708943551230477e-07,
      "loss": 0.3216,
      "step": 58015
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.343900784186914,
      "learning_rate": 9.701654536617972e-07,
      "loss": 0.3365,
      "step": 58020
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.9269734842288497,
      "learning_rate": 9.69436796517546e-07,
      "loss": 0.3285,
      "step": 58025
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.007266108796683,
      "learning_rate": 9.687083837344686e-07,
      "loss": 0.3744,
      "step": 58030
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.3212273849333895,
      "learning_rate": 9.679802153567269e-07,
      "loss": 0.334,
      "step": 58035
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.9806066433569924,
      "learning_rate": 9.672522914284698e-07,
      "loss": 0.335,
      "step": 58040
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.6812938298816436,
      "learning_rate": 9.665246119938255e-07,
      "loss": 0.3446,
      "step": 58045
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.059877876749104,
      "learning_rate": 9.657971770969132e-07,
      "loss": 0.3464,
      "step": 58050
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.710145415616947,
      "learning_rate": 9.650699867818352e-07,
      "loss": 0.3226,
      "step": 58055
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.26468005775309,
      "learning_rate": 9.643430410926796e-07,
      "loss": 0.3404,
      "step": 58060
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.758440218495642,
      "learning_rate": 9.636163400735171e-07,
      "loss": 0.3179,
      "step": 58065
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.176114235529649,
      "learning_rate": 9.628898837684058e-07,
      "loss": 0.34,
      "step": 58070
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.720959094192976,
      "learning_rate": 9.621636722213911e-07,
      "loss": 0.3297,
      "step": 58075
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.198156584241073,
      "learning_rate": 9.614377054764972e-07,
      "loss": 0.3628,
      "step": 58080
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.7058224136965374,
      "learning_rate": 9.607119835777395e-07,
      "loss": 0.3496,
      "step": 58085
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.186234773337296,
      "learning_rate": 9.599865065691177e-07,
      "loss": 0.3472,
      "step": 58090
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.301061000952912,
      "learning_rate": 9.592612744946122e-07,
      "loss": 0.323,
      "step": 58095
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.9779878522924412,
      "learning_rate": 9.585362873981934e-07,
      "loss": 0.3483,
      "step": 58100
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.594277868331492,
      "learning_rate": 9.578115453238146e-07,
      "loss": 0.3302,
      "step": 58105
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.9350733920100303,
      "learning_rate": 9.570870483154166e-07,
      "loss": 0.3312,
      "step": 58110
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.6554657083400315,
      "learning_rate": 9.563627964169208e-07,
      "loss": 0.3166,
      "step": 58115
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.693416019850289,
      "learning_rate": 9.55638789672238e-07,
      "loss": 0.3074,
      "step": 58120
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.463928052541967,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.3326,
      "step": 58125
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.370003220954758,
      "learning_rate": 9.54191511819874e-07,
      "loss": 0.3788,
      "step": 58130
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.208975837731788,
      "learning_rate": 9.534682407999363e-07,
      "loss": 0.3286,
      "step": 58135
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.772740363982837,
      "learning_rate": 9.527452151093003e-07,
      "loss": 0.3068,
      "step": 58140
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.344465173934658,
      "learning_rate": 9.520224347918011e-07,
      "loss": 0.3276,
      "step": 58145
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.7909641611039735,
      "learning_rate": 9.512998998912565e-07,
      "loss": 0.3632,
      "step": 58150
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.020885442292411,
      "learning_rate": 9.50577610451473e-07,
      "loss": 0.3348,
      "step": 58155
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.2546838652340355,
      "learning_rate": 9.498555665162423e-07,
      "loss": 0.3195,
      "step": 58160
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.8440447611500277,
      "learning_rate": 9.491337681293372e-07,
      "loss": 0.3193,
      "step": 58165
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.114406331294704,
      "learning_rate": 9.484122153345193e-07,
      "loss": 0.3688,
      "step": 58170
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.5736438617384545,
      "learning_rate": 9.476909081755348e-07,
      "loss": 0.3284,
      "step": 58175
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.775883518369523,
      "learning_rate": 9.469698466961153e-07,
      "loss": 0.3473,
      "step": 58180
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.1274448455720165,
      "learning_rate": 9.462490309399741e-07,
      "loss": 0.3255,
      "step": 58185
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.95530638738191,
      "learning_rate": 9.455284609508131e-07,
      "loss": 0.3115,
      "step": 58190
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.895430299413554,
      "learning_rate": 9.448081367723205e-07,
      "loss": 0.335,
      "step": 58195
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.455174541729543,
      "learning_rate": 9.440880584481637e-07,
      "loss": 0.3426,
      "step": 58200
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.070652128340231,
      "learning_rate": 9.433682260220011e-07,
      "loss": 0.3334,
      "step": 58205
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.011371457504796,
      "learning_rate": 9.426486395374757e-07,
      "loss": 0.3299,
      "step": 58210
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.8882679379722207,
      "learning_rate": 9.419292990382101e-07,
      "loss": 0.3501,
      "step": 58215
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.4677649919736595,
      "learning_rate": 9.412102045678173e-07,
      "loss": 0.3261,
      "step": 58220
    },
    {
      "epoch": 2.46,
      "grad_norm": 3.9680582153818484,
      "learning_rate": 9.404913561698964e-07,
      "loss": 0.3285,
      "step": 58225
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.122767508663721,
      "learning_rate": 9.397727538880247e-07,
      "loss": 0.3396,
      "step": 58230
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.1124101755867635,
      "learning_rate": 9.390543977657718e-07,
      "loss": 0.323,
      "step": 58235
    },
    {
      "epoch": 2.46,
      "grad_norm": 4.291380443769415,
      "learning_rate": 9.383362878466884e-07,
      "loss": 0.3665,
      "step": 58240
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.7092245249342137,
      "learning_rate": 9.376184241743136e-07,
      "loss": 0.3219,
      "step": 58245
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.687090341626141,
      "learning_rate": 9.369008067921659e-07,
      "loss": 0.3458,
      "step": 58250
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.973464197846426,
      "learning_rate": 9.361834357437544e-07,
      "loss": 0.3368,
      "step": 58255
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.208980161266249,
      "learning_rate": 9.35466311072572e-07,
      "loss": 0.3184,
      "step": 58260
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.6889042926387217,
      "learning_rate": 9.34749432822093e-07,
      "loss": 0.3148,
      "step": 58265
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.180494058250353,
      "learning_rate": 9.340328010357819e-07,
      "loss": 0.3359,
      "step": 58270
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.893159481283603,
      "learning_rate": 9.333164157570862e-07,
      "loss": 0.3124,
      "step": 58275
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.644463092156722,
      "learning_rate": 9.326002770294362e-07,
      "loss": 0.354,
      "step": 58280
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.666878066985316,
      "learning_rate": 9.318843848962506e-07,
      "loss": 0.3507,
      "step": 58285
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.590830100358635,
      "learning_rate": 9.311687394009317e-07,
      "loss": 0.3462,
      "step": 58290
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.595728307105326,
      "learning_rate": 9.304533405868682e-07,
      "loss": 0.3155,
      "step": 58295
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.620419596737383,
      "learning_rate": 9.297381884974305e-07,
      "loss": 0.3509,
      "step": 58300
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.52902342051602,
      "learning_rate": 9.290232831759783e-07,
      "loss": 0.3291,
      "step": 58305
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.831226808705749,
      "learning_rate": 9.283086246658512e-07,
      "loss": 0.3199,
      "step": 58310
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.346870057432061,
      "learning_rate": 9.275942130103805e-07,
      "loss": 0.3573,
      "step": 58315
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.932522509004938,
      "learning_rate": 9.26880048252875e-07,
      "loss": 0.3355,
      "step": 58320
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.47507555533065,
      "learning_rate": 9.261661304366349e-07,
      "loss": 0.3332,
      "step": 58325
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.36904969334475,
      "learning_rate": 9.254524596049436e-07,
      "loss": 0.315,
      "step": 58330
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.066510037309596,
      "learning_rate": 9.247390358010666e-07,
      "loss": 0.3097,
      "step": 58335
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.846180841540975,
      "learning_rate": 9.240258590682577e-07,
      "loss": 0.3436,
      "step": 58340
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.120987707775423,
      "learning_rate": 9.233129294497556e-07,
      "loss": 0.3184,
      "step": 58345
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.4204227091229478,
      "learning_rate": 9.226002469887818e-07,
      "loss": 0.301,
      "step": 58350
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.316774541312869,
      "learning_rate": 9.21887811728544e-07,
      "loss": 0.3174,
      "step": 58355
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.897440013264647,
      "learning_rate": 9.21175623712236e-07,
      "loss": 0.3193,
      "step": 58360
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.558413012499301,
      "learning_rate": 9.204636829830366e-07,
      "loss": 0.3292,
      "step": 58365
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.46212629300126,
      "learning_rate": 9.197519895841062e-07,
      "loss": 0.3283,
      "step": 58370
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.7081494631482386,
      "learning_rate": 9.190405435585942e-07,
      "loss": 0.3201,
      "step": 58375
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.821804232453492,
      "learning_rate": 9.183293449496338e-07,
      "loss": 0.3316,
      "step": 58380
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.008789381846802,
      "learning_rate": 9.176183938003413e-07,
      "loss": 0.3184,
      "step": 58385
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.954433012395198,
      "learning_rate": 9.16907690153821e-07,
      "loss": 0.3239,
      "step": 58390
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.41426769054615,
      "learning_rate": 9.161972340531611e-07,
      "loss": 0.3532,
      "step": 58395
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.485994223007647,
      "learning_rate": 9.154870255414323e-07,
      "loss": 0.3359,
      "step": 58400
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.35881063486017,
      "learning_rate": 9.147770646616938e-07,
      "loss": 0.3144,
      "step": 58405
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.091901587978427,
      "learning_rate": 9.140673514569887e-07,
      "loss": 0.3263,
      "step": 58410
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.7238183652070664,
      "learning_rate": 9.133578859703457e-07,
      "loss": 0.3433,
      "step": 58415
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.432782061707567,
      "learning_rate": 9.126486682447755e-07,
      "loss": 0.3423,
      "step": 58420
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.11916193849522,
      "learning_rate": 9.119396983232765e-07,
      "loss": 0.3232,
      "step": 58425
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.8425905287039237,
      "learning_rate": 9.112309762488331e-07,
      "loss": 0.3469,
      "step": 58430
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.2075187448700415,
      "learning_rate": 9.105225020644109e-07,
      "loss": 0.3435,
      "step": 58435
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.2003156453723935,
      "learning_rate": 9.09814275812963e-07,
      "loss": 0.3452,
      "step": 58440
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.195084230276883,
      "learning_rate": 9.091062975374287e-07,
      "loss": 0.3138,
      "step": 58445
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.1021101967404014,
      "learning_rate": 9.083985672807283e-07,
      "loss": 0.3376,
      "step": 58450
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.8959783252176603,
      "learning_rate": 9.076910850857706e-07,
      "loss": 0.3413,
      "step": 58455
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.798082719633067,
      "learning_rate": 9.069838509954499e-07,
      "loss": 0.2872,
      "step": 58460
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.074653191472527,
      "learning_rate": 9.062768650526399e-07,
      "loss": 0.3615,
      "step": 58465
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.256795681957427,
      "learning_rate": 9.055701273002071e-07,
      "loss": 0.3365,
      "step": 58470
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.007933346692168,
      "learning_rate": 9.04863637780995e-07,
      "loss": 0.3233,
      "step": 58475
    },
    {
      "epoch": 2.47,
      "grad_norm": 4.192174292196861,
      "learning_rate": 9.041573965378381e-07,
      "loss": 0.3163,
      "step": 58480
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.3235963385365705,
      "learning_rate": 9.03451403613555e-07,
      "loss": 0.3533,
      "step": 58485
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.000500032147201,
      "learning_rate": 9.027456590509453e-07,
      "loss": 0.3455,
      "step": 58490
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.551394077317539,
      "learning_rate": 9.020401628927972e-07,
      "loss": 0.332,
      "step": 58495
    },
    {
      "epoch": 2.48,
      "grad_norm": 5.288404178687894,
      "learning_rate": 9.013349151818845e-07,
      "loss": 0.3075,
      "step": 58500
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.104596820504148,
      "learning_rate": 9.006299159609616e-07,
      "loss": 0.3333,
      "step": 58505
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.778141298216287,
      "learning_rate": 8.999251652727714e-07,
      "loss": 0.3337,
      "step": 58510
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.342982785724166,
      "learning_rate": 8.992206631600431e-07,
      "loss": 0.3033,
      "step": 58515
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8216353786389456,
      "learning_rate": 8.985164096654853e-07,
      "loss": 0.3295,
      "step": 58520
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.054447736403844,
      "learning_rate": 8.97812404831796e-07,
      "loss": 0.3246,
      "step": 58525
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.72847822466765,
      "learning_rate": 8.971086487016589e-07,
      "loss": 0.3436,
      "step": 58530
    },
    {
      "epoch": 2.48,
      "grad_norm": 5.505367402810515,
      "learning_rate": 8.964051413177377e-07,
      "loss": 0.3244,
      "step": 58535
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.880147906450789,
      "learning_rate": 8.957018827226849e-07,
      "loss": 0.335,
      "step": 58540
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8977944968886624,
      "learning_rate": 8.949988729591375e-07,
      "loss": 0.3612,
      "step": 58545
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.336580404163079,
      "learning_rate": 8.94296112069718e-07,
      "loss": 0.3239,
      "step": 58550
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8582104762199414,
      "learning_rate": 8.935936000970303e-07,
      "loss": 0.3248,
      "step": 58555
    },
    {
      "epoch": 2.48,
      "grad_norm": 5.076738611723542,
      "learning_rate": 8.928913370836667e-07,
      "loss": 0.317,
      "step": 58560
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.9274700203818815,
      "learning_rate": 8.92189323072205e-07,
      "loss": 0.3261,
      "step": 58565
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.9777906919751045,
      "learning_rate": 8.914875581052029e-07,
      "loss": 0.3443,
      "step": 58570
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.7821678402872556,
      "learning_rate": 8.907860422252085e-07,
      "loss": 0.3316,
      "step": 58575
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.943707057062974,
      "learning_rate": 8.900847754747533e-07,
      "loss": 0.3002,
      "step": 58580
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.1933186838592595,
      "learning_rate": 8.89383757896351e-07,
      "loss": 0.3053,
      "step": 58585
    },
    {
      "epoch": 2.48,
      "grad_norm": 5.446519441990898,
      "learning_rate": 8.886829895325033e-07,
      "loss": 0.334,
      "step": 58590
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.00209566047743,
      "learning_rate": 8.879824704256956e-07,
      "loss": 0.3259,
      "step": 58595
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.935041893481826,
      "learning_rate": 8.872822006183996e-07,
      "loss": 0.3122,
      "step": 58600
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.380111008563508,
      "learning_rate": 8.865821801530677e-07,
      "loss": 0.3192,
      "step": 58605
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8277054381220643,
      "learning_rate": 8.858824090721424e-07,
      "loss": 0.3362,
      "step": 58610
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8100739264799737,
      "learning_rate": 8.851828874180491e-07,
      "loss": 0.3097,
      "step": 58615
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.768931206364091,
      "learning_rate": 8.844836152331954e-07,
      "loss": 0.3316,
      "step": 58620
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.7458001777082153,
      "learning_rate": 8.837845925599775e-07,
      "loss": 0.3182,
      "step": 58625
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.948371023352602,
      "learning_rate": 8.83085819440776e-07,
      "loss": 0.3382,
      "step": 58630
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.135767218043399,
      "learning_rate": 8.823872959179547e-07,
      "loss": 0.3225,
      "step": 58635
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.893008466657662,
      "learning_rate": 8.81689022033862e-07,
      "loss": 0.3167,
      "step": 58640
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.719960746100174,
      "learning_rate": 8.809909978308323e-07,
      "loss": 0.331,
      "step": 58645
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.87942159322901,
      "learning_rate": 8.802932233511874e-07,
      "loss": 0.3454,
      "step": 58650
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8344993831730423,
      "learning_rate": 8.795956986372278e-07,
      "loss": 0.3557,
      "step": 58655
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.054187803961821,
      "learning_rate": 8.788984237312437e-07,
      "loss": 0.33,
      "step": 58660
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.190706683770032,
      "learning_rate": 8.782013986755094e-07,
      "loss": 0.3282,
      "step": 58665
    },
    {
      "epoch": 2.48,
      "grad_norm": 8.16308410311078,
      "learning_rate": 8.775046235122847e-07,
      "loss": 0.3263,
      "step": 58670
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.8777335600294487,
      "learning_rate": 8.768080982838096e-07,
      "loss": 0.3231,
      "step": 58675
    },
    {
      "epoch": 2.48,
      "grad_norm": 5.084343449050413,
      "learning_rate": 8.761118230323146e-07,
      "loss": 0.3229,
      "step": 58680
    },
    {
      "epoch": 2.48,
      "grad_norm": 6.077679419743137,
      "learning_rate": 8.754157978000133e-07,
      "loss": 0.3362,
      "step": 58685
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.277839053969695,
      "learning_rate": 8.747200226291019e-07,
      "loss": 0.3361,
      "step": 58690
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.334987602613929,
      "learning_rate": 8.740244975617639e-07,
      "loss": 0.3436,
      "step": 58695
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.445038226272411,
      "learning_rate": 8.733292226401691e-07,
      "loss": 0.3232,
      "step": 58700
    },
    {
      "epoch": 2.48,
      "grad_norm": 4.08888295856013,
      "learning_rate": 8.726341979064662e-07,
      "loss": 0.3215,
      "step": 58705
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.357013434606589,
      "learning_rate": 8.719394234027945e-07,
      "loss": 0.3073,
      "step": 58710
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.804892504339479,
      "learning_rate": 8.712448991712757e-07,
      "loss": 0.3333,
      "step": 58715
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.6189021674492063,
      "learning_rate": 8.70550625254018e-07,
      "loss": 0.3127,
      "step": 58720
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.138400164281322,
      "learning_rate": 8.698566016931114e-07,
      "loss": 0.3423,
      "step": 58725
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.006190563873847,
      "learning_rate": 8.691628285306336e-07,
      "loss": 0.3462,
      "step": 58730
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.260099285734865,
      "learning_rate": 8.684693058086463e-07,
      "loss": 0.3378,
      "step": 58735
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.981250927784904,
      "learning_rate": 8.677760335691942e-07,
      "loss": 0.3259,
      "step": 58740
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.099213153900137,
      "learning_rate": 8.670830118543094e-07,
      "loss": 0.3342,
      "step": 58745
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.3514594057971925,
      "learning_rate": 8.663902407060093e-07,
      "loss": 0.3037,
      "step": 58750
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.7539554067145207,
      "learning_rate": 8.656977201662908e-07,
      "loss": 0.3225,
      "step": 58755
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.6137390912871874,
      "learning_rate": 8.650054502771421e-07,
      "loss": 0.3421,
      "step": 58760
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.4318814724480315,
      "learning_rate": 8.643134310805346e-07,
      "loss": 0.3131,
      "step": 58765
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.515224671825227,
      "learning_rate": 8.636216626184196e-07,
      "loss": 0.3269,
      "step": 58770
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.23263990082479,
      "learning_rate": 8.629301449327399e-07,
      "loss": 0.36,
      "step": 58775
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.8923881496274224,
      "learning_rate": 8.622388780654183e-07,
      "loss": 0.3457,
      "step": 58780
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.834103722103489,
      "learning_rate": 8.615478620583678e-07,
      "loss": 0.3164,
      "step": 58785
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.7258403880280575,
      "learning_rate": 8.608570969534797e-07,
      "loss": 0.3223,
      "step": 58790
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.192899136925136,
      "learning_rate": 8.60166582792632e-07,
      "loss": 0.3226,
      "step": 58795
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.6977824554796874,
      "learning_rate": 8.594763196176903e-07,
      "loss": 0.3016,
      "step": 58800
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.0325258340937635,
      "learning_rate": 8.587863074705044e-07,
      "loss": 0.3451,
      "step": 58805
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.577778187756743,
      "learning_rate": 8.580965463929053e-07,
      "loss": 0.3273,
      "step": 58810
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.141649443153325,
      "learning_rate": 8.574070364267117e-07,
      "loss": 0.3389,
      "step": 58815
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.2485709666449205,
      "learning_rate": 8.567177776137286e-07,
      "loss": 0.3187,
      "step": 58820
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.964900119715837,
      "learning_rate": 8.560287699957409e-07,
      "loss": 0.3346,
      "step": 58825
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.051902085781984,
      "learning_rate": 8.553400136145224e-07,
      "loss": 0.332,
      "step": 58830
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.8768151973188063,
      "learning_rate": 8.546515085118317e-07,
      "loss": 0.3162,
      "step": 58835
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.117639526938961,
      "learning_rate": 8.539632547294086e-07,
      "loss": 0.3452,
      "step": 58840
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.158855921756477,
      "learning_rate": 8.532752523089799e-07,
      "loss": 0.346,
      "step": 58845
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.731733355289314,
      "learning_rate": 8.525875012922591e-07,
      "loss": 0.3284,
      "step": 58850
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.214768251587851,
      "learning_rate": 8.519000017209422e-07,
      "loss": 0.3315,
      "step": 58855
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.041421851831155,
      "learning_rate": 8.512127536367082e-07,
      "loss": 0.3264,
      "step": 58860
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.995042913375904,
      "learning_rate": 8.505257570812247e-07,
      "loss": 0.3497,
      "step": 58865
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.060052209606571,
      "learning_rate": 8.498390120961425e-07,
      "loss": 0.3695,
      "step": 58870
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.775867109854676,
      "learning_rate": 8.491525187230959e-07,
      "loss": 0.3194,
      "step": 58875
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.038957450063596,
      "learning_rate": 8.484662770037045e-07,
      "loss": 0.3582,
      "step": 58880
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.89067126182076,
      "learning_rate": 8.477802869795759e-07,
      "loss": 0.3292,
      "step": 58885
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.854609023700863,
      "learning_rate": 8.47094548692296e-07,
      "loss": 0.3359,
      "step": 58890
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.966840882256013,
      "learning_rate": 8.464090621834414e-07,
      "loss": 0.3294,
      "step": 58895
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.429258963801496,
      "learning_rate": 8.457238274945701e-07,
      "loss": 0.3444,
      "step": 58900
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.156591986012652,
      "learning_rate": 8.450388446672275e-07,
      "loss": 0.3405,
      "step": 58905
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.067304446123939,
      "learning_rate": 8.443541137429401e-07,
      "loss": 0.313,
      "step": 58910
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.039966795203793,
      "learning_rate": 8.436696347632217e-07,
      "loss": 0.3352,
      "step": 58915
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.228929101556266,
      "learning_rate": 8.429854077695715e-07,
      "loss": 0.3345,
      "step": 58920
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.089246390736494,
      "learning_rate": 8.423014328034701e-07,
      "loss": 0.3427,
      "step": 58925
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.807227350861328,
      "learning_rate": 8.41617709906386e-07,
      "loss": 0.3442,
      "step": 58930
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.8634331730545606,
      "learning_rate": 8.409342391197722e-07,
      "loss": 0.332,
      "step": 58935
    },
    {
      "epoch": 2.49,
      "grad_norm": 3.8915481839237933,
      "learning_rate": 8.402510204850634e-07,
      "loss": 0.3052,
      "step": 58940
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.370919345732317,
      "learning_rate": 8.395680540436824e-07,
      "loss": 0.3541,
      "step": 58945
    },
    {
      "epoch": 2.49,
      "grad_norm": 4.365954224798414,
      "learning_rate": 8.388853398370356e-07,
      "loss": 0.3251,
      "step": 58950
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.7711686927919352,
      "learning_rate": 8.382028779065154e-07,
      "loss": 0.3547,
      "step": 58955
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.825913535969419,
      "learning_rate": 8.375206682934939e-07,
      "loss": 0.3424,
      "step": 58960
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.008513274760299,
      "learning_rate": 8.368387110393334e-07,
      "loss": 0.3438,
      "step": 58965
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.781910330100673,
      "learning_rate": 8.361570061853785e-07,
      "loss": 0.3439,
      "step": 58970
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.9399810172314,
      "learning_rate": 8.354755537729603e-07,
      "loss": 0.3402,
      "step": 58975
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.170777503352065,
      "learning_rate": 8.347943538433916e-07,
      "loss": 0.3364,
      "step": 58980
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.238833583444122,
      "learning_rate": 8.341134064379718e-07,
      "loss": 0.3021,
      "step": 58985
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.8788168354739323,
      "learning_rate": 8.334327115979867e-07,
      "loss": 0.3066,
      "step": 58990
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.555126485950414,
      "learning_rate": 8.327522693647016e-07,
      "loss": 0.3234,
      "step": 58995
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.104557432455638,
      "learning_rate": 8.320720797793718e-07,
      "loss": 0.3456,
      "step": 59000
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.951570563823209,
      "learning_rate": 8.313921428832361e-07,
      "loss": 0.3294,
      "step": 59005
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.173219356345165,
      "learning_rate": 8.307124587175141e-07,
      "loss": 0.3394,
      "step": 59010
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.6202244587136345,
      "learning_rate": 8.300330273234147e-07,
      "loss": 0.3427,
      "step": 59015
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.429648675125979,
      "learning_rate": 8.293538487421299e-07,
      "loss": 0.3133,
      "step": 59020
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.050367722234799,
      "learning_rate": 8.286749230148378e-07,
      "loss": 0.3245,
      "step": 59025
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.0372871422143035,
      "learning_rate": 8.279962501826965e-07,
      "loss": 0.3546,
      "step": 59030
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.4138522404107308,
      "learning_rate": 8.273178302868534e-07,
      "loss": 0.3045,
      "step": 59035
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.306120606640471,
      "learning_rate": 8.26639663368441e-07,
      "loss": 0.3262,
      "step": 59040
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.68966676374268,
      "learning_rate": 8.259617494685706e-07,
      "loss": 0.3307,
      "step": 59045
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.095915542373566,
      "learning_rate": 8.252840886283453e-07,
      "loss": 0.3417,
      "step": 59050
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.49749356746126,
      "learning_rate": 8.246066808888492e-07,
      "loss": 0.3165,
      "step": 59055
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.9747725210497684,
      "learning_rate": 8.239295262911501e-07,
      "loss": 0.3125,
      "step": 59060
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.025760428429545,
      "learning_rate": 8.232526248763028e-07,
      "loss": 0.3277,
      "step": 59065
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.388603525408128,
      "learning_rate": 8.225759766853469e-07,
      "loss": 0.3377,
      "step": 59070
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.961473987771514,
      "learning_rate": 8.218995817593034e-07,
      "loss": 0.3214,
      "step": 59075
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.07421745834712,
      "learning_rate": 8.212234401391817e-07,
      "loss": 0.3318,
      "step": 59080
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.069505368616721,
      "learning_rate": 8.205475518659734e-07,
      "loss": 0.3032,
      "step": 59085
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.7108747535814857,
      "learning_rate": 8.198719169806568e-07,
      "loss": 0.3397,
      "step": 59090
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.96862032196223,
      "learning_rate": 8.191965355241921e-07,
      "loss": 0.3257,
      "step": 59095
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.316274806786525,
      "learning_rate": 8.185214075375269e-07,
      "loss": 0.3519,
      "step": 59100
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.9830456047835145,
      "learning_rate": 8.178465330615926e-07,
      "loss": 0.3157,
      "step": 59105
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.087839912675959,
      "learning_rate": 8.171719121373028e-07,
      "loss": 0.3188,
      "step": 59110
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.843163064948869,
      "learning_rate": 8.164975448055606e-07,
      "loss": 0.3111,
      "step": 59115
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.5514400925150404,
      "learning_rate": 8.158234311072477e-07,
      "loss": 0.3445,
      "step": 59120
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.8350963836785006,
      "learning_rate": 8.151495710832368e-07,
      "loss": 0.318,
      "step": 59125
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.649791019276955,
      "learning_rate": 8.144759647743794e-07,
      "loss": 0.3437,
      "step": 59130
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.7371911867332406,
      "learning_rate": 8.138026122215148e-07,
      "loss": 0.3546,
      "step": 59135
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.860935362757238,
      "learning_rate": 8.131295134654688e-07,
      "loss": 0.3558,
      "step": 59140
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.476222308172504,
      "learning_rate": 8.124566685470458e-07,
      "loss": 0.3368,
      "step": 59145
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.178226295157558,
      "learning_rate": 8.117840775070406e-07,
      "loss": 0.3059,
      "step": 59150
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.239808204584898,
      "learning_rate": 8.111117403862301e-07,
      "loss": 0.3525,
      "step": 59155
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.217491164551992,
      "learning_rate": 8.10439657225377e-07,
      "loss": 0.3346,
      "step": 59160
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.941996504721705,
      "learning_rate": 8.097678280652255e-07,
      "loss": 0.3076,
      "step": 59165
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.9255223873200253,
      "learning_rate": 8.090962529465079e-07,
      "loss": 0.3207,
      "step": 59170
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.647812240194351,
      "learning_rate": 8.084249319099413e-07,
      "loss": 0.3231,
      "step": 59175
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.7071907829865,
      "learning_rate": 8.077538649962224e-07,
      "loss": 0.3184,
      "step": 59180
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.17655956924807,
      "learning_rate": 8.070830522460388e-07,
      "loss": 0.3338,
      "step": 59185
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.2911878007222635,
      "learning_rate": 8.0641249370006e-07,
      "loss": 0.314,
      "step": 59190
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.8579700572713893,
      "learning_rate": 8.057421893989376e-07,
      "loss": 0.3257,
      "step": 59195
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.5675712924895455,
      "learning_rate": 8.050721393833122e-07,
      "loss": 0.3308,
      "step": 59200
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.062517413885641,
      "learning_rate": 8.044023436938059e-07,
      "loss": 0.3216,
      "step": 59205
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.753771273018451,
      "learning_rate": 8.037328023710284e-07,
      "loss": 0.3563,
      "step": 59210
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.263291098167929,
      "learning_rate": 8.030635154555693e-07,
      "loss": 0.3145,
      "step": 59215
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.9372479836477265,
      "learning_rate": 8.02394482988007e-07,
      "loss": 0.2997,
      "step": 59220
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.818830632240378,
      "learning_rate": 8.017257050089039e-07,
      "loss": 0.3232,
      "step": 59225
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.7747768088828217,
      "learning_rate": 8.010571815588037e-07,
      "loss": 0.301,
      "step": 59230
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.660699962398027,
      "learning_rate": 8.003889126782377e-07,
      "loss": 0.334,
      "step": 59235
    },
    {
      "epoch": 2.51,
      "grad_norm": 6.282923676286064,
      "learning_rate": 7.997208984077232e-07,
      "loss": 0.3383,
      "step": 59240
    },
    {
      "epoch": 2.51,
      "grad_norm": 5.386209908148228,
      "learning_rate": 7.99053138787757e-07,
      "loss": 0.3411,
      "step": 59245
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.016248386394783,
      "learning_rate": 7.983856338588247e-07,
      "loss": 0.3381,
      "step": 59250
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.515148939304655,
      "learning_rate": 7.97718383661395e-07,
      "loss": 0.3502,
      "step": 59255
    },
    {
      "epoch": 2.51,
      "grad_norm": 5.41189848342285,
      "learning_rate": 7.970513882359233e-07,
      "loss": 0.3668,
      "step": 59260
    },
    {
      "epoch": 2.51,
      "grad_norm": 5.628244596276695,
      "learning_rate": 7.963846476228438e-07,
      "loss": 0.3408,
      "step": 59265
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.027123714180529,
      "learning_rate": 7.957181618625815e-07,
      "loss": 0.3279,
      "step": 59270
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.746128053515588,
      "learning_rate": 7.95051930995544e-07,
      "loss": 0.3133,
      "step": 59275
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.203785334787506,
      "learning_rate": 7.943859550621213e-07,
      "loss": 0.3359,
      "step": 59280
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.8900463386404294,
      "learning_rate": 7.937202341026896e-07,
      "loss": 0.3364,
      "step": 59285
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.209176673434223,
      "learning_rate": 7.930547681576095e-07,
      "loss": 0.3116,
      "step": 59290
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.538853158616472,
      "learning_rate": 7.923895572672286e-07,
      "loss": 0.3323,
      "step": 59295
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.817678060019137,
      "learning_rate": 7.917246014718738e-07,
      "loss": 0.3158,
      "step": 59300
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.141449220825571,
      "learning_rate": 7.910599008118603e-07,
      "loss": 0.3247,
      "step": 59305
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.601890778492797,
      "learning_rate": 7.903954553274878e-07,
      "loss": 0.3108,
      "step": 59310
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.32745497788374,
      "learning_rate": 7.897312650590383e-07,
      "loss": 0.3374,
      "step": 59315
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.530612648154769,
      "learning_rate": 7.890673300467805e-07,
      "loss": 0.321,
      "step": 59320
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.654232621100237,
      "learning_rate": 7.884036503309673e-07,
      "loss": 0.325,
      "step": 59325
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.604454785358741,
      "learning_rate": 7.877402259518357e-07,
      "loss": 0.3025,
      "step": 59330
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.462611038736459,
      "learning_rate": 7.87077056949605e-07,
      "loss": 0.3387,
      "step": 59335
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.6975954169822334,
      "learning_rate": 7.864141433644834e-07,
      "loss": 0.3188,
      "step": 59340
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.85150669142777,
      "learning_rate": 7.857514852366621e-07,
      "loss": 0.3292,
      "step": 59345
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.1904791124107,
      "learning_rate": 7.850890826063129e-07,
      "loss": 0.3145,
      "step": 59350
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.862036027955601,
      "learning_rate": 7.844269355135975e-07,
      "loss": 0.3405,
      "step": 59355
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.793644227102505,
      "learning_rate": 7.83765043998661e-07,
      "loss": 0.3191,
      "step": 59360
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.654141634392695,
      "learning_rate": 7.831034081016292e-07,
      "loss": 0.3233,
      "step": 59365
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.228846983712715,
      "learning_rate": 7.824420278626166e-07,
      "loss": 0.3392,
      "step": 59370
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.885658199814358,
      "learning_rate": 7.817809033217216e-07,
      "loss": 0.3366,
      "step": 59375
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.7403193210688674,
      "learning_rate": 7.811200345190245e-07,
      "loss": 0.3211,
      "step": 59380
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.06335013644027,
      "learning_rate": 7.804594214945921e-07,
      "loss": 0.3225,
      "step": 59385
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.97689899189851,
      "learning_rate": 7.797990642884762e-07,
      "loss": 0.3485,
      "step": 59390
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.7404999482310384,
      "learning_rate": 7.791389629407131e-07,
      "loss": 0.3408,
      "step": 59395
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.9631399690064155,
      "learning_rate": 7.784791174913209e-07,
      "loss": 0.3429,
      "step": 59400
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.089356190587559,
      "learning_rate": 7.778195279803041e-07,
      "loss": 0.3537,
      "step": 59405
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.9490251066258133,
      "learning_rate": 7.771601944476542e-07,
      "loss": 0.3266,
      "step": 59410
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.7236914292901915,
      "learning_rate": 7.765011169333419e-07,
      "loss": 0.3049,
      "step": 59415
    },
    {
      "epoch": 2.51,
      "grad_norm": 4.017895535717673,
      "learning_rate": 7.758422954773259e-07,
      "loss": 0.3476,
      "step": 59420
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.46868801887529,
      "learning_rate": 7.751837301195503e-07,
      "loss": 0.3081,
      "step": 59425
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.9508890663922847,
      "learning_rate": 7.745254208999392e-07,
      "loss": 0.3129,
      "step": 59430
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.9121141955990617,
      "learning_rate": 7.738673678584052e-07,
      "loss": 0.3261,
      "step": 59435
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.10480361176652,
      "learning_rate": 7.732095710348453e-07,
      "loss": 0.3107,
      "step": 59440
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.762556270795881,
      "learning_rate": 7.725520304691386e-07,
      "loss": 0.3226,
      "step": 59445
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.8332173670209815,
      "learning_rate": 7.718947462011489e-07,
      "loss": 0.3469,
      "step": 59450
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.2459475179733985,
      "learning_rate": 7.712377182707265e-07,
      "loss": 0.3491,
      "step": 59455
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.81613155242911,
      "learning_rate": 7.70580946717705e-07,
      "loss": 0.3293,
      "step": 59460
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.98676709773683,
      "learning_rate": 7.699244315819037e-07,
      "loss": 0.331,
      "step": 59465
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.9048908827302204,
      "learning_rate": 7.692681729031232e-07,
      "loss": 0.3317,
      "step": 59470
    },
    {
      "epoch": 2.52,
      "grad_norm": 6.000047929831132,
      "learning_rate": 7.686121707211508e-07,
      "loss": 0.3131,
      "step": 59475
    },
    {
      "epoch": 2.52,
      "grad_norm": 8.068356453698133,
      "learning_rate": 7.679564250757598e-07,
      "loss": 0.3238,
      "step": 59480
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.566372471123929,
      "learning_rate": 7.673009360067041e-07,
      "loss": 0.3403,
      "step": 59485
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.837527413814982,
      "learning_rate": 7.666457035537245e-07,
      "loss": 0.3701,
      "step": 59490
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.274115320114632,
      "learning_rate": 7.65990727756547e-07,
      "loss": 0.3202,
      "step": 59495
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.737252534009414,
      "learning_rate": 7.653360086548795e-07,
      "loss": 0.3079,
      "step": 59500
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.943777546452165,
      "learning_rate": 7.646815462884155e-07,
      "loss": 0.3459,
      "step": 59505
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.9796797568370885,
      "learning_rate": 7.640273406968346e-07,
      "loss": 0.3401,
      "step": 59510
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.933766080622509,
      "learning_rate": 7.633733919197989e-07,
      "loss": 0.3084,
      "step": 59515
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.7959885694883204,
      "learning_rate": 7.627196999969544e-07,
      "loss": 0.3202,
      "step": 59520
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.7109861952781817,
      "learning_rate": 7.620662649679333e-07,
      "loss": 0.3464,
      "step": 59525
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.971425611942858,
      "learning_rate": 7.614130868723518e-07,
      "loss": 0.3172,
      "step": 59530
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.494658310844765,
      "learning_rate": 7.607601657498092e-07,
      "loss": 0.3266,
      "step": 59535
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.800318541240934,
      "learning_rate": 7.601075016398901e-07,
      "loss": 0.3114,
      "step": 59540
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.143851173815291,
      "learning_rate": 7.59455094582166e-07,
      "loss": 0.3295,
      "step": 59545
    },
    {
      "epoch": 2.52,
      "grad_norm": 5.356375795086613,
      "learning_rate": 7.588029446161871e-07,
      "loss": 0.311,
      "step": 59550
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.886496575388588,
      "learning_rate": 7.581510517814927e-07,
      "loss": 0.3316,
      "step": 59555
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.021070934546926,
      "learning_rate": 7.574994161176053e-07,
      "loss": 0.3194,
      "step": 59560
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.043773580526641,
      "learning_rate": 7.568480376640325e-07,
      "loss": 0.3177,
      "step": 59565
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.9416867631654218,
      "learning_rate": 7.561969164602639e-07,
      "loss": 0.3182,
      "step": 59570
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.8505498081259755,
      "learning_rate": 7.555460525457753e-07,
      "loss": 0.3064,
      "step": 59575
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.29734800243156,
      "learning_rate": 7.548954459600288e-07,
      "loss": 0.3386,
      "step": 59580
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.305180239732112,
      "learning_rate": 7.542450967424653e-07,
      "loss": 0.3274,
      "step": 59585
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.489416697282077,
      "learning_rate": 7.535950049325153e-07,
      "loss": 0.3702,
      "step": 59590
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.838327834158462,
      "learning_rate": 7.529451705695934e-07,
      "loss": 0.3032,
      "step": 59595
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.074615635275481,
      "learning_rate": 7.522955936930959e-07,
      "loss": 0.3484,
      "step": 59600
    },
    {
      "epoch": 2.52,
      "grad_norm": 5.506027377207432,
      "learning_rate": 7.51646274342403e-07,
      "loss": 0.3056,
      "step": 59605
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.5953751044098814,
      "learning_rate": 7.509972125568826e-07,
      "loss": 0.3325,
      "step": 59610
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.8354525936069988,
      "learning_rate": 7.503484083758866e-07,
      "loss": 0.311,
      "step": 59615
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.277667863256289,
      "learning_rate": 7.49699861838748e-07,
      "loss": 0.3285,
      "step": 59620
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.686741883579355,
      "learning_rate": 7.490515729847864e-07,
      "loss": 0.3247,
      "step": 59625
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.878912618454519,
      "learning_rate": 7.48403541853307e-07,
      "loss": 0.3078,
      "step": 59630
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.894726990131706,
      "learning_rate": 7.477557684835979e-07,
      "loss": 0.3269,
      "step": 59635
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.6242062664600985,
      "learning_rate": 7.471082529149304e-07,
      "loss": 0.3132,
      "step": 59640
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.260740747796143,
      "learning_rate": 7.46460995186562e-07,
      "loss": 0.3068,
      "step": 59645
    },
    {
      "epoch": 2.52,
      "grad_norm": 3.8246068797947563,
      "learning_rate": 7.458139953377358e-07,
      "loss": 0.3363,
      "step": 59650
    },
    {
      "epoch": 2.52,
      "grad_norm": 9.066831689625904,
      "learning_rate": 7.451672534076748e-07,
      "loss": 0.3205,
      "step": 59655
    },
    {
      "epoch": 2.52,
      "grad_norm": 4.257545554073458,
      "learning_rate": 7.4452076943559e-07,
      "loss": 0.302,
      "step": 59660
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.939807365835118,
      "learning_rate": 7.43874543460677e-07,
      "loss": 0.3426,
      "step": 59665
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.123730007789098,
      "learning_rate": 7.432285755221125e-07,
      "loss": 0.3031,
      "step": 59670
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.211775339917534,
      "learning_rate": 7.425828656590611e-07,
      "loss": 0.3058,
      "step": 59675
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.220743859827115,
      "learning_rate": 7.419374139106706e-07,
      "loss": 0.3407,
      "step": 59680
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.9959640779286496,
      "learning_rate": 7.41292220316071e-07,
      "loss": 0.3233,
      "step": 59685
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.572689592631259,
      "learning_rate": 7.406472849143797e-07,
      "loss": 0.3314,
      "step": 59690
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.9873545757146704,
      "learning_rate": 7.400026077446975e-07,
      "loss": 0.3104,
      "step": 59695
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.7262835903178937,
      "learning_rate": 7.393581888461099e-07,
      "loss": 0.3364,
      "step": 59700
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.159143267379357,
      "learning_rate": 7.387140282576843e-07,
      "loss": 0.323,
      "step": 59705
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.610566803796951,
      "learning_rate": 7.380701260184748e-07,
      "loss": 0.3545,
      "step": 59710
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.089682756389737,
      "learning_rate": 7.374264821675208e-07,
      "loss": 0.3343,
      "step": 59715
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.487867920004437,
      "learning_rate": 7.367830967438427e-07,
      "loss": 0.3039,
      "step": 59720
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.2051118155817875,
      "learning_rate": 7.361399697864474e-07,
      "loss": 0.3081,
      "step": 59725
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.346930805827001,
      "learning_rate": 7.35497101334327e-07,
      "loss": 0.3387,
      "step": 59730
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.009230476535528,
      "learning_rate": 7.34854491426455e-07,
      "loss": 0.3386,
      "step": 59735
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.609355700472187,
      "learning_rate": 7.342121401017915e-07,
      "loss": 0.2886,
      "step": 59740
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.475290506412041,
      "learning_rate": 7.335700473992808e-07,
      "loss": 0.3414,
      "step": 59745
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.89405597160021,
      "learning_rate": 7.329282133578524e-07,
      "loss": 0.3209,
      "step": 59750
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.59622157491749,
      "learning_rate": 7.322866380164157e-07,
      "loss": 0.3181,
      "step": 59755
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.4941735148486455,
      "learning_rate": 7.316453214138697e-07,
      "loss": 0.3039,
      "step": 59760
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.8936420126764717,
      "learning_rate": 7.310042635890957e-07,
      "loss": 0.3336,
      "step": 59765
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.6232250331611535,
      "learning_rate": 7.303634645809588e-07,
      "loss": 0.3172,
      "step": 59770
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.153706896362985,
      "learning_rate": 7.297229244283072e-07,
      "loss": 0.353,
      "step": 59775
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.603243054085999,
      "learning_rate": 7.29082643169976e-07,
      "loss": 0.3125,
      "step": 59780
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.910458029321549,
      "learning_rate": 7.284426208447848e-07,
      "loss": 0.3393,
      "step": 59785
    },
    {
      "epoch": 2.53,
      "grad_norm": 5.738434661618399,
      "learning_rate": 7.278028574915341e-07,
      "loss": 0.3431,
      "step": 59790
    },
    {
      "epoch": 2.53,
      "grad_norm": 6.37977270422054,
      "learning_rate": 7.271633531490124e-07,
      "loss": 0.3273,
      "step": 59795
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.924155628682069,
      "learning_rate": 7.265241078559915e-07,
      "loss": 0.3412,
      "step": 59800
    },
    {
      "epoch": 2.53,
      "grad_norm": 5.146740524690065,
      "learning_rate": 7.258851216512247e-07,
      "loss": 0.3151,
      "step": 59805
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.765253508446272,
      "learning_rate": 7.252463945734529e-07,
      "loss": 0.3619,
      "step": 59810
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.5618826053339374,
      "learning_rate": 7.246079266614009e-07,
      "loss": 0.3434,
      "step": 59815
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.8401029248856,
      "learning_rate": 7.23969717953778e-07,
      "loss": 0.3263,
      "step": 59820
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.9385262366393894,
      "learning_rate": 7.233317684892744e-07,
      "loss": 0.3128,
      "step": 59825
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.4044455110737815,
      "learning_rate": 7.226940783065678e-07,
      "loss": 0.3334,
      "step": 59830
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.447380879835691,
      "learning_rate": 7.220566474443219e-07,
      "loss": 0.333,
      "step": 59835
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.8620156284140963,
      "learning_rate": 7.214194759411786e-07,
      "loss": 0.3078,
      "step": 59840
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.2676883677847455,
      "learning_rate": 7.207825638357702e-07,
      "loss": 0.344,
      "step": 59845
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.540381203139076,
      "learning_rate": 7.201459111667108e-07,
      "loss": 0.3354,
      "step": 59850
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.203504785503325,
      "learning_rate": 7.19509517972597e-07,
      "loss": 0.3445,
      "step": 59855
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.8016333589226843,
      "learning_rate": 7.188733842920126e-07,
      "loss": 0.3269,
      "step": 59860
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.769434168850976,
      "learning_rate": 7.182375101635241e-07,
      "loss": 0.3251,
      "step": 59865
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.669461653292791,
      "learning_rate": 7.176018956256847e-07,
      "loss": 0.3286,
      "step": 59870
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.020252337155452,
      "learning_rate": 7.169665407170268e-07,
      "loss": 0.3549,
      "step": 59875
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.166632616432261,
      "learning_rate": 7.163314454760711e-07,
      "loss": 0.309,
      "step": 59880
    },
    {
      "epoch": 2.53,
      "grad_norm": 5.002179838041498,
      "learning_rate": 7.156966099413231e-07,
      "loss": 0.3126,
      "step": 59885
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.9766786984249065,
      "learning_rate": 7.150620341512693e-07,
      "loss": 0.3307,
      "step": 59890
    },
    {
      "epoch": 2.53,
      "grad_norm": 4.108463688772603,
      "learning_rate": 7.14427718144382e-07,
      "loss": 0.3342,
      "step": 59895
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.899877322435653,
      "learning_rate": 7.137936619591201e-07,
      "loss": 0.3166,
      "step": 59900
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.124973819499981,
      "learning_rate": 7.131598656339217e-07,
      "loss": 0.3171,
      "step": 59905
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.450194601809639,
      "learning_rate": 7.125263292072132e-07,
      "loss": 0.3271,
      "step": 59910
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.754796725347709,
      "learning_rate": 7.118930527174056e-07,
      "loss": 0.3003,
      "step": 59915
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.481548791741152,
      "learning_rate": 7.112600362028904e-07,
      "loss": 0.3151,
      "step": 59920
    },
    {
      "epoch": 2.54,
      "grad_norm": 5.153841670059852,
      "learning_rate": 7.106272797020469e-07,
      "loss": 0.3541,
      "step": 59925
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.361985739266365,
      "learning_rate": 7.099947832532355e-07,
      "loss": 0.3356,
      "step": 59930
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.5394792780967017,
      "learning_rate": 7.093625468948051e-07,
      "loss": 0.3181,
      "step": 59935
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.8176454656652252,
      "learning_rate": 7.087305706650837e-07,
      "loss": 0.3101,
      "step": 59940
    },
    {
      "epoch": 2.54,
      "grad_norm": 6.190022450319625,
      "learning_rate": 7.080988546023876e-07,
      "loss": 0.3098,
      "step": 59945
    },
    {
      "epoch": 2.54,
      "grad_norm": 5.484092122461223,
      "learning_rate": 7.074673987450159e-07,
      "loss": 0.3036,
      "step": 59950
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.0591760240222134,
      "learning_rate": 7.068362031312526e-07,
      "loss": 0.2872,
      "step": 59955
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.7822418378374296,
      "learning_rate": 7.062052677993636e-07,
      "loss": 0.3316,
      "step": 59960
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.430190769410036,
      "learning_rate": 7.055745927876012e-07,
      "loss": 0.3436,
      "step": 59965
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.29267852134061,
      "learning_rate": 7.049441781342026e-07,
      "loss": 0.3219,
      "step": 59970
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.7735698878523243,
      "learning_rate": 7.04314023877386e-07,
      "loss": 0.3183,
      "step": 59975
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.7169112206559247,
      "learning_rate": 7.036841300553565e-07,
      "loss": 0.3088,
      "step": 59980
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.381551916456472,
      "learning_rate": 7.030544967063047e-07,
      "loss": 0.334,
      "step": 59985
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.5381773910357204,
      "learning_rate": 7.024251238683999e-07,
      "loss": 0.3288,
      "step": 59990
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.044473688624339,
      "learning_rate": 7.017960115798012e-07,
      "loss": 0.2832,
      "step": 59995
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.7722178718590356,
      "learning_rate": 7.011671598786496e-07,
      "loss": 0.3438,
      "step": 60000
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.8971499955084368,
      "learning_rate": 7.005385688030713e-07,
      "loss": 0.3268,
      "step": 60005
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.225998919441519,
      "learning_rate": 6.999102383911744e-07,
      "loss": 0.3316,
      "step": 60010
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.9899434860295386,
      "learning_rate": 6.992821686810536e-07,
      "loss": 0.3079,
      "step": 60015
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.964539794338357,
      "learning_rate": 6.986543597107875e-07,
      "loss": 0.2986,
      "step": 60020
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.230036763899095,
      "learning_rate": 6.980268115184363e-07,
      "loss": 0.3292,
      "step": 60025
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.9557055495543554,
      "learning_rate": 6.973995241420478e-07,
      "loss": 0.3359,
      "step": 60030
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.8611554006504347,
      "learning_rate": 6.967724976196532e-07,
      "loss": 0.302,
      "step": 60035
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.4551201427357174,
      "learning_rate": 6.961457319892656e-07,
      "loss": 0.3282,
      "step": 60040
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.686132249473026,
      "learning_rate": 6.95519227288885e-07,
      "loss": 0.328,
      "step": 60045
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.498405263329421,
      "learning_rate": 6.948929835564944e-07,
      "loss": 0.3385,
      "step": 60050
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.323771158421139,
      "learning_rate": 6.942670008300622e-07,
      "loss": 0.3295,
      "step": 60055
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.985424843898281,
      "learning_rate": 6.936412791475377e-07,
      "loss": 0.3095,
      "step": 60060
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.917734481631152,
      "learning_rate": 6.930158185468578e-07,
      "loss": 0.3202,
      "step": 60065
    },
    {
      "epoch": 2.54,
      "grad_norm": 5.261622182783867,
      "learning_rate": 6.92390619065943e-07,
      "loss": 0.3162,
      "step": 60070
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.7118307254708016,
      "learning_rate": 6.917656807426953e-07,
      "loss": 0.3317,
      "step": 60075
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.819649872673034,
      "learning_rate": 6.911410036150041e-07,
      "loss": 0.3355,
      "step": 60080
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.092356305981138,
      "learning_rate": 6.905165877207426e-07,
      "loss": 0.3375,
      "step": 60085
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.200441489623247,
      "learning_rate": 6.898924330977669e-07,
      "loss": 0.337,
      "step": 60090
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.7366249124796984,
      "learning_rate": 6.892685397839161e-07,
      "loss": 0.3423,
      "step": 60095
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.697677287923659,
      "learning_rate": 6.886449078170154e-07,
      "loss": 0.3209,
      "step": 60100
    },
    {
      "epoch": 2.54,
      "grad_norm": 5.341719427270338,
      "learning_rate": 6.880215372348759e-07,
      "loss": 0.3078,
      "step": 60105
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.178229741452269,
      "learning_rate": 6.873984280752882e-07,
      "loss": 0.339,
      "step": 60110
    },
    {
      "epoch": 2.54,
      "grad_norm": 4.1482601100558245,
      "learning_rate": 6.867755803760306e-07,
      "loss": 0.354,
      "step": 60115
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.5744507712474616,
      "learning_rate": 6.861529941748646e-07,
      "loss": 0.3025,
      "step": 60120
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.913959936649277,
      "learning_rate": 6.855306695095365e-07,
      "loss": 0.3315,
      "step": 60125
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.754446097277773,
      "learning_rate": 6.849086064177745e-07,
      "loss": 0.3268,
      "step": 60130
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.8403367197732172,
      "learning_rate": 6.84286804937293e-07,
      "loss": 0.2969,
      "step": 60135
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.014147750276571,
      "learning_rate": 6.836652651057912e-07,
      "loss": 0.33,
      "step": 60140
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.0223688230333075,
      "learning_rate": 6.83043986960949e-07,
      "loss": 0.3282,
      "step": 60145
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.8905603103908595,
      "learning_rate": 6.824229705404339e-07,
      "loss": 0.3137,
      "step": 60150
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.463797114164437,
      "learning_rate": 6.818022158818976e-07,
      "loss": 0.3246,
      "step": 60155
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.700214886931591,
      "learning_rate": 6.811817230229723e-07,
      "loss": 0.3259,
      "step": 60160
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.245794609574286,
      "learning_rate": 6.805614920012771e-07,
      "loss": 0.3266,
      "step": 60165
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.5247086041512965,
      "learning_rate": 6.799415228544159e-07,
      "loss": 0.334,
      "step": 60170
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.562493537765349,
      "learning_rate": 6.793218156199754e-07,
      "loss": 0.2912,
      "step": 60175
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.93124551508199,
      "learning_rate": 6.787023703355261e-07,
      "loss": 0.3128,
      "step": 60180
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.950683794164515,
      "learning_rate": 6.780831870386223e-07,
      "loss": 0.3315,
      "step": 60185
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.9486667646716604,
      "learning_rate": 6.774642657668063e-07,
      "loss": 0.3162,
      "step": 60190
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.7056213473545903,
      "learning_rate": 6.768456065575973e-07,
      "loss": 0.3262,
      "step": 60195
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.558391670058067,
      "learning_rate": 6.762272094485056e-07,
      "loss": 0.3475,
      "step": 60200
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.948417283770093,
      "learning_rate": 6.756090744770233e-07,
      "loss": 0.3098,
      "step": 60205
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.125841875489312,
      "learning_rate": 6.749912016806231e-07,
      "loss": 0.3268,
      "step": 60210
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.930962517424942,
      "learning_rate": 6.743735910967675e-07,
      "loss": 0.314,
      "step": 60215
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.176779330992543,
      "learning_rate": 6.737562427629002e-07,
      "loss": 0.3311,
      "step": 60220
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.205975105134719,
      "learning_rate": 6.731391567164475e-07,
      "loss": 0.3275,
      "step": 60225
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.6712143285179266,
      "learning_rate": 6.725223329948221e-07,
      "loss": 0.3345,
      "step": 60230
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.359628457642211,
      "learning_rate": 6.719057716354211e-07,
      "loss": 0.352,
      "step": 60235
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.7099754110848906,
      "learning_rate": 6.712894726756253e-07,
      "loss": 0.3243,
      "step": 60240
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.165017573973444,
      "learning_rate": 6.706734361527972e-07,
      "loss": 0.3329,
      "step": 60245
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.8265183801083773,
      "learning_rate": 6.70057662104287e-07,
      "loss": 0.3213,
      "step": 60250
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.133524904188468,
      "learning_rate": 6.694421505674259e-07,
      "loss": 0.3117,
      "step": 60255
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.001863664423481,
      "learning_rate": 6.688269015795318e-07,
      "loss": 0.3244,
      "step": 60260
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.103951122714365,
      "learning_rate": 6.682119151779037e-07,
      "loss": 0.3277,
      "step": 60265
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.3006487258510395,
      "learning_rate": 6.675971913998274e-07,
      "loss": 0.3192,
      "step": 60270
    },
    {
      "epoch": 2.55,
      "grad_norm": 5.069164023364594,
      "learning_rate": 6.669827302825732e-07,
      "loss": 0.3284,
      "step": 60275
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.106503874231339,
      "learning_rate": 6.663685318633917e-07,
      "loss": 0.2937,
      "step": 60280
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.4129892148103975,
      "learning_rate": 6.657545961795209e-07,
      "loss": 0.322,
      "step": 60285
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.835251287329939,
      "learning_rate": 6.651409232681832e-07,
      "loss": 0.3465,
      "step": 60290
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.719365725411388,
      "learning_rate": 6.645275131665812e-07,
      "loss": 0.3107,
      "step": 60295
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.291617929473293,
      "learning_rate": 6.639143659119057e-07,
      "loss": 0.3175,
      "step": 60300
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.181532040920078,
      "learning_rate": 6.633014815413297e-07,
      "loss": 0.3311,
      "step": 60305
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.817873502640602,
      "learning_rate": 6.626888600920118e-07,
      "loss": 0.3037,
      "step": 60310
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.0174940148699445,
      "learning_rate": 6.620765016010916e-07,
      "loss": 0.3261,
      "step": 60315
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.466116156338231,
      "learning_rate": 6.614644061056947e-07,
      "loss": 0.3427,
      "step": 60320
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.102721518392326,
      "learning_rate": 6.608525736429327e-07,
      "loss": 0.3137,
      "step": 60325
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.212619003126551,
      "learning_rate": 6.602410042498969e-07,
      "loss": 0.3274,
      "step": 60330
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.404479645135486,
      "learning_rate": 6.596296979636657e-07,
      "loss": 0.3209,
      "step": 60335
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.9788117396100326,
      "learning_rate": 6.590186548213024e-07,
      "loss": 0.3384,
      "step": 60340
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.192032664025439,
      "learning_rate": 6.584078748598494e-07,
      "loss": 0.3294,
      "step": 60345
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.8187910469163495,
      "learning_rate": 6.577973581163388e-07,
      "loss": 0.317,
      "step": 60350
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.799317871144328,
      "learning_rate": 6.571871046277839e-07,
      "loss": 0.312,
      "step": 60355
    },
    {
      "epoch": 2.55,
      "grad_norm": 4.389165936300695,
      "learning_rate": 6.565771144311839e-07,
      "loss": 0.3158,
      "step": 60360
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.9038180814624943,
      "learning_rate": 6.559673875635181e-07,
      "loss": 0.3445,
      "step": 60365
    },
    {
      "epoch": 2.55,
      "grad_norm": 3.9438443292973284,
      "learning_rate": 6.553579240617542e-07,
      "loss": 0.3311,
      "step": 60370
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.866632270946839,
      "learning_rate": 6.547487239628425e-07,
      "loss": 0.3479,
      "step": 60375
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.571714511295955,
      "learning_rate": 6.541397873037159e-07,
      "loss": 0.3459,
      "step": 60380
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.365948555736455,
      "learning_rate": 6.535311141212924e-07,
      "loss": 0.3493,
      "step": 60385
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.4014852589521976,
      "learning_rate": 6.529227044524755e-07,
      "loss": 0.3168,
      "step": 60390
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.8956098317614547,
      "learning_rate": 6.52314558334149e-07,
      "loss": 0.3168,
      "step": 60395
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.285001775537623,
      "learning_rate": 6.517066758031848e-07,
      "loss": 0.2866,
      "step": 60400
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.948400570147473,
      "learning_rate": 6.510990568964365e-07,
      "loss": 0.3124,
      "step": 60405
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.69799336869235,
      "learning_rate": 6.504917016507439e-07,
      "loss": 0.3291,
      "step": 60410
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.823576171610237,
      "learning_rate": 6.49884610102926e-07,
      "loss": 0.3275,
      "step": 60415
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.18842494997157,
      "learning_rate": 6.4927778228979e-07,
      "loss": 0.3262,
      "step": 60420
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.564782326829167,
      "learning_rate": 6.48671218248127e-07,
      "loss": 0.3576,
      "step": 60425
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.835360849719144,
      "learning_rate": 6.480649180147114e-07,
      "loss": 0.3103,
      "step": 60430
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.884120624893983,
      "learning_rate": 6.474588816263e-07,
      "loss": 0.3305,
      "step": 60435
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.367480096268666,
      "learning_rate": 6.468531091196362e-07,
      "loss": 0.3151,
      "step": 60440
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.441609670349472,
      "learning_rate": 6.462476005314461e-07,
      "loss": 0.3092,
      "step": 60445
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.6918592951755786,
      "learning_rate": 6.456423558984387e-07,
      "loss": 0.3177,
      "step": 60450
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.252760210918517,
      "learning_rate": 6.450373752573091e-07,
      "loss": 0.3366,
      "step": 60455
    },
    {
      "epoch": 2.56,
      "grad_norm": 5.002615904312581,
      "learning_rate": 6.444326586447369e-07,
      "loss": 0.3303,
      "step": 60460
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.951713687011071,
      "learning_rate": 6.438282060973816e-07,
      "loss": 0.3438,
      "step": 60465
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.045528159872285,
      "learning_rate": 6.432240176518906e-07,
      "loss": 0.3103,
      "step": 60470
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.911730912075422,
      "learning_rate": 6.42620093344894e-07,
      "loss": 0.3399,
      "step": 60475
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.246180504770331,
      "learning_rate": 6.420164332130069e-07,
      "loss": 0.3217,
      "step": 60480
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.059635217073218,
      "learning_rate": 6.414130372928263e-07,
      "loss": 0.3253,
      "step": 60485
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.8893273338119156,
      "learning_rate": 6.408099056209338e-07,
      "loss": 0.336,
      "step": 60490
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.3870252524615925,
      "learning_rate": 6.402070382338976e-07,
      "loss": 0.3187,
      "step": 60495
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.078630820728053,
      "learning_rate": 6.396044351682656e-07,
      "loss": 0.3179,
      "step": 60500
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.165571015583123,
      "learning_rate": 6.390020964605725e-07,
      "loss": 0.3171,
      "step": 60505
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.4324970650196125,
      "learning_rate": 6.38400022147338e-07,
      "loss": 0.3259,
      "step": 60510
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.219860905718792,
      "learning_rate": 6.377982122650611e-07,
      "loss": 0.3285,
      "step": 60515
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.8666632629621827,
      "learning_rate": 6.3719666685023e-07,
      "loss": 0.3179,
      "step": 60520
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.281638383729655,
      "learning_rate": 6.365953859393143e-07,
      "loss": 0.3438,
      "step": 60525
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.813085313500639,
      "learning_rate": 6.359943695687665e-07,
      "loss": 0.3076,
      "step": 60530
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.352900150053343,
      "learning_rate": 6.353936177750258e-07,
      "loss": 0.3131,
      "step": 60535
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.055747717232485,
      "learning_rate": 6.347931305945138e-07,
      "loss": 0.3121,
      "step": 60540
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.010891312602815,
      "learning_rate": 6.341929080636366e-07,
      "loss": 0.3227,
      "step": 60545
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.165844590153781,
      "learning_rate": 6.335929502187832e-07,
      "loss": 0.2912,
      "step": 60550
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.9258213587435296,
      "learning_rate": 6.32993257096327e-07,
      "loss": 0.3155,
      "step": 60555
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.7237915316879797,
      "learning_rate": 6.323938287326275e-07,
      "loss": 0.3274,
      "step": 60560
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.237922448911811,
      "learning_rate": 6.317946651640239e-07,
      "loss": 0.3225,
      "step": 60565
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.5828062616197167,
      "learning_rate": 6.311957664268437e-07,
      "loss": 0.3252,
      "step": 60570
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.9295260623137436,
      "learning_rate": 6.30597132557394e-07,
      "loss": 0.3348,
      "step": 60575
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.357972980430638,
      "learning_rate": 6.299987635919713e-07,
      "loss": 0.3287,
      "step": 60580
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.6546876633669587,
      "learning_rate": 6.2940065956685e-07,
      "loss": 0.3234,
      "step": 60585
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.733188457374905,
      "learning_rate": 6.288028205182928e-07,
      "loss": 0.332,
      "step": 60590
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.0485185225083775,
      "learning_rate": 6.282052464825461e-07,
      "loss": 0.3012,
      "step": 60595
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.410965373173383,
      "learning_rate": 6.276079374958361e-07,
      "loss": 0.3165,
      "step": 60600
    },
    {
      "epoch": 2.56,
      "grad_norm": 4.354633261072752,
      "learning_rate": 6.270108935943781e-07,
      "loss": 0.3163,
      "step": 60605
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.741932092281292,
      "learning_rate": 6.264141148143682e-07,
      "loss": 0.3221,
      "step": 60610
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.455577958032577,
      "learning_rate": 6.258176011919886e-07,
      "loss": 0.3111,
      "step": 60615
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.934174313430105,
      "learning_rate": 6.252213527634026e-07,
      "loss": 0.3213,
      "step": 60620
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.9296083249142515,
      "learning_rate": 6.246253695647597e-07,
      "loss": 0.3479,
      "step": 60625
    },
    {
      "epoch": 2.57,
      "grad_norm": 6.046134828368785,
      "learning_rate": 6.240296516321937e-07,
      "loss": 0.3295,
      "step": 60630
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.22772554274616,
      "learning_rate": 6.234341990018184e-07,
      "loss": 0.3084,
      "step": 60635
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.6969121250247663,
      "learning_rate": 6.228390117097366e-07,
      "loss": 0.3001,
      "step": 60640
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.476640043833009,
      "learning_rate": 6.222440897920329e-07,
      "loss": 0.3175,
      "step": 60645
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.711755542255426,
      "learning_rate": 6.216494332847739e-07,
      "loss": 0.2996,
      "step": 60650
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.850102395876614,
      "learning_rate": 6.21055042224013e-07,
      "loss": 0.315,
      "step": 60655
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.7612764079786003,
      "learning_rate": 6.204609166457864e-07,
      "loss": 0.328,
      "step": 60660
    },
    {
      "epoch": 2.57,
      "grad_norm": 5.0275154471976835,
      "learning_rate": 6.198670565861148e-07,
      "loss": 0.3201,
      "step": 60665
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.7683288682066465,
      "learning_rate": 6.192734620810009e-07,
      "loss": 0.3475,
      "step": 60670
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.140457978702257,
      "learning_rate": 6.186801331664327e-07,
      "loss": 0.3212,
      "step": 60675
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.090453073693936,
      "learning_rate": 6.180870698783836e-07,
      "loss": 0.341,
      "step": 60680
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.9042863321041086,
      "learning_rate": 6.174942722528071e-07,
      "loss": 0.3379,
      "step": 60685
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.7792951036320535,
      "learning_rate": 6.169017403256439e-07,
      "loss": 0.3221,
      "step": 60690
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.941908723011603,
      "learning_rate": 6.163094741328179e-07,
      "loss": 0.3131,
      "step": 60695
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.64483322839673,
      "learning_rate": 6.157174737102356e-07,
      "loss": 0.3187,
      "step": 60700
    },
    {
      "epoch": 2.57,
      "grad_norm": 5.310391032095249,
      "learning_rate": 6.151257390937882e-07,
      "loss": 0.3206,
      "step": 60705
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.8853871469043915,
      "learning_rate": 6.145342703193507e-07,
      "loss": 0.3115,
      "step": 60710
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.950711456034995,
      "learning_rate": 6.139430674227842e-07,
      "loss": 0.3108,
      "step": 60715
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.3138678705462388,
      "learning_rate": 6.133521304399287e-07,
      "loss": 0.3127,
      "step": 60720
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.442805937571479,
      "learning_rate": 6.127614594066123e-07,
      "loss": 0.3181,
      "step": 60725
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.26935102954461,
      "learning_rate": 6.121710543586467e-07,
      "loss": 0.3206,
      "step": 60730
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.016064853362011,
      "learning_rate": 6.115809153318253e-07,
      "loss": 0.3367,
      "step": 60735
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.428061863475383,
      "learning_rate": 6.109910423619253e-07,
      "loss": 0.3186,
      "step": 60740
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.70480221007726,
      "learning_rate": 6.104014354847104e-07,
      "loss": 0.3555,
      "step": 60745
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.213475491497634,
      "learning_rate": 6.098120947359271e-07,
      "loss": 0.328,
      "step": 60750
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.7294433310255055,
      "learning_rate": 6.092230201513039e-07,
      "loss": 0.3101,
      "step": 60755
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.584011838725251,
      "learning_rate": 6.086342117665556e-07,
      "loss": 0.2997,
      "step": 60760
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.464617787179817,
      "learning_rate": 6.080456696173808e-07,
      "loss": 0.3495,
      "step": 60765
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.772956784899947,
      "learning_rate": 6.074573937394596e-07,
      "loss": 0.3072,
      "step": 60770
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.500191507016388,
      "learning_rate": 6.068693841684576e-07,
      "loss": 0.3147,
      "step": 60775
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.8066797977423907,
      "learning_rate": 6.062816409400241e-07,
      "loss": 0.3245,
      "step": 60780
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.474648180608742,
      "learning_rate": 6.056941640897939e-07,
      "loss": 0.3099,
      "step": 60785
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.113143690635657,
      "learning_rate": 6.051069536533816e-07,
      "loss": 0.3279,
      "step": 60790
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.7988892753298784,
      "learning_rate": 6.045200096663895e-07,
      "loss": 0.3208,
      "step": 60795
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.447266577623724,
      "learning_rate": 6.039333321644026e-07,
      "loss": 0.3162,
      "step": 60800
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.6227665853562665,
      "learning_rate": 6.033469211829873e-07,
      "loss": 0.3274,
      "step": 60805
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.127141914998595,
      "learning_rate": 6.027607767576982e-07,
      "loss": 0.3228,
      "step": 60810
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.7711423883518465,
      "learning_rate": 6.021748989240712e-07,
      "loss": 0.2997,
      "step": 60815
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.899470346777691,
      "learning_rate": 6.015892877176249e-07,
      "loss": 0.3158,
      "step": 60820
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.865217404064876,
      "learning_rate": 6.01003943173864e-07,
      "loss": 0.2958,
      "step": 60825
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.0068408926721775,
      "learning_rate": 6.004188653282777e-07,
      "loss": 0.3161,
      "step": 60830
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.8812480211843075,
      "learning_rate": 5.99834054216335e-07,
      "loss": 0.3208,
      "step": 60835
    },
    {
      "epoch": 2.57,
      "grad_norm": 3.652016014701967,
      "learning_rate": 5.992495098734929e-07,
      "loss": 0.314,
      "step": 60840
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.9303718850930034,
      "learning_rate": 5.9866523233519e-07,
      "loss": 0.339,
      "step": 60845
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.0378803941598616,
      "learning_rate": 5.980812216368503e-07,
      "loss": 0.3184,
      "step": 60850
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.442180802541784,
      "learning_rate": 5.974974778138787e-07,
      "loss": 0.3321,
      "step": 60855
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.894805942569527,
      "learning_rate": 5.969140009016672e-07,
      "loss": 0.3189,
      "step": 60860
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.596451518289577,
      "learning_rate": 5.963307909355914e-07,
      "loss": 0.3171,
      "step": 60865
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.267617797822112,
      "learning_rate": 5.957478479510071e-07,
      "loss": 0.2916,
      "step": 60870
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.033338327596204,
      "learning_rate": 5.951651719832575e-07,
      "loss": 0.3218,
      "step": 60875
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.035040744182014,
      "learning_rate": 5.945827630676698e-07,
      "loss": 0.2873,
      "step": 60880
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.8642122514784707,
      "learning_rate": 5.940006212395511e-07,
      "loss": 0.3119,
      "step": 60885
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.9148160156933494,
      "learning_rate": 5.934187465341972e-07,
      "loss": 0.3213,
      "step": 60890
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.886421574667727,
      "learning_rate": 5.928371389868853e-07,
      "loss": 0.3073,
      "step": 60895
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.846833879232973,
      "learning_rate": 5.922557986328759e-07,
      "loss": 0.3076,
      "step": 60900
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.8847042948503945,
      "learning_rate": 5.916747255074129e-07,
      "loss": 0.3366,
      "step": 60905
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.039904694906965,
      "learning_rate": 5.910939196457266e-07,
      "loss": 0.3227,
      "step": 60910
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.9748669288165823,
      "learning_rate": 5.905133810830288e-07,
      "loss": 0.3139,
      "step": 60915
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.369120696143879,
      "learning_rate": 5.899331098545174e-07,
      "loss": 0.3174,
      "step": 60920
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.618535040952674,
      "learning_rate": 5.8935310599537e-07,
      "loss": 0.3271,
      "step": 60925
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.663677158971989,
      "learning_rate": 5.887733695407516e-07,
      "loss": 0.3272,
      "step": 60930
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.52064709775734,
      "learning_rate": 5.881939005258114e-07,
      "loss": 0.3122,
      "step": 60935
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.13976370701866,
      "learning_rate": 5.876146989856784e-07,
      "loss": 0.3281,
      "step": 60940
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.5769002043863187,
      "learning_rate": 5.870357649554692e-07,
      "loss": 0.275,
      "step": 60945
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.740256851770981,
      "learning_rate": 5.864570984702838e-07,
      "loss": 0.3206,
      "step": 60950
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.9770483037016437,
      "learning_rate": 5.858786995652033e-07,
      "loss": 0.3071,
      "step": 60955
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.7273865980644203,
      "learning_rate": 5.853005682752949e-07,
      "loss": 0.2977,
      "step": 60960
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.7275253649024123,
      "learning_rate": 5.847227046356085e-07,
      "loss": 0.3604,
      "step": 60965
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.6034041413820472,
      "learning_rate": 5.841451086811806e-07,
      "loss": 0.3144,
      "step": 60970
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.041026484457373,
      "learning_rate": 5.835677804470263e-07,
      "loss": 0.3005,
      "step": 60975
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.738534395948023,
      "learning_rate": 5.829907199681484e-07,
      "loss": 0.3207,
      "step": 60980
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.568046800380762,
      "learning_rate": 5.824139272795332e-07,
      "loss": 0.3177,
      "step": 60985
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.7897766213372606,
      "learning_rate": 5.818374024161483e-07,
      "loss": 0.3435,
      "step": 60990
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.639531932957799,
      "learning_rate": 5.812611454129474e-07,
      "loss": 0.3421,
      "step": 60995
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.028132244094959,
      "learning_rate": 5.806851563048688e-07,
      "loss": 0.3338,
      "step": 61000
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.895145479158649,
      "learning_rate": 5.801094351268305e-07,
      "loss": 0.28,
      "step": 61005
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.803586040843032,
      "learning_rate": 5.795339819137375e-07,
      "loss": 0.3066,
      "step": 61010
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.840450104568355,
      "learning_rate": 5.789587967004789e-07,
      "loss": 0.3223,
      "step": 61015
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.701161083602059,
      "learning_rate": 5.783838795219266e-07,
      "loss": 0.2987,
      "step": 61020
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.012175348374319,
      "learning_rate": 5.778092304129346e-07,
      "loss": 0.3263,
      "step": 61025
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.87569263241528,
      "learning_rate": 5.772348494083424e-07,
      "loss": 0.3188,
      "step": 61030
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.11302596015302,
      "learning_rate": 5.766607365429749e-07,
      "loss": 0.3448,
      "step": 61035
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.9631778006363074,
      "learning_rate": 5.760868918516372e-07,
      "loss": 0.3247,
      "step": 61040
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.127014182169294,
      "learning_rate": 5.755133153691194e-07,
      "loss": 0.3345,
      "step": 61045
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.150535034538587,
      "learning_rate": 5.749400071301975e-07,
      "loss": 0.3033,
      "step": 61050
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.028437573369199,
      "learning_rate": 5.743669671696279e-07,
      "loss": 0.3393,
      "step": 61055
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.039452774202473,
      "learning_rate": 5.737941955221538e-07,
      "loss": 0.3255,
      "step": 61060
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.7849163705972186,
      "learning_rate": 5.732216922224987e-07,
      "loss": 0.3053,
      "step": 61065
    },
    {
      "epoch": 2.58,
      "grad_norm": 4.222992448549588,
      "learning_rate": 5.726494573053742e-07,
      "loss": 0.3319,
      "step": 61070
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.9712264275483493,
      "learning_rate": 5.72077490805471e-07,
      "loss": 0.3053,
      "step": 61075
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.9590354745895087,
      "learning_rate": 5.715057927574663e-07,
      "loss": 0.325,
      "step": 61080
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.070422239819677,
      "learning_rate": 5.709343631960218e-07,
      "loss": 0.3587,
      "step": 61085
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.620214439414734,
      "learning_rate": 5.703632021557797e-07,
      "loss": 0.3028,
      "step": 61090
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.7393405551422014,
      "learning_rate": 5.697923096713687e-07,
      "loss": 0.346,
      "step": 61095
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.0526063955932825,
      "learning_rate": 5.692216857774008e-07,
      "loss": 0.3114,
      "step": 61100
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.360772158423643,
      "learning_rate": 5.686513305084712e-07,
      "loss": 0.3196,
      "step": 61105
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.842795953319503,
      "learning_rate": 5.680812438991578e-07,
      "loss": 0.3238,
      "step": 61110
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.547614277255096,
      "learning_rate": 5.675114259840236e-07,
      "loss": 0.3194,
      "step": 61115
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.8166903139690835,
      "learning_rate": 5.66941876797617e-07,
      "loss": 0.3417,
      "step": 61120
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.235668872700056,
      "learning_rate": 5.663725963744648e-07,
      "loss": 0.318,
      "step": 61125
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.087943928064198,
      "learning_rate": 5.658035847490828e-07,
      "loss": 0.3224,
      "step": 61130
    },
    {
      "epoch": 2.59,
      "grad_norm": 5.822427054717882,
      "learning_rate": 5.652348419559689e-07,
      "loss": 0.329,
      "step": 61135
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.8128348472130504,
      "learning_rate": 5.646663680296022e-07,
      "loss": 0.328,
      "step": 61140
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.6042068546496724,
      "learning_rate": 5.640981630044495e-07,
      "loss": 0.3261,
      "step": 61145
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.6707972949487186,
      "learning_rate": 5.635302269149584e-07,
      "loss": 0.3167,
      "step": 61150
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.930946828205894,
      "learning_rate": 5.629625597955623e-07,
      "loss": 0.3256,
      "step": 61155
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.183062901373014,
      "learning_rate": 5.623951616806761e-07,
      "loss": 0.3083,
      "step": 61160
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.1168953992130115,
      "learning_rate": 5.618280326046999e-07,
      "loss": 0.3468,
      "step": 61165
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.023259171251964,
      "learning_rate": 5.612611726020173e-07,
      "loss": 0.3206,
      "step": 61170
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.5965053580108663,
      "learning_rate": 5.606945817069947e-07,
      "loss": 0.3114,
      "step": 61175
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.490143774222751,
      "learning_rate": 5.601282599539831e-07,
      "loss": 0.3104,
      "step": 61180
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.8279880389206884,
      "learning_rate": 5.595622073773177e-07,
      "loss": 0.3237,
      "step": 61185
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.9198491567649505,
      "learning_rate": 5.589964240113155e-07,
      "loss": 0.3125,
      "step": 61190
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.0852472350888425,
      "learning_rate": 5.584309098902784e-07,
      "loss": 0.3136,
      "step": 61195
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.032288826492062,
      "learning_rate": 5.578656650484921e-07,
      "loss": 0.3324,
      "step": 61200
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.381239856800773,
      "learning_rate": 5.573006895202266e-07,
      "loss": 0.3302,
      "step": 61205
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.817670769410906,
      "learning_rate": 5.567359833397335e-07,
      "loss": 0.3391,
      "step": 61210
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.9761984129207595,
      "learning_rate": 5.561715465412493e-07,
      "loss": 0.3247,
      "step": 61215
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.157059663593086,
      "learning_rate": 5.556073791589955e-07,
      "loss": 0.3281,
      "step": 61220
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.689968599966529,
      "learning_rate": 5.550434812271749e-07,
      "loss": 0.3344,
      "step": 61225
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.435767451859413,
      "learning_rate": 5.544798527799739e-07,
      "loss": 0.3226,
      "step": 61230
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.088049484016113,
      "learning_rate": 5.539164938515646e-07,
      "loss": 0.3157,
      "step": 61235
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.9985024859747322,
      "learning_rate": 5.533534044761035e-07,
      "loss": 0.3096,
      "step": 61240
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.9110266732988332,
      "learning_rate": 5.527905846877257e-07,
      "loss": 0.3188,
      "step": 61245
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.5825818521903154,
      "learning_rate": 5.522280345205556e-07,
      "loss": 0.3104,
      "step": 61250
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.104909381589157,
      "learning_rate": 5.516657540086995e-07,
      "loss": 0.3325,
      "step": 61255
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.0815074274619505,
      "learning_rate": 5.511037431862443e-07,
      "loss": 0.3336,
      "step": 61260
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.020535704740354,
      "learning_rate": 5.505420020872648e-07,
      "loss": 0.3184,
      "step": 61265
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.162962251168534,
      "learning_rate": 5.499805307458173e-07,
      "loss": 0.3078,
      "step": 61270
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.794050945093777,
      "learning_rate": 5.494193291959432e-07,
      "loss": 0.3067,
      "step": 61275
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.455721477503858,
      "learning_rate": 5.488583974716644e-07,
      "loss": 0.3161,
      "step": 61280
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.33710986213934,
      "learning_rate": 5.482977356069902e-07,
      "loss": 0.3105,
      "step": 61285
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.4768296477460545,
      "learning_rate": 5.477373436359118e-07,
      "loss": 0.3034,
      "step": 61290
    },
    {
      "epoch": 2.59,
      "grad_norm": 5.5509595221800785,
      "learning_rate": 5.471772215924032e-07,
      "loss": 0.3076,
      "step": 61295
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.145174398713696,
      "learning_rate": 5.466173695104232e-07,
      "loss": 0.3173,
      "step": 61300
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.839010359210999,
      "learning_rate": 5.460577874239153e-07,
      "loss": 0.3008,
      "step": 61305
    },
    {
      "epoch": 2.59,
      "grad_norm": 4.3915965453445915,
      "learning_rate": 5.45498475366803e-07,
      "loss": 0.3067,
      "step": 61310
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.747774104315277,
      "learning_rate": 5.449394333729974e-07,
      "loss": 0.3275,
      "step": 61315
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.583037674297164,
      "learning_rate": 5.443806614763908e-07,
      "loss": 0.3529,
      "step": 61320
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.143469671257063,
      "learning_rate": 5.438221597108612e-07,
      "loss": 0.3181,
      "step": 61325
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.888963159957272,
      "learning_rate": 5.432639281102669e-07,
      "loss": 0.3103,
      "step": 61330
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.032944697459095,
      "learning_rate": 5.427059667084533e-07,
      "loss": 0.298,
      "step": 61335
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.917368241201488,
      "learning_rate": 5.421482755392482e-07,
      "loss": 0.3171,
      "step": 61340
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.219228617506832,
      "learning_rate": 5.415908546364612e-07,
      "loss": 0.3195,
      "step": 61345
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.730917153037885,
      "learning_rate": 5.410337040338882e-07,
      "loss": 0.316,
      "step": 61350
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.70030821660783,
      "learning_rate": 5.40476823765308e-07,
      "loss": 0.3196,
      "step": 61355
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.961794813221685,
      "learning_rate": 5.399202138644816e-07,
      "loss": 0.3134,
      "step": 61360
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.9567151687211712,
      "learning_rate": 5.393638743651547e-07,
      "loss": 0.2996,
      "step": 61365
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.7629407546595504,
      "learning_rate": 5.388078053010576e-07,
      "loss": 0.3001,
      "step": 61370
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.172999905615919,
      "learning_rate": 5.382520067059017e-07,
      "loss": 0.3212,
      "step": 61375
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.8332256006061374,
      "learning_rate": 5.376964786133854e-07,
      "loss": 0.3279,
      "step": 61380
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.065914314714507,
      "learning_rate": 5.371412210571858e-07,
      "loss": 0.2902,
      "step": 61385
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.871140565630022,
      "learning_rate": 5.365862340709698e-07,
      "loss": 0.3377,
      "step": 61390
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.214846082012037,
      "learning_rate": 5.360315176883818e-07,
      "loss": 0.3163,
      "step": 61395
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.610035993270218,
      "learning_rate": 5.354770719430535e-07,
      "loss": 0.3291,
      "step": 61400
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.056680274821778,
      "learning_rate": 5.349228968685999e-07,
      "loss": 0.307,
      "step": 61405
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.064196285707383,
      "learning_rate": 5.343689924986201e-07,
      "loss": 0.2844,
      "step": 61410
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.713762737796012,
      "learning_rate": 5.338153588666928e-07,
      "loss": 0.3033,
      "step": 61415
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.859532337199477,
      "learning_rate": 5.332619960063851e-07,
      "loss": 0.3321,
      "step": 61420
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.514041757146596,
      "learning_rate": 5.327089039512462e-07,
      "loss": 0.2973,
      "step": 61425
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.955936615061935,
      "learning_rate": 5.321560827348066e-07,
      "loss": 0.3107,
      "step": 61430
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.051360880785011,
      "learning_rate": 5.316035323905827e-07,
      "loss": 0.3334,
      "step": 61435
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.508834619724927,
      "learning_rate": 5.310512529520761e-07,
      "loss": 0.3136,
      "step": 61440
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.582404413204896,
      "learning_rate": 5.304992444527673e-07,
      "loss": 0.3258,
      "step": 61445
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.653254716716395,
      "learning_rate": 5.29947506926124e-07,
      "loss": 0.3258,
      "step": 61450
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.23980459570075,
      "learning_rate": 5.293960404055964e-07,
      "loss": 0.3507,
      "step": 61455
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.9653649188136666,
      "learning_rate": 5.288448449246192e-07,
      "loss": 0.3267,
      "step": 61460
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.439453074226438,
      "learning_rate": 5.282939205166082e-07,
      "loss": 0.34,
      "step": 61465
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.6593351934282534,
      "learning_rate": 5.277432672149646e-07,
      "loss": 0.2928,
      "step": 61470
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.229136717602049,
      "learning_rate": 5.271928850530738e-07,
      "loss": 0.3075,
      "step": 61475
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.011319211210055,
      "learning_rate": 5.266427740643032e-07,
      "loss": 0.3181,
      "step": 61480
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.9782550573193665,
      "learning_rate": 5.260929342820037e-07,
      "loss": 0.3391,
      "step": 61485
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.269395641288928,
      "learning_rate": 5.255433657395126e-07,
      "loss": 0.3107,
      "step": 61490
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.93890289764734,
      "learning_rate": 5.249940684701465e-07,
      "loss": 0.3218,
      "step": 61495
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.976579031612839,
      "learning_rate": 5.244450425072084e-07,
      "loss": 0.3276,
      "step": 61500
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.891678475478426,
      "learning_rate": 5.238962878839838e-07,
      "loss": 0.3103,
      "step": 61505
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.651118941853109,
      "learning_rate": 5.233478046337437e-07,
      "loss": 0.3057,
      "step": 61510
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.078771878913234,
      "learning_rate": 5.227995927897389e-07,
      "loss": 0.3119,
      "step": 61515
    },
    {
      "epoch": 2.6,
      "grad_norm": 5.719823506994609,
      "learning_rate": 5.222516523852067e-07,
      "loss": 0.3334,
      "step": 61520
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.029061818008321,
      "learning_rate": 5.217039834533682e-07,
      "loss": 0.3285,
      "step": 61525
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.251631040390349,
      "learning_rate": 5.211565860274249e-07,
      "loss": 0.3053,
      "step": 61530
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.617490820412938,
      "learning_rate": 5.206094601405647e-07,
      "loss": 0.3193,
      "step": 61535
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.123861243760195,
      "learning_rate": 5.2006260582596e-07,
      "loss": 0.3232,
      "step": 61540
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.258286951590188,
      "learning_rate": 5.195160231167628e-07,
      "loss": 0.2974,
      "step": 61545
    },
    {
      "epoch": 2.6,
      "grad_norm": 4.026935032155439,
      "learning_rate": 5.189697120461107e-07,
      "loss": 0.3156,
      "step": 61550
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.394538596533563,
      "learning_rate": 5.184236726471259e-07,
      "loss": 0.3025,
      "step": 61555
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.3138110842901805,
      "learning_rate": 5.178779049529137e-07,
      "loss": 0.3346,
      "step": 61560
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.082881525756429,
      "learning_rate": 5.173324089965603e-07,
      "loss": 0.3098,
      "step": 61565
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.091433780955253,
      "learning_rate": 5.167871848111395e-07,
      "loss": 0.3039,
      "step": 61570
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.332971502348048,
      "learning_rate": 5.16242232429705e-07,
      "loss": 0.3211,
      "step": 61575
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.853558979089911,
      "learning_rate": 5.156975518852986e-07,
      "loss": 0.3284,
      "step": 61580
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.002871294151511,
      "learning_rate": 5.151531432109391e-07,
      "loss": 0.3071,
      "step": 61585
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.306157195120202,
      "learning_rate": 5.146090064396342e-07,
      "loss": 0.3355,
      "step": 61590
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.278748564462806,
      "learning_rate": 5.140651416043735e-07,
      "loss": 0.3114,
      "step": 61595
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.538938550701009,
      "learning_rate": 5.13521548738129e-07,
      "loss": 0.3364,
      "step": 61600
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.704198544622109,
      "learning_rate": 5.129782278738576e-07,
      "loss": 0.3255,
      "step": 61605
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.2296578085479215,
      "learning_rate": 5.124351790445003e-07,
      "loss": 0.3224,
      "step": 61610
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.052205664895545,
      "learning_rate": 5.118924022829791e-07,
      "loss": 0.2997,
      "step": 61615
    },
    {
      "epoch": 2.61,
      "grad_norm": 5.296221797076509,
      "learning_rate": 5.113498976222004e-07,
      "loss": 0.322,
      "step": 61620
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.486936528064822,
      "learning_rate": 5.108076650950561e-07,
      "loss": 0.3316,
      "step": 61625
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.8211108030274903,
      "learning_rate": 5.102657047344212e-07,
      "loss": 0.3078,
      "step": 61630
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.784780053334359,
      "learning_rate": 5.097240165731499e-07,
      "loss": 0.3067,
      "step": 61635
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.00107472647597,
      "learning_rate": 5.091826006440854e-07,
      "loss": 0.3211,
      "step": 61640
    },
    {
      "epoch": 2.61,
      "grad_norm": 7.819110920355072,
      "learning_rate": 5.086414569800524e-07,
      "loss": 0.344,
      "step": 61645
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.397941167638235,
      "learning_rate": 5.081005856138571e-07,
      "loss": 0.3329,
      "step": 61650
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.7534630370329203,
      "learning_rate": 5.075599865782921e-07,
      "loss": 0.2901,
      "step": 61655
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.8749462115055944,
      "learning_rate": 5.070196599061333e-07,
      "loss": 0.3156,
      "step": 61660
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.606854872207813,
      "learning_rate": 5.064796056301369e-07,
      "loss": 0.2865,
      "step": 61665
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.200039837918513,
      "learning_rate": 5.05939823783046e-07,
      "loss": 0.3222,
      "step": 61670
    },
    {
      "epoch": 2.61,
      "grad_norm": 5.186989163023474,
      "learning_rate": 5.054003143975871e-07,
      "loss": 0.3225,
      "step": 61675
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.357362557753834,
      "learning_rate": 5.048610775064666e-07,
      "loss": 0.3414,
      "step": 61680
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.7122388095019168,
      "learning_rate": 5.043221131423781e-07,
      "loss": 0.3165,
      "step": 61685
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.306950235764491,
      "learning_rate": 5.037834213379972e-07,
      "loss": 0.3,
      "step": 61690
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.1521129283972105,
      "learning_rate": 5.032450021259849e-07,
      "loss": 0.3212,
      "step": 61695
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.9560971402412024,
      "learning_rate": 5.027068555389813e-07,
      "loss": 0.3494,
      "step": 61700
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.7487499340118404,
      "learning_rate": 5.021689816096148e-07,
      "loss": 0.3022,
      "step": 61705
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.715152816470811,
      "learning_rate": 5.016313803704931e-07,
      "loss": 0.3001,
      "step": 61710
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.616085660599245,
      "learning_rate": 5.010940518542112e-07,
      "loss": 0.3374,
      "step": 61715
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.730181293782424,
      "learning_rate": 5.005569960933443e-07,
      "loss": 0.3168,
      "step": 61720
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.202024059702223,
      "learning_rate": 5.000202131204529e-07,
      "loss": 0.2923,
      "step": 61725
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.8509993657846255,
      "learning_rate": 4.994837029680821e-07,
      "loss": 0.2968,
      "step": 61730
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.9658950088168803,
      "learning_rate": 4.989474656687571e-07,
      "loss": 0.3152,
      "step": 61735
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.777446708358338,
      "learning_rate": 4.984115012549884e-07,
      "loss": 0.3259,
      "step": 61740
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.024418791367168,
      "learning_rate": 4.978758097592717e-07,
      "loss": 0.3006,
      "step": 61745
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.1609474324050515,
      "learning_rate": 4.973403912140823e-07,
      "loss": 0.3405,
      "step": 61750
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.02304442653169,
      "learning_rate": 4.968052456518818e-07,
      "loss": 0.3143,
      "step": 61755
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.6323390945395433,
      "learning_rate": 4.962703731051144e-07,
      "loss": 0.3193,
      "step": 61760
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.257151533242243,
      "learning_rate": 4.957357736062097e-07,
      "loss": 0.3005,
      "step": 61765
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.827600218900033,
      "learning_rate": 4.95201447187576e-07,
      "loss": 0.3434,
      "step": 61770
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.9067172250515707,
      "learning_rate": 4.946673938816088e-07,
      "loss": 0.3251,
      "step": 61775
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.5134945546421132,
      "learning_rate": 4.941336137206882e-07,
      "loss": 0.303,
      "step": 61780
    },
    {
      "epoch": 2.61,
      "grad_norm": 3.8412196427141563,
      "learning_rate": 4.936001067371732e-07,
      "loss": 0.3223,
      "step": 61785
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.7360459099370615,
      "learning_rate": 4.930668729634091e-07,
      "loss": 0.3131,
      "step": 61790
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.8565568441411173,
      "learning_rate": 4.925339124317264e-07,
      "loss": 0.3335,
      "step": 61795
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.26537055671759,
      "learning_rate": 4.920012251744344e-07,
      "loss": 0.3063,
      "step": 61800
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.873356813153911,
      "learning_rate": 4.914688112238292e-07,
      "loss": 0.3327,
      "step": 61805
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.843880213304912,
      "learning_rate": 4.909366706121899e-07,
      "loss": 0.3271,
      "step": 61810
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.487797344896647,
      "learning_rate": 4.90404803371779e-07,
      "loss": 0.3279,
      "step": 61815
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.813643277797878,
      "learning_rate": 4.898732095348407e-07,
      "loss": 0.322,
      "step": 61820
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.300010933164801,
      "learning_rate": 4.893418891336043e-07,
      "loss": 0.3063,
      "step": 61825
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.326386207261732,
      "learning_rate": 4.88810842200284e-07,
      "loss": 0.327,
      "step": 61830
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.123818365226369,
      "learning_rate": 4.88280068767073e-07,
      "loss": 0.324,
      "step": 61835
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.8449714680780556,
      "learning_rate": 4.877495688661521e-07,
      "loss": 0.291,
      "step": 61840
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.05860815753287,
      "learning_rate": 4.872193425296839e-07,
      "loss": 0.3342,
      "step": 61845
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.80730820226469,
      "learning_rate": 4.866893897898139e-07,
      "loss": 0.3223,
      "step": 61850
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.802247311625388,
      "learning_rate": 4.861597106786715e-07,
      "loss": 0.3211,
      "step": 61855
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.8366309225363158,
      "learning_rate": 4.856303052283695e-07,
      "loss": 0.3027,
      "step": 61860
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.097382181051562,
      "learning_rate": 4.85101173471006e-07,
      "loss": 0.3263,
      "step": 61865
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.8980764405068578,
      "learning_rate": 4.845723154386589e-07,
      "loss": 0.3338,
      "step": 61870
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.321895141840432,
      "learning_rate": 4.840437311633911e-07,
      "loss": 0.3164,
      "step": 61875
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.754000467843632,
      "learning_rate": 4.835154206772497e-07,
      "loss": 0.3078,
      "step": 61880
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.542726878933037,
      "learning_rate": 4.829873840122651e-07,
      "loss": 0.3162,
      "step": 61885
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.9475740788495264,
      "learning_rate": 4.824596212004495e-07,
      "loss": 0.3065,
      "step": 61890
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.863871108321877,
      "learning_rate": 4.819321322738002e-07,
      "loss": 0.3153,
      "step": 61895
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.9835604851027,
      "learning_rate": 4.814049172642982e-07,
      "loss": 0.2873,
      "step": 61900
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.179191854270572,
      "learning_rate": 4.80877976203905e-07,
      "loss": 0.3248,
      "step": 61905
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.135033390382629,
      "learning_rate": 4.803513091245693e-07,
      "loss": 0.3227,
      "step": 61910
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.050966570963871,
      "learning_rate": 4.798249160582208e-07,
      "loss": 0.3117,
      "step": 61915
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.9028896479163047,
      "learning_rate": 4.792987970367724e-07,
      "loss": 0.3066,
      "step": 61920
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.02754580911077,
      "learning_rate": 4.787729520921219e-07,
      "loss": 0.3389,
      "step": 61925
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.783000975423128,
      "learning_rate": 4.782473812561494e-07,
      "loss": 0.3241,
      "step": 61930
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.9338731064580923,
      "learning_rate": 4.7772208456072e-07,
      "loss": 0.3288,
      "step": 61935
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.48857932430121,
      "learning_rate": 4.771970620376787e-07,
      "loss": 0.304,
      "step": 61940
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.877747580483727,
      "learning_rate": 4.7667231371885724e-07,
      "loss": 0.3267,
      "step": 61945
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.525515266412173,
      "learning_rate": 4.761478396360708e-07,
      "loss": 0.3426,
      "step": 61950
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.8893811360417962,
      "learning_rate": 4.7562363982111394e-07,
      "loss": 0.3163,
      "step": 61955
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.129005840373028,
      "learning_rate": 4.75099714305769e-07,
      "loss": 0.2966,
      "step": 61960
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.081183530936069,
      "learning_rate": 4.745760631218005e-07,
      "loss": 0.3152,
      "step": 61965
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.14841009531825,
      "learning_rate": 4.740526863009548e-07,
      "loss": 0.3092,
      "step": 61970
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.026489780859911,
      "learning_rate": 4.7352958387496264e-07,
      "loss": 0.314,
      "step": 61975
    },
    {
      "epoch": 2.62,
      "grad_norm": 5.434235289941069,
      "learning_rate": 4.730067558755402e-07,
      "loss": 0.2846,
      "step": 61980
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.186504864393021,
      "learning_rate": 4.7248420233438177e-07,
      "loss": 0.31,
      "step": 61985
    },
    {
      "epoch": 2.62,
      "grad_norm": 3.987838830645109,
      "learning_rate": 4.7196192328317067e-07,
      "loss": 0.3193,
      "step": 61990
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.06604996020392,
      "learning_rate": 4.7143991875357007e-07,
      "loss": 0.3044,
      "step": 61995
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.4786265278390625,
      "learning_rate": 4.709181887772285e-07,
      "loss": 0.3143,
      "step": 62000
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.0293313270815645,
      "learning_rate": 4.7039673338577563e-07,
      "loss": 0.287,
      "step": 62005
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.184571140639369,
      "learning_rate": 4.6987555261082673e-07,
      "loss": 0.3172,
      "step": 62010
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.2418463732659095,
      "learning_rate": 4.6935464648397935e-07,
      "loss": 0.3148,
      "step": 62015
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.648134298010513,
      "learning_rate": 4.688340150368142e-07,
      "loss": 0.3159,
      "step": 62020
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.4068946381351,
      "learning_rate": 4.6831365830089514e-07,
      "loss": 0.2947,
      "step": 62025
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.9851286632081186,
      "learning_rate": 4.677935763077718e-07,
      "loss": 0.3148,
      "step": 62030
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.154976559697933,
      "learning_rate": 4.672737690889734e-07,
      "loss": 0.3138,
      "step": 62035
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.367260197703903,
      "learning_rate": 4.6675423667601417e-07,
      "loss": 0.3191,
      "step": 62040
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.7986328193506735,
      "learning_rate": 4.6623497910039227e-07,
      "loss": 0.3235,
      "step": 62045
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.6624905167990454,
      "learning_rate": 4.657159963935903e-07,
      "loss": 0.3849,
      "step": 62050
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.0803212857036435,
      "learning_rate": 4.651972885870698e-07,
      "loss": 0.2947,
      "step": 62055
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.978560572543455,
      "learning_rate": 4.6467885571228057e-07,
      "loss": 0.3035,
      "step": 62060
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.940676377197255,
      "learning_rate": 4.641606978006524e-07,
      "loss": 0.3075,
      "step": 62065
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.15025134143643,
      "learning_rate": 4.63642814883602e-07,
      "loss": 0.3069,
      "step": 62070
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.071293092827934,
      "learning_rate": 4.6312520699252406e-07,
      "loss": 0.3374,
      "step": 62075
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.284741222171689,
      "learning_rate": 4.6260787415880137e-07,
      "loss": 0.3004,
      "step": 62080
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.162697807570442,
      "learning_rate": 4.620908164137988e-07,
      "loss": 0.3184,
      "step": 62085
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.979645905454209,
      "learning_rate": 4.615740337888619e-07,
      "loss": 0.3148,
      "step": 62090
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.891892827953397,
      "learning_rate": 4.610575263153233e-07,
      "loss": 0.3107,
      "step": 62095
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.000297329638286,
      "learning_rate": 4.60541294024498e-07,
      "loss": 0.3228,
      "step": 62100
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.532291744061456,
      "learning_rate": 4.6002533694768147e-07,
      "loss": 0.3406,
      "step": 62105
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.812773912051182,
      "learning_rate": 4.595096551161554e-07,
      "loss": 0.2894,
      "step": 62110
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.7747853287298923,
      "learning_rate": 4.5899424856118533e-07,
      "loss": 0.3076,
      "step": 62115
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.166983895910034,
      "learning_rate": 4.584791173140185e-07,
      "loss": 0.3063,
      "step": 62120
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.213401972157103,
      "learning_rate": 4.579642614058843e-07,
      "loss": 0.3316,
      "step": 62125
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.473739050296158,
      "learning_rate": 4.5744968086799793e-07,
      "loss": 0.312,
      "step": 62130
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.3238967528751817,
      "learning_rate": 4.5693537573155813e-07,
      "loss": 0.2782,
      "step": 62135
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.205583338534464,
      "learning_rate": 4.5642134602774337e-07,
      "loss": 0.3461,
      "step": 62140
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.9367353994930103,
      "learning_rate": 4.5590759178771873e-07,
      "loss": 0.3199,
      "step": 62145
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.6753713253600466,
      "learning_rate": 4.553941130426326e-07,
      "loss": 0.3126,
      "step": 62150
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.7247961111282244,
      "learning_rate": 4.54880909823614e-07,
      "loss": 0.3091,
      "step": 62155
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.6897391032153757,
      "learning_rate": 4.54367982161778e-07,
      "loss": 0.3022,
      "step": 62160
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.6982490355415982,
      "learning_rate": 4.5385533008822146e-07,
      "loss": 0.2961,
      "step": 62165
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.0672506624954865,
      "learning_rate": 4.5334295363402616e-07,
      "loss": 0.3072,
      "step": 62170
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.260266438889107,
      "learning_rate": 4.5283085283025384e-07,
      "loss": 0.3465,
      "step": 62175
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.8681760165837944,
      "learning_rate": 4.523190277079531e-07,
      "loss": 0.2965,
      "step": 62180
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.414103981966088,
      "learning_rate": 4.518074782981546e-07,
      "loss": 0.3062,
      "step": 62185
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.891465929709592,
      "learning_rate": 4.512962046318714e-07,
      "loss": 0.2996,
      "step": 62190
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.317033430660098,
      "learning_rate": 4.507852067401003e-07,
      "loss": 0.2992,
      "step": 62195
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.086105384469176,
      "learning_rate": 4.5027448465382207e-07,
      "loss": 0.3161,
      "step": 62200
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.250221577729695,
      "learning_rate": 4.497640384040003e-07,
      "loss": 0.3432,
      "step": 62205
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.6493664409687074,
      "learning_rate": 4.492538680215813e-07,
      "loss": 0.3026,
      "step": 62210
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.135977039658859,
      "learning_rate": 4.4874397353749597e-07,
      "loss": 0.3065,
      "step": 62215
    },
    {
      "epoch": 2.63,
      "grad_norm": 5.232765021489601,
      "learning_rate": 4.482343549826579e-07,
      "loss": 0.3174,
      "step": 62220
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.463472978189227,
      "learning_rate": 4.477250123879623e-07,
      "loss": 0.308,
      "step": 62225
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.274657271005043,
      "learning_rate": 4.4721594578429005e-07,
      "loss": 0.3294,
      "step": 62230
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.127685598654136,
      "learning_rate": 4.4670715520250484e-07,
      "loss": 0.3092,
      "step": 62235
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.505085828176142,
      "learning_rate": 4.4619864067345255e-07,
      "loss": 0.3208,
      "step": 62240
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.150739956785816,
      "learning_rate": 4.4569040222796234e-07,
      "loss": 0.3365,
      "step": 62245
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.5715358713377645,
      "learning_rate": 4.4518243989684793e-07,
      "loss": 0.3076,
      "step": 62250
    },
    {
      "epoch": 2.63,
      "grad_norm": 4.004925192446083,
      "learning_rate": 4.44674753710907e-07,
      "loss": 0.285,
      "step": 62255
    },
    {
      "epoch": 2.63,
      "grad_norm": 3.880839020667359,
      "learning_rate": 4.441673437009164e-07,
      "loss": 0.2977,
      "step": 62260
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.7802019264526927,
      "learning_rate": 4.4366020989764005e-07,
      "loss": 0.3073,
      "step": 62265
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.3868838708714994,
      "learning_rate": 4.4315335233182487e-07,
      "loss": 0.3323,
      "step": 62270
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.783471177471014,
      "learning_rate": 4.4264677103419863e-07,
      "loss": 0.3217,
      "step": 62275
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.8928707398943856,
      "learning_rate": 4.421404660354739e-07,
      "loss": 0.2982,
      "step": 62280
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.032453782623777,
      "learning_rate": 4.416344373663489e-07,
      "loss": 0.3218,
      "step": 62285
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.576640811038284,
      "learning_rate": 4.4112868505749917e-07,
      "loss": 0.3027,
      "step": 62290
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.6051552283845214,
      "learning_rate": 4.40623209139589e-07,
      "loss": 0.3115,
      "step": 62295
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.218690241005833,
      "learning_rate": 4.401180096432639e-07,
      "loss": 0.3025,
      "step": 62300
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.192772501321442,
      "learning_rate": 4.396130865991527e-07,
      "loss": 0.3263,
      "step": 62305
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.7606415646585747,
      "learning_rate": 4.391084400378665e-07,
      "loss": 0.3391,
      "step": 62310
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.177950437824774,
      "learning_rate": 4.3860406999000074e-07,
      "loss": 0.3063,
      "step": 62315
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.7171736505639994,
      "learning_rate": 4.3809997648613545e-07,
      "loss": 0.3046,
      "step": 62320
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.6741382501235718,
      "learning_rate": 4.3759615955682955e-07,
      "loss": 0.2893,
      "step": 62325
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.270355098223836,
      "learning_rate": 4.3709261923263027e-07,
      "loss": 0.3148,
      "step": 62330
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.915009653193231,
      "learning_rate": 4.3658935554406537e-07,
      "loss": 0.3018,
      "step": 62335
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.720301480927801,
      "learning_rate": 4.3608636852164486e-07,
      "loss": 0.3082,
      "step": 62340
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.87274674045287,
      "learning_rate": 4.355836581958645e-07,
      "loss": 0.299,
      "step": 62345
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.084986336560197,
      "learning_rate": 4.350812245972025e-07,
      "loss": 0.3072,
      "step": 62350
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.216755447824875,
      "learning_rate": 4.345790677561207e-07,
      "loss": 0.3353,
      "step": 62355
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.9103749153226413,
      "learning_rate": 4.3407718770306026e-07,
      "loss": 0.3117,
      "step": 62360
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.064086601364142,
      "learning_rate": 4.3357558446845085e-07,
      "loss": 0.296,
      "step": 62365
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.588534171076538,
      "learning_rate": 4.3307425808270296e-07,
      "loss": 0.3187,
      "step": 62370
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.211333261488694,
      "learning_rate": 4.325732085762113e-07,
      "loss": 0.3282,
      "step": 62375
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.667269424607841,
      "learning_rate": 4.3207243597935153e-07,
      "loss": 0.3167,
      "step": 62380
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.582452102109443,
      "learning_rate": 4.315719403224844e-07,
      "loss": 0.314,
      "step": 62385
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.107428418292896,
      "learning_rate": 4.3107172163595446e-07,
      "loss": 0.2944,
      "step": 62390
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.5976620529439676,
      "learning_rate": 4.305717799500875e-07,
      "loss": 0.3222,
      "step": 62395
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.942093710922755,
      "learning_rate": 4.3007211529519366e-07,
      "loss": 0.3243,
      "step": 62400
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.127776951655821,
      "learning_rate": 4.295727277015671e-07,
      "loss": 0.3204,
      "step": 62405
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.5753943665623837,
      "learning_rate": 4.2907361719948246e-07,
      "loss": 0.2937,
      "step": 62410
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.125975312288947,
      "learning_rate": 4.2857478381920004e-07,
      "loss": 0.3092,
      "step": 62415
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.307459394067735,
      "learning_rate": 4.280762275909633e-07,
      "loss": 0.3401,
      "step": 62420
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.680277844240787,
      "learning_rate": 4.275779485449988e-07,
      "loss": 0.3043,
      "step": 62425
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.8945985928468043,
      "learning_rate": 4.2707994671151334e-07,
      "loss": 0.3194,
      "step": 62430
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.193478380339177,
      "learning_rate": 4.265822221207011e-07,
      "loss": 0.3408,
      "step": 62435
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.9947280895547292,
      "learning_rate": 4.260847748027386e-07,
      "loss": 0.3068,
      "step": 62440
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.706864588725747,
      "learning_rate": 4.255876047877816e-07,
      "loss": 0.3,
      "step": 62445
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.183828890630293,
      "learning_rate": 4.250907121059744e-07,
      "loss": 0.3128,
      "step": 62450
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.145990129283243,
      "learning_rate": 4.245940967874418e-07,
      "loss": 0.3389,
      "step": 62455
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.108575491707238,
      "learning_rate": 4.240977588622913e-07,
      "loss": 0.3046,
      "step": 62460
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.006368525070137,
      "learning_rate": 4.2360169836061504e-07,
      "loss": 0.3299,
      "step": 62465
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.7961833393325795,
      "learning_rate": 4.231059153124872e-07,
      "loss": 0.3277,
      "step": 62470
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.12929652250622,
      "learning_rate": 4.226104097479672e-07,
      "loss": 0.3365,
      "step": 62475
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.0271903468062495,
      "learning_rate": 4.221151816970942e-07,
      "loss": 0.3146,
      "step": 62480
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.777805053933496,
      "learning_rate": 4.216202311898926e-07,
      "loss": 0.2952,
      "step": 62485
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.5376265744807736,
      "learning_rate": 4.2112555825637167e-07,
      "loss": 0.3228,
      "step": 62490
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.8143661922808745,
      "learning_rate": 4.2063116292651974e-07,
      "loss": 0.341,
      "step": 62495
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.9101203009994405,
      "learning_rate": 4.201370452303111e-07,
      "loss": 0.2957,
      "step": 62500
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.5829591916765864,
      "learning_rate": 4.1964320519770395e-07,
      "loss": 0.2879,
      "step": 62505
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.907673467667554,
      "learning_rate": 4.191496428586367e-07,
      "loss": 0.2917,
      "step": 62510
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.035933862704518,
      "learning_rate": 4.1865635824303365e-07,
      "loss": 0.3192,
      "step": 62515
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.188700353930271,
      "learning_rate": 4.1816335138080035e-07,
      "loss": 0.2923,
      "step": 62520
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.9489801907466973,
      "learning_rate": 4.1767062230182784e-07,
      "loss": 0.3267,
      "step": 62525
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.0380631671401135,
      "learning_rate": 4.171781710359868e-07,
      "loss": 0.3062,
      "step": 62530
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.139472944979526,
      "learning_rate": 4.166859976131338e-07,
      "loss": 0.3284,
      "step": 62535
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.447480325181174,
      "learning_rate": 4.161941020631094e-07,
      "loss": 0.3094,
      "step": 62540
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.301603010620997,
      "learning_rate": 4.157024844157337e-07,
      "loss": 0.3178,
      "step": 62545
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.8473966544862317,
      "learning_rate": 4.1521114470081223e-07,
      "loss": 0.3019,
      "step": 62550
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.102813216240773,
      "learning_rate": 4.1472008294813515e-07,
      "loss": 0.3093,
      "step": 62555
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.1421077011691185,
      "learning_rate": 4.1422929918747303e-07,
      "loss": 0.3395,
      "step": 62560
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.447333841541427,
      "learning_rate": 4.1373879344858035e-07,
      "loss": 0.2888,
      "step": 62565
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.9110970265115625,
      "learning_rate": 4.132485657611951e-07,
      "loss": 0.2902,
      "step": 62570
    },
    {
      "epoch": 2.65,
      "grad_norm": 5.009399336059135,
      "learning_rate": 4.1275861615503954e-07,
      "loss": 0.3266,
      "step": 62575
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.6052795707777734,
      "learning_rate": 4.12268944659816e-07,
      "loss": 0.2812,
      "step": 62580
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.252989943955905,
      "learning_rate": 4.117795513052125e-07,
      "loss": 0.328,
      "step": 62585
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.212473609643261,
      "learning_rate": 4.1129043612090137e-07,
      "loss": 0.3145,
      "step": 62590
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.6820735534044653,
      "learning_rate": 4.1080159913653274e-07,
      "loss": 0.3187,
      "step": 62595
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.230818392896048,
      "learning_rate": 4.1031304038174625e-07,
      "loss": 0.3033,
      "step": 62600
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.8304496603345872,
      "learning_rate": 4.098247598861599e-07,
      "loss": 0.2981,
      "step": 62605
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.000488891883417,
      "learning_rate": 4.0933675767937884e-07,
      "loss": 0.3309,
      "step": 62610
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.9304357191152843,
      "learning_rate": 4.0884903379098673e-07,
      "loss": 0.3119,
      "step": 62615
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.10233754043791,
      "learning_rate": 4.083615882505543e-07,
      "loss": 0.3594,
      "step": 62620
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.261915961266912,
      "learning_rate": 4.078744210876345e-07,
      "loss": 0.3436,
      "step": 62625
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.8868360895943073,
      "learning_rate": 4.073875323317617e-07,
      "loss": 0.348,
      "step": 62630
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.018408617594063,
      "learning_rate": 4.069009220124542e-07,
      "loss": 0.3261,
      "step": 62635
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.857177456550557,
      "learning_rate": 4.064145901592154e-07,
      "loss": 0.3114,
      "step": 62640
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.8878469000594262,
      "learning_rate": 4.059285368015281e-07,
      "loss": 0.3083,
      "step": 62645
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.1309302922725335,
      "learning_rate": 4.0544276196886113e-07,
      "loss": 0.3515,
      "step": 62650
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.117796456344412,
      "learning_rate": 4.049572656906664e-07,
      "loss": 0.2964,
      "step": 62655
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.245049819102319,
      "learning_rate": 4.0447204799637765e-07,
      "loss": 0.3412,
      "step": 62660
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.868061169510937,
      "learning_rate": 4.039871089154118e-07,
      "loss": 0.2927,
      "step": 62665
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.9487228630287237,
      "learning_rate": 4.0350244847716936e-07,
      "loss": 0.3212,
      "step": 62670
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.9362236426391983,
      "learning_rate": 4.030180667110351e-07,
      "loss": 0.3151,
      "step": 62675
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.022362415444539,
      "learning_rate": 4.025339636463743e-07,
      "loss": 0.3107,
      "step": 62680
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.657344414422595,
      "learning_rate": 4.020501393125359e-07,
      "loss": 0.3043,
      "step": 62685
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.317213174352339,
      "learning_rate": 4.0156659373885465e-07,
      "loss": 0.3298,
      "step": 62690
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.030750161798941,
      "learning_rate": 4.0108332695464603e-07,
      "loss": 0.3125,
      "step": 62695
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.919019740430411,
      "learning_rate": 4.006003389892077e-07,
      "loss": 0.2974,
      "step": 62700
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.4595626926307785,
      "learning_rate": 4.001176298718229e-07,
      "loss": 0.3342,
      "step": 62705
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.913374992107964,
      "learning_rate": 3.9963519963175825e-07,
      "loss": 0.3183,
      "step": 62710
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.8654010721597842,
      "learning_rate": 3.9915304829825916e-07,
      "loss": 0.2976,
      "step": 62715
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.6358854877686295,
      "learning_rate": 3.986711759005585e-07,
      "loss": 0.3069,
      "step": 62720
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.883880699745248,
      "learning_rate": 3.981895824678711e-07,
      "loss": 0.3011,
      "step": 62725
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.019832031041032,
      "learning_rate": 3.977082680293953e-07,
      "loss": 0.3278,
      "step": 62730
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.964238755025775,
      "learning_rate": 3.972272326143095e-07,
      "loss": 0.3213,
      "step": 62735
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.036516533172175,
      "learning_rate": 3.9674647625177867e-07,
      "loss": 0.3157,
      "step": 62740
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.471826685983796,
      "learning_rate": 3.962659989709505e-07,
      "loss": 0.3112,
      "step": 62745
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.549004953113974,
      "learning_rate": 3.9578580080095296e-07,
      "loss": 0.2975,
      "step": 62750
    },
    {
      "epoch": 2.66,
      "grad_norm": 6.373782728759738,
      "learning_rate": 3.95305881770901e-07,
      "loss": 0.2901,
      "step": 62755
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.2960755236821235,
      "learning_rate": 3.948262419098897e-07,
      "loss": 0.3105,
      "step": 62760
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.458788581526812,
      "learning_rate": 3.9434688124699803e-07,
      "loss": 0.3079,
      "step": 62765
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.007021741412846,
      "learning_rate": 3.938677998112889e-07,
      "loss": 0.3127,
      "step": 62770
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.466896579803888,
      "learning_rate": 3.933889976318067e-07,
      "loss": 0.312,
      "step": 62775
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.223018166272255,
      "learning_rate": 3.929104747375817e-07,
      "loss": 0.33,
      "step": 62780
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.822980934469118,
      "learning_rate": 3.924322311576223e-07,
      "loss": 0.3094,
      "step": 62785
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.7504172250582966,
      "learning_rate": 3.919542669209253e-07,
      "loss": 0.3043,
      "step": 62790
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.3690442015275064,
      "learning_rate": 3.9147658205646867e-07,
      "loss": 0.2906,
      "step": 62795
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.4881108828009033,
      "learning_rate": 3.9099917659321086e-07,
      "loss": 0.2937,
      "step": 62800
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.9622058354861496,
      "learning_rate": 3.9052205056009705e-07,
      "loss": 0.2916,
      "step": 62805
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.8285423444192466,
      "learning_rate": 3.900452039860542e-07,
      "loss": 0.278,
      "step": 62810
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.786064232446165,
      "learning_rate": 3.895686368999907e-07,
      "loss": 0.3251,
      "step": 62815
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.559570286536799,
      "learning_rate": 3.890923493308002e-07,
      "loss": 0.3033,
      "step": 62820
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.613033224864597,
      "learning_rate": 3.8861634130735957e-07,
      "loss": 0.2866,
      "step": 62825
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.328184832963255,
      "learning_rate": 3.881406128585263e-07,
      "loss": 0.3178,
      "step": 62830
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.212968339442758,
      "learning_rate": 3.876651640131429e-07,
      "loss": 0.3337,
      "step": 62835
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.082110291055248,
      "learning_rate": 3.8718999480003516e-07,
      "loss": 0.3207,
      "step": 62840
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.5040417113813787,
      "learning_rate": 3.867151052480106e-07,
      "loss": 0.3092,
      "step": 62845
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.7191139762989667,
      "learning_rate": 3.862404953858595e-07,
      "loss": 0.2999,
      "step": 62850
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.5432629042874906,
      "learning_rate": 3.857661652423572e-07,
      "loss": 0.2978,
      "step": 62855
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.7260448657606315,
      "learning_rate": 3.8529211484626016e-07,
      "loss": 0.3208,
      "step": 62860
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.143889934351701,
      "learning_rate": 3.8481834422631037e-07,
      "loss": 0.3165,
      "step": 62865
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.595413338474495,
      "learning_rate": 3.843448534112287e-07,
      "loss": 0.3279,
      "step": 62870
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.113120506968053,
      "learning_rate": 3.838716424297229e-07,
      "loss": 0.3237,
      "step": 62875
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.366969188580846,
      "learning_rate": 3.8339871131048314e-07,
      "loss": 0.3086,
      "step": 62880
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.849661488717415,
      "learning_rate": 3.8292606008217993e-07,
      "loss": 0.2953,
      "step": 62885
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.13492382678597,
      "learning_rate": 3.824536887734698e-07,
      "loss": 0.3214,
      "step": 62890
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.6930570783544154,
      "learning_rate": 3.8198159741299146e-07,
      "loss": 0.2991,
      "step": 62895
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.652318367684894,
      "learning_rate": 3.8150978602936594e-07,
      "loss": 0.3125,
      "step": 62900
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.965424330921914,
      "learning_rate": 3.81038254651197e-07,
      "loss": 0.3394,
      "step": 62905
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.663769557283041,
      "learning_rate": 3.8056700330707407e-07,
      "loss": 0.3188,
      "step": 62910
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.8631574569840006,
      "learning_rate": 3.80096032025567e-07,
      "loss": 0.3049,
      "step": 62915
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.3772366127057785,
      "learning_rate": 3.7962534083522796e-07,
      "loss": 0.2999,
      "step": 62920
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.700627785795629,
      "learning_rate": 3.7915492976459524e-07,
      "loss": 0.2935,
      "step": 62925
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.769597669478493,
      "learning_rate": 3.786847988421888e-07,
      "loss": 0.3141,
      "step": 62930
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.476550767849943,
      "learning_rate": 3.782149480965097e-07,
      "loss": 0.3171,
      "step": 62935
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.9883081859016882,
      "learning_rate": 3.777453775560441e-07,
      "loss": 0.3319,
      "step": 62940
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.322442805168393,
      "learning_rate": 3.77276087249262e-07,
      "loss": 0.3,
      "step": 62945
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.5312706288985742,
      "learning_rate": 3.768070772046134e-07,
      "loss": 0.3276,
      "step": 62950
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.7259883954948725,
      "learning_rate": 3.763383474505333e-07,
      "loss": 0.3094,
      "step": 62955
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.675865446565234,
      "learning_rate": 3.758698980154396e-07,
      "loss": 0.3445,
      "step": 62960
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.257036900331993,
      "learning_rate": 3.754017289277345e-07,
      "loss": 0.3151,
      "step": 62965
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.141267654892652,
      "learning_rate": 3.7493384021579926e-07,
      "loss": 0.3053,
      "step": 62970
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.517845425439684,
      "learning_rate": 3.7446623190800167e-07,
      "loss": 0.3399,
      "step": 62975
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.547274188659153,
      "learning_rate": 3.7399890403269244e-07,
      "loss": 0.2901,
      "step": 62980
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.9424105440736965,
      "learning_rate": 3.735318566182022e-07,
      "loss": 0.3101,
      "step": 62985
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.71816535066981,
      "learning_rate": 3.730650896928484e-07,
      "loss": 0.3335,
      "step": 62990
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.774555322403104,
      "learning_rate": 3.7259860328492934e-07,
      "loss": 0.3266,
      "step": 62995
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.836153236432657,
      "learning_rate": 3.72132397422727e-07,
      "loss": 0.309,
      "step": 63000
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.660229316598106,
      "learning_rate": 3.7166647213450423e-07,
      "loss": 0.3095,
      "step": 63005
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.079337364428318,
      "learning_rate": 3.7120082744851016e-07,
      "loss": 0.3323,
      "step": 63010
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.388071970382545,
      "learning_rate": 3.707354633929766e-07,
      "loss": 0.2991,
      "step": 63015
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.247074231390211,
      "learning_rate": 3.702703799961149e-07,
      "loss": 0.3331,
      "step": 63020
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.9172282293462968,
      "learning_rate": 3.6980557728612254e-07,
      "loss": 0.299,
      "step": 63025
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.763376038016525,
      "learning_rate": 3.693410552911797e-07,
      "loss": 0.3275,
      "step": 63030
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.7089181044183257,
      "learning_rate": 3.6887681403944885e-07,
      "loss": 0.3016,
      "step": 63035
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.515039029522563,
      "learning_rate": 3.684128535590753e-07,
      "loss": 0.3105,
      "step": 63040
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.823857735537007,
      "learning_rate": 3.6794917387818706e-07,
      "loss": 0.3155,
      "step": 63045
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.638622436570413,
      "learning_rate": 3.6748577502489725e-07,
      "loss": 0.29,
      "step": 63050
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.7307567709265133,
      "learning_rate": 3.670226570272989e-07,
      "loss": 0.3276,
      "step": 63055
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.9707330344181617,
      "learning_rate": 3.6655981991346955e-07,
      "loss": 0.2909,
      "step": 63060
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.25895628680597,
      "learning_rate": 3.6609726371147067e-07,
      "loss": 0.3084,
      "step": 63065
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.6790490376620313,
      "learning_rate": 3.6563498844934485e-07,
      "loss": 0.2935,
      "step": 63070
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.160279738610442,
      "learning_rate": 3.651729941551185e-07,
      "loss": 0.2873,
      "step": 63075
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.290986744296786,
      "learning_rate": 3.6471128085680096e-07,
      "loss": 0.3229,
      "step": 63080
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.5576104524744623,
      "learning_rate": 3.642498485823859e-07,
      "loss": 0.2902,
      "step": 63085
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.7227335358086036,
      "learning_rate": 3.637886973598465e-07,
      "loss": 0.3142,
      "step": 63090
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.347242945399834,
      "learning_rate": 3.63327827217142e-07,
      "loss": 0.3258,
      "step": 63095
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.8316713591159774,
      "learning_rate": 3.628672381822146e-07,
      "loss": 0.3038,
      "step": 63100
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.7631134696392308,
      "learning_rate": 3.6240693028298636e-07,
      "loss": 0.3046,
      "step": 63105
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.629145535827151,
      "learning_rate": 3.619469035473655e-07,
      "loss": 0.3093,
      "step": 63110
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.848838485693634,
      "learning_rate": 3.614871580032425e-07,
      "loss": 0.3139,
      "step": 63115
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.462028056040803,
      "learning_rate": 3.6102769367848953e-07,
      "loss": 0.3112,
      "step": 63120
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.308744281050955,
      "learning_rate": 3.605685106009632e-07,
      "loss": 0.3188,
      "step": 63125
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.050373853463839,
      "learning_rate": 3.601096087985023e-07,
      "loss": 0.3085,
      "step": 63130
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.28989871063134,
      "learning_rate": 3.596509882989285e-07,
      "loss": 0.3254,
      "step": 63135
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.617329027302648,
      "learning_rate": 3.5919264913004624e-07,
      "loss": 0.3167,
      "step": 63140
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.34690333924876,
      "learning_rate": 3.5873459131964383e-07,
      "loss": 0.2956,
      "step": 63145
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.8602412605825775,
      "learning_rate": 3.582768148954924e-07,
      "loss": 0.3126,
      "step": 63150
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.558462619116576,
      "learning_rate": 3.578193198853447e-07,
      "loss": 0.3045,
      "step": 63155
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.089247466433942,
      "learning_rate": 3.573621063169375e-07,
      "loss": 0.3014,
      "step": 63160
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.448606070304286,
      "learning_rate": 3.5690517421799076e-07,
      "loss": 0.3129,
      "step": 63165
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.005898398982488,
      "learning_rate": 3.564485236162074e-07,
      "loss": 0.3048,
      "step": 63170
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.750109998091521,
      "learning_rate": 3.559921545392708e-07,
      "loss": 0.3027,
      "step": 63175
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.8583589959766167,
      "learning_rate": 3.5553606701485055e-07,
      "loss": 0.3218,
      "step": 63180
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.062017841310653,
      "learning_rate": 3.5508026107059836e-07,
      "loss": 0.3051,
      "step": 63185
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.8276387556936613,
      "learning_rate": 3.5462473673414766e-07,
      "loss": 0.3146,
      "step": 63190
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.213257576198637,
      "learning_rate": 3.541694940331153e-07,
      "loss": 0.3366,
      "step": 63195
    },
    {
      "epoch": 2.67,
      "grad_norm": 7.909023961073213,
      "learning_rate": 3.537145329951025e-07,
      "loss": 0.3113,
      "step": 63200
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.5756925536496453,
      "learning_rate": 3.532598536476911e-07,
      "loss": 0.3067,
      "step": 63205
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.7428387732383315,
      "learning_rate": 3.528054560184474e-07,
      "loss": 0.2989,
      "step": 63210
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.387162293554796,
      "learning_rate": 3.523513401349199e-07,
      "loss": 0.3185,
      "step": 63215
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.724452529211642,
      "learning_rate": 3.51897506024641e-07,
      "loss": 0.3233,
      "step": 63220
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.2459704624738634,
      "learning_rate": 3.5144395371512485e-07,
      "loss": 0.2926,
      "step": 63225
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.941283983722673,
      "learning_rate": 3.509906832338683e-07,
      "loss": 0.2853,
      "step": 63230
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.839492283560509,
      "learning_rate": 3.5053769460835385e-07,
      "loss": 0.3072,
      "step": 63235
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.720361842743651,
      "learning_rate": 3.500849878660423e-07,
      "loss": 0.3099,
      "step": 63240
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.7045334456504424,
      "learning_rate": 3.4963256303438167e-07,
      "loss": 0.3233,
      "step": 63245
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.410965785152744,
      "learning_rate": 3.491804201408011e-07,
      "loss": 0.309,
      "step": 63250
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.983955286713831,
      "learning_rate": 3.4872855921271156e-07,
      "loss": 0.329,
      "step": 63255
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.298398960441399,
      "learning_rate": 3.482769802775082e-07,
      "loss": 0.3159,
      "step": 63260
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.444943039756403,
      "learning_rate": 3.4782568336256983e-07,
      "loss": 0.3312,
      "step": 63265
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.202554289875888,
      "learning_rate": 3.473746684952578e-07,
      "loss": 0.3295,
      "step": 63270
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.0608905995582845,
      "learning_rate": 3.469239357029136e-07,
      "loss": 0.3257,
      "step": 63275
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.1242594056199335,
      "learning_rate": 3.4647348501286537e-07,
      "loss": 0.3149,
      "step": 63280
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.037876476697423,
      "learning_rate": 3.4602331645242347e-07,
      "loss": 0.337,
      "step": 63285
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.9340345575230233,
      "learning_rate": 3.4557343004887776e-07,
      "loss": 0.3228,
      "step": 63290
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.161195654822754,
      "learning_rate": 3.451238258295053e-07,
      "loss": 0.3221,
      "step": 63295
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.6699141417986887,
      "learning_rate": 3.4467450382156484e-07,
      "loss": 0.3085,
      "step": 63300
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.6659597530951906,
      "learning_rate": 3.442254640522952e-07,
      "loss": 0.322,
      "step": 63305
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.752612471594262,
      "learning_rate": 3.437767065489223e-07,
      "loss": 0.3128,
      "step": 63310
    },
    {
      "epoch": 2.68,
      "grad_norm": 5.923503439919269,
      "learning_rate": 3.433282313386516e-07,
      "loss": 0.3333,
      "step": 63315
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.5581217984930817,
      "learning_rate": 3.4288003844867467e-07,
      "loss": 0.2896,
      "step": 63320
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.981903419644975,
      "learning_rate": 3.424321279061632e-07,
      "loss": 0.341,
      "step": 63325
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.8096554574919606,
      "learning_rate": 3.4198449973827154e-07,
      "loss": 0.3409,
      "step": 63330
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.15027510271367,
      "learning_rate": 3.4153715397213905e-07,
      "loss": 0.3024,
      "step": 63335
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.869499713850801,
      "learning_rate": 3.410900906348874e-07,
      "loss": 0.297,
      "step": 63340
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.852368621188546,
      "learning_rate": 3.4064330975361993e-07,
      "loss": 0.2935,
      "step": 63345
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.466918705146553,
      "learning_rate": 3.4019681135542383e-07,
      "loss": 0.2983,
      "step": 63350
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.704616046684038,
      "learning_rate": 3.397505954673702e-07,
      "loss": 0.2914,
      "step": 63355
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.632388038180807,
      "learning_rate": 3.393046621165097e-07,
      "loss": 0.3156,
      "step": 63360
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.287795553208689,
      "learning_rate": 3.3885901132987896e-07,
      "loss": 0.295,
      "step": 63365
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.189436966609815,
      "learning_rate": 3.3841364313449756e-07,
      "loss": 0.3345,
      "step": 63370
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.065564924882823,
      "learning_rate": 3.379685575573649e-07,
      "loss": 0.3313,
      "step": 63375
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.7660807688813756,
      "learning_rate": 3.3752375462546615e-07,
      "loss": 0.3046,
      "step": 63380
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.4195753906266653,
      "learning_rate": 3.370792343657692e-07,
      "loss": 0.2795,
      "step": 63385
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.8846844275227705,
      "learning_rate": 3.366349968052224e-07,
      "loss": 0.3082,
      "step": 63390
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.9861153799117064,
      "learning_rate": 3.361910419707592e-07,
      "loss": 0.3087,
      "step": 63395
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.772696980978408,
      "learning_rate": 3.3574736988929545e-07,
      "loss": 0.3099,
      "step": 63400
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.881036085790811,
      "learning_rate": 3.353039805877306e-07,
      "loss": 0.3246,
      "step": 63405
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.541051401992806,
      "learning_rate": 3.348608740929443e-07,
      "loss": 0.329,
      "step": 63410
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.779742961823956,
      "learning_rate": 3.3441805043180175e-07,
      "loss": 0.2922,
      "step": 63415
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.760066285635997,
      "learning_rate": 3.3397550963115034e-07,
      "loss": 0.2974,
      "step": 63420
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.920615279935689,
      "learning_rate": 3.335332517178186e-07,
      "loss": 0.3085,
      "step": 63425
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.7892340448363435,
      "learning_rate": 3.3309127671862063e-07,
      "loss": 0.2864,
      "step": 63430
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.730264375575319,
      "learning_rate": 3.3264958466035226e-07,
      "loss": 0.3324,
      "step": 63435
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.662611820688694,
      "learning_rate": 3.322081755697909e-07,
      "loss": 0.2951,
      "step": 63440
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.747979117619223,
      "learning_rate": 3.317670494736985e-07,
      "loss": 0.3134,
      "step": 63445
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.82233068396079,
      "learning_rate": 3.3132620639881873e-07,
      "loss": 0.3319,
      "step": 63450
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.86455603076855,
      "learning_rate": 3.3088564637188013e-07,
      "loss": 0.2907,
      "step": 63455
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.6702179813193556,
      "learning_rate": 3.3044536941959083e-07,
      "loss": 0.3038,
      "step": 63460
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.726698312871764,
      "learning_rate": 3.3000537556864385e-07,
      "loss": 0.3326,
      "step": 63465
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.8586134768816223,
      "learning_rate": 3.295656648457163e-07,
      "loss": 0.319,
      "step": 63470
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.921426741207301,
      "learning_rate": 3.2912623727746395e-07,
      "loss": 0.3212,
      "step": 63475
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.653790379306705,
      "learning_rate": 3.2868709289052947e-07,
      "loss": 0.2948,
      "step": 63480
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.900115844058006,
      "learning_rate": 3.282482317115376e-07,
      "loss": 0.3087,
      "step": 63485
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.876253816104976,
      "learning_rate": 3.278096537670944e-07,
      "loss": 0.2953,
      "step": 63490
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.001770883004138,
      "learning_rate": 3.2737135908378845e-07,
      "loss": 0.337,
      "step": 63495
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.085340234854928,
      "learning_rate": 3.269333476881936e-07,
      "loss": 0.3229,
      "step": 63500
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.585882067439357,
      "learning_rate": 3.264956196068653e-07,
      "loss": 0.3253,
      "step": 63505
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.9680427948082277,
      "learning_rate": 3.2605817486634105e-07,
      "loss": 0.3261,
      "step": 63510
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.8694908491687556,
      "learning_rate": 3.2562101349314147e-07,
      "loss": 0.2982,
      "step": 63515
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.4312982568659467,
      "learning_rate": 3.2518413551377193e-07,
      "loss": 0.288,
      "step": 63520
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.80816709774872,
      "learning_rate": 3.2474754095471796e-07,
      "loss": 0.3244,
      "step": 63525
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.6847682122815955,
      "learning_rate": 3.243112298424489e-07,
      "loss": 0.3007,
      "step": 63530
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.8848673786384422,
      "learning_rate": 3.2387520220341753e-07,
      "loss": 0.335,
      "step": 63535
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.9410146728325555,
      "learning_rate": 3.2343945806405875e-07,
      "loss": 0.3207,
      "step": 63540
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.90522845987869,
      "learning_rate": 3.2300399745079037e-07,
      "loss": 0.3254,
      "step": 63545
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.26840163288536,
      "learning_rate": 3.2256882039001237e-07,
      "loss": 0.3075,
      "step": 63550
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.065886810523565,
      "learning_rate": 3.221339269081103e-07,
      "loss": 0.2964,
      "step": 63555
    },
    {
      "epoch": 2.69,
      "grad_norm": 5.45058027763528,
      "learning_rate": 3.216993170314481e-07,
      "loss": 0.3121,
      "step": 63560
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.7060357052530755,
      "learning_rate": 3.2126499078637576e-07,
      "loss": 0.2936,
      "step": 63565
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.055878240936127,
      "learning_rate": 3.2083094819922553e-07,
      "loss": 0.3215,
      "step": 63570
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.8228959847870874,
      "learning_rate": 3.2039718929631256e-07,
      "loss": 0.2993,
      "step": 63575
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.9934887468671945,
      "learning_rate": 3.199637141039336e-07,
      "loss": 0.3202,
      "step": 63580
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.804194201095336,
      "learning_rate": 3.195305226483686e-07,
      "loss": 0.3047,
      "step": 63585
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.5671889488205935,
      "learning_rate": 3.190976149558822e-07,
      "loss": 0.3186,
      "step": 63590
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.040845114664043,
      "learning_rate": 3.186649910527184e-07,
      "loss": 0.3369,
      "step": 63595
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.039201313262964,
      "learning_rate": 3.1823265096510737e-07,
      "loss": 0.3007,
      "step": 63600
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.883774306819197,
      "learning_rate": 3.178005947192603e-07,
      "loss": 0.3034,
      "step": 63605
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.800042071543404,
      "learning_rate": 3.173688223413707e-07,
      "loss": 0.3014,
      "step": 63610
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.1273685394802335,
      "learning_rate": 3.169373338576165e-07,
      "loss": 0.3154,
      "step": 63615
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.789935603158744,
      "learning_rate": 3.165061292941568e-07,
      "loss": 0.3107,
      "step": 63620
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.7359632488317276,
      "learning_rate": 3.1607520867713623e-07,
      "loss": 0.322,
      "step": 63625
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.168455923392447,
      "learning_rate": 3.1564457203267784e-07,
      "loss": 0.3198,
      "step": 63630
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.690063748548547,
      "learning_rate": 3.152142193868907e-07,
      "loss": 0.298,
      "step": 63635
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.042027342232877,
      "learning_rate": 3.1478415076586663e-07,
      "loss": 0.3277,
      "step": 63640
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.9779070734197153,
      "learning_rate": 3.143543661956783e-07,
      "loss": 0.3189,
      "step": 63645
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.8850518679923915,
      "learning_rate": 3.1392486570238303e-07,
      "loss": 0.3061,
      "step": 63650
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.69665094964033,
      "learning_rate": 3.134956493120195e-07,
      "loss": 0.3263,
      "step": 63655
    },
    {
      "epoch": 2.69,
      "grad_norm": 3.5573779177008027,
      "learning_rate": 3.1306671705061073e-07,
      "loss": 0.2985,
      "step": 63660
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.113197805901236,
      "learning_rate": 3.1263806894416047e-07,
      "loss": 0.3201,
      "step": 63665
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.17981287054807,
      "learning_rate": 3.1220970501865675e-07,
      "loss": 0.3157,
      "step": 63670
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.263370874042329,
      "learning_rate": 3.117816253000711e-07,
      "loss": 0.3044,
      "step": 63675
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.7341386768023512,
      "learning_rate": 3.113538298143554e-07,
      "loss": 0.3103,
      "step": 63680
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.621293475813652,
      "learning_rate": 3.109263185874462e-07,
      "loss": 0.3096,
      "step": 63685
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.9956075230829926,
      "learning_rate": 3.1049909164526273e-07,
      "loss": 0.3104,
      "step": 63690
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.674613369724306,
      "learning_rate": 3.1007214901370487e-07,
      "loss": 0.3124,
      "step": 63695
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8471102855492756,
      "learning_rate": 3.0964549071865856e-07,
      "loss": 0.2803,
      "step": 63700
    },
    {
      "epoch": 2.7,
      "grad_norm": 5.558912531651892,
      "learning_rate": 3.092191167859898e-07,
      "loss": 0.337,
      "step": 63705
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.679136639045699,
      "learning_rate": 3.0879302724155016e-07,
      "loss": 0.3254,
      "step": 63710
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.9861287914206542,
      "learning_rate": 3.083672221111694e-07,
      "loss": 0.302,
      "step": 63715
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8216308714475997,
      "learning_rate": 3.079417014206648e-07,
      "loss": 0.305,
      "step": 63720
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.315489527546183,
      "learning_rate": 3.075164651958346e-07,
      "loss": 0.2937,
      "step": 63725
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.762818044416501,
      "learning_rate": 3.0709151346245803e-07,
      "loss": 0.3154,
      "step": 63730
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.3883082312070987,
      "learning_rate": 3.0666684624630016e-07,
      "loss": 0.293,
      "step": 63735
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.106198318346511,
      "learning_rate": 3.0624246357310705e-07,
      "loss": 0.2911,
      "step": 63740
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.292953244889915,
      "learning_rate": 3.0581836546860644e-07,
      "loss": 0.3344,
      "step": 63745
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8848897592786416,
      "learning_rate": 3.0539455195851164e-07,
      "loss": 0.3049,
      "step": 63750
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.491949522522433,
      "learning_rate": 3.0497102306851656e-07,
      "loss": 0.2986,
      "step": 63755
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8630005720322527,
      "learning_rate": 3.0454777882429953e-07,
      "loss": 0.3098,
      "step": 63760
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.057622721833372,
      "learning_rate": 3.0412481925151893e-07,
      "loss": 0.3227,
      "step": 63765
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8460417925757944,
      "learning_rate": 3.0370214437581814e-07,
      "loss": 0.3147,
      "step": 63770
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8354884947505252,
      "learning_rate": 3.0327975422282383e-07,
      "loss": 0.3095,
      "step": 63775
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.317436466762637,
      "learning_rate": 3.028576488181428e-07,
      "loss": 0.3009,
      "step": 63780
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.221063406665968,
      "learning_rate": 3.024358281873663e-07,
      "loss": 0.3082,
      "step": 63785
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.6902844225620437,
      "learning_rate": 3.0201429235606983e-07,
      "loss": 0.2949,
      "step": 63790
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.618086436097366,
      "learning_rate": 3.0159304134980693e-07,
      "loss": 0.3236,
      "step": 63795
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.885467391342991,
      "learning_rate": 3.0117207519411883e-07,
      "loss": 0.3374,
      "step": 63800
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.391157750167784,
      "learning_rate": 3.0075139391452677e-07,
      "loss": 0.2929,
      "step": 63805
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.473366924779301,
      "learning_rate": 3.0033099753653704e-07,
      "loss": 0.2993,
      "step": 63810
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.8817423115897958,
      "learning_rate": 2.999108860856342e-07,
      "loss": 0.3119,
      "step": 63815
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.655230148340203,
      "learning_rate": 2.994910595872896e-07,
      "loss": 0.308,
      "step": 63820
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.739944001879175,
      "learning_rate": 2.990715180669562e-07,
      "loss": 0.3047,
      "step": 63825
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.7786235098532286,
      "learning_rate": 2.9865226155007086e-07,
      "loss": 0.3041,
      "step": 63830
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.4775167018495274,
      "learning_rate": 2.9823329006204936e-07,
      "loss": 0.3101,
      "step": 63835
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.915246478684711,
      "learning_rate": 2.9781460362829417e-07,
      "loss": 0.3078,
      "step": 63840
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.965063873760154,
      "learning_rate": 2.9739620227418995e-07,
      "loss": 0.3014,
      "step": 63845
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.291054859735662,
      "learning_rate": 2.969780860251009e-07,
      "loss": 0.3085,
      "step": 63850
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.810278348940488,
      "learning_rate": 2.965602549063773e-07,
      "loss": 0.2977,
      "step": 63855
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.57645031035686,
      "learning_rate": 2.9614270894335164e-07,
      "loss": 0.3113,
      "step": 63860
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.9215543524963117,
      "learning_rate": 2.957254481613375e-07,
      "loss": 0.2989,
      "step": 63865
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.7947664605093596,
      "learning_rate": 2.95308472585632e-07,
      "loss": 0.3032,
      "step": 63870
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.888267428395147,
      "learning_rate": 2.9489178224151595e-07,
      "loss": 0.3177,
      "step": 63875
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.872568078772845,
      "learning_rate": 2.9447537715425247e-07,
      "loss": 0.3018,
      "step": 63880
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.441106636748337,
      "learning_rate": 2.940592573490858e-07,
      "loss": 0.3079,
      "step": 63885
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.2215215301710725,
      "learning_rate": 2.936434228512447e-07,
      "loss": 0.3049,
      "step": 63890
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.267603528270491,
      "learning_rate": 2.932278736859401e-07,
      "loss": 0.3027,
      "step": 63895
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.051962348831266,
      "learning_rate": 2.9281260987836456e-07,
      "loss": 0.3294,
      "step": 63900
    },
    {
      "epoch": 2.7,
      "grad_norm": 3.438990742707771,
      "learning_rate": 2.923976314536953e-07,
      "loss": 0.2922,
      "step": 63905
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.263088562824927,
      "learning_rate": 2.9198293843709146e-07,
      "loss": 0.3219,
      "step": 63910
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.431371288156548,
      "learning_rate": 2.915685308536936e-07,
      "loss": 0.2958,
      "step": 63915
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.875489819547162,
      "learning_rate": 2.911544087286261e-07,
      "loss": 0.3217,
      "step": 63920
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.8687520081945346,
      "learning_rate": 2.9074057208699714e-07,
      "loss": 0.3095,
      "step": 63925
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.772740514612205,
      "learning_rate": 2.903270209538961e-07,
      "loss": 0.3387,
      "step": 63930
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.926257533006478,
      "learning_rate": 2.8991375535439403e-07,
      "loss": 0.3297,
      "step": 63935
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.240385828952048,
      "learning_rate": 2.895007753135476e-07,
      "loss": 0.32,
      "step": 63940
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.230984729912234,
      "learning_rate": 2.8908808085639396e-07,
      "loss": 0.3268,
      "step": 63945
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.21158824567527,
      "learning_rate": 2.886756720079531e-07,
      "loss": 0.3102,
      "step": 63950
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.8929001628293403,
      "learning_rate": 2.882635487932284e-07,
      "loss": 0.312,
      "step": 63955
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.077681018342183,
      "learning_rate": 2.87851711237207e-07,
      "loss": 0.2767,
      "step": 63960
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.0806386262009315,
      "learning_rate": 2.8744015936485513e-07,
      "loss": 0.3158,
      "step": 63965
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.885086719153386,
      "learning_rate": 2.870288932011256e-07,
      "loss": 0.3579,
      "step": 63970
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.149216587724372,
      "learning_rate": 2.8661791277095176e-07,
      "loss": 0.2599,
      "step": 63975
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.56302701011447,
      "learning_rate": 2.862072180992509e-07,
      "loss": 0.2955,
      "step": 63980
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.566227830630732,
      "learning_rate": 2.8579680921092036e-07,
      "loss": 0.3004,
      "step": 63985
    },
    {
      "epoch": 2.71,
      "grad_norm": 12.124329961324188,
      "learning_rate": 2.8538668613084297e-07,
      "loss": 0.334,
      "step": 63990
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.181377837810836,
      "learning_rate": 2.8497684888388444e-07,
      "loss": 0.3674,
      "step": 63995
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.7458388581051283,
      "learning_rate": 2.8456729749489044e-07,
      "loss": 0.3172,
      "step": 64000
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.094997749220291,
      "learning_rate": 2.841580319886911e-07,
      "loss": 0.2993,
      "step": 64005
    },
    {
      "epoch": 2.71,
      "grad_norm": 5.3314382245235725,
      "learning_rate": 2.837490523900993e-07,
      "loss": 0.3216,
      "step": 64010
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.733429516035325,
      "learning_rate": 2.8334035872391085e-07,
      "loss": 0.3118,
      "step": 64015
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.53270812501522,
      "learning_rate": 2.829319510149031e-07,
      "loss": 0.315,
      "step": 64020
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.171464368760084,
      "learning_rate": 2.825238292878357e-07,
      "loss": 0.3366,
      "step": 64025
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.2957495826062075,
      "learning_rate": 2.821159935674539e-07,
      "loss": 0.282,
      "step": 64030
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.749558518378707,
      "learning_rate": 2.817084438784817e-07,
      "loss": 0.2771,
      "step": 64035
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.141913530580697,
      "learning_rate": 2.8130118024562833e-07,
      "loss": 0.3244,
      "step": 64040
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.680250779544887,
      "learning_rate": 2.808942026935857e-07,
      "loss": 0.3179,
      "step": 64045
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.601873442086598,
      "learning_rate": 2.804875112470268e-07,
      "loss": 0.3176,
      "step": 64050
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.8325538786766886,
      "learning_rate": 2.8008110593060745e-07,
      "loss": 0.2939,
      "step": 64055
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.284659842794704,
      "learning_rate": 2.79674986768968e-07,
      "loss": 0.3084,
      "step": 64060
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.835472088122396,
      "learning_rate": 2.792691537867309e-07,
      "loss": 0.3154,
      "step": 64065
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.130691298992677,
      "learning_rate": 2.788636070084988e-07,
      "loss": 0.3084,
      "step": 64070
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.9739039036133477,
      "learning_rate": 2.7845834645885914e-07,
      "loss": 0.3288,
      "step": 64075
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.558133060328146,
      "learning_rate": 2.780533721623835e-07,
      "loss": 0.2906,
      "step": 64080
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.9499472539613554,
      "learning_rate": 2.776486841436221e-07,
      "loss": 0.3029,
      "step": 64085
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.142574396307065,
      "learning_rate": 2.77244282427111e-07,
      "loss": 0.2922,
      "step": 64090
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.312668281480161,
      "learning_rate": 2.768401670373683e-07,
      "loss": 0.2938,
      "step": 64095
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.444615380743153,
      "learning_rate": 2.7643633799889267e-07,
      "loss": 0.3045,
      "step": 64100
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.568110770735827,
      "learning_rate": 2.760327953361686e-07,
      "loss": 0.3121,
      "step": 64105
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.574518144220236,
      "learning_rate": 2.7562953907366075e-07,
      "loss": 0.3422,
      "step": 64110
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.696754118138441,
      "learning_rate": 2.7522656923581917e-07,
      "loss": 0.3015,
      "step": 64115
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.490902359365186,
      "learning_rate": 2.748238858470725e-07,
      "loss": 0.2962,
      "step": 64120
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.9820118415000927,
      "learning_rate": 2.7442148893183475e-07,
      "loss": 0.3385,
      "step": 64125
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.107855796688776,
      "learning_rate": 2.74019378514504e-07,
      "loss": 0.3234,
      "step": 64130
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.5448266934766477,
      "learning_rate": 2.7361755461945696e-07,
      "loss": 0.29,
      "step": 64135
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.27660859517952,
      "learning_rate": 2.7321601727105463e-07,
      "loss": 0.357,
      "step": 64140
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.967608413207996,
      "learning_rate": 2.728147664936426e-07,
      "loss": 0.2865,
      "step": 64145
    },
    {
      "epoch": 2.71,
      "grad_norm": 3.890745686730736,
      "learning_rate": 2.7241380231154747e-07,
      "loss": 0.2926,
      "step": 64150
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9010254805835896,
      "learning_rate": 2.7201312474907704e-07,
      "loss": 0.2956,
      "step": 64155
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.160936068790499,
      "learning_rate": 2.716127338305241e-07,
      "loss": 0.318,
      "step": 64160
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.7563094490898585,
      "learning_rate": 2.712126295801637e-07,
      "loss": 0.2836,
      "step": 64165
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.989752569168498,
      "learning_rate": 2.7081281202225194e-07,
      "loss": 0.296,
      "step": 64170
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.8126943237969533,
      "learning_rate": 2.704132811810295e-07,
      "loss": 0.2916,
      "step": 64175
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.455004424296066,
      "learning_rate": 2.700140370807175e-07,
      "loss": 0.3167,
      "step": 64180
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9453780016771365,
      "learning_rate": 2.6961507974552324e-07,
      "loss": 0.3314,
      "step": 64185
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9898028121243305,
      "learning_rate": 2.6921640919963187e-07,
      "loss": 0.323,
      "step": 64190
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.5783949299413202,
      "learning_rate": 2.6881802546721515e-07,
      "loss": 0.3147,
      "step": 64195
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9061295852107416,
      "learning_rate": 2.6841992857242536e-07,
      "loss": 0.2983,
      "step": 64200
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.287263467894012,
      "learning_rate": 2.680221185393977e-07,
      "loss": 0.2979,
      "step": 64205
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.571390822266338,
      "learning_rate": 2.6762459539225016e-07,
      "loss": 0.294,
      "step": 64210
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.431756108399277,
      "learning_rate": 2.67227359155085e-07,
      "loss": 0.3154,
      "step": 64215
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.402563928373287,
      "learning_rate": 2.668304098519831e-07,
      "loss": 0.3289,
      "step": 64220
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.038070451843523,
      "learning_rate": 2.6643374750701166e-07,
      "loss": 0.2962,
      "step": 64225
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.3328286436348735,
      "learning_rate": 2.6603737214421886e-07,
      "loss": 0.3235,
      "step": 64230
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.69674049914878,
      "learning_rate": 2.6564128378763596e-07,
      "loss": 0.3073,
      "step": 64235
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.842500347970012,
      "learning_rate": 2.6524548246127656e-07,
      "loss": 0.3345,
      "step": 64240
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.8257964238814477,
      "learning_rate": 2.6484996818913646e-07,
      "loss": 0.3273,
      "step": 64245
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.500263008020389,
      "learning_rate": 2.6445474099519585e-07,
      "loss": 0.3208,
      "step": 64250
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.7222006523872215,
      "learning_rate": 2.6405980090341454e-07,
      "loss": 0.3129,
      "step": 64255
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9225322574897126,
      "learning_rate": 2.636651479377378e-07,
      "loss": 0.3389,
      "step": 64260
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.722188585215918,
      "learning_rate": 2.63270782122092e-07,
      "loss": 0.3019,
      "step": 64265
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9322755379967314,
      "learning_rate": 2.628767034803853e-07,
      "loss": 0.3186,
      "step": 64270
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.937987452862827,
      "learning_rate": 2.6248291203651075e-07,
      "loss": 0.3139,
      "step": 64275
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.032752321367245,
      "learning_rate": 2.620894078143432e-07,
      "loss": 0.3148,
      "step": 64280
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.82378609006923,
      "learning_rate": 2.61696190837738e-07,
      "loss": 0.3241,
      "step": 64285
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.259701101212652,
      "learning_rate": 2.613032611305355e-07,
      "loss": 0.3158,
      "step": 64290
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.168396664554637,
      "learning_rate": 2.609106187165589e-07,
      "loss": 0.3012,
      "step": 64295
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9894103113593054,
      "learning_rate": 2.6051826361961186e-07,
      "loss": 0.3032,
      "step": 64300
    },
    {
      "epoch": 2.72,
      "grad_norm": 5.291686441602706,
      "learning_rate": 2.6012619586348154e-07,
      "loss": 0.3263,
      "step": 64305
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.987501901596429,
      "learning_rate": 2.597344154719378e-07,
      "loss": 0.3076,
      "step": 64310
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.9094691174905223,
      "learning_rate": 2.5934292246873385e-07,
      "loss": 0.2997,
      "step": 64315
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.545985202038435,
      "learning_rate": 2.589517168776051e-07,
      "loss": 0.3266,
      "step": 64320
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.8011610258727857,
      "learning_rate": 2.585607987222677e-07,
      "loss": 0.2872,
      "step": 64325
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.183545442755828,
      "learning_rate": 2.5817016802642257e-07,
      "loss": 0.3057,
      "step": 64330
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.194145374456808,
      "learning_rate": 2.577798248137536e-07,
      "loss": 0.2935,
      "step": 64335
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.755203623845972,
      "learning_rate": 2.5738976910792466e-07,
      "loss": 0.333,
      "step": 64340
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.860562504345379,
      "learning_rate": 2.5700000093258405e-07,
      "loss": 0.2935,
      "step": 64345
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.742879755861836,
      "learning_rate": 2.566105203113628e-07,
      "loss": 0.3379,
      "step": 64350
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.006173990072811,
      "learning_rate": 2.562213272678732e-07,
      "loss": 0.3018,
      "step": 64355
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.726284372634893,
      "learning_rate": 2.558324218257113e-07,
      "loss": 0.303,
      "step": 64360
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.054091024148938,
      "learning_rate": 2.5544380400845503e-07,
      "loss": 0.3637,
      "step": 64365
    },
    {
      "epoch": 2.72,
      "grad_norm": 5.088400788246344,
      "learning_rate": 2.55055473839666e-07,
      "loss": 0.3242,
      "step": 64370
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.12334607574293,
      "learning_rate": 2.546674313428865e-07,
      "loss": 0.3186,
      "step": 64375
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.7972906843988308,
      "learning_rate": 2.5427967654164275e-07,
      "loss": 0.2977,
      "step": 64380
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.112105611123452,
      "learning_rate": 2.538922094594437e-07,
      "loss": 0.2937,
      "step": 64385
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.921388123884773,
      "learning_rate": 2.5350503011977943e-07,
      "loss": 0.3287,
      "step": 64390
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.7008722767529925,
      "learning_rate": 2.53118138546124e-07,
      "loss": 0.309,
      "step": 64395
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.864252584028144,
      "learning_rate": 2.527315347619341e-07,
      "loss": 0.3479,
      "step": 64400
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.52684586239443,
      "learning_rate": 2.523452187906467e-07,
      "loss": 0.3011,
      "step": 64405
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.97939542356285,
      "learning_rate": 2.519591906556845e-07,
      "loss": 0.3107,
      "step": 64410
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.821303620613166,
      "learning_rate": 2.5157345038045013e-07,
      "loss": 0.2879,
      "step": 64415
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.071821119369255,
      "learning_rate": 2.511879979883314e-07,
      "loss": 0.3146,
      "step": 64420
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.123936879652426,
      "learning_rate": 2.5080283350269574e-07,
      "loss": 0.3011,
      "step": 64425
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.4123132135864194,
      "learning_rate": 2.504179569468951e-07,
      "loss": 0.3009,
      "step": 64430
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.218098898980717,
      "learning_rate": 2.5003336834426406e-07,
      "loss": 0.3035,
      "step": 64435
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.420353727305965,
      "learning_rate": 2.4964906771811794e-07,
      "loss": 0.3019,
      "step": 64440
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.6111994363737314,
      "learning_rate": 2.492650550917558e-07,
      "loss": 0.2902,
      "step": 64445
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.756554923424346,
      "learning_rate": 2.488813304884602e-07,
      "loss": 0.3225,
      "step": 64450
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.9324622279898627,
      "learning_rate": 2.4849789393149416e-07,
      "loss": 0.2931,
      "step": 64455
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.8141876463297724,
      "learning_rate": 2.4811474544410575e-07,
      "loss": 0.3111,
      "step": 64460
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.564738202726435,
      "learning_rate": 2.477318850495225e-07,
      "loss": 0.3161,
      "step": 64465
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.202979457692879,
      "learning_rate": 2.47349312770957e-07,
      "loss": 0.3103,
      "step": 64470
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.23549596103264,
      "learning_rate": 2.4696702863160284e-07,
      "loss": 0.3237,
      "step": 64475
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.6315628170521697,
      "learning_rate": 2.4658503265463764e-07,
      "loss": 0.3069,
      "step": 64480
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.9198042788291882,
      "learning_rate": 2.4620332486321954e-07,
      "loss": 0.2857,
      "step": 64485
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.794952269989331,
      "learning_rate": 2.458219052804922e-07,
      "loss": 0.3216,
      "step": 64490
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.056557395123123,
      "learning_rate": 2.454407739295783e-07,
      "loss": 0.3018,
      "step": 64495
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.850756434170652,
      "learning_rate": 2.4505993083358493e-07,
      "loss": 0.3389,
      "step": 64500
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.495598643573532,
      "learning_rate": 2.4467937601560246e-07,
      "loss": 0.2858,
      "step": 64505
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.924171363916104,
      "learning_rate": 2.4429910949870185e-07,
      "loss": 0.3061,
      "step": 64510
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.8522674518873288,
      "learning_rate": 2.4391913130593745e-07,
      "loss": 0.316,
      "step": 64515
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.097685024304629,
      "learning_rate": 2.435394414603476e-07,
      "loss": 0.3085,
      "step": 64520
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.8525635612875386,
      "learning_rate": 2.4316003998494987e-07,
      "loss": 0.3005,
      "step": 64525
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.6120401223088727,
      "learning_rate": 2.427809269027476e-07,
      "loss": 0.2933,
      "step": 64530
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.672660795565602,
      "learning_rate": 2.424021022367251e-07,
      "loss": 0.3233,
      "step": 64535
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.445882246179231,
      "learning_rate": 2.4202356600984857e-07,
      "loss": 0.2858,
      "step": 64540
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.8229755763777513,
      "learning_rate": 2.4164531824506845e-07,
      "loss": 0.3087,
      "step": 64545
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.582452960183164,
      "learning_rate": 2.412673589653164e-07,
      "loss": 0.3146,
      "step": 64550
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.221715476220587,
      "learning_rate": 2.4088968819350744e-07,
      "loss": 0.3159,
      "step": 64555
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.8613121891840128,
      "learning_rate": 2.405123059525377e-07,
      "loss": 0.2918,
      "step": 64560
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.9518270360115912,
      "learning_rate": 2.401352122652878e-07,
      "loss": 0.3185,
      "step": 64565
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.8450482787372677,
      "learning_rate": 2.397584071546194e-07,
      "loss": 0.3195,
      "step": 64570
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.024476699641109,
      "learning_rate": 2.3938189064337647e-07,
      "loss": 0.2673,
      "step": 64575
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.292608082018545,
      "learning_rate": 2.3900566275438687e-07,
      "loss": 0.3273,
      "step": 64580
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.567751739106351,
      "learning_rate": 2.3862972351046066e-07,
      "loss": 0.2948,
      "step": 64585
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.673519355701563,
      "learning_rate": 2.3825407293438852e-07,
      "loss": 0.3093,
      "step": 64590
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.120359961131146,
      "learning_rate": 2.3787871104894555e-07,
      "loss": 0.3291,
      "step": 64595
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.067017270540312,
      "learning_rate": 2.3750363787688913e-07,
      "loss": 0.3281,
      "step": 64600
    },
    {
      "epoch": 2.73,
      "grad_norm": 3.6180085399876423,
      "learning_rate": 2.3712885344095937e-07,
      "loss": 0.2988,
      "step": 64605
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.112850684993914,
      "learning_rate": 2.36754357763877e-07,
      "loss": 0.3278,
      "step": 64610
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.225845852020629,
      "learning_rate": 2.3638015086834776e-07,
      "loss": 0.303,
      "step": 64615
    },
    {
      "epoch": 2.73,
      "grad_norm": 4.22012475300299,
      "learning_rate": 2.3600623277705847e-07,
      "loss": 0.3212,
      "step": 64620
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.5061510264771174,
      "learning_rate": 2.356326035126788e-07,
      "loss": 0.3032,
      "step": 64625
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.036888798018336,
      "learning_rate": 2.3525926309785952e-07,
      "loss": 0.2974,
      "step": 64630
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.637103489756905,
      "learning_rate": 2.348862115552364e-07,
      "loss": 0.3076,
      "step": 64635
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.6595033787501676,
      "learning_rate": 2.345134489074269e-07,
      "loss": 0.3082,
      "step": 64640
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.310421670403091,
      "learning_rate": 2.341409751770285e-07,
      "loss": 0.3324,
      "step": 64645
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.5662237564473487,
      "learning_rate": 2.3376879038662537e-07,
      "loss": 0.3265,
      "step": 64650
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.02649492543004,
      "learning_rate": 2.3339689455878112e-07,
      "loss": 0.3254,
      "step": 64655
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.772013746437683,
      "learning_rate": 2.3302528771604215e-07,
      "loss": 0.3189,
      "step": 64660
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.33927275870982,
      "learning_rate": 2.3265396988093825e-07,
      "loss": 0.3043,
      "step": 64665
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.7312077946729194,
      "learning_rate": 2.3228294107598194e-07,
      "loss": 0.3222,
      "step": 64670
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.7443999843563125,
      "learning_rate": 2.3191220132366744e-07,
      "loss": 0.2795,
      "step": 64675
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.9666098795834506,
      "learning_rate": 2.3154175064647068e-07,
      "loss": 0.3107,
      "step": 64680
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.798754018011551,
      "learning_rate": 2.3117158906685143e-07,
      "loss": 0.312,
      "step": 64685
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.98040172538192,
      "learning_rate": 2.3080171660725293e-07,
      "loss": 0.3448,
      "step": 64690
    },
    {
      "epoch": 2.74,
      "grad_norm": 6.35679047414917,
      "learning_rate": 2.3043213329009718e-07,
      "loss": 0.3238,
      "step": 64695
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.9613800380209945,
      "learning_rate": 2.300628391377918e-07,
      "loss": 0.3077,
      "step": 64700
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.833436524968185,
      "learning_rate": 2.2969383417272672e-07,
      "loss": 0.3556,
      "step": 64705
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.7819657438978176,
      "learning_rate": 2.293251184172729e-07,
      "loss": 0.3088,
      "step": 64710
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.70056909962587,
      "learning_rate": 2.2895669189378466e-07,
      "loss": 0.3223,
      "step": 64715
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.172014618854703,
      "learning_rate": 2.2858855462459805e-07,
      "loss": 0.2963,
      "step": 64720
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.601684356983415,
      "learning_rate": 2.282207066320341e-07,
      "loss": 0.2841,
      "step": 64725
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.8215403593735107,
      "learning_rate": 2.278531479383922e-07,
      "loss": 0.333,
      "step": 64730
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.8229177013126985,
      "learning_rate": 2.2748587856595672e-07,
      "loss": 0.2964,
      "step": 64735
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.925452980140284,
      "learning_rate": 2.2711889853699542e-07,
      "loss": 0.2797,
      "step": 64740
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.5425670316783058,
      "learning_rate": 2.2675220787375551e-07,
      "loss": 0.2958,
      "step": 64745
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.876709846869579,
      "learning_rate": 2.263858065984692e-07,
      "loss": 0.3042,
      "step": 64750
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.8287541255722006,
      "learning_rate": 2.2601969473335094e-07,
      "loss": 0.3377,
      "step": 64755
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.6659429320113706,
      "learning_rate": 2.2565387230059577e-07,
      "loss": 0.329,
      "step": 64760
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.085625882507761,
      "learning_rate": 2.2528833932238313e-07,
      "loss": 0.3334,
      "step": 64765
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.582345753272541,
      "learning_rate": 2.2492309582087425e-07,
      "loss": 0.3003,
      "step": 64770
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.9866021643814826,
      "learning_rate": 2.24558141818213e-07,
      "loss": 0.2933,
      "step": 64775
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.087044199785954,
      "learning_rate": 2.241934773365251e-07,
      "loss": 0.2991,
      "step": 64780
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.16247034728151,
      "learning_rate": 2.2382910239791833e-07,
      "loss": 0.2895,
      "step": 64785
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.705156902317562,
      "learning_rate": 2.2346501702448452e-07,
      "loss": 0.2844,
      "step": 64790
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.584680425294526,
      "learning_rate": 2.231012212382977e-07,
      "loss": 0.3188,
      "step": 64795
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.59399050087571,
      "learning_rate": 2.227377150614124e-07,
      "loss": 0.34,
      "step": 64800
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.806615009214411,
      "learning_rate": 2.223744985158671e-07,
      "loss": 0.3034,
      "step": 64805
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.8082261288004027,
      "learning_rate": 2.2201157162368425e-07,
      "loss": 0.3048,
      "step": 64810
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.757452731394143,
      "learning_rate": 2.2164893440686508e-07,
      "loss": 0.3348,
      "step": 64815
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.405462319453972,
      "learning_rate": 2.2128658688739536e-07,
      "loss": 0.3249,
      "step": 64820
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.368184522266987,
      "learning_rate": 2.209245290872447e-07,
      "loss": 0.35,
      "step": 64825
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.1155682278962695,
      "learning_rate": 2.2056276102836228e-07,
      "loss": 0.3342,
      "step": 64830
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.193755988449308,
      "learning_rate": 2.2020128273268106e-07,
      "loss": 0.3227,
      "step": 64835
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.597634350879257,
      "learning_rate": 2.1984009422211738e-07,
      "loss": 0.2899,
      "step": 64840
    },
    {
      "epoch": 2.74,
      "grad_norm": 5.070330020651896,
      "learning_rate": 2.1947919551856767e-07,
      "loss": 0.3221,
      "step": 64845
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.74117832970566,
      "learning_rate": 2.1911858664391272e-07,
      "loss": 0.3093,
      "step": 64850
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.727513319194406,
      "learning_rate": 2.1875826762001562e-07,
      "loss": 0.3076,
      "step": 64855
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.486668113847009,
      "learning_rate": 2.1839823846872165e-07,
      "loss": 0.3041,
      "step": 64860
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.466974263282053,
      "learning_rate": 2.1803849921185728e-07,
      "loss": 0.3075,
      "step": 64865
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.7686271749194256,
      "learning_rate": 2.1767904987123278e-07,
      "loss": 0.2973,
      "step": 64870
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.333694984162197,
      "learning_rate": 2.1731989046864132e-07,
      "loss": 0.2977,
      "step": 64875
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.9172145269028964,
      "learning_rate": 2.1696102102585658e-07,
      "loss": 0.3025,
      "step": 64880
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8777319184268446,
      "learning_rate": 2.1660244156463617e-07,
      "loss": 0.3135,
      "step": 64885
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.7940391432218656,
      "learning_rate": 2.1624415210672044e-07,
      "loss": 0.2825,
      "step": 64890
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.127851168880768,
      "learning_rate": 2.158861526738304e-07,
      "loss": 0.3014,
      "step": 64895
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.819128409808455,
      "learning_rate": 2.155284432876703e-07,
      "loss": 0.2963,
      "step": 64900
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8351265349169963,
      "learning_rate": 2.1517102396992784e-07,
      "loss": 0.296,
      "step": 64905
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.715909265681781,
      "learning_rate": 2.1481389474227233e-07,
      "loss": 0.2951,
      "step": 64910
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8858493772977396,
      "learning_rate": 2.1445705562635478e-07,
      "loss": 0.3117,
      "step": 64915
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8750300151212236,
      "learning_rate": 2.141005066438101e-07,
      "loss": 0.3129,
      "step": 64920
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.95500372894167,
      "learning_rate": 2.1374424781625434e-07,
      "loss": 0.3237,
      "step": 64925
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.142952344537536,
      "learning_rate": 2.1338827916528637e-07,
      "loss": 0.2912,
      "step": 64930
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.6888942906917026,
      "learning_rate": 2.1303260071248722e-07,
      "loss": 0.3043,
      "step": 64935
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.243586291605428,
      "learning_rate": 2.1267721247942186e-07,
      "loss": 0.3156,
      "step": 64940
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8653251124915475,
      "learning_rate": 2.1232211448763528e-07,
      "loss": 0.3046,
      "step": 64945
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.200824789281138,
      "learning_rate": 2.1196730675865584e-07,
      "loss": 0.3299,
      "step": 64950
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.193963979984511,
      "learning_rate": 2.116127893139952e-07,
      "loss": 0.3008,
      "step": 64955
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.9576133520488224,
      "learning_rate": 2.1125856217514672e-07,
      "loss": 0.3116,
      "step": 64960
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.754620424217938,
      "learning_rate": 2.1090462536358546e-07,
      "loss": 0.3351,
      "step": 64965
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.912268731083,
      "learning_rate": 2.1055097890077038e-07,
      "loss": 0.2941,
      "step": 64970
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.9356487192506915,
      "learning_rate": 2.1019762280814148e-07,
      "loss": 0.3755,
      "step": 64975
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8753060000683717,
      "learning_rate": 2.0984455710712226e-07,
      "loss": 0.2954,
      "step": 64980
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.9960872331101203,
      "learning_rate": 2.0949178181911724e-07,
      "loss": 0.3063,
      "step": 64985
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.514092569433876,
      "learning_rate": 2.0913929696551427e-07,
      "loss": 0.2872,
      "step": 64990
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.106368768150705,
      "learning_rate": 2.0878710256768463e-07,
      "loss": 0.2855,
      "step": 64995
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8336283186113413,
      "learning_rate": 2.084351986469796e-07,
      "loss": 0.2803,
      "step": 65000
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.489162892944125,
      "learning_rate": 2.0808358522473425e-07,
      "loss": 0.3688,
      "step": 65005
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.899764268480078,
      "learning_rate": 2.0773226232226662e-07,
      "loss": 0.3455,
      "step": 65010
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.7446596683305065,
      "learning_rate": 2.073812299608752e-07,
      "loss": 0.2892,
      "step": 65015
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.278141365472454,
      "learning_rate": 2.070304881618429e-07,
      "loss": 0.3179,
      "step": 65020
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.886493087314849,
      "learning_rate": 2.066800369464339e-07,
      "loss": 0.3118,
      "step": 65025
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.872922585901304,
      "learning_rate": 2.0632987633589562e-07,
      "loss": 0.3075,
      "step": 65030
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8515226226935613,
      "learning_rate": 2.059800063514561e-07,
      "loss": 0.3153,
      "step": 65035
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.768941911392484,
      "learning_rate": 2.0563042701432723e-07,
      "loss": 0.2825,
      "step": 65040
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.021608317409099,
      "learning_rate": 2.052811383457043e-07,
      "loss": 0.3282,
      "step": 65045
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8046871340060346,
      "learning_rate": 2.0493214036676201e-07,
      "loss": 0.2868,
      "step": 65050
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.9487338933842646,
      "learning_rate": 2.0458343309865957e-07,
      "loss": 0.3142,
      "step": 65055
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.873583777965302,
      "learning_rate": 2.0423501656253897e-07,
      "loss": 0.3093,
      "step": 65060
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.9082790773882095,
      "learning_rate": 2.0388689077952217e-07,
      "loss": 0.3187,
      "step": 65065
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.955452043134012,
      "learning_rate": 2.035390557707162e-07,
      "loss": 0.3173,
      "step": 65070
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.964901306990434,
      "learning_rate": 2.0319151155720806e-07,
      "loss": 0.3058,
      "step": 65075
    },
    {
      "epoch": 2.75,
      "grad_norm": 5.269240709116969,
      "learning_rate": 2.0284425816007035e-07,
      "loss": 0.3127,
      "step": 65080
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.8676326939416543,
      "learning_rate": 2.0249729560035403e-07,
      "loss": 0.3114,
      "step": 65085
    },
    {
      "epoch": 2.75,
      "grad_norm": 3.698020681983677,
      "learning_rate": 2.02150623899095e-07,
      "loss": 0.3054,
      "step": 65090
    },
    {
      "epoch": 2.75,
      "grad_norm": 4.4125433885599525,
      "learning_rate": 2.0180424307731207e-07,
      "loss": 0.3192,
      "step": 65095
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.883717733940784,
      "learning_rate": 2.0145815315600335e-07,
      "loss": 0.3104,
      "step": 65100
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.842755776682679,
      "learning_rate": 2.0111235415615327e-07,
      "loss": 0.2834,
      "step": 65105
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.7927718610114733,
      "learning_rate": 2.0076684609872443e-07,
      "loss": 0.3012,
      "step": 65110
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.4450468139065,
      "learning_rate": 2.0042162900466622e-07,
      "loss": 0.3263,
      "step": 65115
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.245525699240544,
      "learning_rate": 2.0007670289490577e-07,
      "loss": 0.2897,
      "step": 65120
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.032783752746433,
      "learning_rate": 1.9973206779035692e-07,
      "loss": 0.3081,
      "step": 65125
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.9465208964687735,
      "learning_rate": 1.993877237119135e-07,
      "loss": 0.3082,
      "step": 65130
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.165506205537388,
      "learning_rate": 1.9904367068045048e-07,
      "loss": 0.3035,
      "step": 65135
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.718943241187539,
      "learning_rate": 1.9869990871682898e-07,
      "loss": 0.3083,
      "step": 65140
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.9493668626786245,
      "learning_rate": 1.9835643784188953e-07,
      "loss": 0.2886,
      "step": 65145
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.065892744562885,
      "learning_rate": 1.9801325807645489e-07,
      "loss": 0.3044,
      "step": 65150
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.027141162226871,
      "learning_rate": 1.9767036944133178e-07,
      "loss": 0.3158,
      "step": 65155
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.056519483125561,
      "learning_rate": 1.97327771957308e-07,
      "loss": 0.3021,
      "step": 65160
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.8679624695620722,
      "learning_rate": 1.9698546564515586e-07,
      "loss": 0.323,
      "step": 65165
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.848463614312208,
      "learning_rate": 1.966434505256265e-07,
      "loss": 0.2999,
      "step": 65170
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.9796206759044996,
      "learning_rate": 1.9630172661945612e-07,
      "loss": 0.3077,
      "step": 65175
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.05645923036871,
      "learning_rate": 1.959602939473626e-07,
      "loss": 0.2726,
      "step": 65180
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.719844561251929,
      "learning_rate": 1.9561915253004494e-07,
      "loss": 0.3176,
      "step": 65185
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.6207441044336504,
      "learning_rate": 1.9527830238818656e-07,
      "loss": 0.3328,
      "step": 65190
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.9061851623110293,
      "learning_rate": 1.9493774354245265e-07,
      "loss": 0.3066,
      "step": 65195
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.3015199275416025,
      "learning_rate": 1.9459747601348888e-07,
      "loss": 0.3036,
      "step": 65200
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.404087921835684,
      "learning_rate": 1.9425749982192488e-07,
      "loss": 0.2907,
      "step": 65205
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.226711822729783,
      "learning_rate": 1.939178149883736e-07,
      "loss": 0.316,
      "step": 65210
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.287596681112214,
      "learning_rate": 1.9357842153342853e-07,
      "loss": 0.2958,
      "step": 65215
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.002288003678812,
      "learning_rate": 1.932393194776655e-07,
      "loss": 0.2882,
      "step": 65220
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.8234044516707755,
      "learning_rate": 1.9290050884164413e-07,
      "loss": 0.2955,
      "step": 65225
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.7499957524221186,
      "learning_rate": 1.9256198964590523e-07,
      "loss": 0.2914,
      "step": 65230
    },
    {
      "epoch": 2.76,
      "grad_norm": 5.948283643453769,
      "learning_rate": 1.922237619109718e-07,
      "loss": 0.3035,
      "step": 65235
    },
    {
      "epoch": 2.76,
      "grad_norm": 5.329173549498761,
      "learning_rate": 1.918858256573497e-07,
      "loss": 0.3262,
      "step": 65240
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.921624831727337,
      "learning_rate": 1.9154818090552807e-07,
      "loss": 0.3157,
      "step": 65245
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.98626161992799,
      "learning_rate": 1.9121082767597608e-07,
      "loss": 0.32,
      "step": 65250
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.811663425539556,
      "learning_rate": 1.9087376598914632e-07,
      "loss": 0.3005,
      "step": 65255
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.8054003858601866,
      "learning_rate": 1.905369958654746e-07,
      "loss": 0.3029,
      "step": 65260
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.79523018656876,
      "learning_rate": 1.9020051732537904e-07,
      "loss": 0.322,
      "step": 65265
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.9995917920528257,
      "learning_rate": 1.8986433038925777e-07,
      "loss": 0.3022,
      "step": 65270
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.29626271696926,
      "learning_rate": 1.8952843507749287e-07,
      "loss": 0.311,
      "step": 65275
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.409596273534369,
      "learning_rate": 1.8919283141044963e-07,
      "loss": 0.305,
      "step": 65280
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.6632270873831843,
      "learning_rate": 1.8885751940847464e-07,
      "loss": 0.3001,
      "step": 65285
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.954162200154615,
      "learning_rate": 1.8852249909189602e-07,
      "loss": 0.2908,
      "step": 65290
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.9853858249401344,
      "learning_rate": 1.8818777048102588e-07,
      "loss": 0.2976,
      "step": 65295
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.5511247496283236,
      "learning_rate": 1.8785333359615799e-07,
      "loss": 0.2792,
      "step": 65300
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.80671534926733,
      "learning_rate": 1.8751918845756722e-07,
      "loss": 0.3221,
      "step": 65305
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.060698877158665,
      "learning_rate": 1.8718533508551239e-07,
      "loss": 0.3086,
      "step": 65310
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.121792242192534,
      "learning_rate": 1.8685177350023453e-07,
      "loss": 0.3259,
      "step": 65315
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.1658854567631005,
      "learning_rate": 1.865185037219558e-07,
      "loss": 0.307,
      "step": 65320
    },
    {
      "epoch": 2.76,
      "grad_norm": 3.5710311581931364,
      "learning_rate": 1.8618552577088166e-07,
      "loss": 0.3151,
      "step": 65325
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.057308731654288,
      "learning_rate": 1.8585283966719937e-07,
      "loss": 0.306,
      "step": 65330
    },
    {
      "epoch": 2.77,
      "grad_norm": 5.292362732722109,
      "learning_rate": 1.8552044543107938e-07,
      "loss": 0.318,
      "step": 65335
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.00932084644621,
      "learning_rate": 1.851883430826723e-07,
      "loss": 0.3123,
      "step": 65340
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.7318512459579996,
      "learning_rate": 1.8485653264211423e-07,
      "loss": 0.2803,
      "step": 65345
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.9288826764931155,
      "learning_rate": 1.8452501412952128e-07,
      "loss": 0.3091,
      "step": 65350
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.621948399798097,
      "learning_rate": 1.8419378756499184e-07,
      "loss": 0.3066,
      "step": 65355
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.766212039988786,
      "learning_rate": 1.8386285296860762e-07,
      "loss": 0.2955,
      "step": 65360
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8817224663631773,
      "learning_rate": 1.8353221036043256e-07,
      "loss": 0.3032,
      "step": 65365
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8436266193523974,
      "learning_rate": 1.8320185976051175e-07,
      "loss": 0.3068,
      "step": 65370
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8610410563903272,
      "learning_rate": 1.8287180118887415e-07,
      "loss": 0.2979,
      "step": 65375
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.917937626245474,
      "learning_rate": 1.825420346655299e-07,
      "loss": 0.3072,
      "step": 65380
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.9542310461385424,
      "learning_rate": 1.8221256021047184e-07,
      "loss": 0.3187,
      "step": 65385
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.5735932898633,
      "learning_rate": 1.818833778436746e-07,
      "loss": 0.297,
      "step": 65390
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.7262538724710335,
      "learning_rate": 1.8155448758509664e-07,
      "loss": 0.2938,
      "step": 65395
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.9927699038225,
      "learning_rate": 1.8122588945467644e-07,
      "loss": 0.3275,
      "step": 65400
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.153100333832798,
      "learning_rate": 1.8089758347233644e-07,
      "loss": 0.3179,
      "step": 65405
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.60220193009997,
      "learning_rate": 1.8056956965798068e-07,
      "loss": 0.2853,
      "step": 65410
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8211872084671077,
      "learning_rate": 1.802418480314966e-07,
      "loss": 0.3179,
      "step": 65415
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.17942259917213,
      "learning_rate": 1.7991441861275161e-07,
      "loss": 0.3178,
      "step": 65420
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.780293457962462,
      "learning_rate": 1.795872814215971e-07,
      "loss": 0.3106,
      "step": 65425
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8410367166036727,
      "learning_rate": 1.7926043647786716e-07,
      "loss": 0.3299,
      "step": 65430
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.755185057538734,
      "learning_rate": 1.7893388380137766e-07,
      "loss": 0.3006,
      "step": 65435
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.202531761377634,
      "learning_rate": 1.7860762341192495e-07,
      "loss": 0.3002,
      "step": 65440
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.7466652284014947,
      "learning_rate": 1.7828165532928988e-07,
      "loss": 0.3022,
      "step": 65445
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.282603630565211,
      "learning_rate": 1.7795597957323607e-07,
      "loss": 0.3083,
      "step": 65450
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.856652818434062,
      "learning_rate": 1.7763059616350664e-07,
      "loss": 0.3113,
      "step": 65455
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.520296979901218,
      "learning_rate": 1.773055051198297e-07,
      "loss": 0.3241,
      "step": 65460
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.876866998330857,
      "learning_rate": 1.769807064619139e-07,
      "loss": 0.3186,
      "step": 65465
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.9258771650668027,
      "learning_rate": 1.766562002094524e-07,
      "loss": 0.3082,
      "step": 65470
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.8302538695348027,
      "learning_rate": 1.7633198638211668e-07,
      "loss": 0.3311,
      "step": 65475
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.424954356570667,
      "learning_rate": 1.7600806499956435e-07,
      "loss": 0.315,
      "step": 65480
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.050334123571713,
      "learning_rate": 1.756844360814336e-07,
      "loss": 0.2958,
      "step": 65485
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.226016854547005,
      "learning_rate": 1.7536109964734482e-07,
      "loss": 0.3016,
      "step": 65490
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.831013533798978,
      "learning_rate": 1.7503805571690123e-07,
      "loss": 0.3108,
      "step": 65495
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.959873930872243,
      "learning_rate": 1.747153043096883e-07,
      "loss": 0.3144,
      "step": 65500
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.860851315506426,
      "learning_rate": 1.7439284544527314e-07,
      "loss": 0.2884,
      "step": 65505
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.7337144701626515,
      "learning_rate": 1.7407067914320452e-07,
      "loss": 0.3008,
      "step": 65510
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.073854493257005,
      "learning_rate": 1.7374880542301576e-07,
      "loss": 0.293,
      "step": 65515
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.7936794615524163,
      "learning_rate": 1.7342722430422176e-07,
      "loss": 0.3107,
      "step": 65520
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.7129089803122053,
      "learning_rate": 1.7310593580631696e-07,
      "loss": 0.2924,
      "step": 65525
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.775658664583829,
      "learning_rate": 1.727849399487813e-07,
      "loss": 0.3031,
      "step": 65530
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.013941907497206,
      "learning_rate": 1.7246423675107649e-07,
      "loss": 0.3228,
      "step": 65535
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.342091431439759,
      "learning_rate": 1.7214382623264415e-07,
      "loss": 0.3245,
      "step": 65540
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.9285576854575237,
      "learning_rate": 1.7182370841291095e-07,
      "loss": 0.3272,
      "step": 65545
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.919250390237749,
      "learning_rate": 1.7150388331128476e-07,
      "loss": 0.3042,
      "step": 65550
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.135570087473103,
      "learning_rate": 1.7118435094715447e-07,
      "loss": 0.3125,
      "step": 65555
    },
    {
      "epoch": 2.77,
      "grad_norm": 4.172267531330611,
      "learning_rate": 1.7086511133989404e-07,
      "loss": 0.3081,
      "step": 65560
    },
    {
      "epoch": 2.77,
      "grad_norm": 3.9805558251982527,
      "learning_rate": 1.7054616450885685e-07,
      "loss": 0.3088,
      "step": 65565
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.5592148820168257,
      "learning_rate": 1.7022751047338027e-07,
      "loss": 0.3077,
      "step": 65570
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9655760415055767,
      "learning_rate": 1.6990914925278325e-07,
      "loss": 0.3018,
      "step": 65575
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.081779248704726,
      "learning_rate": 1.6959108086636645e-07,
      "loss": 0.3121,
      "step": 65580
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.8991638152931314,
      "learning_rate": 1.6927330533341502e-07,
      "loss": 0.3145,
      "step": 65585
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.8600334961310243,
      "learning_rate": 1.6895582267319355e-07,
      "loss": 0.286,
      "step": 65590
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.563222013298887,
      "learning_rate": 1.6863863290494943e-07,
      "loss": 0.318,
      "step": 65595
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.456289764630795,
      "learning_rate": 1.683217360479139e-07,
      "loss": 0.3167,
      "step": 65600
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.2320522502159115,
      "learning_rate": 1.680051321213e-07,
      "loss": 0.3386,
      "step": 65605
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.066092956413713,
      "learning_rate": 1.676888211443012e-07,
      "loss": 0.3123,
      "step": 65610
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.081743828024168,
      "learning_rate": 1.673728031360955e-07,
      "loss": 0.3325,
      "step": 65615
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.67229114479859,
      "learning_rate": 1.6705707811584203e-07,
      "loss": 0.3187,
      "step": 65620
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.7243555863784104,
      "learning_rate": 1.6674164610268108e-07,
      "loss": 0.2918,
      "step": 65625
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.731336288815053,
      "learning_rate": 1.664265071157378e-07,
      "loss": 0.2963,
      "step": 65630
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.003264025463793,
      "learning_rate": 1.661116611741176e-07,
      "loss": 0.3506,
      "step": 65635
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.009768694030956,
      "learning_rate": 1.6579710829690898e-07,
      "loss": 0.3222,
      "step": 65640
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.827382315581644,
      "learning_rate": 1.6548284850318175e-07,
      "loss": 0.2984,
      "step": 65645
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.873600253682237,
      "learning_rate": 1.6516888181198898e-07,
      "loss": 0.2863,
      "step": 65650
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9141922479332827,
      "learning_rate": 1.6485520824236544e-07,
      "loss": 0.3264,
      "step": 65655
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.064819200594836,
      "learning_rate": 1.6454182781332818e-07,
      "loss": 0.3382,
      "step": 65660
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.930905194379195,
      "learning_rate": 1.6422874054387638e-07,
      "loss": 0.2981,
      "step": 65665
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.859641525842305,
      "learning_rate": 1.639159464529927e-07,
      "loss": 0.2835,
      "step": 65670
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.8706396994931938,
      "learning_rate": 1.6360344555963913e-07,
      "loss": 0.3163,
      "step": 65675
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.867007564868154,
      "learning_rate": 1.632912378827628e-07,
      "loss": 0.3017,
      "step": 65680
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.8487713649721584,
      "learning_rate": 1.6297932344129186e-07,
      "loss": 0.3177,
      "step": 65685
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.153643470848459,
      "learning_rate": 1.6266770225413674e-07,
      "loss": 0.3082,
      "step": 65690
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.211692337238646,
      "learning_rate": 1.623563743401896e-07,
      "loss": 0.2948,
      "step": 65695
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.618626290524091,
      "learning_rate": 1.6204533971832582e-07,
      "loss": 0.2929,
      "step": 65700
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9566671135323013,
      "learning_rate": 1.6173459840740258e-07,
      "loss": 0.3221,
      "step": 65705
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.827249875879951,
      "learning_rate": 1.6142415042625926e-07,
      "loss": 0.3045,
      "step": 65710
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9594287758759457,
      "learning_rate": 1.6111399579371633e-07,
      "loss": 0.3018,
      "step": 65715
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.050963687525933,
      "learning_rate": 1.608041345285799e-07,
      "loss": 0.3087,
      "step": 65720
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.8859261335709756,
      "learning_rate": 1.6049456664963327e-07,
      "loss": 0.2882,
      "step": 65725
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9744083858635153,
      "learning_rate": 1.6018529217564584e-07,
      "loss": 0.3123,
      "step": 65730
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9780436089522957,
      "learning_rate": 1.5987631112536928e-07,
      "loss": 0.2863,
      "step": 65735
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.979728497719229,
      "learning_rate": 1.5956762351753364e-07,
      "loss": 0.3313,
      "step": 65740
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9462244109348212,
      "learning_rate": 1.5925922937085503e-07,
      "loss": 0.3087,
      "step": 65745
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.775861995879866,
      "learning_rate": 1.5895112870403072e-07,
      "loss": 0.2796,
      "step": 65750
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.288523663108847,
      "learning_rate": 1.586433215357408e-07,
      "loss": 0.2976,
      "step": 65755
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.7813724929829045,
      "learning_rate": 1.5833580788464477e-07,
      "loss": 0.2825,
      "step": 65760
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.802954394261705,
      "learning_rate": 1.5802858776938658e-07,
      "loss": 0.3055,
      "step": 65765
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.099488632156402,
      "learning_rate": 1.57721661208593e-07,
      "loss": 0.3045,
      "step": 65770
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.977325451614377,
      "learning_rate": 1.5741502822087195e-07,
      "loss": 0.3127,
      "step": 65775
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.164210107013999,
      "learning_rate": 1.57108688824813e-07,
      "loss": 0.289,
      "step": 65780
    },
    {
      "epoch": 2.78,
      "grad_norm": 4.019580638354942,
      "learning_rate": 1.5680264303898907e-07,
      "loss": 0.3396,
      "step": 65785
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.80698606149582,
      "learning_rate": 1.564968908819553e-07,
      "loss": 0.2983,
      "step": 65790
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.9040511640965962,
      "learning_rate": 1.5619143237224798e-07,
      "loss": 0.3108,
      "step": 65795
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.634743205608102,
      "learning_rate": 1.5588626752838565e-07,
      "loss": 0.2774,
      "step": 65800
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.8882672040483834,
      "learning_rate": 1.555813963688707e-07,
      "loss": 0.3166,
      "step": 65805
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.0784562119060705,
      "learning_rate": 1.552768189121856e-07,
      "loss": 0.3114,
      "step": 65810
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.9722743165088956,
      "learning_rate": 1.549725351767961e-07,
      "loss": 0.3272,
      "step": 65815
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.250603616966226,
      "learning_rate": 1.5466854518115082e-07,
      "loss": 0.3242,
      "step": 65820
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.868944725672153,
      "learning_rate": 1.5436484894367943e-07,
      "loss": 0.2953,
      "step": 65825
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.778503149430003,
      "learning_rate": 1.5406144648279332e-07,
      "loss": 0.3113,
      "step": 65830
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.9215593183432382,
      "learning_rate": 1.5375833781688776e-07,
      "loss": 0.3021,
      "step": 65835
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.7266602745959005,
      "learning_rate": 1.5345552296433918e-07,
      "loss": 0.2923,
      "step": 65840
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.147943985141481,
      "learning_rate": 1.5315300194350624e-07,
      "loss": 0.3376,
      "step": 65845
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.088802083529507,
      "learning_rate": 1.528507747727298e-07,
      "loss": 0.2829,
      "step": 65850
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.625653074701914,
      "learning_rate": 1.5254884147033355e-07,
      "loss": 0.289,
      "step": 65855
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.7981901939653686,
      "learning_rate": 1.522472020546223e-07,
      "loss": 0.3024,
      "step": 65860
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.4331042906156117,
      "learning_rate": 1.519458565438836e-07,
      "loss": 0.2972,
      "step": 65865
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.8856951238989987,
      "learning_rate": 1.516448049563868e-07,
      "loss": 0.2938,
      "step": 65870
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.02024041886062,
      "learning_rate": 1.51344047310385e-07,
      "loss": 0.3137,
      "step": 65875
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.161525451083014,
      "learning_rate": 1.5104358362411032e-07,
      "loss": 0.3009,
      "step": 65880
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.208350316959127,
      "learning_rate": 1.5074341391578095e-07,
      "loss": 0.301,
      "step": 65885
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.246633644909555,
      "learning_rate": 1.504435382035946e-07,
      "loss": 0.3126,
      "step": 65890
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.592633898423814,
      "learning_rate": 1.501439565057311e-07,
      "loss": 0.2825,
      "step": 65895
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.033431196340206,
      "learning_rate": 1.4984466884035376e-07,
      "loss": 0.3091,
      "step": 65900
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.4427903745678985,
      "learning_rate": 1.4954567522560804e-07,
      "loss": 0.3106,
      "step": 65905
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.13457974549506,
      "learning_rate": 1.4924697567961998e-07,
      "loss": 0.2988,
      "step": 65910
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.413833724534987,
      "learning_rate": 1.4894857022050014e-07,
      "loss": 0.3029,
      "step": 65915
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.544897487000302,
      "learning_rate": 1.48650458866339e-07,
      "loss": 0.3283,
      "step": 65920
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.650366655502041,
      "learning_rate": 1.4835264163521047e-07,
      "loss": 0.302,
      "step": 65925
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.834593135339089,
      "learning_rate": 1.4805511854517006e-07,
      "loss": 0.3188,
      "step": 65930
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.7003760460264923,
      "learning_rate": 1.4775788961425563e-07,
      "loss": 0.317,
      "step": 65935
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.97500191055279,
      "learning_rate": 1.4746095486048774e-07,
      "loss": 0.3174,
      "step": 65940
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.864482816397804,
      "learning_rate": 1.4716431430186918e-07,
      "loss": 0.3107,
      "step": 65945
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.1889489305325815,
      "learning_rate": 1.4686796795638337e-07,
      "loss": 0.2941,
      "step": 65950
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.9135850723917853,
      "learning_rate": 1.4657191584199703e-07,
      "loss": 0.314,
      "step": 65955
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.062020471468528,
      "learning_rate": 1.4627615797665974e-07,
      "loss": 0.3186,
      "step": 65960
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.443281254147498,
      "learning_rate": 1.4598069437830097e-07,
      "loss": 0.3119,
      "step": 65965
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.052735967007083,
      "learning_rate": 1.4568552506483537e-07,
      "loss": 0.3122,
      "step": 65970
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.965357707240167,
      "learning_rate": 1.4539065005415742e-07,
      "loss": 0.3072,
      "step": 65975
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.226944156444805,
      "learning_rate": 1.4509606936414455e-07,
      "loss": 0.3092,
      "step": 65980
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.071702724945526,
      "learning_rate": 1.4480178301265635e-07,
      "loss": 0.3117,
      "step": 65985
    },
    {
      "epoch": 2.79,
      "grad_norm": 5.023876085145244,
      "learning_rate": 1.445077910175352e-07,
      "loss": 0.3256,
      "step": 65990
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.8785603258719106,
      "learning_rate": 1.4421409339660353e-07,
      "loss": 0.2957,
      "step": 65995
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.911163729482931,
      "learning_rate": 1.4392069016766874e-07,
      "loss": 0.2872,
      "step": 66000
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.028421561638905,
      "learning_rate": 1.4362758134851772e-07,
      "loss": 0.2926,
      "step": 66005
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.325781685512501,
      "learning_rate": 1.4333476695692296e-07,
      "loss": 0.2872,
      "step": 66010
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.9056340251082333,
      "learning_rate": 1.4304224701063406e-07,
      "loss": 0.3223,
      "step": 66015
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.7310709054637488,
      "learning_rate": 1.427500215273875e-07,
      "loss": 0.2802,
      "step": 66020
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.141366499944626,
      "learning_rate": 1.4245809052490012e-07,
      "loss": 0.306,
      "step": 66025
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.739957568213342,
      "learning_rate": 1.4216645402087003e-07,
      "loss": 0.3028,
      "step": 66030
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.752018530157752,
      "learning_rate": 1.4187511203297865e-07,
      "loss": 0.3123,
      "step": 66035
    },
    {
      "epoch": 2.79,
      "grad_norm": 4.146440336732898,
      "learning_rate": 1.4158406457888908e-07,
      "loss": 0.2967,
      "step": 66040
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.3917037196166975,
      "learning_rate": 1.412933116762466e-07,
      "loss": 0.3172,
      "step": 66045
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.966434214472327,
      "learning_rate": 1.4100285334267883e-07,
      "loss": 0.2909,
      "step": 66050
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.032815589787488,
      "learning_rate": 1.4071268959579554e-07,
      "loss": 0.2995,
      "step": 66055
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.6738601771493418,
      "learning_rate": 1.4042282045318877e-07,
      "loss": 0.3349,
      "step": 66060
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.014133544085371,
      "learning_rate": 1.4013324593243173e-07,
      "loss": 0.296,
      "step": 66065
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.4958918166402295,
      "learning_rate": 1.398439660510803e-07,
      "loss": 0.3116,
      "step": 66070
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.639524101433475,
      "learning_rate": 1.395549808266744e-07,
      "loss": 0.2999,
      "step": 66075
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8281172121019464,
      "learning_rate": 1.3926629027673222e-07,
      "loss": 0.3099,
      "step": 66080
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8134126758102402,
      "learning_rate": 1.3897789441875753e-07,
      "loss": 0.2731,
      "step": 66085
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.665191917005747,
      "learning_rate": 1.3868979327023358e-07,
      "loss": 0.3197,
      "step": 66090
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.7703843925263225,
      "learning_rate": 1.3840198684862915e-07,
      "loss": 0.3023,
      "step": 66095
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8900945243153995,
      "learning_rate": 1.3811447517139086e-07,
      "loss": 0.3041,
      "step": 66100
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.7938056430600655,
      "learning_rate": 1.3782725825595146e-07,
      "loss": 0.2771,
      "step": 66105
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.9852064453687115,
      "learning_rate": 1.375403361197236e-07,
      "loss": 0.3366,
      "step": 66110
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8460223738589363,
      "learning_rate": 1.372537087801018e-07,
      "loss": 0.3116,
      "step": 66115
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.822705939614552,
      "learning_rate": 1.3696737625446376e-07,
      "loss": 0.2989,
      "step": 66120
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.217222978473458,
      "learning_rate": 1.3668133856016953e-07,
      "loss": 0.3085,
      "step": 66125
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.694765075114718,
      "learning_rate": 1.3639559571456074e-07,
      "loss": 0.3018,
      "step": 66130
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8911314611795067,
      "learning_rate": 1.3611014773496024e-07,
      "loss": 0.2812,
      "step": 66135
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.9126210855348806,
      "learning_rate": 1.3582499463867416e-07,
      "loss": 0.3296,
      "step": 66140
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.439518418745524,
      "learning_rate": 1.3554013644299145e-07,
      "loss": 0.3112,
      "step": 66145
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.939802094505862,
      "learning_rate": 1.3525557316518113e-07,
      "loss": 0.3211,
      "step": 66150
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8076097102150306,
      "learning_rate": 1.3497130482249598e-07,
      "loss": 0.2915,
      "step": 66155
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.4456904535647266,
      "learning_rate": 1.346873314321706e-07,
      "loss": 0.2933,
      "step": 66160
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.447273713019155,
      "learning_rate": 1.344036530114201e-07,
      "loss": 0.3326,
      "step": 66165
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.7500368607705625,
      "learning_rate": 1.3412026957744463e-07,
      "loss": 0.3048,
      "step": 66170
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.6303736126457324,
      "learning_rate": 1.3383718114742373e-07,
      "loss": 0.2961,
      "step": 66175
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.719380208876276,
      "learning_rate": 1.3355438773852202e-07,
      "loss": 0.3075,
      "step": 66180
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.6397414587303296,
      "learning_rate": 1.3327188936788194e-07,
      "loss": 0.2915,
      "step": 66185
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.090318334292168,
      "learning_rate": 1.329896860526325e-07,
      "loss": 0.2946,
      "step": 66190
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8903856233679655,
      "learning_rate": 1.327077778098823e-07,
      "loss": 0.3034,
      "step": 66195
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.6919065182351027,
      "learning_rate": 1.324261646567221e-07,
      "loss": 0.3183,
      "step": 66200
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.409039021198208,
      "learning_rate": 1.3214484661022543e-07,
      "loss": 0.304,
      "step": 66205
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.59498819413163,
      "learning_rate": 1.318638236874481e-07,
      "loss": 0.2906,
      "step": 66210
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.9360898193845855,
      "learning_rate": 1.315830959054276e-07,
      "loss": 0.2923,
      "step": 66215
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.924996466679918,
      "learning_rate": 1.313026632811837e-07,
      "loss": 0.3068,
      "step": 66220
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.4859392814785477,
      "learning_rate": 1.310225258317177e-07,
      "loss": 0.3051,
      "step": 66225
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.703131483792007,
      "learning_rate": 1.30742683574015e-07,
      "loss": 0.3141,
      "step": 66230
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.034385462896406,
      "learning_rate": 1.3046313652503918e-07,
      "loss": 0.2771,
      "step": 66235
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.814564600304197,
      "learning_rate": 1.301838847017406e-07,
      "loss": 0.2934,
      "step": 66240
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.406916117055063,
      "learning_rate": 1.299049281210485e-07,
      "loss": 0.3217,
      "step": 66245
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.3186557228139755,
      "learning_rate": 1.2962626679987543e-07,
      "loss": 0.3265,
      "step": 66250
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.8366800017247673,
      "learning_rate": 1.2934790075511516e-07,
      "loss": 0.3018,
      "step": 66255
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.264645999545481,
      "learning_rate": 1.290698300036447e-07,
      "loss": 0.2936,
      "step": 66260
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.7968885158637278,
      "learning_rate": 1.287920545623228e-07,
      "loss": 0.3119,
      "step": 66265
    },
    {
      "epoch": 2.8,
      "grad_norm": 3.6999912154949843,
      "learning_rate": 1.2851457444798988e-07,
      "loss": 0.3171,
      "step": 66270
    },
    {
      "epoch": 2.8,
      "grad_norm": 4.896802516960357,
      "learning_rate": 1.2823738967746912e-07,
      "loss": 0.331,
      "step": 66275
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.49203861117748,
      "learning_rate": 1.2796050026756546e-07,
      "loss": 0.274,
      "step": 66280
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.6950060753647525,
      "learning_rate": 1.2768390623506487e-07,
      "loss": 0.2977,
      "step": 66285
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.522275482796303,
      "learning_rate": 1.274076075967373e-07,
      "loss": 0.3094,
      "step": 66290
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.437475305388126,
      "learning_rate": 1.2713160436933437e-07,
      "loss": 0.2871,
      "step": 66295
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.047906103524821,
      "learning_rate": 1.2685589656958818e-07,
      "loss": 0.34,
      "step": 66300
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.626234582023741,
      "learning_rate": 1.2658048421421487e-07,
      "loss": 0.3084,
      "step": 66305
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.250559563406509,
      "learning_rate": 1.2630536731991106e-07,
      "loss": 0.2964,
      "step": 66310
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.8933911771135854,
      "learning_rate": 1.2603054590335785e-07,
      "loss": 0.3189,
      "step": 66315
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.580270398139942,
      "learning_rate": 1.2575601998121577e-07,
      "loss": 0.2813,
      "step": 66320
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.473838338085486,
      "learning_rate": 1.2548178957012826e-07,
      "loss": 0.3129,
      "step": 66325
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.219292715687132,
      "learning_rate": 1.2520785468672192e-07,
      "loss": 0.3014,
      "step": 66330
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.6997243931386485,
      "learning_rate": 1.2493421534760354e-07,
      "loss": 0.3042,
      "step": 66335
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.6345014932830244,
      "learning_rate": 1.2466087156936368e-07,
      "loss": 0.3027,
      "step": 66340
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.8904124113843737,
      "learning_rate": 1.2438782336857525e-07,
      "loss": 0.3091,
      "step": 66345
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.8721156041205114,
      "learning_rate": 1.2411507076179108e-07,
      "loss": 0.2981,
      "step": 66350
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.8044772393915336,
      "learning_rate": 1.238426137655474e-07,
      "loss": 0.3002,
      "step": 66355
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.9387966053478314,
      "learning_rate": 1.2357045239636323e-07,
      "loss": 0.2817,
      "step": 66360
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.355239811357412,
      "learning_rate": 1.232985866707387e-07,
      "loss": 0.3168,
      "step": 66365
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.0207966569729265,
      "learning_rate": 1.2302701660515558e-07,
      "loss": 0.3299,
      "step": 66370
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.4585672106122747,
      "learning_rate": 1.2275574221607912e-07,
      "loss": 0.2923,
      "step": 66375
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.4851600903793774,
      "learning_rate": 1.2248476351995609e-07,
      "loss": 0.2884,
      "step": 66380
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.6264083475493236,
      "learning_rate": 1.2221408053321392e-07,
      "loss": 0.3101,
      "step": 66385
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.371861768326324,
      "learning_rate": 1.2194369327226397e-07,
      "loss": 0.293,
      "step": 66390
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.6737664898381173,
      "learning_rate": 1.2167360175349975e-07,
      "loss": 0.2697,
      "step": 66395
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.006439456031335,
      "learning_rate": 1.2140380599329537e-07,
      "loss": 0.3001,
      "step": 66400
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.7863118098481534,
      "learning_rate": 1.211343060080078e-07,
      "loss": 0.3427,
      "step": 66405
    },
    {
      "epoch": 2.81,
      "grad_norm": 7.956955039903079,
      "learning_rate": 1.208651018139756e-07,
      "loss": 0.3299,
      "step": 66410
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.740787110781312,
      "learning_rate": 1.2059619342752126e-07,
      "loss": 0.2937,
      "step": 66415
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.710710325984871,
      "learning_rate": 1.2032758086494621e-07,
      "loss": 0.3005,
      "step": 66420
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.17434892568975,
      "learning_rate": 1.2005926414253633e-07,
      "loss": 0.2888,
      "step": 66425
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.156747286732518,
      "learning_rate": 1.1979124327655911e-07,
      "loss": 0.3278,
      "step": 66430
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.844988407388581,
      "learning_rate": 1.195235182832638e-07,
      "loss": 0.2961,
      "step": 66435
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.7061772966782365,
      "learning_rate": 1.192560891788813e-07,
      "loss": 0.2834,
      "step": 66440
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.8501975516257794,
      "learning_rate": 1.189889559796259e-07,
      "loss": 0.3198,
      "step": 66445
    },
    {
      "epoch": 2.81,
      "grad_norm": 5.223992057883741,
      "learning_rate": 1.1872211870169236e-07,
      "loss": 0.3175,
      "step": 66450
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.9418931560435895,
      "learning_rate": 1.1845557736125835e-07,
      "loss": 0.3131,
      "step": 66455
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.246955262915597,
      "learning_rate": 1.1818933197448368e-07,
      "loss": 0.2881,
      "step": 66460
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.138000130925753,
      "learning_rate": 1.1792338255751045e-07,
      "loss": 0.3006,
      "step": 66465
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.282356569696341,
      "learning_rate": 1.1765772912646133e-07,
      "loss": 0.2985,
      "step": 66470
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.965595060283118,
      "learning_rate": 1.1739237169744288e-07,
      "loss": 0.3499,
      "step": 66475
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.510424965104573,
      "learning_rate": 1.1712731028654279e-07,
      "loss": 0.3071,
      "step": 66480
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.641196288338066,
      "learning_rate": 1.1686254490983096e-07,
      "loss": 0.2732,
      "step": 66485
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.7716152403759553,
      "learning_rate": 1.16598075583359e-07,
      "loss": 0.2958,
      "step": 66490
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.7126417140555814,
      "learning_rate": 1.1633390232316132e-07,
      "loss": 0.2912,
      "step": 66495
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.9047309377096298,
      "learning_rate": 1.16070025145254e-07,
      "loss": 0.3091,
      "step": 66500
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.6744445091962383,
      "learning_rate": 1.1580644406563479e-07,
      "loss": 0.3072,
      "step": 66505
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.773190221162525,
      "learning_rate": 1.1554315910028424e-07,
      "loss": 0.2989,
      "step": 66510
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.322302187142827,
      "learning_rate": 1.1528017026516458e-07,
      "loss": 0.2927,
      "step": 66515
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.227083381794769,
      "learning_rate": 1.1501747757621918e-07,
      "loss": 0.3149,
      "step": 66520
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.8731382124985827,
      "learning_rate": 1.1475508104937472e-07,
      "loss": 0.2752,
      "step": 66525
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.051716361737263,
      "learning_rate": 1.1449298070054016e-07,
      "loss": 0.3037,
      "step": 66530
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.025524018018709,
      "learning_rate": 1.1423117654560555e-07,
      "loss": 0.3067,
      "step": 66535
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.814428004942523,
      "learning_rate": 1.1396966860044267e-07,
      "loss": 0.3134,
      "step": 66540
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.8299371845202486,
      "learning_rate": 1.1370845688090715e-07,
      "loss": 0.3122,
      "step": 66545
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.494629189356971,
      "learning_rate": 1.1344754140283464e-07,
      "loss": 0.3023,
      "step": 66550
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.018656238634794,
      "learning_rate": 1.1318692218204363e-07,
      "loss": 0.2883,
      "step": 66555
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.031157030415214,
      "learning_rate": 1.129265992343348e-07,
      "loss": 0.3082,
      "step": 66560
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9132225889676575,
      "learning_rate": 1.1266657257549162e-07,
      "loss": 0.3046,
      "step": 66565
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.916739331239777,
      "learning_rate": 1.1240684222127818e-07,
      "loss": 0.2943,
      "step": 66570
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.719415444070998,
      "learning_rate": 1.1214740818744019e-07,
      "loss": 0.3074,
      "step": 66575
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.08373562402601,
      "learning_rate": 1.118882704897073e-07,
      "loss": 0.28,
      "step": 66580
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.24456280357786,
      "learning_rate": 1.1162942914379083e-07,
      "loss": 0.2924,
      "step": 66585
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.660316792103006,
      "learning_rate": 1.1137088416538267e-07,
      "loss": 0.3135,
      "step": 66590
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.698729900036621,
      "learning_rate": 1.111126355701575e-07,
      "loss": 0.3012,
      "step": 66595
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.167581656179021,
      "learning_rate": 1.1085468337377281e-07,
      "loss": 0.3075,
      "step": 66600
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.8171665310468548,
      "learning_rate": 1.105970275918672e-07,
      "loss": 0.2958,
      "step": 66605
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9681352687333904,
      "learning_rate": 1.1033966824006092e-07,
      "loss": 0.3119,
      "step": 66610
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9633875908972263,
      "learning_rate": 1.100826053339582e-07,
      "loss": 0.3119,
      "step": 66615
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.730272699542559,
      "learning_rate": 1.0982583888914377e-07,
      "loss": 0.3036,
      "step": 66620
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.6601226956168045,
      "learning_rate": 1.0956936892118353e-07,
      "loss": 0.3141,
      "step": 66625
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.190035551682392,
      "learning_rate": 1.0931319544562724e-07,
      "loss": 0.3018,
      "step": 66630
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9608967123520027,
      "learning_rate": 1.0905731847800693e-07,
      "loss": 0.2923,
      "step": 66635
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.8473340876702293,
      "learning_rate": 1.0880173803383354e-07,
      "loss": 0.3096,
      "step": 66640
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.433265141912316,
      "learning_rate": 1.0854645412860353e-07,
      "loss": 0.3059,
      "step": 66645
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.8555058767248305,
      "learning_rate": 1.0829146677779456e-07,
      "loss": 0.299,
      "step": 66650
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.1060316409972,
      "learning_rate": 1.0803677599686369e-07,
      "loss": 0.3112,
      "step": 66655
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.318509897899225,
      "learning_rate": 1.0778238180125411e-07,
      "loss": 0.3058,
      "step": 66660
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.7757600730691934,
      "learning_rate": 1.0752828420638795e-07,
      "loss": 0.29,
      "step": 66665
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.289543254221664,
      "learning_rate": 1.072744832276712e-07,
      "loss": 0.3093,
      "step": 66670
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.714514357589515,
      "learning_rate": 1.070209788804899e-07,
      "loss": 0.3133,
      "step": 66675
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.7879337111108913,
      "learning_rate": 1.0676777118021397e-07,
      "loss": 0.2798,
      "step": 66680
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9285849043468937,
      "learning_rate": 1.0651486014219503e-07,
      "loss": 0.2822,
      "step": 66685
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.968928836829013,
      "learning_rate": 1.0626224578176525e-07,
      "loss": 0.2777,
      "step": 66690
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.948881254646119,
      "learning_rate": 1.0600992811424071e-07,
      "loss": 0.3056,
      "step": 66695
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.89588114156917,
      "learning_rate": 1.0575790715491918e-07,
      "loss": 0.2935,
      "step": 66700
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.277950863043998,
      "learning_rate": 1.0550618291907844e-07,
      "loss": 0.2798,
      "step": 66705
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.991506313733201,
      "learning_rate": 1.0525475542198072e-07,
      "loss": 0.3286,
      "step": 66710
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9552777567751463,
      "learning_rate": 1.0500362467886937e-07,
      "loss": 0.2991,
      "step": 66715
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.6976846786492086,
      "learning_rate": 1.0475279070497001e-07,
      "loss": 0.3166,
      "step": 66720
    },
    {
      "epoch": 2.82,
      "grad_norm": 4.087263916805855,
      "learning_rate": 1.0450225351548993e-07,
      "loss": 0.3147,
      "step": 66725
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.6776440146372,
      "learning_rate": 1.0425201312561695e-07,
      "loss": 0.3208,
      "step": 66730
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.642004041983497,
      "learning_rate": 1.0400206955052395e-07,
      "loss": 0.3248,
      "step": 66735
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.8480691587041855,
      "learning_rate": 1.037524228053638e-07,
      "loss": 0.2957,
      "step": 66740
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.9243982774402513,
      "learning_rate": 1.0350307290527162e-07,
      "loss": 0.3265,
      "step": 66745
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.7087642966822187,
      "learning_rate": 1.0325401986536532e-07,
      "loss": 0.3021,
      "step": 66750
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.304901180263585,
      "learning_rate": 1.0300526370074393e-07,
      "loss": 0.3108,
      "step": 66755
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.640901913698597,
      "learning_rate": 1.0275680442648873e-07,
      "loss": 0.3078,
      "step": 66760
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.97890554650979,
      "learning_rate": 1.025086420576632e-07,
      "loss": 0.3269,
      "step": 66765
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.475593604317089,
      "learning_rate": 1.0226077660931311e-07,
      "loss": 0.2902,
      "step": 66770
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.923594932682321,
      "learning_rate": 1.0201320809646476e-07,
      "loss": 0.3,
      "step": 66775
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.472157445624218,
      "learning_rate": 1.0176593653412837e-07,
      "loss": 0.2782,
      "step": 66780
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.5476683857251676,
      "learning_rate": 1.0151896193729471e-07,
      "loss": 0.3155,
      "step": 66785
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.793003852099486,
      "learning_rate": 1.0127228432093794e-07,
      "loss": 0.2847,
      "step": 66790
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.987219964957159,
      "learning_rate": 1.0102590370001275e-07,
      "loss": 0.2955,
      "step": 66795
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.233337728231237,
      "learning_rate": 1.0077982008945608e-07,
      "loss": 0.2966,
      "step": 66800
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.154253761603253,
      "learning_rate": 1.0053403350418877e-07,
      "loss": 0.2884,
      "step": 66805
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.7180060022390804,
      "learning_rate": 1.0028854395911059e-07,
      "loss": 0.3111,
      "step": 66810
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.230618273330698,
      "learning_rate": 1.0004335146910516e-07,
      "loss": 0.3244,
      "step": 66815
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.9089877683429024,
      "learning_rate": 9.979845604903837e-08,
      "loss": 0.2684,
      "step": 66820
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.2535574983620545,
      "learning_rate": 9.955385771375725e-08,
      "loss": 0.31,
      "step": 66825
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.804029814543924,
      "learning_rate": 9.930955647809049e-08,
      "loss": 0.2864,
      "step": 66830
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.067805896880169,
      "learning_rate": 9.906555235685066e-08,
      "loss": 0.3078,
      "step": 66835
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.181333226734484,
      "learning_rate": 9.882184536482987e-08,
      "loss": 0.3165,
      "step": 66840
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.0228709019557405,
      "learning_rate": 9.857843551680346e-08,
      "loss": 0.3166,
      "step": 66845
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.9351911751079607,
      "learning_rate": 9.833532282752912e-08,
      "loss": 0.3183,
      "step": 66850
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.8229410262220975,
      "learning_rate": 9.809250731174558e-08,
      "loss": 0.3053,
      "step": 66855
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.169923979475677,
      "learning_rate": 9.784998898417441e-08,
      "loss": 0.28,
      "step": 66860
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.780588783471478,
      "learning_rate": 9.760776785951886e-08,
      "loss": 0.3093,
      "step": 66865
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.7373309534321053,
      "learning_rate": 9.736584395246384e-08,
      "loss": 0.325,
      "step": 66870
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.718656870711661,
      "learning_rate": 9.712421727767651e-08,
      "loss": 0.2828,
      "step": 66875
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.6361435903433663,
      "learning_rate": 9.68828878498057e-08,
      "loss": 0.2988,
      "step": 66880
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.198849493646334,
      "learning_rate": 9.664185568348305e-08,
      "loss": 0.3024,
      "step": 66885
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.833330956417512,
      "learning_rate": 9.640112079332186e-08,
      "loss": 0.3067,
      "step": 66890
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.604323995141905,
      "learning_rate": 9.616068319391603e-08,
      "loss": 0.3095,
      "step": 66895
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.084264266208825,
      "learning_rate": 9.592054289984332e-08,
      "loss": 0.3288,
      "step": 66900
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.8986920321468745,
      "learning_rate": 9.568069992566265e-08,
      "loss": 0.3095,
      "step": 66905
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.471322968063947,
      "learning_rate": 9.544115428591515e-08,
      "loss": 0.3116,
      "step": 66910
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.922905062612245,
      "learning_rate": 9.520190599512314e-08,
      "loss": 0.2908,
      "step": 66915
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.014577940712433,
      "learning_rate": 9.496295506779219e-08,
      "loss": 0.3122,
      "step": 66920
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.290409916522706,
      "learning_rate": 9.472430151840962e-08,
      "loss": 0.2965,
      "step": 66925
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.086715247383836,
      "learning_rate": 9.44859453614433e-08,
      "loss": 0.2938,
      "step": 66930
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.761564399289158,
      "learning_rate": 9.424788661134444e-08,
      "loss": 0.2985,
      "step": 66935
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.987361858351176,
      "learning_rate": 9.401012528254594e-08,
      "loss": 0.2968,
      "step": 66940
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.1126567584095595,
      "learning_rate": 9.377266138946239e-08,
      "loss": 0.3273,
      "step": 66945
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.24645724976815,
      "learning_rate": 9.353549494649006e-08,
      "loss": 0.3007,
      "step": 66950
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.843980853445743,
      "learning_rate": 9.329862596800909e-08,
      "loss": 0.3169,
      "step": 66955
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.9311125650960994,
      "learning_rate": 9.306205446837912e-08,
      "loss": 0.291,
      "step": 66960
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.421728961437795,
      "learning_rate": 9.282578046194257e-08,
      "loss": 0.3164,
      "step": 66965
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.8054452522309687,
      "learning_rate": 9.258980396302409e-08,
      "loss": 0.2996,
      "step": 66970
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.9598817795196797,
      "learning_rate": 9.235412498593165e-08,
      "loss": 0.3023,
      "step": 66975
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.6019334100173337,
      "learning_rate": 9.211874354495164e-08,
      "loss": 0.3101,
      "step": 66980
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.8374474195160357,
      "learning_rate": 9.188365965435598e-08,
      "loss": 0.2989,
      "step": 66985
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.382311891476272,
      "learning_rate": 9.164887332839656e-08,
      "loss": 0.306,
      "step": 66990
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.232989210342415,
      "learning_rate": 9.141438458130814e-08,
      "loss": 0.3057,
      "step": 66995
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9487858540544245,
      "learning_rate": 9.118019342730655e-08,
      "loss": 0.3332,
      "step": 67000
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9949049276147126,
      "learning_rate": 9.0946299880591e-08,
      "loss": 0.3063,
      "step": 67005
    },
    {
      "epoch": 2.84,
      "grad_norm": 5.4523534242535945,
      "learning_rate": 9.07127039553407e-08,
      "loss": 0.3016,
      "step": 67010
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.965588192676178,
      "learning_rate": 9.047940566571823e-08,
      "loss": 0.2986,
      "step": 67015
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.5922554495644334,
      "learning_rate": 9.024640502586835e-08,
      "loss": 0.2834,
      "step": 67020
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.5436549591077435,
      "learning_rate": 9.0013702049917e-08,
      "loss": 0.2961,
      "step": 67025
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.97567368738804,
      "learning_rate": 8.978129675197122e-08,
      "loss": 0.3012,
      "step": 67030
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.913069323928701,
      "learning_rate": 8.954918914612254e-08,
      "loss": 0.2815,
      "step": 67035
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.897313176201929,
      "learning_rate": 8.931737924644247e-08,
      "loss": 0.3037,
      "step": 67040
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.8299068606769406,
      "learning_rate": 8.908586706698419e-08,
      "loss": 0.3211,
      "step": 67045
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.591180751762232,
      "learning_rate": 8.885465262178483e-08,
      "loss": 0.3101,
      "step": 67050
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.2310044345013775,
      "learning_rate": 8.862373592486152e-08,
      "loss": 0.3116,
      "step": 67055
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.048123222347952,
      "learning_rate": 8.839311699021414e-08,
      "loss": 0.3159,
      "step": 67060
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.606003615309165,
      "learning_rate": 8.816279583182375e-08,
      "loss": 0.3017,
      "step": 67065
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.7274266699363734,
      "learning_rate": 8.793277246365528e-08,
      "loss": 0.3036,
      "step": 67070
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9281172227640746,
      "learning_rate": 8.770304689965426e-08,
      "loss": 0.3169,
      "step": 67075
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.7628051387493926,
      "learning_rate": 8.747361915374731e-08,
      "loss": 0.3256,
      "step": 67080
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.5778257401188416,
      "learning_rate": 8.72444892398444e-08,
      "loss": 0.308,
      "step": 67085
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.129571152398946,
      "learning_rate": 8.701565717183725e-08,
      "loss": 0.297,
      "step": 67090
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.49191748976813,
      "learning_rate": 8.67871229635997e-08,
      "loss": 0.2798,
      "step": 67095
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.352438169831849,
      "learning_rate": 8.65588866289857e-08,
      "loss": 0.3018,
      "step": 67100
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.074233368999392,
      "learning_rate": 8.633094818183418e-08,
      "loss": 0.2965,
      "step": 67105
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9033117807989948,
      "learning_rate": 8.610330763596298e-08,
      "loss": 0.2981,
      "step": 67110
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.883689386157709,
      "learning_rate": 8.587596500517437e-08,
      "loss": 0.2747,
      "step": 67115
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.924283114694407,
      "learning_rate": 8.564892030325068e-08,
      "loss": 0.3189,
      "step": 67120
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9881209026213793,
      "learning_rate": 8.542217354395755e-08,
      "loss": 0.3048,
      "step": 67125
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.6083239392812256,
      "learning_rate": 8.519572474104121e-08,
      "loss": 0.2921,
      "step": 67130
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.553843084523379,
      "learning_rate": 8.496957390823068e-08,
      "loss": 0.2847,
      "step": 67135
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.49659366333283,
      "learning_rate": 8.474372105923834e-08,
      "loss": 0.2958,
      "step": 67140
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9245769426870196,
      "learning_rate": 8.451816620775488e-08,
      "loss": 0.2984,
      "step": 67145
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.909139265601154,
      "learning_rate": 8.42929093674566e-08,
      "loss": 0.3151,
      "step": 67150
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.754452870833101,
      "learning_rate": 8.406795055199868e-08,
      "loss": 0.2966,
      "step": 67155
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.896211401292076,
      "learning_rate": 8.384328977502188e-08,
      "loss": 0.313,
      "step": 67160
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9652700647286125,
      "learning_rate": 8.361892705014418e-08,
      "loss": 0.2965,
      "step": 67165
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.932440722120992,
      "learning_rate": 8.339486239096972e-08,
      "loss": 0.3188,
      "step": 67170
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.026457700814128,
      "learning_rate": 8.317109581108263e-08,
      "loss": 0.2954,
      "step": 67175
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.179536205621041,
      "learning_rate": 8.294762732404871e-08,
      "loss": 0.2791,
      "step": 67180
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.331706014121443,
      "learning_rate": 8.272445694341658e-08,
      "loss": 0.3181,
      "step": 67185
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.110975468232174,
      "learning_rate": 8.250158468271652e-08,
      "loss": 0.3083,
      "step": 67190
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.661368634763695,
      "learning_rate": 8.227901055545995e-08,
      "loss": 0.2927,
      "step": 67195
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.127516834305215,
      "learning_rate": 8.205673457514162e-08,
      "loss": 0.3107,
      "step": 67200
    },
    {
      "epoch": 2.84,
      "grad_norm": 4.036782015352208,
      "learning_rate": 8.183475675523744e-08,
      "loss": 0.292,
      "step": 67205
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.666382375396397,
      "learning_rate": 8.161307710920552e-08,
      "loss": 0.3156,
      "step": 67210
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.8317792185015334,
      "learning_rate": 8.139169565048399e-08,
      "loss": 0.2946,
      "step": 67215
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.9666831557572815,
      "learning_rate": 8.117061239249602e-08,
      "loss": 0.2888,
      "step": 67220
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.7162140116190088,
      "learning_rate": 8.094982734864476e-08,
      "loss": 0.2838,
      "step": 67225
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.4369378363985605,
      "learning_rate": 8.072934053231674e-08,
      "loss": 0.2822,
      "step": 67230
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.8473237105543006,
      "learning_rate": 8.050915195687792e-08,
      "loss": 0.3293,
      "step": 67235
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.063928807524676,
      "learning_rate": 8.028926163567874e-08,
      "loss": 0.304,
      "step": 67240
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.8961441480635766,
      "learning_rate": 8.00696695820502e-08,
      "loss": 0.3277,
      "step": 67245
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.104839097838789,
      "learning_rate": 7.985037580930499e-08,
      "loss": 0.3086,
      "step": 67250
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.7561862366570544,
      "learning_rate": 7.963138033073858e-08,
      "loss": 0.2823,
      "step": 67255
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.032227328781345,
      "learning_rate": 7.941268315962925e-08,
      "loss": 0.2955,
      "step": 67260
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.666582286084504,
      "learning_rate": 7.919428430923415e-08,
      "loss": 0.3058,
      "step": 67265
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.2208130023421315,
      "learning_rate": 7.897618379279436e-08,
      "loss": 0.3088,
      "step": 67270
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.155545487274384,
      "learning_rate": 7.875838162353378e-08,
      "loss": 0.2827,
      "step": 67275
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.581677797445036,
      "learning_rate": 7.85408778146568e-08,
      "loss": 0.2984,
      "step": 67280
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.298966167329686,
      "learning_rate": 7.832367237934956e-08,
      "loss": 0.2965,
      "step": 67285
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.9096560830894274,
      "learning_rate": 7.810676533078043e-08,
      "loss": 0.3074,
      "step": 67290
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.525608092844694,
      "learning_rate": 7.789015668210053e-08,
      "loss": 0.3188,
      "step": 67295
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.968339560893488,
      "learning_rate": 7.76738464464416e-08,
      "loss": 0.3048,
      "step": 67300
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.041956996513647,
      "learning_rate": 7.745783463691869e-08,
      "loss": 0.3322,
      "step": 67305
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.9109044858291866,
      "learning_rate": 7.724212126662744e-08,
      "loss": 0.3129,
      "step": 67310
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.095252809311223,
      "learning_rate": 7.702670634864518e-08,
      "loss": 0.2884,
      "step": 67315
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.018334635389714,
      "learning_rate": 7.68115898960331e-08,
      "loss": 0.3268,
      "step": 67320
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.8834332328254075,
      "learning_rate": 7.659677192183245e-08,
      "loss": 0.3217,
      "step": 67325
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.7482215232131466,
      "learning_rate": 7.638225243906727e-08,
      "loss": 0.2986,
      "step": 67330
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.259989580802478,
      "learning_rate": 7.616803146074326e-08,
      "loss": 0.295,
      "step": 67335
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.040747440323012,
      "learning_rate": 7.595410899984779e-08,
      "loss": 0.3212,
      "step": 67340
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.930933398956746,
      "learning_rate": 7.57404850693505e-08,
      "loss": 0.2845,
      "step": 67345
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.038288472412129,
      "learning_rate": 7.552715968220214e-08,
      "loss": 0.3144,
      "step": 67350
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.276393039838735,
      "learning_rate": 7.531413285133737e-08,
      "loss": 0.3137,
      "step": 67355
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.6448938374224613,
      "learning_rate": 7.51014045896703e-08,
      "loss": 0.2923,
      "step": 67360
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.7237640698104713,
      "learning_rate": 7.488897491009839e-08,
      "loss": 0.2926,
      "step": 67365
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.225816973028743,
      "learning_rate": 7.467684382550022e-08,
      "loss": 0.3039,
      "step": 67370
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.7796898809835207,
      "learning_rate": 7.446501134873719e-08,
      "loss": 0.3167,
      "step": 67375
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.057451796765932,
      "learning_rate": 7.425347749265233e-08,
      "loss": 0.3266,
      "step": 67380
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.8978002623192625,
      "learning_rate": 7.404224227006928e-08,
      "loss": 0.3101,
      "step": 67385
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.234921223837073,
      "learning_rate": 7.383130569379504e-08,
      "loss": 0.2961,
      "step": 67390
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.605006233651287,
      "learning_rate": 7.36206677766188e-08,
      "loss": 0.3243,
      "step": 67395
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.8396152618131727,
      "learning_rate": 7.341032853131036e-08,
      "loss": 0.2923,
      "step": 67400
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.197094253155551,
      "learning_rate": 7.320028797062117e-08,
      "loss": 0.3123,
      "step": 67405
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.325509709024995,
      "learning_rate": 7.299054610728662e-08,
      "loss": 0.3071,
      "step": 67410
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.120089212305107,
      "learning_rate": 7.278110295402263e-08,
      "loss": 0.2915,
      "step": 67415
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.001738714254425,
      "learning_rate": 7.257195852352628e-08,
      "loss": 0.2858,
      "step": 67420
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.8368123057226837,
      "learning_rate": 7.236311282847797e-08,
      "loss": 0.3194,
      "step": 67425
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.7713647258107335,
      "learning_rate": 7.21545658815398e-08,
      "loss": 0.2956,
      "step": 67430
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.831142116085687,
      "learning_rate": 7.194631769535443e-08,
      "loss": 0.3012,
      "step": 67435
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.794940382091522,
      "learning_rate": 7.173836828254732e-08,
      "loss": 0.3006,
      "step": 67440
    },
    {
      "epoch": 2.85,
      "grad_norm": 5.121171423024304,
      "learning_rate": 7.153071765572727e-08,
      "loss": 0.3408,
      "step": 67445
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.970861748522601,
      "learning_rate": 7.132336582748201e-08,
      "loss": 0.2998,
      "step": 67450
    },
    {
      "epoch": 2.85,
      "grad_norm": 4.390808862202137,
      "learning_rate": 7.111631281038257e-08,
      "loss": 0.2821,
      "step": 67455
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.146049709044729,
      "learning_rate": 7.090955861698334e-08,
      "loss": 0.2673,
      "step": 67460
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.623546080779092,
      "learning_rate": 7.070310325981822e-08,
      "loss": 0.3049,
      "step": 67465
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.9327759616211826,
      "learning_rate": 7.04969467514044e-08,
      "loss": 0.2955,
      "step": 67470
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.0420158407377995,
      "learning_rate": 7.029108910424021e-08,
      "loss": 0.3057,
      "step": 67475
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.298644760191139,
      "learning_rate": 7.008553033080678e-08,
      "loss": 0.2857,
      "step": 67480
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.320236665572237,
      "learning_rate": 6.988027044356527e-08,
      "loss": 0.2797,
      "step": 67485
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.993607925555911,
      "learning_rate": 6.967530945496126e-08,
      "loss": 0.2974,
      "step": 67490
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.902493176651077,
      "learning_rate": 6.947064737742093e-08,
      "loss": 0.32,
      "step": 67495
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.1351472616892835,
      "learning_rate": 6.926628422335103e-08,
      "loss": 0.3097,
      "step": 67500
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.7985499954401676,
      "learning_rate": 6.906222000514274e-08,
      "loss": 0.3283,
      "step": 67505
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.000809056287292,
      "learning_rate": 6.885845473516783e-08,
      "loss": 0.2762,
      "step": 67510
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.052511167455435,
      "learning_rate": 6.865498842577923e-08,
      "loss": 0.3007,
      "step": 67515
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.345118827410775,
      "learning_rate": 6.845182108931315e-08,
      "loss": 0.2875,
      "step": 67520
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.943753358572154,
      "learning_rate": 6.824895273808641e-08,
      "loss": 0.3167,
      "step": 67525
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.009374474150269,
      "learning_rate": 6.804638338439972e-08,
      "loss": 0.3214,
      "step": 67530
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.372306066023339,
      "learning_rate": 6.784411304053273e-08,
      "loss": 0.3088,
      "step": 67535
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.103184726385116,
      "learning_rate": 6.764214171874895e-08,
      "loss": 0.2969,
      "step": 67540
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.3183564963653405,
      "learning_rate": 6.744046943129301e-08,
      "loss": 0.2961,
      "step": 67545
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.403772313516793,
      "learning_rate": 6.723909619039237e-08,
      "loss": 0.3548,
      "step": 67550
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.129769649073681,
      "learning_rate": 6.70380220082556e-08,
      "loss": 0.3007,
      "step": 67555
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.4857746091169934,
      "learning_rate": 6.683724689707238e-08,
      "loss": 0.3164,
      "step": 67560
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.5568505654827196,
      "learning_rate": 6.663677086901687e-08,
      "loss": 0.3034,
      "step": 67565
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.059550441839261,
      "learning_rate": 6.643659393624158e-08,
      "loss": 0.2834,
      "step": 67570
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.765808617531961,
      "learning_rate": 6.623671611088289e-08,
      "loss": 0.2958,
      "step": 67575
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.643079583414973,
      "learning_rate": 6.603713740506001e-08,
      "loss": 0.2918,
      "step": 67580
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.956742881121114,
      "learning_rate": 6.583785783087159e-08,
      "loss": 0.3094,
      "step": 67585
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.7142177989822303,
      "learning_rate": 6.56388774004002e-08,
      "loss": 0.3,
      "step": 67590
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.6569637025348833,
      "learning_rate": 6.54401961257084e-08,
      "loss": 0.293,
      "step": 67595
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.113461126702002,
      "learning_rate": 6.524181401884267e-08,
      "loss": 0.3142,
      "step": 67600
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.509535201185449,
      "learning_rate": 6.504373109183004e-08,
      "loss": 0.3632,
      "step": 67605
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.013390228681255,
      "learning_rate": 6.484594735667926e-08,
      "loss": 0.294,
      "step": 67610
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.107401201002257,
      "learning_rate": 6.46484628253824e-08,
      "loss": 0.2892,
      "step": 67615
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.007312675983092,
      "learning_rate": 6.445127750991099e-08,
      "loss": 0.3196,
      "step": 67620
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.0246456107255835,
      "learning_rate": 6.425439142222045e-08,
      "loss": 0.3147,
      "step": 67625
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.7929694953876356,
      "learning_rate": 6.405780457424793e-08,
      "loss": 0.3114,
      "step": 67630
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.827221937244473,
      "learning_rate": 6.38615169779111e-08,
      "loss": 0.304,
      "step": 67635
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.017811275385251,
      "learning_rate": 6.366552864511099e-08,
      "loss": 0.2988,
      "step": 67640
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.254243161932273,
      "learning_rate": 6.346983958772867e-08,
      "loss": 0.3092,
      "step": 67645
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.235618035852096,
      "learning_rate": 6.327444981762964e-08,
      "loss": 0.3317,
      "step": 67650
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.649658289948641,
      "learning_rate": 6.307935934665887e-08,
      "loss": 0.3027,
      "step": 67655
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.212879270549319,
      "learning_rate": 6.288456818664468e-08,
      "loss": 0.3035,
      "step": 67660
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.40303196951868,
      "learning_rate": 6.269007634939594e-08,
      "loss": 0.3016,
      "step": 67665
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.763644277122364,
      "learning_rate": 6.249588384670491e-08,
      "loss": 0.3089,
      "step": 67670
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.604362611174508,
      "learning_rate": 6.230199069034437e-08,
      "loss": 0.2903,
      "step": 67675
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.348688564146852,
      "learning_rate": 6.210839689206938e-08,
      "loss": 0.2995,
      "step": 67680
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.110360691380537,
      "learning_rate": 6.191510246361776e-08,
      "loss": 0.2925,
      "step": 67685
    },
    {
      "epoch": 2.86,
      "grad_norm": 4.13776166576163,
      "learning_rate": 6.172210741670737e-08,
      "loss": 0.3171,
      "step": 67690
    },
    {
      "epoch": 2.86,
      "grad_norm": 3.619616213327034,
      "learning_rate": 6.152941176303995e-08,
      "loss": 0.2922,
      "step": 67695
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.9368302678334337,
      "learning_rate": 6.133701551429727e-08,
      "loss": 0.3457,
      "step": 67700
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.13122736886974,
      "learning_rate": 6.114491868214445e-08,
      "loss": 0.2926,
      "step": 67705
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.939790121370067,
      "learning_rate": 6.095312127822661e-08,
      "loss": 0.284,
      "step": 67710
    },
    {
      "epoch": 2.87,
      "grad_norm": 5.278099854757711,
      "learning_rate": 6.076162331417279e-08,
      "loss": 0.3234,
      "step": 67715
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.9564192143498924,
      "learning_rate": 6.057042480159315e-08,
      "loss": 0.2896,
      "step": 67720
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7215124084459044,
      "learning_rate": 6.037952575207895e-08,
      "loss": 0.2969,
      "step": 67725
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.6967953692561095,
      "learning_rate": 6.018892617720373e-08,
      "loss": 0.3069,
      "step": 67730
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.309509381670668,
      "learning_rate": 5.999862608852325e-08,
      "loss": 0.3088,
      "step": 67735
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.079767653835693,
      "learning_rate": 5.980862549757494e-08,
      "loss": 0.294,
      "step": 67740
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7731058588156903,
      "learning_rate": 5.96189244158779e-08,
      "loss": 0.3094,
      "step": 67745
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.520293549013381,
      "learning_rate": 5.9429522854932975e-08,
      "loss": 0.281,
      "step": 67750
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7976274879106366,
      "learning_rate": 5.924042082622317e-08,
      "loss": 0.3162,
      "step": 67755
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.012785851612927,
      "learning_rate": 5.90516183412132e-08,
      "loss": 0.309,
      "step": 67760
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7214814162351324,
      "learning_rate": 5.886311541134948e-08,
      "loss": 0.3075,
      "step": 67765
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7586493796062586,
      "learning_rate": 5.867491204806064e-08,
      "loss": 0.2982,
      "step": 67770
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.592946941566493,
      "learning_rate": 5.848700826275644e-08,
      "loss": 0.3012,
      "step": 67775
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.803601483320265,
      "learning_rate": 5.829940406682943e-08,
      "loss": 0.3036,
      "step": 67780
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.6753659553407987,
      "learning_rate": 5.811209947165386e-08,
      "loss": 0.269,
      "step": 67785
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.249177348010491,
      "learning_rate": 5.7925094488583964e-08,
      "loss": 0.3258,
      "step": 67790
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7360552671367717,
      "learning_rate": 5.773838912895846e-08,
      "loss": 0.3157,
      "step": 67795
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.6380251834841344,
      "learning_rate": 5.755198340409718e-08,
      "loss": 0.3045,
      "step": 67800
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.072445060656571,
      "learning_rate": 5.736587732529997e-08,
      "loss": 0.2807,
      "step": 67805
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.136005460084404,
      "learning_rate": 5.718007090385058e-08,
      "loss": 0.2782,
      "step": 67810
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.1103986138869315,
      "learning_rate": 5.69945641510139e-08,
      "loss": 0.3188,
      "step": 67815
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.6584240467300493,
      "learning_rate": 5.6809357078037034e-08,
      "loss": 0.2844,
      "step": 67820
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.0610861071155915,
      "learning_rate": 5.662444969614822e-08,
      "loss": 0.314,
      "step": 67825
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.753442711285399,
      "learning_rate": 5.643984201655794e-08,
      "loss": 0.291,
      "step": 67830
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.077994111311488,
      "learning_rate": 5.625553405045781e-08,
      "loss": 0.2938,
      "step": 67835
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7582268884852432,
      "learning_rate": 5.607152580902275e-08,
      "loss": 0.3189,
      "step": 67840
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.8401190108814585,
      "learning_rate": 5.588781730340831e-08,
      "loss": 0.2909,
      "step": 67845
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.243334684485353,
      "learning_rate": 5.570440854475223e-08,
      "loss": 0.3252,
      "step": 67850
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.1548263105078655,
      "learning_rate": 5.55212995441734e-08,
      "loss": 0.328,
      "step": 67855
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.749605414916934,
      "learning_rate": 5.5338490312774605e-08,
      "loss": 0.291,
      "step": 67860
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.435419386219366,
      "learning_rate": 5.515598086163754e-08,
      "loss": 0.2955,
      "step": 67865
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.397463558775545,
      "learning_rate": 5.4973771201827784e-08,
      "loss": 0.3271,
      "step": 67870
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.523892991725451,
      "learning_rate": 5.479186134439263e-08,
      "loss": 0.3337,
      "step": 67875
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.6840816927081423,
      "learning_rate": 5.461025130035991e-08,
      "loss": 0.2909,
      "step": 67880
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.9481938107116066,
      "learning_rate": 5.442894108074026e-08,
      "loss": 0.2847,
      "step": 67885
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.039275267035997,
      "learning_rate": 5.424793069652712e-08,
      "loss": 0.2917,
      "step": 67890
    },
    {
      "epoch": 2.87,
      "grad_norm": 4.0444081524366595,
      "learning_rate": 5.406722015869337e-08,
      "loss": 0.3163,
      "step": 67895
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7104128521835618,
      "learning_rate": 5.388680947819524e-08,
      "loss": 0.2813,
      "step": 67900
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.949874376163575,
      "learning_rate": 5.370669866597067e-08,
      "loss": 0.2806,
      "step": 67905
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.9228068709652106,
      "learning_rate": 5.352688773293924e-08,
      "loss": 0.3309,
      "step": 67910
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.8337707097222875,
      "learning_rate": 5.334737669000167e-08,
      "loss": 0.3244,
      "step": 67915
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.7026248893058886,
      "learning_rate": 5.3168165548042605e-08,
      "loss": 0.2998,
      "step": 67920
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.9338322418427825,
      "learning_rate": 5.298925431792612e-08,
      "loss": 0.3018,
      "step": 67925
    },
    {
      "epoch": 2.87,
      "grad_norm": 3.6065154851992363,
      "learning_rate": 5.28106430104991e-08,
      "loss": 0.3138,
      "step": 67930
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.878376198378528,
      "learning_rate": 5.2632331636590094e-08,
      "loss": 0.303,
      "step": 67935
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.135505471397503,
      "learning_rate": 5.245432020701047e-08,
      "loss": 0.2979,
      "step": 67940
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.655834798753016,
      "learning_rate": 5.227660873255158e-08,
      "loss": 0.3165,
      "step": 67945
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.229688844348593,
      "learning_rate": 5.209919722398815e-08,
      "loss": 0.318,
      "step": 67950
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.9280026145734683,
      "learning_rate": 5.1922085692076016e-08,
      "loss": 0.3234,
      "step": 67955
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.694440796313576,
      "learning_rate": 5.1745274147552146e-08,
      "loss": 0.3336,
      "step": 67960
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.967528818820221,
      "learning_rate": 5.156876260113741e-08,
      "loss": 0.3003,
      "step": 67965
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.071088634268772,
      "learning_rate": 5.139255106353214e-08,
      "loss": 0.2832,
      "step": 67970
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.114427385116013,
      "learning_rate": 5.121663954542e-08,
      "loss": 0.3065,
      "step": 67975
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.110623937700965,
      "learning_rate": 5.10410280574658e-08,
      "loss": 0.289,
      "step": 67980
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.8918191297522853,
      "learning_rate": 5.086571661031658e-08,
      "loss": 0.3137,
      "step": 67985
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.03231752129029,
      "learning_rate": 5.069070521460051e-08,
      "loss": 0.31,
      "step": 67990
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.4176898448604147,
      "learning_rate": 5.051599388092854e-08,
      "loss": 0.3079,
      "step": 67995
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.954439025992426,
      "learning_rate": 5.0341582619892196e-08,
      "loss": 0.2985,
      "step": 68000
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.4728957709554305,
      "learning_rate": 5.016747144206691e-08,
      "loss": 0.3245,
      "step": 68005
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.040433767784017,
      "learning_rate": 4.9993660358007013e-08,
      "loss": 0.3241,
      "step": 68010
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.8812577908943213,
      "learning_rate": 4.982014937825075e-08,
      "loss": 0.3119,
      "step": 68015
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.23459266462151,
      "learning_rate": 4.9646938513318034e-08,
      "loss": 0.2992,
      "step": 68020
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.188721997192261,
      "learning_rate": 4.9474027773709906e-08,
      "loss": 0.2986,
      "step": 68025
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.9604939135563164,
      "learning_rate": 4.930141716990855e-08,
      "loss": 0.3271,
      "step": 68030
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.7733903782639002,
      "learning_rate": 4.912910671237947e-08,
      "loss": 0.3125,
      "step": 68035
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.695846733823223,
      "learning_rate": 4.895709641156987e-08,
      "loss": 0.3219,
      "step": 68040
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.953916411472082,
      "learning_rate": 4.878538627790752e-08,
      "loss": 0.3027,
      "step": 68045
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.49525420624861,
      "learning_rate": 4.861397632180243e-08,
      "loss": 0.2831,
      "step": 68050
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.8251415167948046,
      "learning_rate": 4.844286655364794e-08,
      "loss": 0.2996,
      "step": 68055
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.536336657712159,
      "learning_rate": 4.827205698381687e-08,
      "loss": 0.3086,
      "step": 68060
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.662488379414853,
      "learning_rate": 4.810154762266483e-08,
      "loss": 0.3164,
      "step": 68065
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.105437912824385,
      "learning_rate": 4.793133848053022e-08,
      "loss": 0.3091,
      "step": 68070
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.976310772728087,
      "learning_rate": 4.7761429567731444e-08,
      "loss": 0.333,
      "step": 68075
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.655764157848068,
      "learning_rate": 4.759182089456971e-08,
      "loss": 0.3132,
      "step": 68080
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.160088455234844,
      "learning_rate": 4.742251247132845e-08,
      "loss": 0.3188,
      "step": 68085
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.6052639029227476,
      "learning_rate": 4.725350430827225e-08,
      "loss": 0.3171,
      "step": 68090
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.939240023965245,
      "learning_rate": 4.7084796415646785e-08,
      "loss": 0.3316,
      "step": 68095
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.057461533580917,
      "learning_rate": 4.691638880368054e-08,
      "loss": 0.286,
      "step": 68100
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.9405878143644353,
      "learning_rate": 4.674828148258481e-08,
      "loss": 0.3055,
      "step": 68105
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.7936523463053398,
      "learning_rate": 4.658047446254976e-08,
      "loss": 0.3141,
      "step": 68110
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.6367126046531073,
      "learning_rate": 4.6412967753750035e-08,
      "loss": 0.2918,
      "step": 68115
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.43078455807695,
      "learning_rate": 4.624576136634029e-08,
      "loss": 0.2894,
      "step": 68120
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.636939188101196,
      "learning_rate": 4.6078855310459084e-08,
      "loss": 0.2827,
      "step": 68125
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.818867269756552,
      "learning_rate": 4.591224959622387e-08,
      "loss": 0.3114,
      "step": 68130
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.046523286048708,
      "learning_rate": 4.574594423373657e-08,
      "loss": 0.2978,
      "step": 68135
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.707557155336955,
      "learning_rate": 4.557993923307968e-08,
      "loss": 0.2904,
      "step": 68140
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.538963217447223,
      "learning_rate": 4.541423460431738e-08,
      "loss": 0.3441,
      "step": 68145
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.8956082662197864,
      "learning_rate": 4.52488303574955e-08,
      "loss": 0.3008,
      "step": 68150
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.128782649758923,
      "learning_rate": 4.508372650264214e-08,
      "loss": 0.3189,
      "step": 68155
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.099664076186443,
      "learning_rate": 4.491892304976764e-08,
      "loss": 0.3163,
      "step": 68160
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.606471482014266,
      "learning_rate": 4.4754420008862876e-08,
      "loss": 0.2862,
      "step": 68165
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.365835569663179,
      "learning_rate": 4.4590217389901545e-08,
      "loss": 0.2981,
      "step": 68170
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.8880025185480425,
      "learning_rate": 4.442631520283902e-08,
      "loss": 0.2907,
      "step": 68175
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.9339411742608488,
      "learning_rate": 4.4262713457611795e-08,
      "loss": 0.3187,
      "step": 68180
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.637726328749896,
      "learning_rate": 4.40994121641386e-08,
      "loss": 0.2993,
      "step": 68185
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.426330745195855,
      "learning_rate": 4.393641133231985e-08,
      "loss": 0.2843,
      "step": 68190
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.571466892785576,
      "learning_rate": 4.3773710972038194e-08,
      "loss": 0.285,
      "step": 68195
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.124900144647663,
      "learning_rate": 4.361131109315686e-08,
      "loss": 0.2893,
      "step": 68200
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.178267785277212,
      "learning_rate": 4.344921170552241e-08,
      "loss": 0.3025,
      "step": 68205
    },
    {
      "epoch": 2.89,
      "grad_norm": 5.833346353458756,
      "learning_rate": 4.3287412818962535e-08,
      "loss": 0.3035,
      "step": 68210
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.956584964131083,
      "learning_rate": 4.312591444328607e-08,
      "loss": 0.2689,
      "step": 68215
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.6898939086712077,
      "learning_rate": 4.296471658828461e-08,
      "loss": 0.2873,
      "step": 68220
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.8102327435330277,
      "learning_rate": 4.28038192637309e-08,
      "loss": 0.2941,
      "step": 68225
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.832324182113011,
      "learning_rate": 4.264322247937991e-08,
      "loss": 0.2961,
      "step": 68230
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.165332839781083,
      "learning_rate": 4.2482926244967745e-08,
      "loss": 0.306,
      "step": 68235
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.800336400847749,
      "learning_rate": 4.232293057021275e-08,
      "loss": 0.295,
      "step": 68240
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.3388994629932185,
      "learning_rate": 4.216323546481549e-08,
      "loss": 0.2739,
      "step": 68245
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.145580320385166,
      "learning_rate": 4.200384093845711e-08,
      "loss": 0.2809,
      "step": 68250
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.253355485403765,
      "learning_rate": 4.18447470008021e-08,
      "loss": 0.3169,
      "step": 68255
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.99507858763937,
      "learning_rate": 4.1685953661494974e-08,
      "loss": 0.2938,
      "step": 68260
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.525424363631607,
      "learning_rate": 4.152746093016358e-08,
      "loss": 0.2797,
      "step": 68265
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.891564080415869,
      "learning_rate": 4.136926881641634e-08,
      "loss": 0.3034,
      "step": 68270
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.9829505297849526,
      "learning_rate": 4.1211377329844485e-08,
      "loss": 0.285,
      "step": 68275
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.680944982884361,
      "learning_rate": 4.105378648001979e-08,
      "loss": 0.2968,
      "step": 68280
    },
    {
      "epoch": 2.89,
      "grad_norm": 5.936474565118909,
      "learning_rate": 4.0896496276496836e-08,
      "loss": 0.2878,
      "step": 68285
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.8885327228097024,
      "learning_rate": 4.073950672881244e-08,
      "loss": 0.3222,
      "step": 68290
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.863223541092429,
      "learning_rate": 4.058281784648343e-08,
      "loss": 0.2749,
      "step": 68295
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.070152976423319,
      "learning_rate": 4.0426429639009446e-08,
      "loss": 0.3059,
      "step": 68300
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.671708676789781,
      "learning_rate": 4.027034211587233e-08,
      "loss": 0.2955,
      "step": 68305
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.8120203794677217,
      "learning_rate": 4.011455528653563e-08,
      "loss": 0.3222,
      "step": 68310
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.3692484527773665,
      "learning_rate": 3.9959069160442896e-08,
      "loss": 0.3078,
      "step": 68315
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.4630689025834958,
      "learning_rate": 3.98038837470216e-08,
      "loss": 0.2977,
      "step": 68320
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.8004478493882146,
      "learning_rate": 3.9648999055680316e-08,
      "loss": 0.3249,
      "step": 68325
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.955375230998622,
      "learning_rate": 3.949441509580931e-08,
      "loss": 0.3289,
      "step": 68330
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.838179596990951,
      "learning_rate": 3.934013187677943e-08,
      "loss": 0.3276,
      "step": 68335
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.615824165264497,
      "learning_rate": 3.918614940794596e-08,
      "loss": 0.3002,
      "step": 68340
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.6692548039891446,
      "learning_rate": 3.903246769864366e-08,
      "loss": 0.3151,
      "step": 68345
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.017934018585339,
      "learning_rate": 3.887908675818952e-08,
      "loss": 0.3241,
      "step": 68350
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.26999622719143,
      "learning_rate": 3.8726006595883325e-08,
      "loss": 0.3229,
      "step": 68355
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.800516814896526,
      "learning_rate": 3.857322722100487e-08,
      "loss": 0.2897,
      "step": 68360
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.3168952602091473,
      "learning_rate": 3.842074864281786e-08,
      "loss": 0.2962,
      "step": 68365
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.862070172714224,
      "learning_rate": 3.826857087056546e-08,
      "loss": 0.305,
      "step": 68370
    },
    {
      "epoch": 2.89,
      "grad_norm": 4.8556357294495065,
      "learning_rate": 3.811669391347472e-08,
      "loss": 0.309,
      "step": 68375
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.718769022098439,
      "learning_rate": 3.7965117780752735e-08,
      "loss": 0.293,
      "step": 68380
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.6890938660963575,
      "learning_rate": 3.7813842481589926e-08,
      "loss": 0.3101,
      "step": 68385
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.985691098986346,
      "learning_rate": 3.7662868025157284e-08,
      "loss": 0.2941,
      "step": 68390
    },
    {
      "epoch": 2.89,
      "grad_norm": 5.04899522587584,
      "learning_rate": 3.751219442060749e-08,
      "loss": 0.3059,
      "step": 68395
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.9318863193804563,
      "learning_rate": 3.7361821677076024e-08,
      "loss": 0.2895,
      "step": 68400
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.0831677984328945,
      "learning_rate": 3.721174980367948e-08,
      "loss": 0.2987,
      "step": 68405
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.928370033903755,
      "learning_rate": 3.706197880951612e-08,
      "loss": 0.2746,
      "step": 68410
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.636775568518349,
      "learning_rate": 3.691250870366647e-08,
      "loss": 0.2998,
      "step": 68415
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.959377033082534,
      "learning_rate": 3.6763339495191616e-08,
      "loss": 0.3164,
      "step": 68420
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.309048549000451,
      "learning_rate": 3.661447119313599e-08,
      "loss": 0.2983,
      "step": 68425
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.815767290860579,
      "learning_rate": 3.646590380652515e-08,
      "loss": 0.3526,
      "step": 68430
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.8075273010348276,
      "learning_rate": 3.631763734436577e-08,
      "loss": 0.2889,
      "step": 68435
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.4820351964858527,
      "learning_rate": 3.6169671815646787e-08,
      "loss": 0.2809,
      "step": 68440
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.6963708563811273,
      "learning_rate": 3.60220072293399e-08,
      "loss": 0.2844,
      "step": 68445
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.482981463138997,
      "learning_rate": 3.5874643594396296e-08,
      "loss": 0.2877,
      "step": 68450
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.8192170577883817,
      "learning_rate": 3.572758091975048e-08,
      "loss": 0.3183,
      "step": 68455
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.178411123961507,
      "learning_rate": 3.558081921431922e-08,
      "loss": 0.2991,
      "step": 68460
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.159345213807564,
      "learning_rate": 3.543435848699928e-08,
      "loss": 0.3404,
      "step": 68465
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.6826306813228604,
      "learning_rate": 3.5288198746670774e-08,
      "loss": 0.3028,
      "step": 68470
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.9362373809152778,
      "learning_rate": 3.514234000219496e-08,
      "loss": 0.3113,
      "step": 68475
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.550782741689252,
      "learning_rate": 3.499678226241421e-08,
      "loss": 0.2776,
      "step": 68480
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.9203915631799466,
      "learning_rate": 3.4851525536153675e-08,
      "loss": 0.3144,
      "step": 68485
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.8585669134026475,
      "learning_rate": 3.4706569832220206e-08,
      "loss": 0.3336,
      "step": 68490
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.164082168953407,
      "learning_rate": 3.456191515940177e-08,
      "loss": 0.3148,
      "step": 68495
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.8835248821734143,
      "learning_rate": 3.441756152646747e-08,
      "loss": 0.2927,
      "step": 68500
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.372668368564303,
      "learning_rate": 3.427350894217085e-08,
      "loss": 0.3269,
      "step": 68505
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.8110977839958133,
      "learning_rate": 3.412975741524327e-08,
      "loss": 0.3032,
      "step": 68510
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.2603150604024265,
      "learning_rate": 3.3986306954401635e-08,
      "loss": 0.3142,
      "step": 68515
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.9767729319615404,
      "learning_rate": 3.3843157568342333e-08,
      "loss": 0.2896,
      "step": 68520
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.524685702552604,
      "learning_rate": 3.370030926574397e-08,
      "loss": 0.2769,
      "step": 68525
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.561871077630698,
      "learning_rate": 3.3557762055267396e-08,
      "loss": 0.2983,
      "step": 68530
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.144976601722943,
      "learning_rate": 3.3415515945554037e-08,
      "loss": 0.32,
      "step": 68535
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.830215380064149,
      "learning_rate": 3.327357094522865e-08,
      "loss": 0.2793,
      "step": 68540
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.300268188756827,
      "learning_rate": 3.3131927062896585e-08,
      "loss": 0.3009,
      "step": 68545
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.8675226492865487,
      "learning_rate": 3.2990584307145966e-08,
      "loss": 0.3336,
      "step": 68550
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.008794901490101,
      "learning_rate": 3.2849542686544946e-08,
      "loss": 0.3048,
      "step": 68555
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.938411351144604,
      "learning_rate": 3.270880220964501e-08,
      "loss": 0.3131,
      "step": 68560
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.7019420233312235,
      "learning_rate": 3.256836288497878e-08,
      "loss": 0.2871,
      "step": 68565
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.168885781507564,
      "learning_rate": 3.2428224721060556e-08,
      "loss": 0.3184,
      "step": 68570
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.502478347289495,
      "learning_rate": 3.228838772638632e-08,
      "loss": 0.3164,
      "step": 68575
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.311865579528974,
      "learning_rate": 3.2148851909434845e-08,
      "loss": 0.3122,
      "step": 68580
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.672104511879892,
      "learning_rate": 3.200961727866492e-08,
      "loss": 0.3064,
      "step": 68585
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.1320454978654455,
      "learning_rate": 3.1870683842518126e-08,
      "loss": 0.3185,
      "step": 68590
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.537664079395672,
      "learning_rate": 3.173205160941828e-08,
      "loss": 0.3054,
      "step": 68595
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.038821187842253,
      "learning_rate": 3.1593720587768664e-08,
      "loss": 0.293,
      "step": 68600
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.877998533666184,
      "learning_rate": 3.145569078595756e-08,
      "loss": 0.3044,
      "step": 68605
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.643096592138452,
      "learning_rate": 3.131796221235217e-08,
      "loss": 0.3083,
      "step": 68610
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.982477064880714,
      "learning_rate": 3.118053487530304e-08,
      "loss": 0.3331,
      "step": 68615
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.806037529784277,
      "learning_rate": 3.1043408783142383e-08,
      "loss": 0.3017,
      "step": 68620
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.811151265361242,
      "learning_rate": 3.090658394418245e-08,
      "loss": 0.3,
      "step": 68625
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.767268645171214,
      "learning_rate": 3.0770060366720476e-08,
      "loss": 0.2962,
      "step": 68630
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.5548393994982024,
      "learning_rate": 3.063383805903153e-08,
      "loss": 0.2927,
      "step": 68635
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.465699227101694,
      "learning_rate": 3.0497917029375655e-08,
      "loss": 0.2955,
      "step": 68640
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.9169814682200492,
      "learning_rate": 3.0362297285992935e-08,
      "loss": 0.282,
      "step": 68645
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.028086862289432,
      "learning_rate": 3.022697883710513e-08,
      "loss": 0.3031,
      "step": 68650
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.201700573605033,
      "learning_rate": 3.0091961690917324e-08,
      "loss": 0.321,
      "step": 68655
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.5429219029263437,
      "learning_rate": 2.99572458556141e-08,
      "loss": 0.2997,
      "step": 68660
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.8126499059716075,
      "learning_rate": 2.9822831339363364e-08,
      "loss": 0.287,
      "step": 68665
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.048774630790299,
      "learning_rate": 2.968871815031471e-08,
      "loss": 0.3126,
      "step": 68670
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.7712885210146294,
      "learning_rate": 2.955490629659774e-08,
      "loss": 0.2922,
      "step": 68675
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.7139500851331158,
      "learning_rate": 2.9421395786326522e-08,
      "loss": 0.2918,
      "step": 68680
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.8007764773611212,
      "learning_rate": 2.9288186627594584e-08,
      "loss": 0.2952,
      "step": 68685
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.5992752072775716,
      "learning_rate": 2.9155278828477686e-08,
      "loss": 0.3125,
      "step": 68690
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.7670752312504683,
      "learning_rate": 2.9022672397034378e-08,
      "loss": 0.2988,
      "step": 68695
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.017223261357962,
      "learning_rate": 2.889036734130435e-08,
      "loss": 0.3077,
      "step": 68700
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.2384545886906855,
      "learning_rate": 2.8758363669308397e-08,
      "loss": 0.2952,
      "step": 68705
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.676779163365829,
      "learning_rate": 2.8626661389049016e-08,
      "loss": 0.3248,
      "step": 68710
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.5404653850526087,
      "learning_rate": 2.8495260508512036e-08,
      "loss": 0.3257,
      "step": 68715
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.794028023869922,
      "learning_rate": 2.8364161035663307e-08,
      "loss": 0.3269,
      "step": 68720
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.026428704437321,
      "learning_rate": 2.8233362978450918e-08,
      "loss": 0.3126,
      "step": 68725
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.5761846931181487,
      "learning_rate": 2.810286634480519e-08,
      "loss": 0.3174,
      "step": 68730
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.335833305973485,
      "learning_rate": 2.797267114263702e-08,
      "loss": 0.3051,
      "step": 68735
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.833484952058422,
      "learning_rate": 2.7842777379840647e-08,
      "loss": 0.2769,
      "step": 68740
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.317548014082353,
      "learning_rate": 2.7713185064290328e-08,
      "loss": 0.3293,
      "step": 68745
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.579661563467071,
      "learning_rate": 2.7583894203843664e-08,
      "loss": 0.3334,
      "step": 68750
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.9621202681466015,
      "learning_rate": 2.7454904806338277e-08,
      "loss": 0.3024,
      "step": 68755
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.8544583587616192,
      "learning_rate": 2.7326216879595136e-08,
      "loss": 0.2946,
      "step": 68760
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.0146741418993095,
      "learning_rate": 2.7197830431415772e-08,
      "loss": 0.2904,
      "step": 68765
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.241244226482878,
      "learning_rate": 2.706974546958452e-08,
      "loss": 0.284,
      "step": 68770
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.498336471688792,
      "learning_rate": 2.6941962001865718e-08,
      "loss": 0.2746,
      "step": 68775
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.94161705685398,
      "learning_rate": 2.6814480036007617e-08,
      "loss": 0.3009,
      "step": 68780
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.6808520872538018,
      "learning_rate": 2.6687299579739034e-08,
      "loss": 0.3153,
      "step": 68785
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.318316742835231,
      "learning_rate": 2.6560420640769356e-08,
      "loss": 0.2821,
      "step": 68790
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.946723742234188,
      "learning_rate": 2.6433843226791877e-08,
      "loss": 0.31,
      "step": 68795
    },
    {
      "epoch": 2.91,
      "grad_norm": 5.058627236064796,
      "learning_rate": 2.6307567345480457e-08,
      "loss": 0.3214,
      "step": 68800
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.6815746238629887,
      "learning_rate": 2.6181593004490636e-08,
      "loss": 0.3222,
      "step": 68805
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.147851876763575,
      "learning_rate": 2.605592021145964e-08,
      "loss": 0.3001,
      "step": 68810
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.7776485170453227,
      "learning_rate": 2.5930548974007485e-08,
      "loss": 0.3028,
      "step": 68815
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.8478125280383972,
      "learning_rate": 2.5805479299734204e-08,
      "loss": 0.2895,
      "step": 68820
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.5077753251310035,
      "learning_rate": 2.5680711196222618e-08,
      "loss": 0.2964,
      "step": 68825
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.742860252139474,
      "learning_rate": 2.5556244671037787e-08,
      "loss": 0.3063,
      "step": 68830
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.228677553724897,
      "learning_rate": 2.543207973172479e-08,
      "loss": 0.3318,
      "step": 68835
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.8349712403665177,
      "learning_rate": 2.530821638581149e-08,
      "loss": 0.2846,
      "step": 68840
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.776124507539609,
      "learning_rate": 2.5184654640807992e-08,
      "loss": 0.2938,
      "step": 68845
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.311411118780991,
      "learning_rate": 2.5061394504204973e-08,
      "loss": 0.3311,
      "step": 68850
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.5077521269130503,
      "learning_rate": 2.4938435983475338e-08,
      "loss": 0.2958,
      "step": 68855
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.6744346529510157,
      "learning_rate": 2.481577908607369e-08,
      "loss": 0.294,
      "step": 68860
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.8865190295125815,
      "learning_rate": 2.4693423819436846e-08,
      "loss": 0.2894,
      "step": 68865
    },
    {
      "epoch": 2.91,
      "grad_norm": 3.954940760338175,
      "learning_rate": 2.4571370190982214e-08,
      "loss": 0.3076,
      "step": 68870
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.118453112399337,
      "learning_rate": 2.444961820810998e-08,
      "loss": 0.3028,
      "step": 68875
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.064592390970007,
      "learning_rate": 2.4328167878200916e-08,
      "loss": 0.3044,
      "step": 68880
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.180993353802298,
      "learning_rate": 2.4207019208619674e-08,
      "loss": 0.3384,
      "step": 68885
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.392095763137633,
      "learning_rate": 2.408617220670928e-08,
      "loss": 0.3039,
      "step": 68890
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.772506295636982,
      "learning_rate": 2.3965626879797754e-08,
      "loss": 0.3036,
      "step": 68895
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.5832293888868754,
      "learning_rate": 2.3845383235192587e-08,
      "loss": 0.3002,
      "step": 68900
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.562603661030853,
      "learning_rate": 2.3725441280184057e-08,
      "loss": 0.2979,
      "step": 68905
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.5511021122904136,
      "learning_rate": 2.3605801022044128e-08,
      "loss": 0.2974,
      "step": 68910
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.964719018619828,
      "learning_rate": 2.3486462468025882e-08,
      "loss": 0.3138,
      "step": 68915
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.6642424568558845,
      "learning_rate": 2.3367425625364647e-08,
      "loss": 0.3086,
      "step": 68920
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.833275552405031,
      "learning_rate": 2.3248690501277426e-08,
      "loss": 0.3238,
      "step": 68925
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.554982207288336,
      "learning_rate": 2.313025710296235e-08,
      "loss": 0.2713,
      "step": 68930
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.803714723639175,
      "learning_rate": 2.3012125437599787e-08,
      "loss": 0.3123,
      "step": 68935
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.516503240739867,
      "learning_rate": 2.2894295512352337e-08,
      "loss": 0.3027,
      "step": 68940
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.725971668701787,
      "learning_rate": 2.2776767334362627e-08,
      "loss": 0.3119,
      "step": 68945
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.91101511074878,
      "learning_rate": 2.2659540910757173e-08,
      "loss": 0.3213,
      "step": 68950
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.580277042074699,
      "learning_rate": 2.2542616248642514e-08,
      "loss": 0.3171,
      "step": 68955
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.058421775636477,
      "learning_rate": 2.2425993355106866e-08,
      "loss": 0.2942,
      "step": 68960
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.9027736599066487,
      "learning_rate": 2.23096722372218e-08,
      "loss": 0.3027,
      "step": 68965
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.0634263265857955,
      "learning_rate": 2.2193652902038898e-08,
      "loss": 0.2908,
      "step": 68970
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.108222895679821,
      "learning_rate": 2.207793535659253e-08,
      "loss": 0.3183,
      "step": 68975
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.7491728915013462,
      "learning_rate": 2.1962519607897638e-08,
      "loss": 0.2822,
      "step": 68980
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.783634297216293,
      "learning_rate": 2.184740566295196e-08,
      "loss": 0.2915,
      "step": 68985
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.850384182711361,
      "learning_rate": 2.1732593528734912e-08,
      "loss": 0.3107,
      "step": 68990
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.9553156520898023,
      "learning_rate": 2.1618083212205932e-08,
      "loss": 0.3376,
      "step": 68995
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.973669423882026,
      "learning_rate": 2.150387472030835e-08,
      "loss": 0.2969,
      "step": 69000
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.975234145323945,
      "learning_rate": 2.1389968059966625e-08,
      "loss": 0.2958,
      "step": 69005
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.6755181208108856,
      "learning_rate": 2.127636323808635e-08,
      "loss": 0.3089,
      "step": 69010
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.937416320476416,
      "learning_rate": 2.1163060261554237e-08,
      "loss": 0.3108,
      "step": 69015
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.9698618108124495,
      "learning_rate": 2.1050059137240343e-08,
      "loss": 0.2994,
      "step": 69020
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.281887091340919,
      "learning_rate": 2.0937359871995855e-08,
      "loss": 0.2963,
      "step": 69025
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.8023490505716215,
      "learning_rate": 2.082496247265198e-08,
      "loss": 0.2874,
      "step": 69030
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.418863103343946,
      "learning_rate": 2.0712866946024922e-08,
      "loss": 0.302,
      "step": 69035
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.5946670112554844,
      "learning_rate": 2.060107329890926e-08,
      "loss": 0.3152,
      "step": 69040
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.797259351212995,
      "learning_rate": 2.0489581538083446e-08,
      "loss": 0.3171,
      "step": 69045
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.032791676822362,
      "learning_rate": 2.037839167030653e-08,
      "loss": 0.3008,
      "step": 69050
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.069340185753086,
      "learning_rate": 2.026750370231978e-08,
      "loss": 0.2943,
      "step": 69055
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.2778729337520724,
      "learning_rate": 2.0156917640846152e-08,
      "loss": 0.2931,
      "step": 69060
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.938323937829891,
      "learning_rate": 2.004663349258973e-08,
      "loss": 0.301,
      "step": 69065
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.2776538753526,
      "learning_rate": 1.9936651264237384e-08,
      "loss": 0.3039,
      "step": 69070
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.278914080949418,
      "learning_rate": 1.9826970962456006e-08,
      "loss": 0.297,
      "step": 69075
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.9373754191968917,
      "learning_rate": 1.9717592593896384e-08,
      "loss": 0.294,
      "step": 69080
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.7047491550322413,
      "learning_rate": 1.9608516165188774e-08,
      "loss": 0.3022,
      "step": 69085
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.590363145597009,
      "learning_rate": 1.949974168294677e-08,
      "loss": 0.308,
      "step": 69090
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.67948306767533,
      "learning_rate": 1.93912691537651e-08,
      "loss": 0.294,
      "step": 69095
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.756401723490771,
      "learning_rate": 1.9283098584219617e-08,
      "loss": 0.3155,
      "step": 69100
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.670150619972396,
      "learning_rate": 1.917522998086896e-08,
      "loss": 0.2968,
      "step": 69105
    },
    {
      "epoch": 2.92,
      "grad_norm": 5.244452084678132,
      "learning_rate": 1.906766335025234e-08,
      "loss": 0.321,
      "step": 69110
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.236146000764753,
      "learning_rate": 1.896039869889177e-08,
      "loss": 0.3371,
      "step": 69115
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.9737098962402206,
      "learning_rate": 1.8853436033289818e-08,
      "loss": 0.3235,
      "step": 69120
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.68691947709129,
      "learning_rate": 1.8746775359931858e-08,
      "loss": 0.3023,
      "step": 69125
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.572242257399972,
      "learning_rate": 1.8640416685283824e-08,
      "loss": 0.2954,
      "step": 69130
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.8642255241899517,
      "learning_rate": 1.853436001579445e-08,
      "loss": 0.3138,
      "step": 69135
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.573897828752867,
      "learning_rate": 1.842860535789359e-08,
      "loss": 0.3125,
      "step": 69140
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.5674988929690343,
      "learning_rate": 1.8323152717992233e-08,
      "loss": 0.2777,
      "step": 69145
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.7738493782206657,
      "learning_rate": 1.82180021024847e-08,
      "loss": 0.2959,
      "step": 69150
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.115233207672061,
      "learning_rate": 1.8113153517744787e-08,
      "loss": 0.3114,
      "step": 69155
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.8923182051584897,
      "learning_rate": 1.8008606970130184e-08,
      "loss": 0.3001,
      "step": 69160
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.7537788036210427,
      "learning_rate": 1.7904362465978598e-08,
      "loss": 0.2928,
      "step": 69165
    },
    {
      "epoch": 2.93,
      "grad_norm": 5.026913644297368,
      "learning_rate": 1.780042001161053e-08,
      "loss": 0.3019,
      "step": 69170
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.327557905748171,
      "learning_rate": 1.7696779613327053e-08,
      "loss": 0.2951,
      "step": 69175
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.653596385605482,
      "learning_rate": 1.7593441277412028e-08,
      "loss": 0.3108,
      "step": 69180
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.7794504415823154,
      "learning_rate": 1.7490405010130994e-08,
      "loss": 0.31,
      "step": 69185
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.079691325939507,
      "learning_rate": 1.738767081772952e-08,
      "loss": 0.2985,
      "step": 69190
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.4603248499999957,
      "learning_rate": 1.7285238706437613e-08,
      "loss": 0.2929,
      "step": 69195
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.932217135225119,
      "learning_rate": 1.7183108682463644e-08,
      "loss": 0.3269,
      "step": 69200
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.817301561769997,
      "learning_rate": 1.7081280752000995e-08,
      "loss": 0.3016,
      "step": 69205
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.6481721626099834,
      "learning_rate": 1.6979754921222503e-08,
      "loss": 0.2808,
      "step": 69210
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.281323146373273,
      "learning_rate": 1.68785311962838e-08,
      "loss": 0.3473,
      "step": 69215
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.8704246587731785,
      "learning_rate": 1.6777609583321085e-08,
      "loss": 0.2759,
      "step": 69220
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.681892559292812,
      "learning_rate": 1.6676990088453915e-08,
      "loss": 0.2876,
      "step": 69225
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.8056899471265604,
      "learning_rate": 1.6576672717781296e-08,
      "loss": 0.2995,
      "step": 69230
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.0175952184236134,
      "learning_rate": 1.6476657477386137e-08,
      "loss": 0.2815,
      "step": 69235
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.01308490210128,
      "learning_rate": 1.6376944373331926e-08,
      "loss": 0.2945,
      "step": 69240
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.789007634418479,
      "learning_rate": 1.6277533411663272e-08,
      "loss": 0.3056,
      "step": 69245
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.067278453391076,
      "learning_rate": 1.6178424598408128e-08,
      "loss": 0.3047,
      "step": 69250
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.380044302466743,
      "learning_rate": 1.6079617939575022e-08,
      "loss": 0.3121,
      "step": 69255
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.736047962883004,
      "learning_rate": 1.5981113441153608e-08,
      "loss": 0.3034,
      "step": 69260
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.8368476335647292,
      "learning_rate": 1.5882911109116884e-08,
      "loss": 0.2705,
      "step": 69265
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.588978058696701,
      "learning_rate": 1.5785010949417868e-08,
      "loss": 0.3193,
      "step": 69270
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.219559886938127,
      "learning_rate": 1.5687412967991812e-08,
      "loss": 0.3223,
      "step": 69275
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.976212154360366,
      "learning_rate": 1.5590117170756204e-08,
      "loss": 0.2881,
      "step": 69280
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.7494904366071196,
      "learning_rate": 1.5493123563610212e-08,
      "loss": 0.3126,
      "step": 69285
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.9506167151953298,
      "learning_rate": 1.539643215243358e-08,
      "loss": 0.3194,
      "step": 69290
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.248027628814416,
      "learning_rate": 1.5300042943088288e-08,
      "loss": 0.3029,
      "step": 69295
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.3778353984149656,
      "learning_rate": 1.5203955941418547e-08,
      "loss": 0.3069,
      "step": 69300
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.317202795306347,
      "learning_rate": 1.510817115325025e-08,
      "loss": 0.3049,
      "step": 69305
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.8649890827610784,
      "learning_rate": 1.5012688584389868e-08,
      "loss": 0.313,
      "step": 69310
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.123870003595725,
      "learning_rate": 1.49175082406261e-08,
      "loss": 0.3259,
      "step": 69315
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.147584144428297,
      "learning_rate": 1.4822630127729886e-08,
      "loss": 0.327,
      "step": 69320
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.049729660311289,
      "learning_rate": 1.4728054251453849e-08,
      "loss": 0.3164,
      "step": 69325
    },
    {
      "epoch": 2.93,
      "grad_norm": 5.027307755566849,
      "learning_rate": 1.4633780617530624e-08,
      "loss": 0.3027,
      "step": 69330
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.265599407284306,
      "learning_rate": 1.4539809231676194e-08,
      "loss": 0.3027,
      "step": 69335
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.9182774952137804,
      "learning_rate": 1.4446140099588224e-08,
      "loss": 0.2944,
      "step": 69340
    },
    {
      "epoch": 2.93,
      "grad_norm": 4.121998798297339,
      "learning_rate": 1.435277322694606e-08,
      "loss": 0.303,
      "step": 69345
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.861249893732703,
      "learning_rate": 1.4259708619408507e-08,
      "loss": 0.3005,
      "step": 69350
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.874154709892759,
      "learning_rate": 1.4166946282619387e-08,
      "loss": 0.2794,
      "step": 69355
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.787112512513606,
      "learning_rate": 1.4074486222201978e-08,
      "loss": 0.2815,
      "step": 69360
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.931507758482797,
      "learning_rate": 1.3982328443761795e-08,
      "loss": 0.3014,
      "step": 69365
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.973964992026249,
      "learning_rate": 1.389047295288659e-08,
      "loss": 0.322,
      "step": 69370
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.842996014133042,
      "learning_rate": 1.3798919755144691e-08,
      "loss": 0.3091,
      "step": 69375
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.761891792967513,
      "learning_rate": 1.3707668856087208e-08,
      "loss": 0.3107,
      "step": 69380
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.12190106774251,
      "learning_rate": 1.3616720261245831e-08,
      "loss": 0.3397,
      "step": 69385
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.9791680016344486,
      "learning_rate": 1.3526073976135034e-08,
      "loss": 0.3128,
      "step": 69390
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.6165125449599858,
      "learning_rate": 1.343573000624987e-08,
      "loss": 0.2972,
      "step": 69395
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.952145508270869,
      "learning_rate": 1.3345688357068176e-08,
      "loss": 0.3297,
      "step": 69400
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.518136034238327,
      "learning_rate": 1.3255949034048921e-08,
      "loss": 0.3155,
      "step": 69405
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.6404407113721824,
      "learning_rate": 1.3166512042632751e-08,
      "loss": 0.294,
      "step": 69410
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.9755185847325256,
      "learning_rate": 1.307737738824144e-08,
      "loss": 0.3104,
      "step": 69415
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.290243292668716,
      "learning_rate": 1.2988545076279001e-08,
      "loss": 0.3134,
      "step": 69420
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.17867803795504,
      "learning_rate": 1.290001511213168e-08,
      "loss": 0.3022,
      "step": 69425
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.8650072216925864,
      "learning_rate": 1.2811787501166296e-08,
      "loss": 0.3128,
      "step": 69430
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.051197051307589,
      "learning_rate": 1.27238622487319e-08,
      "loss": 0.3228,
      "step": 69435
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.9562608580082426,
      "learning_rate": 1.2636239360159785e-08,
      "loss": 0.2825,
      "step": 69440
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.860547114328133,
      "learning_rate": 1.254891884076126e-08,
      "loss": 0.3024,
      "step": 69445
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.011063822615833,
      "learning_rate": 1.2461900695830976e-08,
      "loss": 0.3207,
      "step": 69450
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.8483114025959955,
      "learning_rate": 1.2375184930644157e-08,
      "loss": 0.2928,
      "step": 69455
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.81927638127644,
      "learning_rate": 1.2288771550458267e-08,
      "loss": 0.3094,
      "step": 69460
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.824218051767776,
      "learning_rate": 1.2202660560512447e-08,
      "loss": 0.2972,
      "step": 69465
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.5645699710634555,
      "learning_rate": 1.2116851966027521e-08,
      "loss": 0.2938,
      "step": 69470
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.828378480856656,
      "learning_rate": 1.2031345772205439e-08,
      "loss": 0.3138,
      "step": 69475
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.119706230422643,
      "learning_rate": 1.1946141984230386e-08,
      "loss": 0.3114,
      "step": 69480
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.9064513654158826,
      "learning_rate": 1.1861240607268231e-08,
      "loss": 0.315,
      "step": 69485
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.614202044562058,
      "learning_rate": 1.1776641646465413e-08,
      "loss": 0.3033,
      "step": 69490
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.6173604711194085,
      "learning_rate": 1.1692345106951719e-08,
      "loss": 0.3226,
      "step": 69495
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.123306587793768,
      "learning_rate": 1.1608350993837503e-08,
      "loss": 0.3178,
      "step": 69500
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.877722454433807,
      "learning_rate": 1.1524659312215358e-08,
      "loss": 0.2859,
      "step": 69505
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.543136355383514,
      "learning_rate": 1.1441270067159005e-08,
      "loss": 0.3104,
      "step": 69510
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.7687363127494504,
      "learning_rate": 1.1358183263724398e-08,
      "loss": 0.2909,
      "step": 69515
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.7110283912520132,
      "learning_rate": 1.127539890694862e-08,
      "loss": 0.2988,
      "step": 69520
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.9257603308165203,
      "learning_rate": 1.1192917001849878e-08,
      "loss": 0.306,
      "step": 69525
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.313500218721749,
      "learning_rate": 1.1110737553430284e-08,
      "loss": 0.3075,
      "step": 69530
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.613242898220054,
      "learning_rate": 1.1028860566671407e-08,
      "loss": 0.3083,
      "step": 69535
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.395869119550324,
      "learning_rate": 1.0947286046537053e-08,
      "loss": 0.279,
      "step": 69540
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.6036246881065566,
      "learning_rate": 1.0866013997973268e-08,
      "loss": 0.2897,
      "step": 69545
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.157828308995961,
      "learning_rate": 1.0785044425906666e-08,
      "loss": 0.3215,
      "step": 69550
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.8275269253015995,
      "learning_rate": 1.0704377335246651e-08,
      "loss": 0.3141,
      "step": 69555
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.8753952234786415,
      "learning_rate": 1.062401273088376e-08,
      "loss": 0.2764,
      "step": 69560
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.574976864770501,
      "learning_rate": 1.0543950617690757e-08,
      "loss": 0.3163,
      "step": 69565
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.927866906070172,
      "learning_rate": 1.0464191000520429e-08,
      "loss": 0.2931,
      "step": 69570
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.902797342203316,
      "learning_rate": 1.0384733884209464e-08,
      "loss": 0.2911,
      "step": 69575
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.6219406944057884,
      "learning_rate": 1.0305579273574562e-08,
      "loss": 0.2834,
      "step": 69580
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.056491882381841,
      "learning_rate": 1.0226727173415218e-08,
      "loss": 0.3005,
      "step": 69585
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.818981345711652,
      "learning_rate": 1.0148177588510944e-08,
      "loss": 0.2994,
      "step": 69590
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.5204218446326014,
      "learning_rate": 1.0069930523624594e-08,
      "loss": 0.3004,
      "step": 69595
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.667151481668382,
      "learning_rate": 9.991985983500152e-09,
      "loss": 0.3105,
      "step": 69600
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.9494083497805264,
      "learning_rate": 9.914343972863284e-09,
      "loss": 0.3089,
      "step": 69605
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.251667582732663,
      "learning_rate": 9.837004496420777e-09,
      "loss": 0.2791,
      "step": 69610
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.8720616213283,
      "learning_rate": 9.75996755886166e-09,
      "loss": 0.3342,
      "step": 69615
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.019648431242003,
      "learning_rate": 9.683233164856642e-09,
      "loss": 0.3115,
      "step": 69620
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.218351982800034,
      "learning_rate": 9.606801319057557e-09,
      "loss": 0.2932,
      "step": 69625
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7110717126337334,
      "learning_rate": 9.530672026098476e-09,
      "loss": 0.3147,
      "step": 69630
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.232999458684634,
      "learning_rate": 9.454845290594594e-09,
      "loss": 0.3157,
      "step": 69635
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.984150720128455,
      "learning_rate": 9.379321117143903e-09,
      "loss": 0.3105,
      "step": 69640
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7238860231144164,
      "learning_rate": 9.304099510324404e-09,
      "loss": 0.307,
      "step": 69645
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.515108910699393,
      "learning_rate": 9.229180474696342e-09,
      "loss": 0.3245,
      "step": 69650
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.006913544022434,
      "learning_rate": 9.154564014802748e-09,
      "loss": 0.3072,
      "step": 69655
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.9624418311113576,
      "learning_rate": 9.080250135166668e-09,
      "loss": 0.3102,
      "step": 69660
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7953230271194762,
      "learning_rate": 9.006238840293946e-09,
      "loss": 0.3116,
      "step": 69665
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.294578477132442,
      "learning_rate": 8.932530134671547e-09,
      "loss": 0.3172,
      "step": 69670
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.5860679718543684,
      "learning_rate": 8.859124022767562e-09,
      "loss": 0.3087,
      "step": 69675
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.417211765666118,
      "learning_rate": 8.786020509033989e-09,
      "loss": 0.2921,
      "step": 69680
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.833154357142647,
      "learning_rate": 8.713219597901169e-09,
      "loss": 0.339,
      "step": 69685
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7856616273749824,
      "learning_rate": 8.640721293783349e-09,
      "loss": 0.2923,
      "step": 69690
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.2772075547614135,
      "learning_rate": 8.568525601076461e-09,
      "loss": 0.2996,
      "step": 69695
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.8329761801228943,
      "learning_rate": 8.496632524157555e-09,
      "loss": 0.3057,
      "step": 69700
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.444841685895909,
      "learning_rate": 8.42504206738426e-09,
      "loss": 0.3006,
      "step": 69705
    },
    {
      "epoch": 2.95,
      "grad_norm": 6.888675108817265,
      "learning_rate": 8.353754235097544e-09,
      "loss": 0.3017,
      "step": 69710
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.4768943416676015,
      "learning_rate": 8.282769031619508e-09,
      "loss": 0.2969,
      "step": 69715
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7400112767247617,
      "learning_rate": 8.212086461253377e-09,
      "loss": 0.2887,
      "step": 69720
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.8697640362474366,
      "learning_rate": 8.141706528285165e-09,
      "loss": 0.2567,
      "step": 69725
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.8104736168045092,
      "learning_rate": 8.071629236980905e-09,
      "loss": 0.3104,
      "step": 69730
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.050060493585424,
      "learning_rate": 8.001854591589419e-09,
      "loss": 0.2906,
      "step": 69735
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.4199351512564786,
      "learning_rate": 7.932382596341215e-09,
      "loss": 0.3144,
      "step": 69740
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.792928152768424,
      "learning_rate": 7.863213255448477e-09,
      "loss": 0.2983,
      "step": 69745
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.479626224597136,
      "learning_rate": 7.794346573103407e-09,
      "loss": 0.2756,
      "step": 69750
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.490161382552908,
      "learning_rate": 7.725782553482664e-09,
      "loss": 0.3428,
      "step": 69755
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.6957980013622054,
      "learning_rate": 7.657521200742368e-09,
      "loss": 0.3251,
      "step": 69760
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.010704039418363,
      "learning_rate": 7.589562519021432e-09,
      "loss": 0.3172,
      "step": 69765
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7259156914324674,
      "learning_rate": 7.521906512439337e-09,
      "loss": 0.2986,
      "step": 69770
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.5170654959355,
      "learning_rate": 7.454553185098357e-09,
      "loss": 0.2944,
      "step": 69775
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.023642729993144,
      "learning_rate": 7.387502541081892e-09,
      "loss": 0.3031,
      "step": 69780
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.279968846870562,
      "learning_rate": 7.320754584454471e-09,
      "loss": 0.3121,
      "step": 69785
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7660069010111576,
      "learning_rate": 7.254309319263409e-09,
      "loss": 0.3027,
      "step": 69790
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.006814122837297,
      "learning_rate": 7.188166749537151e-09,
      "loss": 0.3002,
      "step": 69795
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.100640144034374,
      "learning_rate": 7.122326879285823e-09,
      "loss": 0.3046,
      "step": 69800
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.9970810059400868,
      "learning_rate": 7.0567897125006776e-09,
      "loss": 0.3087,
      "step": 69805
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.6903928157348114,
      "learning_rate": 6.991555253154647e-09,
      "loss": 0.2911,
      "step": 69810
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.8697044706871493,
      "learning_rate": 6.926623505204011e-09,
      "loss": 0.3021,
      "step": 69815
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.7077036560136936,
      "learning_rate": 6.861994472583955e-09,
      "loss": 0.2905,
      "step": 69820
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.427155678616963,
      "learning_rate": 6.797668159213566e-09,
      "loss": 0.3276,
      "step": 69825
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.656688913084691,
      "learning_rate": 6.733644568992504e-09,
      "loss": 0.286,
      "step": 69830
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.810381098672481,
      "learning_rate": 6.669923705802661e-09,
      "loss": 0.3355,
      "step": 69835
    },
    {
      "epoch": 2.96,
      "grad_norm": 5.73562289250491,
      "learning_rate": 6.606505573507061e-09,
      "loss": 0.3352,
      "step": 69840
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.162131414677388,
      "learning_rate": 6.543390175950403e-09,
      "loss": 0.2928,
      "step": 69845
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.180381988029838,
      "learning_rate": 6.480577516959075e-09,
      "loss": 0.3311,
      "step": 69850
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.390160790673757,
      "learning_rate": 6.4180676003416934e-09,
      "loss": 0.3147,
      "step": 69855
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.94940857299616,
      "learning_rate": 6.355860429888005e-09,
      "loss": 0.2972,
      "step": 69860
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.119850657798943,
      "learning_rate": 6.293956009368884e-09,
      "loss": 0.324,
      "step": 69865
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.404943789149675,
      "learning_rate": 6.232354342537994e-09,
      "loss": 0.2992,
      "step": 69870
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.9180760997053916,
      "learning_rate": 6.171055433129569e-09,
      "loss": 0.2868,
      "step": 69875
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.049237073605573,
      "learning_rate": 6.1100592848600815e-09,
      "loss": 0.306,
      "step": 69880
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.5391050094942385,
      "learning_rate": 6.049365901428239e-09,
      "loss": 0.3345,
      "step": 69885
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.9878637425883547,
      "learning_rate": 5.988975286512766e-09,
      "loss": 0.2884,
      "step": 69890
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.940702798714988,
      "learning_rate": 5.9288874437757324e-09,
      "loss": 0.2929,
      "step": 69895
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.21448847756659,
      "learning_rate": 5.869102376859781e-09,
      "loss": 0.2813,
      "step": 69900
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.7555692821779125,
      "learning_rate": 5.809620089388679e-09,
      "loss": 0.3041,
      "step": 69905
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.7219221498414776,
      "learning_rate": 5.750440584970096e-09,
      "loss": 0.2812,
      "step": 69910
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.650666089125777,
      "learning_rate": 5.691563867191163e-09,
      "loss": 0.3111,
      "step": 69915
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.874403218996133,
      "learning_rate": 5.6329899396218024e-09,
      "loss": 0.3038,
      "step": 69920
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.121525418993446,
      "learning_rate": 5.574718805812507e-09,
      "loss": 0.3037,
      "step": 69925
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.7098169244371006,
      "learning_rate": 5.516750469296561e-09,
      "loss": 0.2875,
      "step": 69930
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.29946152871183,
      "learning_rate": 5.459084933587822e-09,
      "loss": 0.3102,
      "step": 69935
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.307436220250806,
      "learning_rate": 5.4017222021834906e-09,
      "loss": 0.3029,
      "step": 69940
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.8625532152485755,
      "learning_rate": 5.344662278560231e-09,
      "loss": 0.3005,
      "step": 69945
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.93674031209828,
      "learning_rate": 5.287905166178053e-09,
      "loss": 0.3102,
      "step": 69950
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.608890140294212,
      "learning_rate": 5.231450868478094e-09,
      "loss": 0.2946,
      "step": 69955
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.9659831383934794,
      "learning_rate": 5.17529938888206e-09,
      "loss": 0.3024,
      "step": 69960
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.769673161955406,
      "learning_rate": 5.119450730795561e-09,
      "loss": 0.3019,
      "step": 69965
    },
    {
      "epoch": 2.96,
      "grad_norm": 5.009075825119275,
      "learning_rate": 5.063904897603666e-09,
      "loss": 0.3052,
      "step": 69970
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.7758522555049785,
      "learning_rate": 5.008661892673683e-09,
      "loss": 0.2945,
      "step": 69975
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.8244012773921563,
      "learning_rate": 4.9537217193557085e-09,
      "loss": 0.2857,
      "step": 69980
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.232909999497438,
      "learning_rate": 4.899084380980412e-09,
      "loss": 0.3078,
      "step": 69985
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.318898272923827,
      "learning_rate": 4.844749880859589e-09,
      "loss": 0.3069,
      "step": 69990
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.8983974547088533,
      "learning_rate": 4.790718222287827e-09,
      "loss": 0.2935,
      "step": 69995
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.8579075172114634,
      "learning_rate": 4.736989408541392e-09,
      "loss": 0.2882,
      "step": 70000
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.177278943671146,
      "learning_rate": 4.683563442877126e-09,
      "loss": 0.3075,
      "step": 70005
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.345862178897866,
      "learning_rate": 4.630440328534103e-09,
      "loss": 0.3034,
      "step": 70010
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.839523640507528,
      "learning_rate": 4.577620068733079e-09,
      "loss": 0.2986,
      "step": 70015
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.087982010295637,
      "learning_rate": 4.525102666676495e-09,
      "loss": 0.3092,
      "step": 70020
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.8927770936361377,
      "learning_rate": 4.472888125548469e-09,
      "loss": 0.2869,
      "step": 70025
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.9048928132621006,
      "learning_rate": 4.4209764485142474e-09,
      "loss": 0.2857,
      "step": 70030
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.818364739173185,
      "learning_rate": 4.369367638721311e-09,
      "loss": 0.3068,
      "step": 70035
    },
    {
      "epoch": 2.96,
      "grad_norm": 3.7925570106816706,
      "learning_rate": 4.3180616992988255e-09,
      "loss": 0.2796,
      "step": 70040
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.31552668830435,
      "learning_rate": 4.267058633356525e-09,
      "loss": 0.2887,
      "step": 70045
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.524636109612765,
      "learning_rate": 4.216358443986934e-09,
      "loss": 0.3232,
      "step": 70050
    },
    {
      "epoch": 2.96,
      "grad_norm": 4.214091939910933,
      "learning_rate": 4.165961134264263e-09,
      "loss": 0.3072,
      "step": 70055
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.81128564753025,
      "learning_rate": 4.115866707243843e-09,
      "loss": 0.2762,
      "step": 70060
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.8822172031713955,
      "learning_rate": 4.0660751659621355e-09,
      "loss": 0.3037,
      "step": 70065
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.107891610460311,
      "learning_rate": 4.016586513438392e-09,
      "loss": 0.3002,
      "step": 70070
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.6077395429264008,
      "learning_rate": 3.96740075267299e-09,
      "loss": 0.3093,
      "step": 70075
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.8219143428417977,
      "learning_rate": 3.918517886647988e-09,
      "loss": 0.3128,
      "step": 70080
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.14871013347077,
      "learning_rate": 3.869937918326572e-09,
      "loss": 0.3155,
      "step": 70085
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.473790800532902,
      "learning_rate": 3.821660850654163e-09,
      "loss": 0.3217,
      "step": 70090
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.1128048361330425,
      "learning_rate": 3.773686686557865e-09,
      "loss": 0.3005,
      "step": 70095
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.867898629450061,
      "learning_rate": 3.726015428945906e-09,
      "loss": 0.2873,
      "step": 70100
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.6937437124934593,
      "learning_rate": 3.6786470807087525e-09,
      "loss": 0.2965,
      "step": 70105
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.174590403505056,
      "learning_rate": 3.6315816447179965e-09,
      "loss": 0.3127,
      "step": 70110
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.8031941633379276,
      "learning_rate": 3.5848191238274654e-09,
      "loss": 0.2737,
      "step": 70115
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.8379302537153204,
      "learning_rate": 3.5383595208715594e-09,
      "loss": 0.2918,
      "step": 70120
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.29588383567088,
      "learning_rate": 3.492202838667469e-09,
      "loss": 0.3313,
      "step": 70125
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.041925819341384,
      "learning_rate": 3.4463490800135114e-09,
      "loss": 0.3034,
      "step": 70130
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.560977162463394,
      "learning_rate": 3.400798247689685e-09,
      "loss": 0.3006,
      "step": 70135
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.475642592477241,
      "learning_rate": 3.3555503444571146e-09,
      "loss": 0.3306,
      "step": 70140
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.634522028690794,
      "learning_rate": 3.3106053730597165e-09,
      "loss": 0.3116,
      "step": 70145
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.597345494503631,
      "learning_rate": 3.2659633362219776e-09,
      "loss": 0.3139,
      "step": 70150
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.777776404919409,
      "learning_rate": 3.221624236650622e-09,
      "loss": 0.3198,
      "step": 70155
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.8467046863581977,
      "learning_rate": 3.1775880770335e-09,
      "loss": 0.2871,
      "step": 70160
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.566164710985339,
      "learning_rate": 3.133854860040697e-09,
      "loss": 0.2811,
      "step": 70165
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.92608062355234,
      "learning_rate": 3.0904245883234262e-09,
      "loss": 0.3403,
      "step": 70170
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.446020332177124,
      "learning_rate": 3.0472972645151366e-09,
      "loss": 0.3349,
      "step": 70175
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.291256179270392,
      "learning_rate": 3.0044728912298483e-09,
      "loss": 0.2996,
      "step": 70180
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.9132330612977824,
      "learning_rate": 2.961951471064373e-09,
      "loss": 0.3061,
      "step": 70185
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.553729955094563,
      "learning_rate": 2.919733006596648e-09,
      "loss": 0.2759,
      "step": 70190
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.4775475580804556,
      "learning_rate": 2.877817500385738e-09,
      "loss": 0.294,
      "step": 70195
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.9211644211674526,
      "learning_rate": 2.8362049549734984e-09,
      "loss": 0.3105,
      "step": 70200
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.7123482755314425,
      "learning_rate": 2.7948953728829107e-09,
      "loss": 0.2922,
      "step": 70205
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.5129504433481875,
      "learning_rate": 2.7538887566175288e-09,
      "loss": 0.3139,
      "step": 70210
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.472807759825269,
      "learning_rate": 2.7131851086642515e-09,
      "loss": 0.3273,
      "step": 70215
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.115108764837657,
      "learning_rate": 2.6727844314905493e-09,
      "loss": 0.3171,
      "step": 70220
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.9345346837315036,
      "learning_rate": 2.632686727545575e-09,
      "loss": 0.2842,
      "step": 70225
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.4657165417155484,
      "learning_rate": 2.592891999260716e-09,
      "loss": 0.304,
      "step": 70230
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.6797386700992107,
      "learning_rate": 2.553400249048488e-09,
      "loss": 0.2895,
      "step": 70235
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.5680159930287014,
      "learning_rate": 2.514211479303086e-09,
      "loss": 0.2937,
      "step": 70240
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.2110032538823585,
      "learning_rate": 2.4753256924003876e-09,
      "loss": 0.3058,
      "step": 70245
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.7420552814800745,
      "learning_rate": 2.4367428906985067e-09,
      "loss": 0.3185,
      "step": 70250
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.475126468660208,
      "learning_rate": 2.3984630765355733e-09,
      "loss": 0.3011,
      "step": 70255
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.71884856987223,
      "learning_rate": 2.360486252233063e-09,
      "loss": 0.3268,
      "step": 70260
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.834630901361669,
      "learning_rate": 2.3228124200930237e-09,
      "loss": 0.3095,
      "step": 70265
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.308866744044423,
      "learning_rate": 2.285441582400294e-09,
      "loss": 0.3086,
      "step": 70270
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.508513445780656,
      "learning_rate": 2.2483737414197292e-09,
      "loss": 0.292,
      "step": 70275
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.197499560277858,
      "learning_rate": 2.211608899398421e-09,
      "loss": 0.3177,
      "step": 70280
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.033360743806799,
      "learning_rate": 2.1751470585662516e-09,
      "loss": 0.3177,
      "step": 70285
    },
    {
      "epoch": 2.97,
      "grad_norm": 4.046989229747052,
      "learning_rate": 2.1389882211336752e-09,
      "loss": 0.319,
      "step": 70290
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.8764286349236907,
      "learning_rate": 2.103132389292273e-09,
      "loss": 0.293,
      "step": 70295
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.5827714786633127,
      "learning_rate": 2.0675795652158603e-09,
      "loss": 0.322,
      "step": 70300
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.893423812906061,
      "learning_rate": 2.032329751060491e-09,
      "loss": 0.3038,
      "step": 70305
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.8297518258550536,
      "learning_rate": 1.9973829489627892e-09,
      "loss": 0.3056,
      "step": 70310
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.5853639957448995,
      "learning_rate": 1.9627391610416158e-09,
      "loss": 0.2799,
      "step": 70315
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.907066856644201,
      "learning_rate": 1.9283983893975124e-09,
      "loss": 0.3252,
      "step": 70320
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.801238706118709,
      "learning_rate": 1.894360636112702e-09,
      "loss": 0.279,
      "step": 70325
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.9912177166410925,
      "learning_rate": 1.8606259032499795e-09,
      "loss": 0.2974,
      "step": 70330
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.770227814867083,
      "learning_rate": 1.8271941928549308e-09,
      "loss": 0.2839,
      "step": 70335
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.6590503174626754,
      "learning_rate": 1.794065506954823e-09,
      "loss": 0.3117,
      "step": 70340
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.7252534169698523,
      "learning_rate": 1.7612398475574944e-09,
      "loss": 0.3315,
      "step": 70345
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.209130209125072,
      "learning_rate": 1.7287172166535749e-09,
      "loss": 0.3135,
      "step": 70350
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.976178900451546,
      "learning_rate": 1.6964976162142655e-09,
      "loss": 0.3134,
      "step": 70355
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.9863273489235915,
      "learning_rate": 1.6645810481935587e-09,
      "loss": 0.2841,
      "step": 70360
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.869330246996778,
      "learning_rate": 1.6329675145265733e-09,
      "loss": 0.3158,
      "step": 70365
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.35833733446752,
      "learning_rate": 1.6016570171289992e-09,
      "loss": 0.3101,
      "step": 70370
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.073039873960955,
      "learning_rate": 1.5706495579004277e-09,
      "loss": 0.3016,
      "step": 70375
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.881411117558849,
      "learning_rate": 1.5399451387193564e-09,
      "loss": 0.3121,
      "step": 70380
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.572663016631878,
      "learning_rate": 1.5095437614481845e-09,
      "loss": 0.3114,
      "step": 70385
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.8428242356451223,
      "learning_rate": 1.479445427929882e-09,
      "loss": 0.2839,
      "step": 70390
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.7496785901952943,
      "learning_rate": 1.4496501399891005e-09,
      "loss": 0.3187,
      "step": 70395
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.228475694261304,
      "learning_rate": 1.4201578994327282e-09,
      "loss": 0.3016,
      "step": 70400
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.8454342868258684,
      "learning_rate": 1.3909687080476685e-09,
      "loss": 0.3038,
      "step": 70405
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.019206989000221,
      "learning_rate": 1.3620825676047277e-09,
      "loss": 0.3154,
      "step": 70410
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.6259399780735695,
      "learning_rate": 1.3334994798547274e-09,
      "loss": 0.31,
      "step": 70415
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.9160310875698108,
      "learning_rate": 1.3052194465307255e-09,
      "loss": 0.3067,
      "step": 70420
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.5944950592481257,
      "learning_rate": 1.2772424693469064e-09,
      "loss": 0.3032,
      "step": 70425
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.061033428497966,
      "learning_rate": 1.2495685499996913e-09,
      "loss": 0.2841,
      "step": 70430
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.416399802931741,
      "learning_rate": 1.2221976901666267e-09,
      "loss": 0.3004,
      "step": 70435
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.348335494132366,
      "learning_rate": 1.195129891508051e-09,
      "loss": 0.3018,
      "step": 70440
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.033180939001871,
      "learning_rate": 1.1683651556637643e-09,
      "loss": 0.2999,
      "step": 70445
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.6444732632307626,
      "learning_rate": 1.141903484256912e-09,
      "loss": 0.2747,
      "step": 70450
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.6395846932311007,
      "learning_rate": 1.115744878891767e-09,
      "loss": 0.2529,
      "step": 70455
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.072445173695659,
      "learning_rate": 1.0898893411542822e-09,
      "loss": 0.3001,
      "step": 70460
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.1117817542443404,
      "learning_rate": 1.0643368726126479e-09,
      "loss": 0.3257,
      "step": 70465
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.964724979286409,
      "learning_rate": 1.0390874748145153e-09,
      "loss": 0.3136,
      "step": 70470
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.242072867810214,
      "learning_rate": 1.0141411492919916e-09,
      "loss": 0.3266,
      "step": 70475
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.910976360279106,
      "learning_rate": 9.894978975572011e-10,
      "loss": 0.3071,
      "step": 70480
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.9953559913131897,
      "learning_rate": 9.651577211039486e-10,
      "loss": 0.3016,
      "step": 70485
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.8049983620172707,
      "learning_rate": 9.411206214082758e-10,
      "loss": 0.3099,
      "step": 70490
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.9207910434479607,
      "learning_rate": 9.173865999273501e-10,
      "loss": 0.2862,
      "step": 70495
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.29459758305129,
      "learning_rate": 8.939556580994657e-10,
      "loss": 0.3042,
      "step": 70500
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.752689334246048,
      "learning_rate": 8.708277973462631e-10,
      "loss": 0.2877,
      "step": 70505
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.4257914947500425,
      "learning_rate": 8.480030190688438e-10,
      "loss": 0.3088,
      "step": 70510
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.9799008207804905,
      "learning_rate": 8.254813246516558e-10,
      "loss": 0.293,
      "step": 70515
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.8469021372194026,
      "learning_rate": 8.032627154602735e-10,
      "loss": 0.3099,
      "step": 70520
    },
    {
      "epoch": 2.98,
      "grad_norm": 4.470557347177352,
      "learning_rate": 7.813471928413973e-10,
      "loss": 0.2897,
      "step": 70525
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.6365721921530403,
      "learning_rate": 7.597347581234094e-10,
      "loss": 0.2916,
      "step": 70530
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.8579595490789367,
      "learning_rate": 7.384254126169277e-10,
      "loss": 0.3067,
      "step": 70535
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.04075518314832,
      "learning_rate": 7.17419157614252e-10,
      "loss": 0.2903,
      "step": 70540
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.753720291366955,
      "learning_rate": 6.96715994388808e-10,
      "loss": 0.2924,
      "step": 70545
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.006901122571129,
      "learning_rate": 6.763159241951478e-10,
      "loss": 0.3253,
      "step": 70550
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.757779331675829,
      "learning_rate": 6.56218948270615e-10,
      "loss": 0.2922,
      "step": 70555
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.732034919181486,
      "learning_rate": 6.364250678331241e-10,
      "loss": 0.298,
      "step": 70560
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.323401385440728,
      "learning_rate": 6.169342840833814e-10,
      "loss": 0.2831,
      "step": 70565
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.7095818812359633,
      "learning_rate": 5.977465982026642e-10,
      "loss": 0.3079,
      "step": 70570
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.4984531606362586,
      "learning_rate": 5.788620113544863e-10,
      "loss": 0.3125,
      "step": 70575
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.9580511462286676,
      "learning_rate": 5.602805246834875e-10,
      "loss": 0.3245,
      "step": 70580
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.59066109035715,
      "learning_rate": 5.420021393165442e-10,
      "loss": 0.2978,
      "step": 70585
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.665580764167094,
      "learning_rate": 5.240268563611039e-10,
      "loss": 0.2773,
      "step": 70590
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.7402880381343766,
      "learning_rate": 5.063546769079608e-10,
      "loss": 0.3026,
      "step": 70595
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.038462855292636,
      "learning_rate": 4.889856020279249e-10,
      "loss": 0.299,
      "step": 70600
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.8905543612946873,
      "learning_rate": 4.719196327740427e-10,
      "loss": 0.2878,
      "step": 70605
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.078318973530521,
      "learning_rate": 4.551567701810422e-10,
      "loss": 0.3115,
      "step": 70610
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.047447722816276,
      "learning_rate": 4.386970152653325e-10,
      "loss": 0.3015,
      "step": 70615
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.6460754802776254,
      "learning_rate": 4.2254036902500404e-10,
      "loss": 0.3162,
      "step": 70620
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.8134927567707453,
      "learning_rate": 4.0668683243927364e-10,
      "loss": 0.3091,
      "step": 70625
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.9807379984830185,
      "learning_rate": 3.911364064690393e-10,
      "loss": 0.3061,
      "step": 70630
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.252773462840397,
      "learning_rate": 3.758890920574354e-10,
      "loss": 0.2951,
      "step": 70635
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.5370726243380965,
      "learning_rate": 3.609448901287227e-10,
      "loss": 0.2892,
      "step": 70640
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.080216421472142,
      "learning_rate": 3.4630380158939824e-10,
      "loss": 0.3099,
      "step": 70645
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.9873977651378483,
      "learning_rate": 3.3196582732653026e-10,
      "loss": 0.2763,
      "step": 70650
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.05471264951897,
      "learning_rate": 3.1793096820942336e-10,
      "loss": 0.3037,
      "step": 70655
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.888577614060574,
      "learning_rate": 3.041992250890635e-10,
      "loss": 0.2928,
      "step": 70660
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.247596436763597,
      "learning_rate": 2.90770598798118e-10,
      "loss": 0.3139,
      "step": 70665
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.316810081793586,
      "learning_rate": 2.776450901503802e-10,
      "loss": 0.3065,
      "step": 70670
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.943195477001331,
      "learning_rate": 2.6482269994243526e-10,
      "loss": 0.2867,
      "step": 70675
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.7491797138481315,
      "learning_rate": 2.52303428950329e-10,
      "loss": 0.2909,
      "step": 70680
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.1464391544718096,
      "learning_rate": 2.400872779345642e-10,
      "loss": 0.305,
      "step": 70685
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.851233546529492,
      "learning_rate": 2.2817424763454943e-10,
      "loss": 0.2901,
      "step": 70690
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.076400367655774,
      "learning_rate": 2.1656433877303984e-10,
      "loss": 0.3076,
      "step": 70695
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.7909089798085005,
      "learning_rate": 2.0525755205447194e-10,
      "loss": 0.3248,
      "step": 70700
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.037527495697327,
      "learning_rate": 1.9425388816329827e-10,
      "loss": 0.3162,
      "step": 70705
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.7089612532748295,
      "learning_rate": 1.8355334776676283e-10,
      "loss": 0.2941,
      "step": 70710
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.100527845485909,
      "learning_rate": 1.7315593151434606e-10,
      "loss": 0.2922,
      "step": 70715
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.605224103909242,
      "learning_rate": 1.6306164003609958e-10,
      "loss": 0.2957,
      "step": 70720
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.453543295072825,
      "learning_rate": 1.5327047394375627e-10,
      "loss": 0.2985,
      "step": 70725
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.8118576540502134,
      "learning_rate": 1.4378243383128544e-10,
      "loss": 0.3098,
      "step": 70730
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.147331427768618,
      "learning_rate": 1.345975202737826e-10,
      "loss": 0.3117,
      "step": 70735
    },
    {
      "epoch": 2.99,
      "grad_norm": 4.602903411192248,
      "learning_rate": 1.257157338274695e-10,
      "loss": 0.3134,
      "step": 70740
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.6822477208693494,
      "learning_rate": 1.171370750319145e-10,
      "loss": 0.2922,
      "step": 70745
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.380180310217267,
      "learning_rate": 1.0886154440670205e-10,
      "loss": 0.2834,
      "step": 70750
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.403085142580083,
      "learning_rate": 1.0088914245309778e-10,
      "loss": 0.3142,
      "step": 70755
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.713778823815118,
      "learning_rate": 9.321986965515894e-11,
      "loss": 0.2854,
      "step": 70760
    },
    {
      "epoch": 2.99,
      "grad_norm": 3.9908660392139175,
      "learning_rate": 8.5853726478069e-11,
      "loss": 0.2888,
      "step": 70765
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.7075799983838698,
      "learning_rate": 7.879071336702738e-11,
      "loss": 0.2976,
      "step": 70770
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.982746880921896,
      "learning_rate": 7.203083075169037e-11,
      "loss": 0.2981,
      "step": 70775
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.82154141933062,
      "learning_rate": 6.55740790417303e-11,
      "loss": 0.3133,
      "step": 70780
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.869583535042438,
      "learning_rate": 5.942045862739055e-11,
      "loss": 0.2852,
      "step": 70785
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.918289435262239,
      "learning_rate": 5.3569969883371376e-11,
      "loss": 0.287,
      "step": 70790
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.846698890372857,
      "learning_rate": 4.802261316272372e-11,
      "loss": 0.3282,
      "step": 70795
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.969683298009337,
      "learning_rate": 4.2778388803510484e-11,
      "loss": 0.2916,
      "step": 70800
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.767344262973072,
      "learning_rate": 3.7837297122145236e-11,
      "loss": 0.3059,
      "step": 70805
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.612186813738805,
      "learning_rate": 3.3199338418943294e-11,
      "loss": 0.2826,
      "step": 70810
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.513144080364581,
      "learning_rate": 2.8864512975346204e-11,
      "loss": 0.3038,
      "step": 70815
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.4086362770214045,
      "learning_rate": 2.4832821053366597e-11,
      "loss": 0.3084,
      "step": 70820
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.614494861743323,
      "learning_rate": 2.1104262898363758e-11,
      "loss": 0.3175,
      "step": 70825
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.819530642605835,
      "learning_rate": 1.7678838735157854e-11,
      "loss": 0.2838,
      "step": 70830
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.7057447575294216,
      "learning_rate": 1.4556548772470813e-11,
      "loss": 0.3005,
      "step": 70835
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.777074680632009,
      "learning_rate": 1.173739319959566e-11,
      "loss": 0.2786,
      "step": 70840
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.9879301558525824,
      "learning_rate": 9.221372186951627e-12,
      "loss": 0.2998,
      "step": 70845
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.61498691886502,
      "learning_rate": 7.008485887194383e-12,
      "loss": 0.3116,
      "step": 70850
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.217555785581664,
      "learning_rate": 5.098734434660912e-12,
      "loss": 0.3335,
      "step": 70855
    },
    {
      "epoch": 3.0,
      "grad_norm": 4.419913126093161,
      "learning_rate": 3.4921179448144103e-12,
      "loss": 0.3083,
      "step": 70860
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.7644958113238682,
      "learning_rate": 2.1886365153545032e-12,
      "loss": 0.3034,
      "step": 70865
    },
    {
      "epoch": 3.0,
      "grad_norm": 5.072162291473488,
      "learning_rate": 1.1882902251070249e-12,
      "loss": 0.278,
      "step": 70870
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.9881709860328476,
      "learning_rate": 4.910791351342425e-13,
      "loss": 0.2982,
      "step": 70875
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.951610246893262,
      "learning_rate": 9.700328706951923e-14,
      "loss": 0.2976,
      "step": 70880
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.8875364065170288,
      "eval_runtime": 37.0526,
      "eval_samples_per_second": 31.55,
      "eval_steps_per_second": 3.967,
      "step": 70884
    },
    {
      "epoch": 3.0,
      "step": 70884,
      "total_flos": 2440295425769472.0,
      "train_loss": 0.81574390725434,
      "train_runtime": 170799.3559,
      "train_samples_per_second": 6.64,
      "train_steps_per_second": 0.415
    }
  ],
  "logging_steps": 5,
  "max_steps": 70884,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2440295425769472.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}