{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 8734,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0002862837859599274,
      "grad_norm": 19.151697158813477,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 516.7791,
      "step": 5
    },
    {
      "epoch": 0.0005725675719198549,
      "grad_norm": 16.986482620239258,
      "learning_rate": 9e-07,
      "loss": 116.6236,
      "step": 10
    },
    {
      "epoch": 0.0008588513578797823,
      "grad_norm": 11.296030044555664,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 176.0494,
      "step": 15
    },
    {
      "epoch": 0.0011451351438397097,
      "grad_norm": 6.591270923614502,
      "learning_rate": 1.9e-06,
      "loss": 34.2581,
      "step": 20
    },
    {
      "epoch": 0.001431418929799637,
      "grad_norm": 3.7718400955200195,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 97.8785,
      "step": 25
    },
    {
      "epoch": 0.0017177027157595647,
      "grad_norm": 2.6732568740844727,
      "learning_rate": 2.9e-06,
      "loss": 56.9053,
      "step": 30
    },
    {
      "epoch": 0.002003986501719492,
      "grad_norm": 1.7744698524475098,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 112.3654,
      "step": 35
    },
    {
      "epoch": 0.0022902702876794194,
      "grad_norm": 1.9627138376235962,
      "learning_rate": 3.9e-06,
      "loss": 319.8193,
      "step": 40
    },
    {
      "epoch": 0.002576554073639347,
      "grad_norm": 3.067491054534912,
      "learning_rate": 4.4e-06,
      "loss": 271.8014,
      "step": 45
    },
    {
      "epoch": 0.002862837859599274,
      "grad_norm": 2.7929775714874268,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 51.7764,
      "step": 50
    },
    {
      "epoch": 0.0031491216455592015,
      "grad_norm": 7.4539971351623535,
      "learning_rate": 5.4e-06,
      "loss": 86.3225,
      "step": 55
    },
    {
      "epoch": 0.0034354054315191293,
      "grad_norm": 6.343064308166504,
      "learning_rate": 5.9e-06,
      "loss": 55.7213,
      "step": 60
    },
    {
      "epoch": 0.0037216892174790567,
      "grad_norm": 12.474454879760742,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 184.1954,
      "step": 65
    },
    {
      "epoch": 0.004007973003438984,
      "grad_norm": 12.984466552734375,
      "learning_rate": 6.900000000000001e-06,
      "loss": 169.4968,
      "step": 70
    },
    {
      "epoch": 0.0042942567893989115,
      "grad_norm": 7.857387542724609,
      "learning_rate": 7.4e-06,
      "loss": 261.5822,
      "step": 75
    },
    {
      "epoch": 0.004580540575358839,
      "grad_norm": 9.837823867797852,
      "learning_rate": 7.9e-06,
      "loss": 74.4514,
      "step": 80
    },
    {
      "epoch": 0.004866824361318766,
      "grad_norm": 8.058916091918945,
      "learning_rate": 8.400000000000001e-06,
      "loss": 124.711,
      "step": 85
    },
    {
      "epoch": 0.005153108147278694,
      "grad_norm": 9.69068431854248,
      "learning_rate": 8.9e-06,
      "loss": 119.5081,
      "step": 90
    },
    {
      "epoch": 0.005439391933238621,
      "grad_norm": 8.338385581970215,
      "learning_rate": 9.4e-06,
      "loss": 163.0948,
      "step": 95
    },
    {
      "epoch": 0.005725675719198548,
      "grad_norm": 12.555005073547363,
      "learning_rate": 9.900000000000002e-06,
      "loss": 223.8742,
      "step": 100
    },
    {
      "epoch": 0.006011959505158476,
      "grad_norm": 7.753326892852783,
      "learning_rate": 1.04e-05,
      "loss": 162.0142,
      "step": 105
    },
    {
      "epoch": 0.006298243291118403,
      "grad_norm": 7.663868427276611,
      "learning_rate": 1.09e-05,
      "loss": 44.9881,
      "step": 110
    },
    {
      "epoch": 0.0065845270770783304,
      "grad_norm": 7.713599681854248,
      "learning_rate": 1.1400000000000001e-05,
      "loss": 59.8376,
      "step": 115
    },
    {
      "epoch": 0.006870810863038259,
      "grad_norm": 6.707058906555176,
      "learning_rate": 1.19e-05,
      "loss": 147.0191,
      "step": 120
    },
    {
      "epoch": 0.007157094648998186,
      "grad_norm": 3.721087694168091,
      "learning_rate": 1.24e-05,
      "loss": 54.5157,
      "step": 125
    },
    {
      "epoch": 0.007443378434958113,
      "grad_norm": 5.96558952331543,
      "learning_rate": 1.29e-05,
      "loss": 114.7495,
      "step": 130
    },
    {
      "epoch": 0.007729662220918041,
      "grad_norm": 3.8625457286834717,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 90.1746,
      "step": 135
    },
    {
      "epoch": 0.008015946006877968,
      "grad_norm": 6.123764991760254,
      "learning_rate": 1.3900000000000002e-05,
      "loss": 41.6194,
      "step": 140
    },
    {
      "epoch": 0.008302229792837896,
      "grad_norm": 5.954704284667969,
      "learning_rate": 1.44e-05,
      "loss": 151.4757,
      "step": 145
    },
    {
      "epoch": 0.008588513578797823,
      "grad_norm": 3.687113046646118,
      "learning_rate": 1.49e-05,
      "loss": 68.2852,
      "step": 150
    },
    {
      "epoch": 0.00887479736475775,
      "grad_norm": 3.733111619949341,
      "learning_rate": 1.54e-05,
      "loss": 44.1476,
      "step": 155
    },
    {
      "epoch": 0.03664261994732623,
      "grad_norm": 4.615285873413086,
      "learning_rate": 1.59e-05,
      "loss": 0.9683,
      "step": 160
    },
    {
      "epoch": 0.037787701820680177,
      "grad_norm": 7.621520519256592,
      "learning_rate": 1.6400000000000002e-05,
      "loss": 0.9631,
      "step": 165
    },
    {
      "epoch": 0.03893278369403412,
      "grad_norm": 12.825485229492188,
      "learning_rate": 1.69e-05,
      "loss": 0.9485,
      "step": 170
    },
    {
      "epoch": 0.04007786556738807,
      "grad_norm": 9.484743118286133,
      "learning_rate": 1.74e-05,
      "loss": 0.9404,
      "step": 175
    },
    {
      "epoch": 0.041222947440742015,
      "grad_norm": 5.518902778625488,
      "learning_rate": 1.79e-05,
      "loss": 0.9197,
      "step": 180
    },
    {
      "epoch": 0.04236802931409596,
      "grad_norm": 3.3146023750305176,
      "learning_rate": 1.84e-05,
      "loss": 0.9093,
      "step": 185
    },
    {
      "epoch": 0.0435131111874499,
      "grad_norm": 2.7806248664855957,
      "learning_rate": 1.8900000000000002e-05,
      "loss": 0.9032,
      "step": 190
    },
    {
      "epoch": 0.04465819306080385,
      "grad_norm": 1.950194239616394,
      "learning_rate": 1.94e-05,
      "loss": 0.8957,
      "step": 195
    },
    {
      "epoch": 0.04580327493415779,
      "grad_norm": 2.105093479156494,
      "learning_rate": 1.9900000000000003e-05,
      "loss": 0.8921,
      "step": 200
    },
    {
      "epoch": 0.046948356807511735,
      "grad_norm": 2.0000460147857666,
      "learning_rate": 2.04e-05,
      "loss": 0.8877,
      "step": 205
    },
    {
      "epoch": 0.04809343868086568,
      "grad_norm": 1.6276265382766724,
      "learning_rate": 2.09e-05,
      "loss": 0.8862,
      "step": 210
    },
    {
      "epoch": 0.04923852055421963,
      "grad_norm": 5.631216049194336,
      "learning_rate": 2.1400000000000002e-05,
      "loss": 0.8805,
      "step": 215
    },
    {
      "epoch": 0.05038360242757357,
      "grad_norm": 2.450659990310669,
      "learning_rate": 2.19e-05,
      "loss": 0.881,
      "step": 220
    },
    {
      "epoch": 0.05152868430092752,
      "grad_norm": 2.5550310611724854,
      "learning_rate": 2.2400000000000002e-05,
      "loss": 0.8737,
      "step": 225
    },
    {
      "epoch": 0.05267376617428146,
      "grad_norm": 2.168761730194092,
      "learning_rate": 2.29e-05,
      "loss": 0.8656,
      "step": 230
    },
    {
      "epoch": 0.053818848047635405,
      "grad_norm": 2.1524722576141357,
      "learning_rate": 2.3400000000000003e-05,
      "loss": 0.8583,
      "step": 235
    },
    {
      "epoch": 0.05496392992098935,
      "grad_norm": 2.131282091140747,
      "learning_rate": 2.39e-05,
      "loss": 0.8554,
      "step": 240
    },
    {
      "epoch": 0.05610901179434329,
      "grad_norm": 2.4136056900024414,
      "learning_rate": 2.44e-05,
      "loss": 0.8504,
      "step": 245
    },
    {
      "epoch": 0.057254093667697244,
      "grad_norm": 2.4667770862579346,
      "learning_rate": 2.4900000000000002e-05,
      "loss": 0.8459,
      "step": 250
    },
    {
      "epoch": 0.05839917554105119,
      "grad_norm": 1.7110178470611572,
      "learning_rate": 2.54e-05,
      "loss": 0.842,
      "step": 255
    },
    {
      "epoch": 0.05954425741440513,
      "grad_norm": 1.4909886121749878,
      "learning_rate": 2.5900000000000003e-05,
      "loss": 0.8408,
      "step": 260
    },
    {
      "epoch": 0.060689339287759075,
      "grad_norm": 1.4522689580917358,
      "learning_rate": 2.64e-05,
      "loss": 0.8371,
      "step": 265
    },
    {
      "epoch": 0.06183442116111302,
      "grad_norm": 1.8193100690841675,
      "learning_rate": 2.6900000000000003e-05,
      "loss": 0.831,
      "step": 270
    },
    {
      "epoch": 0.06297950303446696,
      "grad_norm": 1.8061556816101074,
      "learning_rate": 2.7400000000000002e-05,
      "loss": 0.8288,
      "step": 275
    },
    {
      "epoch": 0.06412458490782091,
      "grad_norm": 2.727900743484497,
      "learning_rate": 2.7900000000000004e-05,
      "loss": 0.829,
      "step": 280
    },
    {
      "epoch": 0.06526966678117485,
      "grad_norm": 2.1262223720550537,
      "learning_rate": 2.84e-05,
      "loss": 0.8282,
      "step": 285
    },
    {
      "epoch": 0.0664147486545288,
      "grad_norm": 1.948807716369629,
      "learning_rate": 2.8899999999999998e-05,
      "loss": 0.8253,
      "step": 290
    },
    {
      "epoch": 0.06755983052788274,
      "grad_norm": 2.3872830867767334,
      "learning_rate": 2.94e-05,
      "loss": 0.8235,
      "step": 295
    },
    {
      "epoch": 0.06870491240123669,
      "grad_norm": 1.0653102397918701,
      "learning_rate": 2.9900000000000002e-05,
      "loss": 0.8204,
      "step": 300
    },
    {
      "epoch": 0.06984999427459063,
      "grad_norm": 1.2086213827133179,
      "learning_rate": 3.04e-05,
      "loss": 0.8216,
      "step": 305
    },
    {
      "epoch": 0.07099507614794458,
      "grad_norm": 1.2478430271148682,
      "learning_rate": 3.09e-05,
      "loss": 0.8218,
      "step": 310
    },
    {
      "epoch": 0.07214015802129853,
      "grad_norm": 1.4686928987503052,
      "learning_rate": 3.1400000000000004e-05,
      "loss": 0.8175,
      "step": 315
    },
    {
      "epoch": 0.07328523989465247,
      "grad_norm": 1.691828966140747,
      "learning_rate": 3.19e-05,
      "loss": 0.8194,
      "step": 320
    },
    {
      "epoch": 0.07443032176800642,
      "grad_norm": 2.0422606468200684,
      "learning_rate": 3.24e-05,
      "loss": 0.8181,
      "step": 325
    },
    {
      "epoch": 0.07557540364136035,
      "grad_norm": 1.3948687314987183,
      "learning_rate": 3.29e-05,
      "loss": 0.8212,
      "step": 330
    },
    {
      "epoch": 0.0767204855147143,
      "grad_norm": 1.1443554162979126,
      "learning_rate": 3.3400000000000005e-05,
      "loss": 0.8169,
      "step": 335
    },
    {
      "epoch": 0.07786556738806824,
      "grad_norm": 1.0854233503341675,
      "learning_rate": 3.3900000000000004e-05,
      "loss": 0.8134,
      "step": 340
    },
    {
      "epoch": 0.07901064926142219,
      "grad_norm": 1.8612085580825806,
      "learning_rate": 3.4399999999999996e-05,
      "loss": 0.8133,
      "step": 345
    },
    {
      "epoch": 0.08015573113477614,
      "grad_norm": 0.5761221647262573,
      "learning_rate": 3.49e-05,
      "loss": 0.8101,
      "step": 350
    },
    {
      "epoch": 0.08130081300813008,
      "grad_norm": 0.7849775552749634,
      "learning_rate": 3.54e-05,
      "loss": 0.8096,
      "step": 355
    },
    {
      "epoch": 0.08244589488148403,
      "grad_norm": 0.47798988223075867,
      "learning_rate": 3.59e-05,
      "loss": 0.8089,
      "step": 360
    },
    {
      "epoch": 0.08359097675483797,
      "grad_norm": 0.9662396311759949,
      "learning_rate": 3.6400000000000004e-05,
      "loss": 0.8073,
      "step": 365
    },
    {
      "epoch": 0.08473605862819192,
      "grad_norm": 0.5217769742012024,
      "learning_rate": 3.69e-05,
      "loss": 0.8079,
      "step": 370
    },
    {
      "epoch": 0.08588114050154586,
      "grad_norm": 1.0022109746932983,
      "learning_rate": 3.74e-05,
      "loss": 0.8041,
      "step": 375
    },
    {
      "epoch": 0.0870262223748998,
      "grad_norm": 1.0508103370666504,
      "learning_rate": 3.79e-05,
      "loss": 0.8045,
      "step": 380
    },
    {
      "epoch": 0.08817130424825376,
      "grad_norm": 0.9267056584358215,
      "learning_rate": 3.8400000000000005e-05,
      "loss": 0.8071,
      "step": 385
    },
    {
      "epoch": 0.0893163861216077,
      "grad_norm": 1.0763800144195557,
      "learning_rate": 3.8900000000000004e-05,
      "loss": 0.8022,
      "step": 390
    },
    {
      "epoch": 0.09046146799496164,
      "grad_norm": 0.9646713733673096,
      "learning_rate": 3.94e-05,
      "loss": 0.8037,
      "step": 395
    },
    {
      "epoch": 0.09160654986831558,
      "grad_norm": 0.7930561900138855,
      "learning_rate": 3.99e-05,
      "loss": 0.8033,
      "step": 400
    },
    {
      "epoch": 0.09275163174166953,
      "grad_norm": 0.7772406339645386,
      "learning_rate": 4.0400000000000006e-05,
      "loss": 0.8014,
      "step": 405
    },
    {
      "epoch": 0.09389671361502347,
      "grad_norm": 0.7040679454803467,
      "learning_rate": 4.09e-05,
      "loss": 0.8016,
      "step": 410
    },
    {
      "epoch": 0.09504179548837742,
      "grad_norm": 0.6085620522499084,
      "learning_rate": 4.14e-05,
      "loss": 0.8009,
      "step": 415
    },
    {
      "epoch": 0.09618687736173136,
      "grad_norm": 0.9182414412498474,
      "learning_rate": 4.19e-05,
      "loss": 0.8023,
      "step": 420
    },
    {
      "epoch": 0.09733195923508531,
      "grad_norm": 0.9242359399795532,
      "learning_rate": 4.24e-05,
      "loss": 0.8014,
      "step": 425
    },
    {
      "epoch": 0.09847704110843926,
      "grad_norm": 1.0517164468765259,
      "learning_rate": 4.29e-05,
      "loss": 0.7987,
      "step": 430
    },
    {
      "epoch": 0.0996221229817932,
      "grad_norm": 0.9291065335273743,
      "learning_rate": 4.3400000000000005e-05,
      "loss": 0.8005,
      "step": 435
    },
    {
      "epoch": 0.10076720485514715,
      "grad_norm": 0.7035262584686279,
      "learning_rate": 4.39e-05,
      "loss": 0.7961,
      "step": 440
    },
    {
      "epoch": 0.10191228672850108,
      "grad_norm": 0.947799801826477,
      "learning_rate": 4.44e-05,
      "loss": 0.7991,
      "step": 445
    },
    {
      "epoch": 0.10305736860185503,
      "grad_norm": 1.345821738243103,
      "learning_rate": 4.49e-05,
      "loss": 0.7948,
      "step": 450
    },
    {
      "epoch": 0.10420245047520897,
      "grad_norm": 0.511234700679779,
      "learning_rate": 4.5400000000000006e-05,
      "loss": 0.7959,
      "step": 455
    },
    {
      "epoch": 0.10534753234856292,
      "grad_norm": 0.8376046419143677,
      "learning_rate": 4.5900000000000004e-05,
      "loss": 0.796,
      "step": 460
    },
    {
      "epoch": 0.10649261422191687,
      "grad_norm": 0.9656692147254944,
      "learning_rate": 4.64e-05,
      "loss": 0.7961,
      "step": 465
    },
    {
      "epoch": 0.10763769609527081,
      "grad_norm": 0.7987990379333496,
      "learning_rate": 4.69e-05,
      "loss": 0.794,
      "step": 470
    },
    {
      "epoch": 0.10878277796862476,
      "grad_norm": 0.9489181637763977,
      "learning_rate": 4.74e-05,
      "loss": 0.7928,
      "step": 475
    },
    {
      "epoch": 0.1099278598419787,
      "grad_norm": 1.0666885375976562,
      "learning_rate": 4.79e-05,
      "loss": 0.796,
      "step": 480
    },
    {
      "epoch": 0.11107294171533265,
      "grad_norm": 0.8832483887672424,
      "learning_rate": 4.8400000000000004e-05,
      "loss": 0.7921,
      "step": 485
    },
    {
      "epoch": 0.11221802358868659,
      "grad_norm": 0.5236397385597229,
      "learning_rate": 4.89e-05,
      "loss": 0.7967,
      "step": 490
    },
    {
      "epoch": 0.11336310546204054,
      "grad_norm": 0.6618067026138306,
      "learning_rate": 4.94e-05,
      "loss": 0.7922,
      "step": 495
    },
    {
      "epoch": 0.11450818733539449,
      "grad_norm": 0.5967982411384583,
      "learning_rate": 4.99e-05,
      "loss": 0.7907,
      "step": 500
    },
    {
      "epoch": 0.11565326920874842,
      "grad_norm": 0.9039576053619385,
      "learning_rate": 5.0400000000000005e-05,
      "loss": 0.7916,
      "step": 505
    },
    {
      "epoch": 0.11679835108210238,
      "grad_norm": 0.7755367159843445,
      "learning_rate": 5.0900000000000004e-05,
      "loss": 0.7916,
      "step": 510
    },
    {
      "epoch": 0.11794343295545631,
      "grad_norm": 0.6333810091018677,
      "learning_rate": 5.14e-05,
      "loss": 0.7917,
      "step": 515
    },
    {
      "epoch": 0.11908851482881026,
      "grad_norm": 0.5068004131317139,
      "learning_rate": 5.19e-05,
      "loss": 0.7936,
      "step": 520
    },
    {
      "epoch": 0.1202335967021642,
      "grad_norm": 0.8217163681983948,
      "learning_rate": 5.2400000000000007e-05,
      "loss": 0.7946,
      "step": 525
    },
    {
      "epoch": 0.12137867857551815,
      "grad_norm": 0.433635413646698,
      "learning_rate": 5.2900000000000005e-05,
      "loss": 0.7947,
      "step": 530
    },
    {
      "epoch": 0.12252376044887209,
      "grad_norm": 0.5382624864578247,
      "learning_rate": 5.3400000000000004e-05,
      "loss": 0.7937,
      "step": 535
    },
    {
      "epoch": 0.12366884232222604,
      "grad_norm": 0.43913567066192627,
      "learning_rate": 5.390000000000001e-05,
      "loss": 0.7929,
      "step": 540
    },
    {
      "epoch": 0.12481392419557999,
      "grad_norm": 0.6911773085594177,
      "learning_rate": 5.440000000000001e-05,
      "loss": 0.7916,
      "step": 545
    },
    {
      "epoch": 0.12595900606893393,
      "grad_norm": 0.4433671236038208,
      "learning_rate": 5.4900000000000006e-05,
      "loss": 0.7912,
      "step": 550
    },
    {
      "epoch": 0.12710408794228786,
      "grad_norm": 0.5388156175613403,
      "learning_rate": 5.5400000000000005e-05,
      "loss": 0.7867,
      "step": 555
    },
    {
      "epoch": 0.12824916981564183,
      "grad_norm": 0.6417338848114014,
      "learning_rate": 5.590000000000001e-05,
      "loss": 0.7916,
      "step": 560
    },
    {
      "epoch": 0.12939425168899577,
      "grad_norm": 0.7721633911132812,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 0.7931,
      "step": 565
    },
    {
      "epoch": 0.1305393335623497,
      "grad_norm": 0.47057852149009705,
      "learning_rate": 5.69e-05,
      "loss": 0.7926,
      "step": 570
    },
    {
      "epoch": 0.13168441543570364,
      "grad_norm": 0.7483288645744324,
      "learning_rate": 5.74e-05,
      "loss": 0.7907,
      "step": 575
    },
    {
      "epoch": 0.1328294973090576,
      "grad_norm": 0.6444791555404663,
      "learning_rate": 5.79e-05,
      "loss": 0.7923,
      "step": 580
    },
    {
      "epoch": 0.13397457918241154,
      "grad_norm": 0.6855381727218628,
      "learning_rate": 5.8399999999999997e-05,
      "loss": 0.7898,
      "step": 585
    },
    {
      "epoch": 0.13511966105576548,
      "grad_norm": 0.4442330300807953,
      "learning_rate": 5.89e-05,
      "loss": 0.7905,
      "step": 590
    },
    {
      "epoch": 0.13626474292911944,
      "grad_norm": 0.6380087733268738,
      "learning_rate": 5.94e-05,
      "loss": 0.7908,
      "step": 595
    },
    {
      "epoch": 0.13740982480247338,
      "grad_norm": 0.6280926465988159,
      "learning_rate": 5.99e-05,
      "loss": 0.7912,
      "step": 600
    },
    {
      "epoch": 0.13855490667582732,
      "grad_norm": 0.3598725497722626,
      "learning_rate": 6.04e-05,
      "loss": 0.7891,
      "step": 605
    },
    {
      "epoch": 0.13969998854918125,
      "grad_norm": 0.5610477924346924,
      "learning_rate": 6.09e-05,
      "loss": 0.7904,
      "step": 610
    },
    {
      "epoch": 0.14084507042253522,
      "grad_norm": 0.5610749125480652,
      "learning_rate": 6.14e-05,
      "loss": 0.7889,
      "step": 615
    },
    {
      "epoch": 0.14199015229588915,
      "grad_norm": 0.5461752414703369,
      "learning_rate": 6.19e-05,
      "loss": 0.7901,
      "step": 620
    },
    {
      "epoch": 0.1431352341692431,
      "grad_norm": 0.4764970541000366,
      "learning_rate": 6.24e-05,
      "loss": 0.7878,
      "step": 625
    },
    {
      "epoch": 0.14428031604259706,
      "grad_norm": 0.40810781717300415,
      "learning_rate": 6.29e-05,
      "loss": 0.789,
      "step": 630
    },
    {
      "epoch": 0.145425397915951,
      "grad_norm": 0.818898618221283,
      "learning_rate": 6.340000000000001e-05,
      "loss": 0.7877,
      "step": 635
    },
    {
      "epoch": 0.14657047978930493,
      "grad_norm": 0.6349887847900391,
      "learning_rate": 6.390000000000001e-05,
      "loss": 0.7902,
      "step": 640
    },
    {
      "epoch": 0.14771556166265887,
      "grad_norm": 0.34319865703582764,
      "learning_rate": 6.440000000000001e-05,
      "loss": 0.7899,
      "step": 645
    },
    {
      "epoch": 0.14886064353601283,
      "grad_norm": 0.6926677823066711,
      "learning_rate": 6.49e-05,
      "loss": 0.7863,
      "step": 650
    },
    {
      "epoch": 0.15000572540936677,
      "grad_norm": 0.6303408145904541,
      "learning_rate": 6.54e-05,
      "loss": 0.7854,
      "step": 655
    },
    {
      "epoch": 0.1511508072827207,
      "grad_norm": 0.5530775189399719,
      "learning_rate": 6.59e-05,
      "loss": 0.7894,
      "step": 660
    },
    {
      "epoch": 0.15229588915607467,
      "grad_norm": 0.5497018098831177,
      "learning_rate": 6.64e-05,
      "loss": 0.7873,
      "step": 665
    },
    {
      "epoch": 0.1534409710294286,
      "grad_norm": 0.6323860883712769,
      "learning_rate": 6.690000000000001e-05,
      "loss": 0.7886,
      "step": 670
    },
    {
      "epoch": 0.15458605290278254,
      "grad_norm": 0.40749162435531616,
      "learning_rate": 6.740000000000001e-05,
      "loss": 0.7887,
      "step": 675
    },
    {
      "epoch": 0.15573113477613648,
      "grad_norm": 0.4536402225494385,
      "learning_rate": 6.790000000000001e-05,
      "loss": 0.7867,
      "step": 680
    },
    {
      "epoch": 0.15687621664949045,
      "grad_norm": 0.5276408195495605,
      "learning_rate": 6.840000000000001e-05,
      "loss": 0.7871,
      "step": 685
    },
    {
      "epoch": 0.15802129852284438,
      "grad_norm": 0.4989128112792969,
      "learning_rate": 6.89e-05,
      "loss": 0.7895,
      "step": 690
    },
    {
      "epoch": 0.15916638039619832,
      "grad_norm": 0.7024186253547668,
      "learning_rate": 6.939999999999999e-05,
      "loss": 0.7847,
      "step": 695
    },
    {
      "epoch": 0.16031146226955229,
      "grad_norm": 0.37541908025741577,
      "learning_rate": 6.99e-05,
      "loss": 0.7833,
      "step": 700
    },
    {
      "epoch": 0.16145654414290622,
      "grad_norm": 0.5957688689231873,
      "learning_rate": 7.04e-05,
      "loss": 0.7864,
      "step": 705
    },
    {
      "epoch": 0.16260162601626016,
      "grad_norm": 0.8745490908622742,
      "learning_rate": 7.09e-05,
      "loss": 0.7851,
      "step": 710
    },
    {
      "epoch": 0.1637467078896141,
      "grad_norm": 0.6420118808746338,
      "learning_rate": 7.14e-05,
      "loss": 0.7879,
      "step": 715
    },
    {
      "epoch": 0.16489178976296806,
      "grad_norm": 0.5130624175071716,
      "learning_rate": 7.19e-05,
      "loss": 0.7864,
      "step": 720
    },
    {
      "epoch": 0.166036871636322,
      "grad_norm": 0.3838486671447754,
      "learning_rate": 7.24e-05,
      "loss": 0.7884,
      "step": 725
    },
    {
      "epoch": 0.16718195350967593,
      "grad_norm": 0.34363555908203125,
      "learning_rate": 7.29e-05,
      "loss": 0.7812,
      "step": 730
    },
    {
      "epoch": 0.1683270353830299,
      "grad_norm": 0.4423808455467224,
      "learning_rate": 7.340000000000001e-05,
      "loss": 0.7881,
      "step": 735
    },
    {
      "epoch": 0.16947211725638384,
      "grad_norm": 0.4986163377761841,
      "learning_rate": 7.390000000000001e-05,
      "loss": 0.7877,
      "step": 740
    },
    {
      "epoch": 0.17061719912973777,
      "grad_norm": 0.651222288608551,
      "learning_rate": 7.44e-05,
      "loss": 0.7871,
      "step": 745
    },
    {
      "epoch": 0.1717622810030917,
      "grad_norm": 0.6739994883537292,
      "learning_rate": 7.49e-05,
      "loss": 0.7854,
      "step": 750
    },
    {
      "epoch": 0.17290736287644567,
      "grad_norm": 0.6694870591163635,
      "learning_rate": 7.54e-05,
      "loss": 0.7858,
      "step": 755
    },
    {
      "epoch": 0.1740524447497996,
      "grad_norm": 0.4669223725795746,
      "learning_rate": 7.59e-05,
      "loss": 0.791,
      "step": 760
    },
    {
      "epoch": 0.17519752662315355,
      "grad_norm": 0.4475642144680023,
      "learning_rate": 7.64e-05,
      "loss": 0.7862,
      "step": 765
    },
    {
      "epoch": 0.1763426084965075,
      "grad_norm": 0.41131269931793213,
      "learning_rate": 7.69e-05,
      "loss": 0.7846,
      "step": 770
    },
    {
      "epoch": 0.17748769036986145,
      "grad_norm": 0.46661683917045593,
      "learning_rate": 7.740000000000001e-05,
      "loss": 0.7849,
      "step": 775
    },
    {
      "epoch": 0.1786327722432154,
      "grad_norm": 0.6970285773277283,
      "learning_rate": 7.790000000000001e-05,
      "loss": 0.7856,
      "step": 780
    },
    {
      "epoch": 0.17977785411656932,
      "grad_norm": 0.4303015470504761,
      "learning_rate": 7.840000000000001e-05,
      "loss": 0.7848,
      "step": 785
    },
    {
      "epoch": 0.1809229359899233,
      "grad_norm": 0.6247970461845398,
      "learning_rate": 7.890000000000001e-05,
      "loss": 0.7871,
      "step": 790
    },
    {
      "epoch": 0.18206801786327723,
      "grad_norm": 0.6847257614135742,
      "learning_rate": 7.94e-05,
      "loss": 0.7898,
      "step": 795
    },
    {
      "epoch": 0.18321309973663116,
      "grad_norm": 0.5924108624458313,
      "learning_rate": 7.99e-05,
      "loss": 0.7809,
      "step": 800
    },
    {
      "epoch": 0.1843581816099851,
      "grad_norm": 0.44637331366539,
      "learning_rate": 8.04e-05,
      "loss": 0.7858,
      "step": 805
    },
    {
      "epoch": 0.18550326348333906,
      "grad_norm": 0.4461766183376312,
      "learning_rate": 8.090000000000001e-05,
      "loss": 0.7844,
      "step": 810
    },
    {
      "epoch": 0.186648345356693,
      "grad_norm": 0.6761696934700012,
      "learning_rate": 8.14e-05,
      "loss": 0.7885,
      "step": 815
    },
    {
      "epoch": 0.18779342723004694,
      "grad_norm": 0.4983709752559662,
      "learning_rate": 8.19e-05,
      "loss": 0.7838,
      "step": 820
    },
    {
      "epoch": 0.1889385091034009,
      "grad_norm": 0.4978698790073395,
      "learning_rate": 8.24e-05,
      "loss": 0.7891,
      "step": 825
    },
    {
      "epoch": 0.19008359097675484,
      "grad_norm": 0.7057047486305237,
      "learning_rate": 8.29e-05,
      "loss": 0.7847,
      "step": 830
    },
    {
      "epoch": 0.19122867285010878,
      "grad_norm": 0.4118855595588684,
      "learning_rate": 8.34e-05,
      "loss": 0.7838,
      "step": 835
    },
    {
      "epoch": 0.19237375472346271,
      "grad_norm": 0.7234144806861877,
      "learning_rate": 8.39e-05,
      "loss": 0.7865,
      "step": 840
    },
    {
      "epoch": 0.19351883659681668,
      "grad_norm": 0.3011474907398224,
      "learning_rate": 8.44e-05,
      "loss": 0.7842,
      "step": 845
    },
    {
      "epoch": 0.19466391847017062,
      "grad_norm": 0.4353237450122833,
      "learning_rate": 8.49e-05,
      "loss": 0.7856,
      "step": 850
    },
    {
      "epoch": 0.19580900034352455,
      "grad_norm": 0.3795527219772339,
      "learning_rate": 8.54e-05,
      "loss": 0.7849,
      "step": 855
    },
    {
      "epoch": 0.19695408221687852,
      "grad_norm": 0.5458865165710449,
      "learning_rate": 8.59e-05,
      "loss": 0.7878,
      "step": 860
    },
    {
      "epoch": 0.19809916409023245,
      "grad_norm": 0.4245418310165405,
      "learning_rate": 8.64e-05,
      "loss": 0.7842,
      "step": 865
    },
    {
      "epoch": 0.1992442459635864,
      "grad_norm": 0.5682182908058167,
      "learning_rate": 8.69e-05,
      "loss": 0.7836,
      "step": 870
    },
    {
      "epoch": 0.20038932783694033,
      "grad_norm": 0.27713704109191895,
      "learning_rate": 8.740000000000001e-05,
      "loss": 0.7856,
      "step": 875
    },
    {
      "epoch": 0.2015344097102943,
      "grad_norm": 0.48992612957954407,
      "learning_rate": 8.790000000000001e-05,
      "loss": 0.7806,
      "step": 880
    },
    {
      "epoch": 0.20267949158364823,
      "grad_norm": 0.7062950134277344,
      "learning_rate": 8.840000000000001e-05,
      "loss": 0.784,
      "step": 885
    },
    {
      "epoch": 0.20382457345700217,
      "grad_norm": 0.5864018797874451,
      "learning_rate": 8.89e-05,
      "loss": 0.7823,
      "step": 890
    },
    {
      "epoch": 0.20496965533035613,
      "grad_norm": 0.5565068125724792,
      "learning_rate": 8.94e-05,
      "loss": 0.7835,
      "step": 895
    },
    {
      "epoch": 0.20611473720371007,
      "grad_norm": 0.3518352806568146,
      "learning_rate": 8.99e-05,
      "loss": 0.7833,
      "step": 900
    },
    {
      "epoch": 0.207259819077064,
      "grad_norm": 0.31621456146240234,
      "learning_rate": 9.04e-05,
      "loss": 0.7837,
      "step": 905
    },
    {
      "epoch": 0.20840490095041794,
      "grad_norm": 0.4907382130622864,
      "learning_rate": 9.090000000000001e-05,
      "loss": 0.7854,
      "step": 910
    },
    {
      "epoch": 0.2095499828237719,
      "grad_norm": 0.3738730549812317,
      "learning_rate": 9.140000000000001e-05,
      "loss": 0.7816,
      "step": 915
    },
    {
      "epoch": 0.21069506469712584,
      "grad_norm": 0.589777946472168,
      "learning_rate": 9.190000000000001e-05,
      "loss": 0.7826,
      "step": 920
    },
    {
      "epoch": 0.21184014657047978,
      "grad_norm": 0.3376914858818054,
      "learning_rate": 9.240000000000001e-05,
      "loss": 0.7851,
      "step": 925
    },
    {
      "epoch": 0.21298522844383375,
      "grad_norm": 0.34315842390060425,
      "learning_rate": 9.290000000000001e-05,
      "loss": 0.7853,
      "step": 930
    },
    {
      "epoch": 0.21413031031718768,
      "grad_norm": 0.34507983922958374,
      "learning_rate": 9.340000000000001e-05,
      "loss": 0.7882,
      "step": 935
    },
    {
      "epoch": 0.21527539219054162,
      "grad_norm": 0.33713966608047485,
      "learning_rate": 9.39e-05,
      "loss": 0.7824,
      "step": 940
    },
    {
      "epoch": 0.21642047406389556,
      "grad_norm": 0.4617432653903961,
      "learning_rate": 9.44e-05,
      "loss": 0.7837,
      "step": 945
    },
    {
      "epoch": 0.21756555593724952,
      "grad_norm": 0.3514646589756012,
      "learning_rate": 9.49e-05,
      "loss": 0.7816,
      "step": 950
    },
    {
      "epoch": 0.21871063781060346,
      "grad_norm": 0.3907168209552765,
      "learning_rate": 9.54e-05,
      "loss": 0.781,
      "step": 955
    },
    {
      "epoch": 0.2198557196839574,
      "grad_norm": 0.4069294333457947,
      "learning_rate": 9.59e-05,
      "loss": 0.7819,
      "step": 960
    },
    {
      "epoch": 0.22100080155731136,
      "grad_norm": 0.4836900532245636,
      "learning_rate": 9.64e-05,
      "loss": 0.7826,
      "step": 965
    },
    {
      "epoch": 0.2221458834306653,
      "grad_norm": 0.387287437915802,
      "learning_rate": 9.69e-05,
      "loss": 0.7819,
      "step": 970
    },
    {
      "epoch": 0.22329096530401923,
      "grad_norm": 0.3079049289226532,
      "learning_rate": 9.74e-05,
      "loss": 0.7808,
      "step": 975
    },
    {
      "epoch": 0.22443604717737317,
      "grad_norm": 0.3615763187408447,
      "learning_rate": 9.790000000000001e-05,
      "loss": 0.7844,
      "step": 980
    },
    {
      "epoch": 0.22558112905072714,
      "grad_norm": 0.4044090509414673,
      "learning_rate": 9.84e-05,
      "loss": 0.7804,
      "step": 985
    },
    {
      "epoch": 0.22672621092408107,
      "grad_norm": 0.2885280251502991,
      "learning_rate": 9.89e-05,
      "loss": 0.7827,
      "step": 990
    },
    {
      "epoch": 0.227871292797435,
      "grad_norm": 0.41335442662239075,
      "learning_rate": 9.94e-05,
      "loss": 0.7798,
      "step": 995
    },
    {
      "epoch": 0.22901637467078897,
      "grad_norm": 0.35099464654922485,
      "learning_rate": 9.99e-05,
      "loss": 0.781,
      "step": 1000
    },
    {
      "epoch": 0.2301614565441429,
      "grad_norm": 0.3341158330440521,
      "learning_rate": 9.994828032066201e-05,
      "loss": 0.7832,
      "step": 1005
    },
    {
      "epoch": 0.23130653841749685,
      "grad_norm": 0.5810456275939941,
      "learning_rate": 9.988363072148953e-05,
      "loss": 0.7852,
      "step": 1010
    },
    {
      "epoch": 0.23245162029085079,
      "grad_norm": 0.487460732460022,
      "learning_rate": 9.981898112231704e-05,
      "loss": 0.7856,
      "step": 1015
    },
    {
      "epoch": 0.23359670216420475,
      "grad_norm": 0.6533329486846924,
      "learning_rate": 9.975433152314456e-05,
      "loss": 0.7842,
      "step": 1020
    },
    {
      "epoch": 0.2347417840375587,
      "grad_norm": 0.2424434870481491,
      "learning_rate": 9.968968192397208e-05,
      "loss": 0.7829,
      "step": 1025
    },
    {
      "epoch": 0.23588686591091262,
      "grad_norm": 0.4848748743534088,
      "learning_rate": 9.96250323247996e-05,
      "loss": 0.7802,
      "step": 1030
    },
    {
      "epoch": 0.23703194778426656,
      "grad_norm": 0.38208287954330444,
      "learning_rate": 9.956038272562711e-05,
      "loss": 0.7855,
      "step": 1035
    },
    {
      "epoch": 0.23817702965762053,
      "grad_norm": 0.5554670095443726,
      "learning_rate": 9.949573312645461e-05,
      "loss": 0.7829,
      "step": 1040
    },
    {
      "epoch": 0.23932211153097446,
      "grad_norm": 0.3737400770187378,
      "learning_rate": 9.943108352728213e-05,
      "loss": 0.7815,
      "step": 1045
    },
    {
      "epoch": 0.2404671934043284,
      "grad_norm": 0.287233829498291,
      "learning_rate": 9.936643392810964e-05,
      "loss": 0.7868,
      "step": 1050
    },
    {
      "epoch": 0.24161227527768236,
      "grad_norm": 0.2772092819213867,
      "learning_rate": 9.930178432893717e-05,
      "loss": 0.7774,
      "step": 1055
    },
    {
      "epoch": 0.2427573571510363,
      "grad_norm": 0.35240355134010315,
      "learning_rate": 9.923713472976468e-05,
      "loss": 0.781,
      "step": 1060
    },
    {
      "epoch": 0.24390243902439024,
      "grad_norm": 0.21047982573509216,
      "learning_rate": 9.91724851305922e-05,
      "loss": 0.7837,
      "step": 1065
    },
    {
      "epoch": 0.24504752089774418,
      "grad_norm": 0.37012019753456116,
      "learning_rate": 9.910783553141971e-05,
      "loss": 0.7809,
      "step": 1070
    },
    {
      "epoch": 0.24619260277109814,
      "grad_norm": 0.5411164164543152,
      "learning_rate": 9.904318593224723e-05,
      "loss": 0.7828,
      "step": 1075
    },
    {
      "epoch": 0.24733768464445208,
      "grad_norm": 0.5926540493965149,
      "learning_rate": 9.897853633307474e-05,
      "loss": 0.7856,
      "step": 1080
    },
    {
      "epoch": 0.24848276651780601,
      "grad_norm": 0.31834328174591064,
      "learning_rate": 9.891388673390225e-05,
      "loss": 0.7848,
      "step": 1085
    },
    {
      "epoch": 0.24962784839115998,
      "grad_norm": 0.9762104153633118,
      "learning_rate": 9.884923713472977e-05,
      "loss": 0.7852,
      "step": 1090
    },
    {
      "epoch": 0.2507729302645139,
      "grad_norm": 0.5515075325965881,
      "learning_rate": 9.878458753555728e-05,
      "loss": 0.7801,
      "step": 1095
    },
    {
      "epoch": 0.25191801213786785,
      "grad_norm": 0.5665395855903625,
      "learning_rate": 9.87199379363848e-05,
      "loss": 0.7808,
      "step": 1100
    },
    {
      "epoch": 0.2530630940112218,
      "grad_norm": 0.2942984104156494,
      "learning_rate": 9.865528833721231e-05,
      "loss": 0.78,
      "step": 1105
    },
    {
      "epoch": 0.2542081758845757,
      "grad_norm": 0.2915126085281372,
      "learning_rate": 9.859063873803983e-05,
      "loss": 0.7799,
      "step": 1110
    },
    {
      "epoch": 0.2553532577579297,
      "grad_norm": 0.39388832449913025,
      "learning_rate": 9.852598913886735e-05,
      "loss": 0.7801,
      "step": 1115
    },
    {
      "epoch": 0.25649833963128366,
      "grad_norm": 0.3185393810272217,
      "learning_rate": 9.846133953969487e-05,
      "loss": 0.7802,
      "step": 1120
    },
    {
      "epoch": 0.2576434215046376,
      "grad_norm": 0.4271395802497864,
      "learning_rate": 9.839668994052237e-05,
      "loss": 0.7819,
      "step": 1125
    },
    {
      "epoch": 0.25878850337799153,
      "grad_norm": 0.3662160336971283,
      "learning_rate": 9.833204034134988e-05,
      "loss": 0.7821,
      "step": 1130
    },
    {
      "epoch": 0.25993358525134547,
      "grad_norm": 0.38755324482917786,
      "learning_rate": 9.82673907421774e-05,
      "loss": 0.7803,
      "step": 1135
    },
    {
      "epoch": 0.2610786671246994,
      "grad_norm": 0.2804664671421051,
      "learning_rate": 9.820274114300491e-05,
      "loss": 0.778,
      "step": 1140
    },
    {
      "epoch": 0.26222374899805334,
      "grad_norm": 0.5177549719810486,
      "learning_rate": 9.813809154383244e-05,
      "loss": 0.7789,
      "step": 1145
    },
    {
      "epoch": 0.2633688308714073,
      "grad_norm": 0.3463110029697418,
      "learning_rate": 9.807344194465995e-05,
      "loss": 0.7794,
      "step": 1150
    },
    {
      "epoch": 0.26451391274476127,
      "grad_norm": 0.6302537322044373,
      "learning_rate": 9.800879234548747e-05,
      "loss": 0.7809,
      "step": 1155
    },
    {
      "epoch": 0.2656589946181152,
      "grad_norm": 0.3459393382072449,
      "learning_rate": 9.794414274631498e-05,
      "loss": 0.7789,
      "step": 1160
    },
    {
      "epoch": 0.26680407649146914,
      "grad_norm": 0.2957208454608917,
      "learning_rate": 9.787949314714248e-05,
      "loss": 0.7818,
      "step": 1165
    },
    {
      "epoch": 0.2679491583648231,
      "grad_norm": 0.3508155941963196,
      "learning_rate": 9.781484354797001e-05,
      "loss": 0.7812,
      "step": 1170
    },
    {
      "epoch": 0.269094240238177,
      "grad_norm": 0.21779291331768036,
      "learning_rate": 9.775019394879752e-05,
      "loss": 0.7805,
      "step": 1175
    },
    {
      "epoch": 0.27023932211153096,
      "grad_norm": 0.42935284972190857,
      "learning_rate": 9.768554434962504e-05,
      "loss": 0.7779,
      "step": 1180
    },
    {
      "epoch": 0.2713844039848849,
      "grad_norm": 0.37435927987098694,
      "learning_rate": 9.762089475045255e-05,
      "loss": 0.7796,
      "step": 1185
    },
    {
      "epoch": 0.2725294858582389,
      "grad_norm": 0.3577134311199188,
      "learning_rate": 9.755624515128007e-05,
      "loss": 0.7779,
      "step": 1190
    },
    {
      "epoch": 0.2736745677315928,
      "grad_norm": 0.36118873953819275,
      "learning_rate": 9.749159555210758e-05,
      "loss": 0.7775,
      "step": 1195
    },
    {
      "epoch": 0.27481964960494676,
      "grad_norm": 0.38497358560562134,
      "learning_rate": 9.74269459529351e-05,
      "loss": 0.776,
      "step": 1200
    },
    {
      "epoch": 0.2759647314783007,
      "grad_norm": 0.3275431990623474,
      "learning_rate": 9.736229635376262e-05,
      "loss": 0.7828,
      "step": 1205
    },
    {
      "epoch": 0.27710981335165463,
      "grad_norm": 0.3107474148273468,
      "learning_rate": 9.729764675459012e-05,
      "loss": 0.7777,
      "step": 1210
    },
    {
      "epoch": 0.27825489522500857,
      "grad_norm": 0.3608294129371643,
      "learning_rate": 9.723299715541764e-05,
      "loss": 0.7791,
      "step": 1215
    },
    {
      "epoch": 0.2793999770983625,
      "grad_norm": 0.3778846859931946,
      "learning_rate": 9.716834755624515e-05,
      "loss": 0.7844,
      "step": 1220
    },
    {
      "epoch": 0.2805450589717165,
      "grad_norm": 0.3609561622142792,
      "learning_rate": 9.710369795707267e-05,
      "loss": 0.7776,
      "step": 1225
    },
    {
      "epoch": 0.28169014084507044,
      "grad_norm": 0.2806609272956848,
      "learning_rate": 9.703904835790018e-05,
      "loss": 0.7797,
      "step": 1230
    },
    {
      "epoch": 0.2828352227184244,
      "grad_norm": 0.3381900191307068,
      "learning_rate": 9.697439875872771e-05,
      "loss": 0.7813,
      "step": 1235
    },
    {
      "epoch": 0.2839803045917783,
      "grad_norm": 0.25639185309410095,
      "learning_rate": 9.690974915955522e-05,
      "loss": 0.7802,
      "step": 1240
    },
    {
      "epoch": 0.28512538646513225,
      "grad_norm": 0.3533225357532501,
      "learning_rate": 9.684509956038274e-05,
      "loss": 0.7773,
      "step": 1245
    },
    {
      "epoch": 0.2862704683384862,
      "grad_norm": 0.4826221466064453,
      "learning_rate": 9.678044996121024e-05,
      "loss": 0.7789,
      "step": 1250
    },
    {
      "epoch": 0.2874155502118401,
      "grad_norm": 0.20925547182559967,
      "learning_rate": 9.671580036203775e-05,
      "loss": 0.7864,
      "step": 1255
    },
    {
      "epoch": 0.2885606320851941,
      "grad_norm": 0.3238098919391632,
      "learning_rate": 9.665115076286527e-05,
      "loss": 0.7782,
      "step": 1260
    },
    {
      "epoch": 0.28970571395854805,
      "grad_norm": 0.451593279838562,
      "learning_rate": 9.65865011636928e-05,
      "loss": 0.7817,
      "step": 1265
    },
    {
      "epoch": 0.290850795831902,
      "grad_norm": 0.3764170706272125,
      "learning_rate": 9.652185156452031e-05,
      "loss": 0.7807,
      "step": 1270
    },
    {
      "epoch": 0.2919958777052559,
      "grad_norm": 0.40236976742744446,
      "learning_rate": 9.645720196534782e-05,
      "loss": 0.7804,
      "step": 1275
    },
    {
      "epoch": 0.29314095957860986,
      "grad_norm": 0.4926491677761078,
      "learning_rate": 9.639255236617534e-05,
      "loss": 0.776,
      "step": 1280
    },
    {
      "epoch": 0.2942860414519638,
      "grad_norm": 0.4249913692474365,
      "learning_rate": 9.632790276700285e-05,
      "loss": 0.7793,
      "step": 1285
    },
    {
      "epoch": 0.29543112332531773,
      "grad_norm": 0.45594412088394165,
      "learning_rate": 9.626325316783036e-05,
      "loss": 0.7791,
      "step": 1290
    },
    {
      "epoch": 0.2965762051986717,
      "grad_norm": 0.2865663170814514,
      "learning_rate": 9.619860356865788e-05,
      "loss": 0.7801,
      "step": 1295
    },
    {
      "epoch": 0.29772128707202566,
      "grad_norm": 0.4077535569667816,
      "learning_rate": 9.613395396948539e-05,
      "loss": 0.7756,
      "step": 1300
    },
    {
      "epoch": 0.2988663689453796,
      "grad_norm": 0.43570780754089355,
      "learning_rate": 9.606930437031291e-05,
      "loss": 0.7796,
      "step": 1305
    },
    {
      "epoch": 0.30001145081873354,
      "grad_norm": 0.304936945438385,
      "learning_rate": 9.600465477114042e-05,
      "loss": 0.7785,
      "step": 1310
    },
    {
      "epoch": 0.3011565326920875,
      "grad_norm": 0.39299318194389343,
      "learning_rate": 9.594000517196794e-05,
      "loss": 0.7809,
      "step": 1315
    },
    {
      "epoch": 0.3023016145654414,
      "grad_norm": 0.4470611810684204,
      "learning_rate": 9.587535557279545e-05,
      "loss": 0.78,
      "step": 1320
    },
    {
      "epoch": 0.30344669643879535,
      "grad_norm": 0.9563407301902771,
      "learning_rate": 9.581070597362298e-05,
      "loss": 0.7793,
      "step": 1325
    },
    {
      "epoch": 0.30459177831214934,
      "grad_norm": 0.6523124575614929,
      "learning_rate": 9.574605637445049e-05,
      "loss": 0.7848,
      "step": 1330
    },
    {
      "epoch": 0.3057368601855033,
      "grad_norm": 0.4043123424053192,
      "learning_rate": 9.568140677527799e-05,
      "loss": 0.783,
      "step": 1335
    },
    {
      "epoch": 0.3068819420588572,
      "grad_norm": 0.2784610986709595,
      "learning_rate": 9.561675717610551e-05,
      "loss": 0.7784,
      "step": 1340
    },
    {
      "epoch": 0.30802702393221115,
      "grad_norm": 0.34219038486480713,
      "learning_rate": 9.555210757693302e-05,
      "loss": 0.7752,
      "step": 1345
    },
    {
      "epoch": 0.3091721058055651,
      "grad_norm": 0.25641149282455444,
      "learning_rate": 9.548745797776054e-05,
      "loss": 0.7767,
      "step": 1350
    },
    {
      "epoch": 0.310317187678919,
      "grad_norm": 0.2744545340538025,
      "learning_rate": 9.542280837858806e-05,
      "loss": 0.779,
      "step": 1355
    },
    {
      "epoch": 0.31146226955227296,
      "grad_norm": 0.3130733072757721,
      "learning_rate": 9.535815877941558e-05,
      "loss": 0.7789,
      "step": 1360
    },
    {
      "epoch": 0.31260735142562696,
      "grad_norm": 0.27104058861732483,
      "learning_rate": 9.529350918024309e-05,
      "loss": 0.7764,
      "step": 1365
    },
    {
      "epoch": 0.3137524332989809,
      "grad_norm": 0.33596792817115784,
      "learning_rate": 9.52288595810706e-05,
      "loss": 0.7742,
      "step": 1370
    },
    {
      "epoch": 0.31489751517233483,
      "grad_norm": 0.23721632361412048,
      "learning_rate": 9.51642099818981e-05,
      "loss": 0.776,
      "step": 1375
    },
    {
      "epoch": 0.31604259704568877,
      "grad_norm": 0.29659345746040344,
      "learning_rate": 9.509956038272563e-05,
      "loss": 0.7764,
      "step": 1380
    },
    {
      "epoch": 0.3171876789190427,
      "grad_norm": 0.27600282430648804,
      "learning_rate": 9.503491078355315e-05,
      "loss": 0.7756,
      "step": 1385
    },
    {
      "epoch": 0.31833276079239664,
      "grad_norm": 0.38308611512184143,
      "learning_rate": 9.497026118438066e-05,
      "loss": 0.7783,
      "step": 1390
    },
    {
      "epoch": 0.3194778426657506,
      "grad_norm": 0.22261874377727509,
      "learning_rate": 9.490561158520818e-05,
      "loss": 0.7762,
      "step": 1395
    },
    {
      "epoch": 0.32062292453910457,
      "grad_norm": 0.3498833179473877,
      "learning_rate": 9.484096198603569e-05,
      "loss": 0.7771,
      "step": 1400
    },
    {
      "epoch": 0.3217680064124585,
      "grad_norm": 0.33357107639312744,
      "learning_rate": 9.47763123868632e-05,
      "loss": 0.7781,
      "step": 1405
    },
    {
      "epoch": 0.32291308828581244,
      "grad_norm": 0.2920812666416168,
      "learning_rate": 9.471166278769072e-05,
      "loss": 0.776,
      "step": 1410
    },
    {
      "epoch": 0.3240581701591664,
      "grad_norm": 0.48737823963165283,
      "learning_rate": 9.464701318851825e-05,
      "loss": 0.7775,
      "step": 1415
    },
    {
      "epoch": 0.3252032520325203,
      "grad_norm": 0.3747755289077759,
      "learning_rate": 9.458236358934575e-05,
      "loss": 0.7749,
      "step": 1420
    },
    {
      "epoch": 0.32634833390587425,
      "grad_norm": 0.2682911455631256,
      "learning_rate": 9.451771399017326e-05,
      "loss": 0.7768,
      "step": 1425
    },
    {
      "epoch": 0.3274934157792282,
      "grad_norm": 0.21982483565807343,
      "learning_rate": 9.445306439100078e-05,
      "loss": 0.7795,
      "step": 1430
    },
    {
      "epoch": 0.3286384976525822,
      "grad_norm": 0.39821624755859375,
      "learning_rate": 9.438841479182829e-05,
      "loss": 0.7797,
      "step": 1435
    },
    {
      "epoch": 0.3297835795259361,
      "grad_norm": 0.3070865571498871,
      "learning_rate": 9.43237651926558e-05,
      "loss": 0.7743,
      "step": 1440
    },
    {
      "epoch": 0.33092866139929006,
      "grad_norm": 0.46173593401908875,
      "learning_rate": 9.425911559348333e-05,
      "loss": 0.7776,
      "step": 1445
    },
    {
      "epoch": 0.332073743272644,
      "grad_norm": 0.66759192943573,
      "learning_rate": 9.419446599431085e-05,
      "loss": 0.7743,
      "step": 1450
    },
    {
      "epoch": 0.33321882514599793,
      "grad_norm": 0.30952057242393494,
      "learning_rate": 9.412981639513836e-05,
      "loss": 0.7757,
      "step": 1455
    },
    {
      "epoch": 0.33436390701935187,
      "grad_norm": 0.433117538690567,
      "learning_rate": 9.406516679596586e-05,
      "loss": 0.7757,
      "step": 1460
    },
    {
      "epoch": 0.3355089888927058,
      "grad_norm": 0.4374425709247589,
      "learning_rate": 9.400051719679338e-05,
      "loss": 0.7765,
      "step": 1465
    },
    {
      "epoch": 0.3366540707660598,
      "grad_norm": 0.3277287781238556,
      "learning_rate": 9.393586759762089e-05,
      "loss": 0.7772,
      "step": 1470
    },
    {
      "epoch": 0.33779915263941374,
      "grad_norm": 0.3934798240661621,
      "learning_rate": 9.387121799844842e-05,
      "loss": 0.7723,
      "step": 1475
    },
    {
      "epoch": 0.3389442345127677,
      "grad_norm": 0.23350226879119873,
      "learning_rate": 9.380656839927593e-05,
      "loss": 0.7771,
      "step": 1480
    },
    {
      "epoch": 0.3400893163861216,
      "grad_norm": 0.3555000126361847,
      "learning_rate": 9.374191880010345e-05,
      "loss": 0.7791,
      "step": 1485
    },
    {
      "epoch": 0.34123439825947555,
      "grad_norm": 0.26215437054634094,
      "learning_rate": 9.367726920093096e-05,
      "loss": 0.7726,
      "step": 1490
    },
    {
      "epoch": 0.3423794801328295,
      "grad_norm": 0.3617369532585144,
      "learning_rate": 9.361261960175847e-05,
      "loss": 0.7724,
      "step": 1495
    },
    {
      "epoch": 0.3435245620061834,
      "grad_norm": 0.2677263617515564,
      "learning_rate": 9.354797000258599e-05,
      "loss": 0.7762,
      "step": 1500
    },
    {
      "epoch": 0.3446696438795374,
      "grad_norm": 0.38829365372657776,
      "learning_rate": 9.34833204034135e-05,
      "loss": 0.7782,
      "step": 1505
    },
    {
      "epoch": 0.34581472575289135,
      "grad_norm": 0.30776435136795044,
      "learning_rate": 9.341867080424102e-05,
      "loss": 0.7762,
      "step": 1510
    },
    {
      "epoch": 0.3469598076262453,
      "grad_norm": 0.4308106005191803,
      "learning_rate": 9.335402120506853e-05,
      "loss": 0.7716,
      "step": 1515
    },
    {
      "epoch": 0.3481048894995992,
      "grad_norm": 0.358279287815094,
      "learning_rate": 9.328937160589605e-05,
      "loss": 0.7774,
      "step": 1520
    },
    {
      "epoch": 0.34924997137295316,
      "grad_norm": 0.32716208696365356,
      "learning_rate": 9.322472200672356e-05,
      "loss": 0.7793,
      "step": 1525
    },
    {
      "epoch": 0.3503950532463071,
      "grad_norm": 0.35342416167259216,
      "learning_rate": 9.316007240755107e-05,
      "loss": 0.775,
      "step": 1530
    },
    {
      "epoch": 0.35154013511966103,
      "grad_norm": 0.4312511682510376,
      "learning_rate": 9.30954228083786e-05,
      "loss": 0.7707,
      "step": 1535
    },
    {
      "epoch": 0.352685216993015,
      "grad_norm": 0.38912200927734375,
      "learning_rate": 9.303077320920612e-05,
      "loss": 0.7769,
      "step": 1540
    },
    {
      "epoch": 0.35383029886636896,
      "grad_norm": 0.3942466974258423,
      "learning_rate": 9.296612361003362e-05,
      "loss": 0.7788,
      "step": 1545
    },
    {
      "epoch": 0.3549753807397229,
      "grad_norm": 0.28408968448638916,
      "learning_rate": 9.290147401086113e-05,
      "loss": 0.7762,
      "step": 1550
    },
    {
      "epoch": 0.35612046261307684,
      "grad_norm": 0.2877650260925293,
      "learning_rate": 9.283682441168865e-05,
      "loss": 0.776,
      "step": 1555
    },
    {
      "epoch": 0.3572655444864308,
      "grad_norm": 0.22159777581691742,
      "learning_rate": 9.277217481251616e-05,
      "loss": 0.7717,
      "step": 1560
    },
    {
      "epoch": 0.3584106263597847,
      "grad_norm": 0.2480212152004242,
      "learning_rate": 9.270752521334369e-05,
      "loss": 0.7755,
      "step": 1565
    },
    {
      "epoch": 0.35955570823313865,
      "grad_norm": 0.3367151916027069,
      "learning_rate": 9.26428756141712e-05,
      "loss": 0.7744,
      "step": 1570
    },
    {
      "epoch": 0.36070079010649264,
      "grad_norm": 0.38849833607673645,
      "learning_rate": 9.257822601499872e-05,
      "loss": 0.7741,
      "step": 1575
    },
    {
      "epoch": 0.3618458719798466,
      "grad_norm": 0.43399858474731445,
      "learning_rate": 9.251357641582623e-05,
      "loss": 0.7716,
      "step": 1580
    },
    {
      "epoch": 0.3629909538532005,
      "grad_norm": 0.4505549371242523,
      "learning_rate": 9.244892681665373e-05,
      "loss": 0.7744,
      "step": 1585
    },
    {
      "epoch": 0.36413603572655445,
      "grad_norm": 0.4170837104320526,
      "learning_rate": 9.238427721748126e-05,
      "loss": 0.7746,
      "step": 1590
    },
    {
      "epoch": 0.3652811175999084,
      "grad_norm": 0.3157612085342407,
      "learning_rate": 9.231962761830877e-05,
      "loss": 0.7772,
      "step": 1595
    },
    {
      "epoch": 0.3664261994732623,
      "grad_norm": 0.2665364742279053,
      "learning_rate": 9.225497801913629e-05,
      "loss": 0.7769,
      "step": 1600
    },
    {
      "epoch": 0.36757128134661626,
      "grad_norm": 0.32196542620658875,
      "learning_rate": 9.21903284199638e-05,
      "loss": 0.7757,
      "step": 1605
    },
    {
      "epoch": 0.3687163632199702,
      "grad_norm": 0.3406280279159546,
      "learning_rate": 9.212567882079132e-05,
      "loss": 0.777,
      "step": 1610
    },
    {
      "epoch": 0.3698614450933242,
      "grad_norm": 0.22886425256729126,
      "learning_rate": 9.206102922161883e-05,
      "loss": 0.7706,
      "step": 1615
    },
    {
      "epoch": 0.37100652696667813,
      "grad_norm": 0.3251831829547882,
      "learning_rate": 9.199637962244634e-05,
      "loss": 0.7775,
      "step": 1620
    },
    {
      "epoch": 0.37215160884003207,
      "grad_norm": 0.5250852704048157,
      "learning_rate": 9.193173002327387e-05,
      "loss": 0.7769,
      "step": 1625
    },
    {
      "epoch": 0.373296690713386,
      "grad_norm": 0.28842002153396606,
      "learning_rate": 9.186708042410137e-05,
      "loss": 0.7739,
      "step": 1630
    },
    {
      "epoch": 0.37444177258673994,
      "grad_norm": 0.35593077540397644,
      "learning_rate": 9.180243082492889e-05,
      "loss": 0.7739,
      "step": 1635
    },
    {
      "epoch": 0.3755868544600939,
      "grad_norm": 0.23034408688545227,
      "learning_rate": 9.17377812257564e-05,
      "loss": 0.7711,
      "step": 1640
    },
    {
      "epoch": 0.3767319363334478,
      "grad_norm": 0.4300076365470886,
      "learning_rate": 9.167313162658391e-05,
      "loss": 0.7723,
      "step": 1645
    },
    {
      "epoch": 0.3778770182068018,
      "grad_norm": 0.805209219455719,
      "learning_rate": 9.160848202741143e-05,
      "loss": 0.7742,
      "step": 1650
    },
    {
      "epoch": 0.37902210008015574,
      "grad_norm": 0.30290111899375916,
      "learning_rate": 9.154383242823896e-05,
      "loss": 0.7715,
      "step": 1655
    },
    {
      "epoch": 0.3801671819535097,
      "grad_norm": 0.3359617590904236,
      "learning_rate": 9.147918282906647e-05,
      "loss": 0.7764,
      "step": 1660
    },
    {
      "epoch": 0.3813122638268636,
      "grad_norm": 0.23703722655773163,
      "learning_rate": 9.141453322989399e-05,
      "loss": 0.7745,
      "step": 1665
    },
    {
      "epoch": 0.38245734570021755,
      "grad_norm": 0.4906832277774811,
      "learning_rate": 9.134988363072149e-05,
      "loss": 0.7743,
      "step": 1670
    },
    {
      "epoch": 0.3836024275735715,
      "grad_norm": 0.4271215498447418,
      "learning_rate": 9.1285234031549e-05,
      "loss": 0.7748,
      "step": 1675
    },
    {
      "epoch": 0.38474750944692543,
      "grad_norm": 0.3383249044418335,
      "learning_rate": 9.122058443237653e-05,
      "loss": 0.7763,
      "step": 1680
    },
    {
      "epoch": 0.3858925913202794,
      "grad_norm": 0.45870041847229004,
      "learning_rate": 9.115593483320404e-05,
      "loss": 0.7738,
      "step": 1685
    },
    {
      "epoch": 0.38703767319363336,
      "grad_norm": 0.3419291079044342,
      "learning_rate": 9.109128523403156e-05,
      "loss": 0.7727,
      "step": 1690
    },
    {
      "epoch": 0.3881827550669873,
      "grad_norm": 0.27392908930778503,
      "learning_rate": 9.102663563485907e-05,
      "loss": 0.7749,
      "step": 1695
    },
    {
      "epoch": 0.38932783694034123,
      "grad_norm": 0.36585626006126404,
      "learning_rate": 9.096198603568658e-05,
      "loss": 0.7731,
      "step": 1700
    },
    {
      "epoch": 0.39047291881369517,
      "grad_norm": 0.31979432702064514,
      "learning_rate": 9.08973364365141e-05,
      "loss": 0.7744,
      "step": 1705
    },
    {
      "epoch": 0.3916180006870491,
      "grad_norm": 0.6172612309455872,
      "learning_rate": 9.083268683734161e-05,
      "loss": 0.7718,
      "step": 1710
    },
    {
      "epoch": 0.39276308256040304,
      "grad_norm": 0.5099327564239502,
      "learning_rate": 9.076803723816913e-05,
      "loss": 0.7745,
      "step": 1715
    },
    {
      "epoch": 0.39390816443375704,
      "grad_norm": 0.2649715840816498,
      "learning_rate": 9.070338763899664e-05,
      "loss": 0.7723,
      "step": 1720
    },
    {
      "epoch": 0.395053246307111,
      "grad_norm": 0.3346767723560333,
      "learning_rate": 9.063873803982416e-05,
      "loss": 0.7754,
      "step": 1725
    },
    {
      "epoch": 0.3961983281804649,
      "grad_norm": 0.36208733916282654,
      "learning_rate": 9.057408844065167e-05,
      "loss": 0.7716,
      "step": 1730
    },
    {
      "epoch": 0.39734341005381885,
      "grad_norm": 0.5012935996055603,
      "learning_rate": 9.050943884147918e-05,
      "loss": 0.7741,
      "step": 1735
    },
    {
      "epoch": 0.3984884919271728,
      "grad_norm": 0.25386548042297363,
      "learning_rate": 9.04447892423067e-05,
      "loss": 0.7722,
      "step": 1740
    },
    {
      "epoch": 0.3996335738005267,
      "grad_norm": 0.5945486426353455,
      "learning_rate": 9.038013964313423e-05,
      "loss": 0.7735,
      "step": 1745
    },
    {
      "epoch": 0.40077865567388066,
      "grad_norm": 0.3223184049129486,
      "learning_rate": 9.031549004396174e-05,
      "loss": 0.7729,
      "step": 1750
    },
    {
      "epoch": 0.40192373754723465,
      "grad_norm": 0.30719512701034546,
      "learning_rate": 9.025084044478924e-05,
      "loss": 0.7776,
      "step": 1755
    },
    {
      "epoch": 0.4030688194205886,
      "grad_norm": 0.2536080777645111,
      "learning_rate": 9.018619084561676e-05,
      "loss": 0.7723,
      "step": 1760
    },
    {
      "epoch": 0.4042139012939425,
      "grad_norm": 0.31377112865448,
      "learning_rate": 9.012154124644427e-05,
      "loss": 0.7722,
      "step": 1765
    },
    {
      "epoch": 0.40535898316729646,
      "grad_norm": 0.5046892762184143,
      "learning_rate": 9.005689164727178e-05,
      "loss": 0.77,
      "step": 1770
    },
    {
      "epoch": 0.4065040650406504,
      "grad_norm": 0.4742714464664459,
      "learning_rate": 8.999224204809931e-05,
      "loss": 0.7738,
      "step": 1775
    },
    {
      "epoch": 0.40764914691400433,
      "grad_norm": 0.39765727519989014,
      "learning_rate": 8.992759244892683e-05,
      "loss": 0.7733,
      "step": 1780
    },
    {
      "epoch": 0.40879422878735827,
      "grad_norm": 0.28733646869659424,
      "learning_rate": 8.986294284975434e-05,
      "loss": 0.773,
      "step": 1785
    },
    {
      "epoch": 0.40993931066071226,
      "grad_norm": 0.38625991344451904,
      "learning_rate": 8.979829325058185e-05,
      "loss": 0.772,
      "step": 1790
    },
    {
      "epoch": 0.4110843925340662,
      "grad_norm": 0.23286648094654083,
      "learning_rate": 8.973364365140936e-05,
      "loss": 0.7687,
      "step": 1795
    },
    {
      "epoch": 0.41222947440742014,
      "grad_norm": 0.36121290922164917,
      "learning_rate": 8.966899405223688e-05,
      "loss": 0.7728,
      "step": 1800
    },
    {
      "epoch": 0.4133745562807741,
      "grad_norm": 0.4618297219276428,
      "learning_rate": 8.96043444530644e-05,
      "loss": 0.7719,
      "step": 1805
    },
    {
      "epoch": 0.414519638154128,
      "grad_norm": 0.3385610282421112,
      "learning_rate": 8.953969485389191e-05,
      "loss": 0.7791,
      "step": 1810
    },
    {
      "epoch": 0.41566472002748195,
      "grad_norm": 0.45700961351394653,
      "learning_rate": 8.947504525471943e-05,
      "loss": 0.7751,
      "step": 1815
    },
    {
      "epoch": 0.4168098019008359,
      "grad_norm": 0.2775494158267975,
      "learning_rate": 8.941039565554694e-05,
      "loss": 0.7705,
      "step": 1820
    },
    {
      "epoch": 0.4179548837741899,
      "grad_norm": 0.32360318303108215,
      "learning_rate": 8.934574605637445e-05,
      "loss": 0.7743,
      "step": 1825
    },
    {
      "epoch": 0.4190999656475438,
      "grad_norm": 0.2116226702928543,
      "learning_rate": 8.928109645720197e-05,
      "loss": 0.7702,
      "step": 1830
    },
    {
      "epoch": 0.42024504752089775,
      "grad_norm": 0.1969475895166397,
      "learning_rate": 8.92164468580295e-05,
      "loss": 0.7767,
      "step": 1835
    },
    {
      "epoch": 0.4213901293942517,
      "grad_norm": 0.2624218165874481,
      "learning_rate": 8.9151797258857e-05,
      "loss": 0.7718,
      "step": 1840
    },
    {
      "epoch": 0.4225352112676056,
      "grad_norm": 0.29143357276916504,
      "learning_rate": 8.908714765968451e-05,
      "loss": 0.7695,
      "step": 1845
    },
    {
      "epoch": 0.42368029314095956,
      "grad_norm": 0.3152911067008972,
      "learning_rate": 8.902249806051203e-05,
      "loss": 0.774,
      "step": 1850
    },
    {
      "epoch": 0.4248253750143135,
      "grad_norm": 0.4107597768306732,
      "learning_rate": 8.895784846133954e-05,
      "loss": 0.7754,
      "step": 1855
    },
    {
      "epoch": 0.4259704568876675,
      "grad_norm": 0.39836370944976807,
      "learning_rate": 8.889319886216705e-05,
      "loss": 0.7706,
      "step": 1860
    },
    {
      "epoch": 0.42711553876102143,
      "grad_norm": 0.44928601384162903,
      "learning_rate": 8.882854926299458e-05,
      "loss": 0.7735,
      "step": 1865
    },
    {
      "epoch": 0.42826062063437537,
      "grad_norm": 0.2496773898601532,
      "learning_rate": 8.87638996638221e-05,
      "loss": 0.7725,
      "step": 1870
    },
    {
      "epoch": 0.4294057025077293,
      "grad_norm": 0.4287888705730438,
      "learning_rate": 8.869925006464961e-05,
      "loss": 0.7733,
      "step": 1875
    },
    {
      "epoch": 0.43055078438108324,
      "grad_norm": 0.18999287486076355,
      "learning_rate": 8.863460046547711e-05,
      "loss": 0.7728,
      "step": 1880
    },
    {
      "epoch": 0.4316958662544372,
      "grad_norm": 0.4098029136657715,
      "learning_rate": 8.856995086630462e-05,
      "loss": 0.7731,
      "step": 1885
    },
    {
      "epoch": 0.4328409481277911,
      "grad_norm": 0.3076467216014862,
      "learning_rate": 8.850530126713215e-05,
      "loss": 0.7762,
      "step": 1890
    },
    {
      "epoch": 0.4339860300011451,
      "grad_norm": 0.3346705436706543,
      "learning_rate": 8.844065166795967e-05,
      "loss": 0.7738,
      "step": 1895
    },
    {
      "epoch": 0.43513111187449904,
      "grad_norm": 0.4677494764328003,
      "learning_rate": 8.837600206878718e-05,
      "loss": 0.7704,
      "step": 1900
    },
    {
      "epoch": 0.436276193747853,
      "grad_norm": 0.3557366132736206,
      "learning_rate": 8.83113524696147e-05,
      "loss": 0.7727,
      "step": 1905
    },
    {
      "epoch": 0.4374212756212069,
      "grad_norm": 0.29481977224349976,
      "learning_rate": 8.824670287044221e-05,
      "loss": 0.7718,
      "step": 1910
    },
    {
      "epoch": 0.43856635749456085,
      "grad_norm": 0.48542648553848267,
      "learning_rate": 8.818205327126972e-05,
      "loss": 0.7716,
      "step": 1915
    },
    {
      "epoch": 0.4397114393679148,
      "grad_norm": 0.29684439301490784,
      "learning_rate": 8.811740367209724e-05,
      "loss": 0.771,
      "step": 1920
    },
    {
      "epoch": 0.44085652124126873,
      "grad_norm": 0.2121712863445282,
      "learning_rate": 8.805275407292475e-05,
      "loss": 0.7709,
      "step": 1925
    },
    {
      "epoch": 0.4420016031146227,
      "grad_norm": 0.221273273229599,
      "learning_rate": 8.798810447375227e-05,
      "loss": 0.7705,
      "step": 1930
    },
    {
      "epoch": 0.44314668498797666,
      "grad_norm": 0.21169264614582062,
      "learning_rate": 8.792345487457978e-05,
      "loss": 0.7696,
      "step": 1935
    },
    {
      "epoch": 0.4442917668613306,
      "grad_norm": 0.33791452646255493,
      "learning_rate": 8.78588052754073e-05,
      "loss": 0.773,
      "step": 1940
    },
    {
      "epoch": 0.44543684873468453,
      "grad_norm": 0.2366342693567276,
      "learning_rate": 8.779415567623481e-05,
      "loss": 0.77,
      "step": 1945
    },
    {
      "epoch": 0.44658193060803847,
      "grad_norm": 0.3381491005420685,
      "learning_rate": 8.772950607706232e-05,
      "loss": 0.7736,
      "step": 1950
    },
    {
      "epoch": 0.4477270124813924,
      "grad_norm": 0.2792139947414398,
      "learning_rate": 8.766485647788985e-05,
      "loss": 0.7698,
      "step": 1955
    },
    {
      "epoch": 0.44887209435474634,
      "grad_norm": 0.2960784435272217,
      "learning_rate": 8.760020687871736e-05,
      "loss": 0.7697,
      "step": 1960
    },
    {
      "epoch": 0.45001717622810034,
      "grad_norm": 0.3192628026008606,
      "learning_rate": 8.753555727954487e-05,
      "loss": 0.7711,
      "step": 1965
    },
    {
      "epoch": 0.45116225810145427,
      "grad_norm": 0.33469468355178833,
      "learning_rate": 8.747090768037238e-05,
      "loss": 0.7679,
      "step": 1970
    },
    {
      "epoch": 0.4523073399748082,
      "grad_norm": 0.8246955871582031,
      "learning_rate": 8.74062580811999e-05,
      "loss": 0.7693,
      "step": 1975
    },
    {
      "epoch": 0.45345242184816215,
      "grad_norm": 0.4532000720500946,
      "learning_rate": 8.734160848202741e-05,
      "loss": 0.7736,
      "step": 1980
    },
    {
      "epoch": 0.4545975037215161,
      "grad_norm": 0.26574060320854187,
      "learning_rate": 8.727695888285494e-05,
      "loss": 0.7727,
      "step": 1985
    },
    {
      "epoch": 0.45574258559487,
      "grad_norm": 0.33447420597076416,
      "learning_rate": 8.721230928368245e-05,
      "loss": 0.772,
      "step": 1990
    },
    {
      "epoch": 0.45688766746822396,
      "grad_norm": 0.2761179804801941,
      "learning_rate": 8.714765968450996e-05,
      "loss": 0.773,
      "step": 1995
    },
    {
      "epoch": 0.45803274934157795,
      "grad_norm": 0.24412183463573456,
      "learning_rate": 8.708301008533748e-05,
      "loss": 0.7695,
      "step": 2000
    },
    {
      "epoch": 0.4591778312149319,
      "grad_norm": 0.20122234523296356,
      "learning_rate": 8.701836048616498e-05,
      "loss": 0.7721,
      "step": 2005
    },
    {
      "epoch": 0.4603229130882858,
      "grad_norm": 0.3519915044307709,
      "learning_rate": 8.695371088699251e-05,
      "loss": 0.7676,
      "step": 2010
    },
    {
      "epoch": 0.46146799496163976,
      "grad_norm": 0.34082722663879395,
      "learning_rate": 8.688906128782002e-05,
      "loss": 0.7682,
      "step": 2015
    },
    {
      "epoch": 0.4626130768349937,
      "grad_norm": 0.4101792573928833,
      "learning_rate": 8.682441168864754e-05,
      "loss": 0.772,
      "step": 2020
    },
    {
      "epoch": 0.46375815870834763,
      "grad_norm": 0.3329888880252838,
      "learning_rate": 8.675976208947505e-05,
      "loss": 0.77,
      "step": 2025
    },
    {
      "epoch": 0.46490324058170157,
      "grad_norm": 0.4946768283843994,
      "learning_rate": 8.669511249030256e-05,
      "loss": 0.7693,
      "step": 2030
    },
    {
      "epoch": 0.46604832245505556,
      "grad_norm": 0.26925185322761536,
      "learning_rate": 8.663046289113008e-05,
      "loss": 0.7708,
      "step": 2035
    },
    {
      "epoch": 0.4671934043284095,
      "grad_norm": 0.23793742060661316,
      "learning_rate": 8.656581329195759e-05,
      "loss": 0.7718,
      "step": 2040
    },
    {
      "epoch": 0.46833848620176344,
      "grad_norm": 0.1948954164981842,
      "learning_rate": 8.650116369278512e-05,
      "loss": 0.7702,
      "step": 2045
    },
    {
      "epoch": 0.4694835680751174,
      "grad_norm": 0.32897040247917175,
      "learning_rate": 8.643651409361262e-05,
      "loss": 0.7686,
      "step": 2050
    },
    {
      "epoch": 0.4706286499484713,
      "grad_norm": 0.2682906687259674,
      "learning_rate": 8.637186449444014e-05,
      "loss": 0.7716,
      "step": 2055
    },
    {
      "epoch": 0.47177373182182525,
      "grad_norm": 0.3296032249927521,
      "learning_rate": 8.630721489526765e-05,
      "loss": 0.7699,
      "step": 2060
    },
    {
      "epoch": 0.4729188136951792,
      "grad_norm": 0.32693296670913696,
      "learning_rate": 8.624256529609516e-05,
      "loss": 0.771,
      "step": 2065
    },
    {
      "epoch": 0.4740638955685331,
      "grad_norm": 0.36149802803993225,
      "learning_rate": 8.617791569692268e-05,
      "loss": 0.7709,
      "step": 2070
    },
    {
      "epoch": 0.4752089774418871,
      "grad_norm": 0.28551197052001953,
      "learning_rate": 8.61132660977502e-05,
      "loss": 0.7676,
      "step": 2075
    },
    {
      "epoch": 0.47635405931524105,
      "grad_norm": 0.19539989531040192,
      "learning_rate": 8.604861649857772e-05,
      "loss": 0.7724,
      "step": 2080
    },
    {
      "epoch": 0.477499141188595,
      "grad_norm": 0.28068146109580994,
      "learning_rate": 8.598396689940523e-05,
      "loss": 0.7717,
      "step": 2085
    },
    {
      "epoch": 0.4786442230619489,
      "grad_norm": 0.40226539969444275,
      "learning_rate": 8.591931730023273e-05,
      "loss": 0.7749,
      "step": 2090
    },
    {
      "epoch": 0.47978930493530286,
      "grad_norm": 0.3484259843826294,
      "learning_rate": 8.585466770106025e-05,
      "loss": 0.7701,
      "step": 2095
    },
    {
      "epoch": 0.4809343868086568,
      "grad_norm": 0.3907545208930969,
      "learning_rate": 8.579001810188778e-05,
      "loss": 0.7713,
      "step": 2100
    },
    {
      "epoch": 0.48207946868201074,
      "grad_norm": 0.25463706254959106,
      "learning_rate": 8.572536850271529e-05,
      "loss": 0.7706,
      "step": 2105
    },
    {
      "epoch": 0.48322455055536473,
      "grad_norm": 0.3591324985027313,
      "learning_rate": 8.56607189035428e-05,
      "loss": 0.7699,
      "step": 2110
    },
    {
      "epoch": 0.48436963242871867,
      "grad_norm": 0.38342562317848206,
      "learning_rate": 8.559606930437032e-05,
      "loss": 0.7679,
      "step": 2115
    },
    {
      "epoch": 0.4855147143020726,
      "grad_norm": 0.2707025408744812,
      "learning_rate": 8.553141970519783e-05,
      "loss": 0.7663,
      "step": 2120
    },
    {
      "epoch": 0.48665979617542654,
      "grad_norm": 0.27740100026130676,
      "learning_rate": 8.546677010602535e-05,
      "loss": 0.7702,
      "step": 2125
    },
    {
      "epoch": 0.4878048780487805,
      "grad_norm": 0.38322174549102783,
      "learning_rate": 8.540212050685286e-05,
      "loss": 0.771,
      "step": 2130
    },
    {
      "epoch": 0.4889499599221344,
      "grad_norm": 0.25964078307151794,
      "learning_rate": 8.533747090768038e-05,
      "loss": 0.77,
      "step": 2135
    },
    {
      "epoch": 0.49009504179548835,
      "grad_norm": 0.3346869647502899,
      "learning_rate": 8.527282130850789e-05,
      "loss": 0.7695,
      "step": 2140
    },
    {
      "epoch": 0.49124012366884234,
      "grad_norm": 0.2086460143327713,
      "learning_rate": 8.52081717093354e-05,
      "loss": 0.7703,
      "step": 2145
    },
    {
      "epoch": 0.4923852055421963,
      "grad_norm": 0.18562249839305878,
      "learning_rate": 8.514352211016292e-05,
      "loss": 0.7691,
      "step": 2150
    },
    {
      "epoch": 0.4935302874155502,
      "grad_norm": 0.27760863304138184,
      "learning_rate": 8.507887251099043e-05,
      "loss": 0.7736,
      "step": 2155
    },
    {
      "epoch": 0.49467536928890415,
      "grad_norm": 0.33482494950294495,
      "learning_rate": 8.501422291181795e-05,
      "loss": 0.7681,
      "step": 2160
    },
    {
      "epoch": 0.4958204511622581,
      "grad_norm": 0.2479313164949417,
      "learning_rate": 8.494957331264548e-05,
      "loss": 0.7714,
      "step": 2165
    },
    {
      "epoch": 0.49696553303561203,
      "grad_norm": 0.31260478496551514,
      "learning_rate": 8.488492371347299e-05,
      "loss": 0.7683,
      "step": 2170
    },
    {
      "epoch": 0.49811061490896597,
      "grad_norm": 0.25391116738319397,
      "learning_rate": 8.482027411430049e-05,
      "loss": 0.7666,
      "step": 2175
    },
    {
      "epoch": 0.49925569678231996,
      "grad_norm": 0.27383050322532654,
      "learning_rate": 8.4755624515128e-05,
      "loss": 0.7678,
      "step": 2180
    },
    {
      "epoch": 0.5004007786556739,
      "grad_norm": 0.4296813905239105,
      "learning_rate": 8.469097491595552e-05,
      "loss": 0.7699,
      "step": 2185
    },
    {
      "epoch": 0.5015458605290278,
      "grad_norm": 0.26294124126434326,
      "learning_rate": 8.462632531678305e-05,
      "loss": 0.7698,
      "step": 2190
    },
    {
      "epoch": 0.5026909424023818,
      "grad_norm": 0.34626004099845886,
      "learning_rate": 8.456167571761056e-05,
      "loss": 0.7709,
      "step": 2195
    },
    {
      "epoch": 0.5038360242757357,
      "grad_norm": 0.44097933173179626,
      "learning_rate": 8.449702611843807e-05,
      "loss": 0.7644,
      "step": 2200
    },
    {
      "epoch": 0.5049811061490896,
      "grad_norm": 0.2886388897895813,
      "learning_rate": 8.443237651926559e-05,
      "loss": 0.7667,
      "step": 2205
    },
    {
      "epoch": 0.5061261880224436,
      "grad_norm": 0.20610256493091583,
      "learning_rate": 8.43677269200931e-05,
      "loss": 0.7723,
      "step": 2210
    },
    {
      "epoch": 0.5072712698957975,
      "grad_norm": 0.17685441672801971,
      "learning_rate": 8.43030773209206e-05,
      "loss": 0.769,
      "step": 2215
    },
    {
      "epoch": 0.5084163517691515,
      "grad_norm": 0.2608635723590851,
      "learning_rate": 8.423842772174813e-05,
      "loss": 0.7676,
      "step": 2220
    },
    {
      "epoch": 0.5095614336425054,
      "grad_norm": 0.19678382575511932,
      "learning_rate": 8.417377812257565e-05,
      "loss": 0.7656,
      "step": 2225
    },
    {
      "epoch": 0.5107065155158594,
      "grad_norm": 0.2524316906929016,
      "learning_rate": 8.410912852340316e-05,
      "loss": 0.7693,
      "step": 2230
    },
    {
      "epoch": 0.5118515973892134,
      "grad_norm": 0.1823377162218094,
      "learning_rate": 8.404447892423067e-05,
      "loss": 0.7675,
      "step": 2235
    },
    {
      "epoch": 0.5129966792625673,
      "grad_norm": 0.3295440375804901,
      "learning_rate": 8.397982932505819e-05,
      "loss": 0.7702,
      "step": 2240
    },
    {
      "epoch": 0.5141417611359212,
      "grad_norm": 0.2945457398891449,
      "learning_rate": 8.39151797258857e-05,
      "loss": 0.7735,
      "step": 2245
    },
    {
      "epoch": 0.5152868430092752,
      "grad_norm": 0.2396729290485382,
      "learning_rate": 8.385053012671322e-05,
      "loss": 0.7695,
      "step": 2250
    },
    {
      "epoch": 0.5164319248826291,
      "grad_norm": 0.4043848514556885,
      "learning_rate": 8.378588052754074e-05,
      "loss": 0.7689,
      "step": 2255
    },
    {
      "epoch": 0.5175770067559831,
      "grad_norm": 0.4132694602012634,
      "learning_rate": 8.372123092836825e-05,
      "loss": 0.7709,
      "step": 2260
    },
    {
      "epoch": 0.518722088629337,
      "grad_norm": 0.3979206681251526,
      "learning_rate": 8.365658132919576e-05,
      "loss": 0.7683,
      "step": 2265
    },
    {
      "epoch": 0.5198671705026909,
      "grad_norm": 0.23621103167533875,
      "learning_rate": 8.359193173002327e-05,
      "loss": 0.7675,
      "step": 2270
    },
    {
      "epoch": 0.5210122523760449,
      "grad_norm": 0.22756105661392212,
      "learning_rate": 8.352728213085079e-05,
      "loss": 0.7729,
      "step": 2275
    },
    {
      "epoch": 0.5221573342493988,
      "grad_norm": 0.2720366418361664,
      "learning_rate": 8.34626325316783e-05,
      "loss": 0.7705,
      "step": 2280
    },
    {
      "epoch": 0.5233024161227527,
      "grad_norm": 0.34250500798225403,
      "learning_rate": 8.339798293250583e-05,
      "loss": 0.7696,
      "step": 2285
    },
    {
      "epoch": 0.5244474979961067,
      "grad_norm": 0.38787877559661865,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.7664,
      "step": 2290
    },
    {
      "epoch": 0.5255925798694606,
      "grad_norm": 0.23861998319625854,
      "learning_rate": 8.326868373416086e-05,
      "loss": 0.7715,
      "step": 2295
    },
    {
      "epoch": 0.5267376617428146,
      "grad_norm": 0.3382185995578766,
      "learning_rate": 8.320403413498836e-05,
      "loss": 0.7689,
      "step": 2300
    },
    {
      "epoch": 0.5278827436161686,
      "grad_norm": 0.29420629143714905,
      "learning_rate": 8.313938453581587e-05,
      "loss": 0.7649,
      "step": 2305
    },
    {
      "epoch": 0.5290278254895225,
      "grad_norm": 0.26455843448638916,
      "learning_rate": 8.30747349366434e-05,
      "loss": 0.7704,
      "step": 2310
    },
    {
      "epoch": 0.5301729073628765,
      "grad_norm": 0.3979088366031647,
      "learning_rate": 8.301008533747092e-05,
      "loss": 0.7679,
      "step": 2315
    },
    {
      "epoch": 0.5313179892362304,
      "grad_norm": 0.24187666177749634,
      "learning_rate": 8.294543573829843e-05,
      "loss": 0.7648,
      "step": 2320
    },
    {
      "epoch": 0.5324630711095844,
      "grad_norm": 0.21819165349006653,
      "learning_rate": 8.288078613912594e-05,
      "loss": 0.7666,
      "step": 2325
    },
    {
      "epoch": 0.5336081529829383,
      "grad_norm": 0.44462597370147705,
      "learning_rate": 8.281613653995346e-05,
      "loss": 0.7653,
      "step": 2330
    },
    {
      "epoch": 0.5347532348562922,
      "grad_norm": 0.2796482443809509,
      "learning_rate": 8.275148694078097e-05,
      "loss": 0.7733,
      "step": 2335
    },
    {
      "epoch": 0.5358983167296462,
      "grad_norm": 0.28863418102264404,
      "learning_rate": 8.268683734160849e-05,
      "loss": 0.7672,
      "step": 2340
    },
    {
      "epoch": 0.5370433986030001,
      "grad_norm": 0.38829681277275085,
      "learning_rate": 8.2622187742436e-05,
      "loss": 0.7717,
      "step": 2345
    },
    {
      "epoch": 0.538188480476354,
      "grad_norm": 0.26142099499702454,
      "learning_rate": 8.255753814326351e-05,
      "loss": 0.7703,
      "step": 2350
    },
    {
      "epoch": 0.539333562349708,
      "grad_norm": 0.32155346870422363,
      "learning_rate": 8.249288854409103e-05,
      "loss": 0.77,
      "step": 2355
    },
    {
      "epoch": 0.5404786442230619,
      "grad_norm": 0.25289323925971985,
      "learning_rate": 8.242823894491854e-05,
      "loss": 0.7712,
      "step": 2360
    },
    {
      "epoch": 0.5416237260964158,
      "grad_norm": 0.4167221188545227,
      "learning_rate": 8.236358934574606e-05,
      "loss": 0.7683,
      "step": 2365
    },
    {
      "epoch": 0.5427688079697698,
      "grad_norm": 0.3492532968521118,
      "learning_rate": 8.229893974657357e-05,
      "loss": 0.7678,
      "step": 2370
    },
    {
      "epoch": 0.5439138898431238,
      "grad_norm": 0.27711644768714905,
      "learning_rate": 8.22342901474011e-05,
      "loss": 0.7659,
      "step": 2375
    },
    {
      "epoch": 0.5450589717164778,
      "grad_norm": 0.2010398656129837,
      "learning_rate": 8.216964054822861e-05,
      "loss": 0.768,
      "step": 2380
    },
    {
      "epoch": 0.5462040535898317,
      "grad_norm": 0.33034494519233704,
      "learning_rate": 8.210499094905611e-05,
      "loss": 0.7723,
      "step": 2385
    },
    {
      "epoch": 0.5473491354631856,
      "grad_norm": 0.251040518283844,
      "learning_rate": 8.204034134988363e-05,
      "loss": 0.7636,
      "step": 2390
    },
    {
      "epoch": 0.5484942173365396,
      "grad_norm": 0.20830363035202026,
      "learning_rate": 8.197569175071114e-05,
      "loss": 0.7698,
      "step": 2395
    },
    {
      "epoch": 0.5496392992098935,
      "grad_norm": 0.21820737421512604,
      "learning_rate": 8.191104215153867e-05,
      "loss": 0.7663,
      "step": 2400
    },
    {
      "epoch": 0.5507843810832475,
      "grad_norm": 0.24281048774719238,
      "learning_rate": 8.184639255236618e-05,
      "loss": 0.7667,
      "step": 2405
    },
    {
      "epoch": 0.5519294629566014,
      "grad_norm": 0.2047927975654602,
      "learning_rate": 8.17817429531937e-05,
      "loss": 0.7688,
      "step": 2410
    },
    {
      "epoch": 0.5530745448299553,
      "grad_norm": 0.3286665081977844,
      "learning_rate": 8.171709335402121e-05,
      "loss": 0.7657,
      "step": 2415
    },
    {
      "epoch": 0.5542196267033093,
      "grad_norm": 0.234399676322937,
      "learning_rate": 8.165244375484873e-05,
      "loss": 0.7672,
      "step": 2420
    },
    {
      "epoch": 0.5553647085766632,
      "grad_norm": 0.27283400297164917,
      "learning_rate": 8.158779415567623e-05,
      "loss": 0.767,
      "step": 2425
    },
    {
      "epoch": 0.5565097904500171,
      "grad_norm": 0.2366446703672409,
      "learning_rate": 8.152314455650376e-05,
      "loss": 0.7654,
      "step": 2430
    },
    {
      "epoch": 0.5576548723233711,
      "grad_norm": 0.37923479080200195,
      "learning_rate": 8.145849495733127e-05,
      "loss": 0.7687,
      "step": 2435
    },
    {
      "epoch": 0.558799954196725,
      "grad_norm": 0.2888481020927429,
      "learning_rate": 8.139384535815878e-05,
      "loss": 0.7667,
      "step": 2440
    },
    {
      "epoch": 0.5599450360700791,
      "grad_norm": 0.2126360982656479,
      "learning_rate": 8.13291957589863e-05,
      "loss": 0.7707,
      "step": 2445
    },
    {
      "epoch": 0.561090117943433,
      "grad_norm": 0.13646523654460907,
      "learning_rate": 8.126454615981381e-05,
      "loss": 0.7697,
      "step": 2450
    },
    {
      "epoch": 0.5622351998167869,
      "grad_norm": 0.18953712284564972,
      "learning_rate": 8.119989656064133e-05,
      "loss": 0.7601,
      "step": 2455
    },
    {
      "epoch": 0.5633802816901409,
      "grad_norm": 0.661413311958313,
      "learning_rate": 8.113524696146884e-05,
      "loss": 0.7687,
      "step": 2460
    },
    {
      "epoch": 0.5645253635634948,
      "grad_norm": 0.39190343022346497,
      "learning_rate": 8.107059736229636e-05,
      "loss": 0.7708,
      "step": 2465
    },
    {
      "epoch": 0.5656704454368487,
      "grad_norm": 0.1970270425081253,
      "learning_rate": 8.100594776312387e-05,
      "loss": 0.7708,
      "step": 2470
    },
    {
      "epoch": 0.5668155273102027,
      "grad_norm": 0.3270746171474457,
      "learning_rate": 8.094129816395138e-05,
      "loss": 0.7685,
      "step": 2475
    },
    {
      "epoch": 0.5679606091835566,
      "grad_norm": 0.19802868366241455,
      "learning_rate": 8.08766485647789e-05,
      "loss": 0.7671,
      "step": 2480
    },
    {
      "epoch": 0.5691056910569106,
      "grad_norm": 0.2073948085308075,
      "learning_rate": 8.081199896560641e-05,
      "loss": 0.7704,
      "step": 2485
    },
    {
      "epoch": 0.5702507729302645,
      "grad_norm": 0.292240172624588,
      "learning_rate": 8.074734936643393e-05,
      "loss": 0.768,
      "step": 2490
    },
    {
      "epoch": 0.5713958548036184,
      "grad_norm": 0.2657301723957062,
      "learning_rate": 8.068269976726145e-05,
      "loss": 0.7699,
      "step": 2495
    },
    {
      "epoch": 0.5725409366769724,
      "grad_norm": 0.30449584126472473,
      "learning_rate": 8.061805016808897e-05,
      "loss": 0.7678,
      "step": 2500
    },
    {
      "epoch": 0.5736860185503263,
      "grad_norm": 0.3515103757381439,
      "learning_rate": 8.055340056891648e-05,
      "loss": 0.7662,
      "step": 2505
    },
    {
      "epoch": 0.5748311004236802,
      "grad_norm": 0.4272383153438568,
      "learning_rate": 8.048875096974398e-05,
      "loss": 0.7664,
      "step": 2510
    },
    {
      "epoch": 0.5759761822970343,
      "grad_norm": 0.2849995791912079,
      "learning_rate": 8.04241013705715e-05,
      "loss": 0.7648,
      "step": 2515
    },
    {
      "epoch": 0.5771212641703882,
      "grad_norm": 0.26403772830963135,
      "learning_rate": 8.035945177139903e-05,
      "loss": 0.7634,
      "step": 2520
    },
    {
      "epoch": 0.5782663460437422,
      "grad_norm": 0.28480133414268494,
      "learning_rate": 8.029480217222654e-05,
      "loss": 0.7665,
      "step": 2525
    },
    {
      "epoch": 0.5794114279170961,
      "grad_norm": 0.3533156216144562,
      "learning_rate": 8.023015257305405e-05,
      "loss": 0.7692,
      "step": 2530
    },
    {
      "epoch": 0.58055650979045,
      "grad_norm": 0.35114097595214844,
      "learning_rate": 8.016550297388157e-05,
      "loss": 0.7657,
      "step": 2535
    },
    {
      "epoch": 0.581701591663804,
      "grad_norm": 0.3216884434223175,
      "learning_rate": 8.010085337470908e-05,
      "loss": 0.7667,
      "step": 2540
    },
    {
      "epoch": 0.5828466735371579,
      "grad_norm": 0.2756839990615845,
      "learning_rate": 8.00362037755366e-05,
      "loss": 0.7683,
      "step": 2545
    },
    {
      "epoch": 0.5839917554105118,
      "grad_norm": 0.24576738476753235,
      "learning_rate": 7.997155417636411e-05,
      "loss": 0.7654,
      "step": 2550
    },
    {
      "epoch": 0.5851368372838658,
      "grad_norm": 0.25817248225212097,
      "learning_rate": 7.990690457719162e-05,
      "loss": 0.7616,
      "step": 2555
    },
    {
      "epoch": 0.5862819191572197,
      "grad_norm": 0.3255434036254883,
      "learning_rate": 7.984225497801914e-05,
      "loss": 0.7684,
      "step": 2560
    },
    {
      "epoch": 0.5874270010305737,
      "grad_norm": 0.3148951232433319,
      "learning_rate": 7.977760537884665e-05,
      "loss": 0.7661,
      "step": 2565
    },
    {
      "epoch": 0.5885720829039276,
      "grad_norm": 0.3079426884651184,
      "learning_rate": 7.971295577967417e-05,
      "loss": 0.7623,
      "step": 2570
    },
    {
      "epoch": 0.5897171647772815,
      "grad_norm": 0.39346790313720703,
      "learning_rate": 7.964830618050168e-05,
      "loss": 0.7656,
      "step": 2575
    },
    {
      "epoch": 0.5908622466506355,
      "grad_norm": 0.22193998098373413,
      "learning_rate": 7.95836565813292e-05,
      "loss": 0.7695,
      "step": 2580
    },
    {
      "epoch": 0.5920073285239895,
      "grad_norm": 0.25602176785469055,
      "learning_rate": 7.951900698215672e-05,
      "loss": 0.7639,
      "step": 2585
    },
    {
      "epoch": 0.5931524103973435,
      "grad_norm": 0.3182067275047302,
      "learning_rate": 7.945435738298424e-05,
      "loss": 0.7667,
      "step": 2590
    },
    {
      "epoch": 0.5942974922706974,
      "grad_norm": 0.2605539858341217,
      "learning_rate": 7.938970778381174e-05,
      "loss": 0.7668,
      "step": 2595
    },
    {
      "epoch": 0.5954425741440513,
      "grad_norm": 0.39896488189697266,
      "learning_rate": 7.932505818463925e-05,
      "loss": 0.767,
      "step": 2600
    },
    {
      "epoch": 0.5965876560174053,
      "grad_norm": 0.32872942090034485,
      "learning_rate": 7.926040858546677e-05,
      "loss": 0.7638,
      "step": 2605
    },
    {
      "epoch": 0.5977327378907592,
      "grad_norm": 0.3192463219165802,
      "learning_rate": 7.91957589862943e-05,
      "loss": 0.7708,
      "step": 2610
    },
    {
      "epoch": 0.5988778197641131,
      "grad_norm": 0.4065992534160614,
      "learning_rate": 7.913110938712181e-05,
      "loss": 0.7669,
      "step": 2615
    },
    {
      "epoch": 0.6000229016374671,
      "grad_norm": 0.24374541640281677,
      "learning_rate": 7.906645978794932e-05,
      "loss": 0.7668,
      "step": 2620
    },
    {
      "epoch": 0.601167983510821,
      "grad_norm": 0.1961938738822937,
      "learning_rate": 7.900181018877684e-05,
      "loss": 0.7628,
      "step": 2625
    },
    {
      "epoch": 0.602313065384175,
      "grad_norm": 0.32636597752571106,
      "learning_rate": 7.893716058960435e-05,
      "loss": 0.7671,
      "step": 2630
    },
    {
      "epoch": 0.6034581472575289,
      "grad_norm": 0.5156142711639404,
      "learning_rate": 7.887251099043185e-05,
      "loss": 0.7643,
      "step": 2635
    },
    {
      "epoch": 0.6046032291308828,
      "grad_norm": 0.4189930856227875,
      "learning_rate": 7.880786139125938e-05,
      "loss": 0.7699,
      "step": 2640
    },
    {
      "epoch": 0.6057483110042368,
      "grad_norm": 0.26843327283859253,
      "learning_rate": 7.87432117920869e-05,
      "loss": 0.7655,
      "step": 2645
    },
    {
      "epoch": 0.6068933928775907,
      "grad_norm": 0.2084437906742096,
      "learning_rate": 7.867856219291441e-05,
      "loss": 0.7642,
      "step": 2650
    },
    {
      "epoch": 0.6080384747509447,
      "grad_norm": 0.23720037937164307,
      "learning_rate": 7.861391259374192e-05,
      "loss": 0.7652,
      "step": 2655
    },
    {
      "epoch": 0.6091835566242987,
      "grad_norm": 0.2192964255809784,
      "learning_rate": 7.854926299456944e-05,
      "loss": 0.7658,
      "step": 2660
    },
    {
      "epoch": 0.6103286384976526,
      "grad_norm": 0.23451314866542816,
      "learning_rate": 7.848461339539695e-05,
      "loss": 0.7622,
      "step": 2665
    },
    {
      "epoch": 0.6114737203710066,
      "grad_norm": 0.2981168031692505,
      "learning_rate": 7.841996379622447e-05,
      "loss": 0.7648,
      "step": 2670
    },
    {
      "epoch": 0.6126188022443605,
      "grad_norm": 0.2158365696668625,
      "learning_rate": 7.835531419705198e-05,
      "loss": 0.7635,
      "step": 2675
    },
    {
      "epoch": 0.6137638841177144,
      "grad_norm": 0.20274029672145844,
      "learning_rate": 7.82906645978795e-05,
      "loss": 0.7642,
      "step": 2680
    },
    {
      "epoch": 0.6149089659910684,
      "grad_norm": 0.35747575759887695,
      "learning_rate": 7.822601499870701e-05,
      "loss": 0.7652,
      "step": 2685
    },
    {
      "epoch": 0.6160540478644223,
      "grad_norm": 0.23654219508171082,
      "learning_rate": 7.816136539953452e-05,
      "loss": 0.7636,
      "step": 2690
    },
    {
      "epoch": 0.6171991297377762,
      "grad_norm": 0.22896522283554077,
      "learning_rate": 7.809671580036204e-05,
      "loss": 0.7629,
      "step": 2695
    },
    {
      "epoch": 0.6183442116111302,
      "grad_norm": 0.2906895577907562,
      "learning_rate": 7.803206620118956e-05,
      "loss": 0.762,
      "step": 2700
    },
    {
      "epoch": 0.6194892934844841,
      "grad_norm": 0.1701405942440033,
      "learning_rate": 7.796741660201708e-05,
      "loss": 0.7646,
      "step": 2705
    },
    {
      "epoch": 0.620634375357838,
      "grad_norm": 0.44711700081825256,
      "learning_rate": 7.790276700284459e-05,
      "loss": 0.7684,
      "step": 2710
    },
    {
      "epoch": 0.621779457231192,
      "grad_norm": 0.2111397385597229,
      "learning_rate": 7.783811740367211e-05,
      "loss": 0.7655,
      "step": 2715
    },
    {
      "epoch": 0.6229245391045459,
      "grad_norm": 0.23328475654125214,
      "learning_rate": 7.777346780449961e-05,
      "loss": 0.7666,
      "step": 2720
    },
    {
      "epoch": 0.6240696209779,
      "grad_norm": 0.2624497711658478,
      "learning_rate": 7.770881820532712e-05,
      "loss": 0.7604,
      "step": 2725
    },
    {
      "epoch": 0.6252147028512539,
      "grad_norm": 0.3210087716579437,
      "learning_rate": 7.764416860615465e-05,
      "loss": 0.7615,
      "step": 2730
    },
    {
      "epoch": 0.6263597847246078,
      "grad_norm": 0.21020106971263885,
      "learning_rate": 7.757951900698216e-05,
      "loss": 0.7629,
      "step": 2735
    },
    {
      "epoch": 0.6275048665979618,
      "grad_norm": 0.23509849607944489,
      "learning_rate": 7.751486940780968e-05,
      "loss": 0.7668,
      "step": 2740
    },
    {
      "epoch": 0.6286499484713157,
      "grad_norm": 0.42521828413009644,
      "learning_rate": 7.745021980863719e-05,
      "loss": 0.7662,
      "step": 2745
    },
    {
      "epoch": 0.6297950303446697,
      "grad_norm": 0.2406284660100937,
      "learning_rate": 7.73855702094647e-05,
      "loss": 0.7631,
      "step": 2750
    },
    {
      "epoch": 0.6309401122180236,
      "grad_norm": 0.25343960523605347,
      "learning_rate": 7.732092061029222e-05,
      "loss": 0.7645,
      "step": 2755
    },
    {
      "epoch": 0.6320851940913775,
      "grad_norm": 0.2309281826019287,
      "learning_rate": 7.725627101111973e-05,
      "loss": 0.7609,
      "step": 2760
    },
    {
      "epoch": 0.6332302759647315,
      "grad_norm": 0.34525343775749207,
      "learning_rate": 7.719162141194725e-05,
      "loss": 0.7657,
      "step": 2765
    },
    {
      "epoch": 0.6343753578380854,
      "grad_norm": 0.42397892475128174,
      "learning_rate": 7.712697181277476e-05,
      "loss": 0.7618,
      "step": 2770
    },
    {
      "epoch": 0.6355204397114393,
      "grad_norm": 0.3545205295085907,
      "learning_rate": 7.706232221360228e-05,
      "loss": 0.7632,
      "step": 2775
    },
    {
      "epoch": 0.6366655215847933,
      "grad_norm": 0.2760995626449585,
      "learning_rate": 7.699767261442979e-05,
      "loss": 0.762,
      "step": 2780
    },
    {
      "epoch": 0.6378106034581472,
      "grad_norm": 0.252374529838562,
      "learning_rate": 7.69330230152573e-05,
      "loss": 0.7627,
      "step": 2785
    },
    {
      "epoch": 0.6389556853315012,
      "grad_norm": 0.24971815943717957,
      "learning_rate": 7.686837341608482e-05,
      "loss": 0.7647,
      "step": 2790
    },
    {
      "epoch": 0.6401007672048551,
      "grad_norm": 0.39055272936820984,
      "learning_rate": 7.680372381691235e-05,
      "loss": 0.7648,
      "step": 2795
    },
    {
      "epoch": 0.6412458490782091,
      "grad_norm": 0.5789263844490051,
      "learning_rate": 7.673907421773986e-05,
      "loss": 0.7631,
      "step": 2800
    },
    {
      "epoch": 0.6423909309515631,
      "grad_norm": 0.34116217494010925,
      "learning_rate": 7.667442461856736e-05,
      "loss": 0.7627,
      "step": 2805
    },
    {
      "epoch": 0.643536012824917,
      "grad_norm": 0.17528630793094635,
      "learning_rate": 7.660977501939488e-05,
      "loss": 0.7662,
      "step": 2810
    },
    {
      "epoch": 0.644681094698271,
      "grad_norm": 0.34103506803512573,
      "learning_rate": 7.654512542022239e-05,
      "loss": 0.7675,
      "step": 2815
    },
    {
      "epoch": 0.6458261765716249,
      "grad_norm": 0.38298407196998596,
      "learning_rate": 7.648047582104992e-05,
      "loss": 0.7608,
      "step": 2820
    },
    {
      "epoch": 0.6469712584449788,
      "grad_norm": 0.34761834144592285,
      "learning_rate": 7.641582622187743e-05,
      "loss": 0.7671,
      "step": 2825
    },
    {
      "epoch": 0.6481163403183328,
      "grad_norm": 0.21851806342601776,
      "learning_rate": 7.635117662270495e-05,
      "loss": 0.7595,
      "step": 2830
    },
    {
      "epoch": 0.6492614221916867,
      "grad_norm": 0.19289985299110413,
      "learning_rate": 7.628652702353246e-05,
      "loss": 0.7606,
      "step": 2835
    },
    {
      "epoch": 0.6504065040650406,
      "grad_norm": 0.17358209192752838,
      "learning_rate": 7.622187742435998e-05,
      "loss": 0.7599,
      "step": 2840
    },
    {
      "epoch": 0.6515515859383946,
      "grad_norm": 0.3359616696834564,
      "learning_rate": 7.615722782518748e-05,
      "loss": 0.7665,
      "step": 2845
    },
    {
      "epoch": 0.6526966678117485,
      "grad_norm": 0.4290224313735962,
      "learning_rate": 7.6092578226015e-05,
      "loss": 0.7643,
      "step": 2850
    },
    {
      "epoch": 0.6538417496851024,
      "grad_norm": 0.24706508219242096,
      "learning_rate": 7.602792862684252e-05,
      "loss": 0.7652,
      "step": 2855
    },
    {
      "epoch": 0.6549868315584564,
      "grad_norm": 0.23143522441387177,
      "learning_rate": 7.596327902767003e-05,
      "loss": 0.7618,
      "step": 2860
    },
    {
      "epoch": 0.6561319134318103,
      "grad_norm": 0.2622555196285248,
      "learning_rate": 7.589862942849755e-05,
      "loss": 0.7599,
      "step": 2865
    },
    {
      "epoch": 0.6572769953051644,
      "grad_norm": 0.30878838896751404,
      "learning_rate": 7.583397982932506e-05,
      "loss": 0.7608,
      "step": 2870
    },
    {
      "epoch": 0.6584220771785183,
      "grad_norm": 0.2817178964614868,
      "learning_rate": 7.576933023015258e-05,
      "loss": 0.764,
      "step": 2875
    },
    {
      "epoch": 0.6595671590518722,
      "grad_norm": 0.23512059450149536,
      "learning_rate": 7.570468063098009e-05,
      "loss": 0.7662,
      "step": 2880
    },
    {
      "epoch": 0.6607122409252262,
      "grad_norm": 0.2094719558954239,
      "learning_rate": 7.56400310318076e-05,
      "loss": 0.7669,
      "step": 2885
    },
    {
      "epoch": 0.6618573227985801,
      "grad_norm": 0.26047632098197937,
      "learning_rate": 7.557538143263512e-05,
      "loss": 0.7613,
      "step": 2890
    },
    {
      "epoch": 0.663002404671934,
      "grad_norm": 0.2346109300851822,
      "learning_rate": 7.551073183346263e-05,
      "loss": 0.7627,
      "step": 2895
    },
    {
      "epoch": 0.664147486545288,
      "grad_norm": 0.2940877676010132,
      "learning_rate": 7.544608223429015e-05,
      "loss": 0.7625,
      "step": 2900
    },
    {
      "epoch": 0.6652925684186419,
      "grad_norm": 0.5073947906494141,
      "learning_rate": 7.538143263511766e-05,
      "loss": 0.761,
      "step": 2905
    },
    {
      "epoch": 0.6664376502919959,
      "grad_norm": 0.28100213408470154,
      "learning_rate": 7.531678303594519e-05,
      "loss": 0.7677,
      "step": 2910
    },
    {
      "epoch": 0.6675827321653498,
      "grad_norm": 0.18574713170528412,
      "learning_rate": 7.52521334367727e-05,
      "loss": 0.7636,
      "step": 2915
    },
    {
      "epoch": 0.6687278140387037,
      "grad_norm": 0.2935812473297119,
      "learning_rate": 7.518748383760022e-05,
      "loss": 0.7658,
      "step": 2920
    },
    {
      "epoch": 0.6698728959120577,
      "grad_norm": 0.1969953179359436,
      "learning_rate": 7.512283423842773e-05,
      "loss": 0.7634,
      "step": 2925
    },
    {
      "epoch": 0.6710179777854116,
      "grad_norm": 0.17374487221240997,
      "learning_rate": 7.505818463925523e-05,
      "loss": 0.7593,
      "step": 2930
    },
    {
      "epoch": 0.6721630596587655,
      "grad_norm": 0.28035348653793335,
      "learning_rate": 7.499353504008275e-05,
      "loss": 0.7637,
      "step": 2935
    },
    {
      "epoch": 0.6733081415321196,
      "grad_norm": 0.23470938205718994,
      "learning_rate": 7.492888544091027e-05,
      "loss": 0.7615,
      "step": 2940
    },
    {
      "epoch": 0.6744532234054735,
      "grad_norm": 0.27259859442710876,
      "learning_rate": 7.486423584173779e-05,
      "loss": 0.7582,
      "step": 2945
    },
    {
      "epoch": 0.6755983052788275,
      "grad_norm": 0.3640976846218109,
      "learning_rate": 7.47995862425653e-05,
      "loss": 0.7662,
      "step": 2950
    },
    {
      "epoch": 0.6767433871521814,
      "grad_norm": 0.1931801736354828,
      "learning_rate": 7.473493664339282e-05,
      "loss": 0.759,
      "step": 2955
    },
    {
      "epoch": 0.6778884690255353,
      "grad_norm": 0.47389793395996094,
      "learning_rate": 7.467028704422033e-05,
      "loss": 0.7636,
      "step": 2960
    },
    {
      "epoch": 0.6790335508988893,
      "grad_norm": 0.34375831484794617,
      "learning_rate": 7.460563744504785e-05,
      "loss": 0.7625,
      "step": 2965
    },
    {
      "epoch": 0.6801786327722432,
      "grad_norm": 0.19540467858314514,
      "learning_rate": 7.454098784587536e-05,
      "loss": 0.759,
      "step": 2970
    },
    {
      "epoch": 0.6813237146455972,
      "grad_norm": 0.33635178208351135,
      "learning_rate": 7.447633824670287e-05,
      "loss": 0.7596,
      "step": 2975
    },
    {
      "epoch": 0.6824687965189511,
      "grad_norm": 0.2803053557872772,
      "learning_rate": 7.441168864753039e-05,
      "loss": 0.7613,
      "step": 2980
    },
    {
      "epoch": 0.683613878392305,
      "grad_norm": 0.21707181632518768,
      "learning_rate": 7.43470390483579e-05,
      "loss": 0.7584,
      "step": 2985
    },
    {
      "epoch": 0.684758960265659,
      "grad_norm": 0.25607773661613464,
      "learning_rate": 7.428238944918542e-05,
      "loss": 0.7617,
      "step": 2990
    },
    {
      "epoch": 0.6859040421390129,
      "grad_norm": 0.3066530227661133,
      "learning_rate": 7.421773985001293e-05,
      "loss": 0.7619,
      "step": 2995
    },
    {
      "epoch": 0.6870491240123668,
      "grad_norm": 0.2540838122367859,
      "learning_rate": 7.415309025084044e-05,
      "loss": 0.7616,
      "step": 3000
    },
    {
      "epoch": 0.6881942058857208,
      "grad_norm": 0.2732105553150177,
      "learning_rate": 7.408844065166797e-05,
      "loss": 0.7638,
      "step": 3005
    },
    {
      "epoch": 0.6893392877590748,
      "grad_norm": 0.23296260833740234,
      "learning_rate": 7.402379105249547e-05,
      "loss": 0.7639,
      "step": 3010
    },
    {
      "epoch": 0.6904843696324288,
      "grad_norm": 0.2918176054954529,
      "learning_rate": 7.395914145332299e-05,
      "loss": 0.7631,
      "step": 3015
    },
    {
      "epoch": 0.6916294515057827,
      "grad_norm": 0.20227240025997162,
      "learning_rate": 7.38944918541505e-05,
      "loss": 0.7624,
      "step": 3020
    },
    {
      "epoch": 0.6927745333791366,
      "grad_norm": 0.2289716601371765,
      "learning_rate": 7.382984225497802e-05,
      "loss": 0.76,
      "step": 3025
    },
    {
      "epoch": 0.6939196152524906,
      "grad_norm": 0.4011261761188507,
      "learning_rate": 7.376519265580554e-05,
      "loss": 0.7639,
      "step": 3030
    },
    {
      "epoch": 0.6950646971258445,
      "grad_norm": 0.16025088727474213,
      "learning_rate": 7.370054305663306e-05,
      "loss": 0.7602,
      "step": 3035
    },
    {
      "epoch": 0.6962097789991984,
      "grad_norm": 0.21587808430194855,
      "learning_rate": 7.363589345746057e-05,
      "loss": 0.7647,
      "step": 3040
    },
    {
      "epoch": 0.6973548608725524,
      "grad_norm": 0.2196759283542633,
      "learning_rate": 7.357124385828809e-05,
      "loss": 0.7657,
      "step": 3045
    },
    {
      "epoch": 0.6984999427459063,
      "grad_norm": 0.34964117407798767,
      "learning_rate": 7.35065942591156e-05,
      "loss": 0.7582,
      "step": 3050
    },
    {
      "epoch": 0.6996450246192603,
      "grad_norm": 0.34376025199890137,
      "learning_rate": 7.34419446599431e-05,
      "loss": 0.7627,
      "step": 3055
    },
    {
      "epoch": 0.7007901064926142,
      "grad_norm": 0.26398709416389465,
      "learning_rate": 7.337729506077063e-05,
      "loss": 0.7624,
      "step": 3060
    },
    {
      "epoch": 0.7019351883659681,
      "grad_norm": 0.24627701938152313,
      "learning_rate": 7.331264546159814e-05,
      "loss": 0.7619,
      "step": 3065
    },
    {
      "epoch": 0.7030802702393221,
      "grad_norm": 0.3965471386909485,
      "learning_rate": 7.324799586242566e-05,
      "loss": 0.7635,
      "step": 3070
    },
    {
      "epoch": 0.704225352112676,
      "grad_norm": 0.2562287747859955,
      "learning_rate": 7.318334626325317e-05,
      "loss": 0.7624,
      "step": 3075
    },
    {
      "epoch": 0.70537043398603,
      "grad_norm": 0.20183491706848145,
      "learning_rate": 7.311869666408069e-05,
      "loss": 0.7617,
      "step": 3080
    },
    {
      "epoch": 0.706515515859384,
      "grad_norm": 0.23248891532421112,
      "learning_rate": 7.30540470649082e-05,
      "loss": 0.7627,
      "step": 3085
    },
    {
      "epoch": 0.7076605977327379,
      "grad_norm": 0.4082355797290802,
      "learning_rate": 7.298939746573571e-05,
      "loss": 0.7649,
      "step": 3090
    },
    {
      "epoch": 0.7088056796060919,
      "grad_norm": 0.3321089744567871,
      "learning_rate": 7.292474786656323e-05,
      "loss": 0.7615,
      "step": 3095
    },
    {
      "epoch": 0.7099507614794458,
      "grad_norm": 0.3166889250278473,
      "learning_rate": 7.286009826739074e-05,
      "loss": 0.7639,
      "step": 3100
    },
    {
      "epoch": 0.7110958433527997,
      "grad_norm": 0.3378044068813324,
      "learning_rate": 7.279544866821826e-05,
      "loss": 0.7617,
      "step": 3105
    },
    {
      "epoch": 0.7122409252261537,
      "grad_norm": 0.24754200875759125,
      "learning_rate": 7.273079906904577e-05,
      "loss": 0.7603,
      "step": 3110
    },
    {
      "epoch": 0.7133860070995076,
      "grad_norm": 0.20655082166194916,
      "learning_rate": 7.266614946987329e-05,
      "loss": 0.7652,
      "step": 3115
    },
    {
      "epoch": 0.7145310889728615,
      "grad_norm": 0.1733807623386383,
      "learning_rate": 7.260149987070081e-05,
      "loss": 0.7638,
      "step": 3120
    },
    {
      "epoch": 0.7156761708462155,
      "grad_norm": 0.20328563451766968,
      "learning_rate": 7.253685027152833e-05,
      "loss": 0.7632,
      "step": 3125
    },
    {
      "epoch": 0.7168212527195694,
      "grad_norm": 0.28387951850891113,
      "learning_rate": 7.247220067235584e-05,
      "loss": 0.7607,
      "step": 3130
    },
    {
      "epoch": 0.7179663345929234,
      "grad_norm": 0.21539588272571564,
      "learning_rate": 7.240755107318336e-05,
      "loss": 0.7624,
      "step": 3135
    },
    {
      "epoch": 0.7191114164662773,
      "grad_norm": 0.20758353173732758,
      "learning_rate": 7.234290147401086e-05,
      "loss": 0.7636,
      "step": 3140
    },
    {
      "epoch": 0.7202564983396312,
      "grad_norm": 0.2486894130706787,
      "learning_rate": 7.227825187483837e-05,
      "loss": 0.766,
      "step": 3145
    },
    {
      "epoch": 0.7214015802129853,
      "grad_norm": 0.2833589017391205,
      "learning_rate": 7.22136022756659e-05,
      "loss": 0.7606,
      "step": 3150
    },
    {
      "epoch": 0.7225466620863392,
      "grad_norm": 0.2237931787967682,
      "learning_rate": 7.214895267649341e-05,
      "loss": 0.7631,
      "step": 3155
    },
    {
      "epoch": 0.7236917439596932,
      "grad_norm": 0.4406035244464874,
      "learning_rate": 7.208430307732093e-05,
      "loss": 0.7629,
      "step": 3160
    },
    {
      "epoch": 0.7248368258330471,
      "grad_norm": 0.32733210921287537,
      "learning_rate": 7.201965347814844e-05,
      "loss": 0.7581,
      "step": 3165
    },
    {
      "epoch": 0.725981907706401,
      "grad_norm": 0.1655232459306717,
      "learning_rate": 7.195500387897596e-05,
      "loss": 0.7621,
      "step": 3170
    },
    {
      "epoch": 0.727126989579755,
      "grad_norm": 0.17573896050453186,
      "learning_rate": 7.189035427980347e-05,
      "loss": 0.7606,
      "step": 3175
    },
    {
      "epoch": 0.7282720714531089,
      "grad_norm": 0.364986777305603,
      "learning_rate": 7.182570468063098e-05,
      "loss": 0.7621,
      "step": 3180
    },
    {
      "epoch": 0.7294171533264628,
      "grad_norm": 0.3647826611995697,
      "learning_rate": 7.17610550814585e-05,
      "loss": 0.7611,
      "step": 3185
    },
    {
      "epoch": 0.7305622351998168,
      "grad_norm": 0.23403432965278625,
      "learning_rate": 7.169640548228601e-05,
      "loss": 0.759,
      "step": 3190
    },
    {
      "epoch": 0.7317073170731707,
      "grad_norm": 0.33898410201072693,
      "learning_rate": 7.163175588311353e-05,
      "loss": 0.7621,
      "step": 3195
    },
    {
      "epoch": 0.7328523989465247,
      "grad_norm": 0.23586343228816986,
      "learning_rate": 7.156710628394104e-05,
      "loss": 0.7608,
      "step": 3200
    },
    {
      "epoch": 0.7339974808198786,
      "grad_norm": 0.2086014598608017,
      "learning_rate": 7.150245668476855e-05,
      "loss": 0.7611,
      "step": 3205
    },
    {
      "epoch": 0.7351425626932325,
      "grad_norm": 0.23983906209468842,
      "learning_rate": 7.143780708559607e-05,
      "loss": 0.7606,
      "step": 3210
    },
    {
      "epoch": 0.7362876445665865,
      "grad_norm": 0.19357101619243622,
      "learning_rate": 7.13731574864236e-05,
      "loss": 0.7623,
      "step": 3215
    },
    {
      "epoch": 0.7374327264399404,
      "grad_norm": 0.24940311908721924,
      "learning_rate": 7.13085078872511e-05,
      "loss": 0.7607,
      "step": 3220
    },
    {
      "epoch": 0.7385778083132944,
      "grad_norm": 0.19958077371120453,
      "learning_rate": 7.124385828807861e-05,
      "loss": 0.7639,
      "step": 3225
    },
    {
      "epoch": 0.7397228901866484,
      "grad_norm": 0.1807526797056198,
      "learning_rate": 7.117920868890613e-05,
      "loss": 0.7597,
      "step": 3230
    },
    {
      "epoch": 0.7408679720600023,
      "grad_norm": 0.3189755082130432,
      "learning_rate": 7.111455908973364e-05,
      "loss": 0.7584,
      "step": 3235
    },
    {
      "epoch": 0.7420130539333563,
      "grad_norm": 0.2481878697872162,
      "learning_rate": 7.104990949056117e-05,
      "loss": 0.7593,
      "step": 3240
    },
    {
      "epoch": 0.7431581358067102,
      "grad_norm": 0.2721337676048279,
      "learning_rate": 7.098525989138868e-05,
      "loss": 0.7606,
      "step": 3245
    },
    {
      "epoch": 0.7443032176800641,
      "grad_norm": 0.23429130017757416,
      "learning_rate": 7.09206102922162e-05,
      "loss": 0.7587,
      "step": 3250
    },
    {
      "epoch": 0.7454482995534181,
      "grad_norm": 0.30629605054855347,
      "learning_rate": 7.085596069304371e-05,
      "loss": 0.7631,
      "step": 3255
    },
    {
      "epoch": 0.746593381426772,
      "grad_norm": 0.3316456973552704,
      "learning_rate": 7.079131109387122e-05,
      "loss": 0.759,
      "step": 3260
    },
    {
      "epoch": 0.7477384633001259,
      "grad_norm": 0.41559523344039917,
      "learning_rate": 7.072666149469873e-05,
      "loss": 0.7592,
      "step": 3265
    },
    {
      "epoch": 0.7488835451734799,
      "grad_norm": 0.28401699662208557,
      "learning_rate": 7.066201189552625e-05,
      "loss": 0.7615,
      "step": 3270
    },
    {
      "epoch": 0.7500286270468338,
      "grad_norm": 0.25793886184692383,
      "learning_rate": 7.059736229635377e-05,
      "loss": 0.763,
      "step": 3275
    },
    {
      "epoch": 0.7511737089201878,
      "grad_norm": 0.19388380646705627,
      "learning_rate": 7.053271269718128e-05,
      "loss": 0.7578,
      "step": 3280
    },
    {
      "epoch": 0.7523187907935417,
      "grad_norm": 0.2703215479850769,
      "learning_rate": 7.04680630980088e-05,
      "loss": 0.7602,
      "step": 3285
    },
    {
      "epoch": 0.7534638726668956,
      "grad_norm": 0.21396110951900482,
      "learning_rate": 7.040341349883631e-05,
      "loss": 0.7593,
      "step": 3290
    },
    {
      "epoch": 0.7546089545402497,
      "grad_norm": 0.2268245965242386,
      "learning_rate": 7.033876389966382e-05,
      "loss": 0.7567,
      "step": 3295
    },
    {
      "epoch": 0.7557540364136036,
      "grad_norm": 0.17969952523708344,
      "learning_rate": 7.027411430049134e-05,
      "loss": 0.7612,
      "step": 3300
    },
    {
      "epoch": 0.7568991182869576,
      "grad_norm": 0.27984392642974854,
      "learning_rate": 7.020946470131885e-05,
      "loss": 0.7593,
      "step": 3305
    },
    {
      "epoch": 0.7580442001603115,
      "grad_norm": 0.22135871648788452,
      "learning_rate": 7.014481510214637e-05,
      "loss": 0.759,
      "step": 3310
    },
    {
      "epoch": 0.7591892820336654,
      "grad_norm": 0.26906076073646545,
      "learning_rate": 7.008016550297388e-05,
      "loss": 0.7618,
      "step": 3315
    },
    {
      "epoch": 0.7603343639070194,
      "grad_norm": 0.26624733209609985,
      "learning_rate": 7.00155159038014e-05,
      "loss": 0.7609,
      "step": 3320
    },
    {
      "epoch": 0.7614794457803733,
      "grad_norm": 0.4906958043575287,
      "learning_rate": 6.995086630462891e-05,
      "loss": 0.7602,
      "step": 3325
    },
    {
      "epoch": 0.7626245276537272,
      "grad_norm": 0.3430747985839844,
      "learning_rate": 6.988621670545644e-05,
      "loss": 0.7596,
      "step": 3330
    },
    {
      "epoch": 0.7637696095270812,
      "grad_norm": 0.2605244219303131,
      "learning_rate": 6.982156710628395e-05,
      "loss": 0.762,
      "step": 3335
    },
    {
      "epoch": 0.7649146914004351,
      "grad_norm": 0.21872663497924805,
      "learning_rate": 6.975691750711147e-05,
      "loss": 0.7603,
      "step": 3340
    },
    {
      "epoch": 0.766059773273789,
      "grad_norm": 0.3047766089439392,
      "learning_rate": 6.969226790793898e-05,
      "loss": 0.7592,
      "step": 3345
    },
    {
      "epoch": 0.767204855147143,
      "grad_norm": 0.317523717880249,
      "learning_rate": 6.962761830876648e-05,
      "loss": 0.7595,
      "step": 3350
    },
    {
      "epoch": 0.7683499370204969,
      "grad_norm": 0.19987279176712036,
      "learning_rate": 6.9562968709594e-05,
      "loss": 0.7557,
      "step": 3355
    },
    {
      "epoch": 0.7694950188938509,
      "grad_norm": 0.17979387938976288,
      "learning_rate": 6.949831911042152e-05,
      "loss": 0.7642,
      "step": 3360
    },
    {
      "epoch": 0.7706401007672049,
      "grad_norm": 0.18759754300117493,
      "learning_rate": 6.943366951124904e-05,
      "loss": 0.76,
      "step": 3365
    },
    {
      "epoch": 0.7717851826405588,
      "grad_norm": 0.25630974769592285,
      "learning_rate": 6.936901991207655e-05,
      "loss": 0.7558,
      "step": 3370
    },
    {
      "epoch": 0.7729302645139128,
      "grad_norm": 0.19636793434619904,
      "learning_rate": 6.930437031290407e-05,
      "loss": 0.7592,
      "step": 3375
    },
    {
      "epoch": 0.7740753463872667,
      "grad_norm": 0.1815565526485443,
      "learning_rate": 6.923972071373158e-05,
      "loss": 0.7553,
      "step": 3380
    },
    {
      "epoch": 0.7752204282606207,
      "grad_norm": 0.28360649943351746,
      "learning_rate": 6.91750711145591e-05,
      "loss": 0.7593,
      "step": 3385
    },
    {
      "epoch": 0.7763655101339746,
      "grad_norm": 0.2512878477573395,
      "learning_rate": 6.911042151538661e-05,
      "loss": 0.7566,
      "step": 3390
    },
    {
      "epoch": 0.7775105920073285,
      "grad_norm": 0.23916715383529663,
      "learning_rate": 6.904577191621412e-05,
      "loss": 0.7604,
      "step": 3395
    },
    {
      "epoch": 0.7786556738806825,
      "grad_norm": 0.2430179864168167,
      "learning_rate": 6.898112231704164e-05,
      "loss": 0.7618,
      "step": 3400
    },
    {
      "epoch": 0.7798007557540364,
      "grad_norm": 0.32661014795303345,
      "learning_rate": 6.891647271786915e-05,
      "loss": 0.7611,
      "step": 3405
    },
    {
      "epoch": 0.7809458376273903,
      "grad_norm": 0.26453500986099243,
      "learning_rate": 6.885182311869666e-05,
      "loss": 0.7612,
      "step": 3410
    },
    {
      "epoch": 0.7820909195007443,
      "grad_norm": 0.19719307124614716,
      "learning_rate": 6.878717351952418e-05,
      "loss": 0.7596,
      "step": 3415
    },
    {
      "epoch": 0.7832360013740982,
      "grad_norm": 0.1745481789112091,
      "learning_rate": 6.872252392035171e-05,
      "loss": 0.7562,
      "step": 3420
    },
    {
      "epoch": 0.7843810832474521,
      "grad_norm": 0.3306420147418976,
      "learning_rate": 6.865787432117922e-05,
      "loss": 0.7589,
      "step": 3425
    },
    {
      "epoch": 0.7855261651208061,
      "grad_norm": 0.3108626902103424,
      "learning_rate": 6.859322472200672e-05,
      "loss": 0.7598,
      "step": 3430
    },
    {
      "epoch": 0.7866712469941601,
      "grad_norm": 0.24207472801208496,
      "learning_rate": 6.852857512283424e-05,
      "loss": 0.7623,
      "step": 3435
    },
    {
      "epoch": 0.7878163288675141,
      "grad_norm": 0.31846627593040466,
      "learning_rate": 6.846392552366175e-05,
      "loss": 0.7595,
      "step": 3440
    },
    {
      "epoch": 0.788961410740868,
      "grad_norm": 0.34010037779808044,
      "learning_rate": 6.839927592448926e-05,
      "loss": 0.7598,
      "step": 3445
    },
    {
      "epoch": 0.790106492614222,
      "grad_norm": 0.2276770919561386,
      "learning_rate": 6.833462632531679e-05,
      "loss": 0.7612,
      "step": 3450
    },
    {
      "epoch": 0.7912515744875759,
      "grad_norm": 0.20850631594657898,
      "learning_rate": 6.82699767261443e-05,
      "loss": 0.7643,
      "step": 3455
    },
    {
      "epoch": 0.7923966563609298,
      "grad_norm": 0.2672816216945648,
      "learning_rate": 6.820532712697182e-05,
      "loss": 0.7585,
      "step": 3460
    },
    {
      "epoch": 0.7935417382342838,
      "grad_norm": 0.22202163934707642,
      "learning_rate": 6.814067752779933e-05,
      "loss": 0.7554,
      "step": 3465
    },
    {
      "epoch": 0.7946868201076377,
      "grad_norm": 0.2591758072376251,
      "learning_rate": 6.807602792862685e-05,
      "loss": 0.7597,
      "step": 3470
    },
    {
      "epoch": 0.7958319019809916,
      "grad_norm": 0.23392333090305328,
      "learning_rate": 6.801137832945435e-05,
      "loss": 0.7648,
      "step": 3475
    },
    {
      "epoch": 0.7969769838543456,
      "grad_norm": 0.2861672639846802,
      "learning_rate": 6.794672873028188e-05,
      "loss": 0.7598,
      "step": 3480
    },
    {
      "epoch": 0.7981220657276995,
      "grad_norm": 0.2735273241996765,
      "learning_rate": 6.788207913110939e-05,
      "loss": 0.759,
      "step": 3485
    },
    {
      "epoch": 0.7992671476010534,
      "grad_norm": 0.2500370442867279,
      "learning_rate": 6.78174295319369e-05,
      "loss": 0.7634,
      "step": 3490
    },
    {
      "epoch": 0.8004122294744074,
      "grad_norm": 0.21133625507354736,
      "learning_rate": 6.775277993276442e-05,
      "loss": 0.7558,
      "step": 3495
    },
    {
      "epoch": 0.8015573113477613,
      "grad_norm": 0.29856279492378235,
      "learning_rate": 6.768813033359193e-05,
      "loss": 0.7598,
      "step": 3500
    },
    {
      "epoch": 0.8027023932211154,
      "grad_norm": 0.29944127798080444,
      "learning_rate": 6.762348073441945e-05,
      "loss": 0.7556,
      "step": 3505
    },
    {
      "epoch": 0.8038474750944693,
      "grad_norm": 0.2628041207790375,
      "learning_rate": 6.755883113524696e-05,
      "loss": 0.7551,
      "step": 3510
    },
    {
      "epoch": 0.8049925569678232,
      "grad_norm": 0.30071425437927246,
      "learning_rate": 6.749418153607448e-05,
      "loss": 0.7613,
      "step": 3515
    },
    {
      "epoch": 0.8061376388411772,
      "grad_norm": 0.19781585037708282,
      "learning_rate": 6.742953193690199e-05,
      "loss": 0.7584,
      "step": 3520
    },
    {
      "epoch": 0.8072827207145311,
      "grad_norm": 0.22515876591205597,
      "learning_rate": 6.73648823377295e-05,
      "loss": 0.7602,
      "step": 3525
    },
    {
      "epoch": 0.808427802587885,
      "grad_norm": 0.18484292924404144,
      "learning_rate": 6.730023273855702e-05,
      "loss": 0.7587,
      "step": 3530
    },
    {
      "epoch": 0.809572884461239,
      "grad_norm": 0.18081125617027283,
      "learning_rate": 6.723558313938453e-05,
      "loss": 0.7571,
      "step": 3535
    },
    {
      "epoch": 0.8107179663345929,
      "grad_norm": 0.27940741181373596,
      "learning_rate": 6.717093354021206e-05,
      "loss": 0.763,
      "step": 3540
    },
    {
      "epoch": 0.8118630482079469,
      "grad_norm": 0.26946401596069336,
      "learning_rate": 6.710628394103958e-05,
      "loss": 0.7608,
      "step": 3545
    },
    {
      "epoch": 0.8130081300813008,
      "grad_norm": 0.2804325222969055,
      "learning_rate": 6.704163434186709e-05,
      "loss": 0.758,
      "step": 3550
    },
    {
      "epoch": 0.8141532119546547,
      "grad_norm": 0.3489839732646942,
      "learning_rate": 6.69769847426946e-05,
      "loss": 0.758,
      "step": 3555
    },
    {
      "epoch": 0.8152982938280087,
      "grad_norm": 0.17796777188777924,
      "learning_rate": 6.69123351435221e-05,
      "loss": 0.7592,
      "step": 3560
    },
    {
      "epoch": 0.8164433757013626,
      "grad_norm": 0.1994534283876419,
      "learning_rate": 6.684768554434962e-05,
      "loss": 0.7555,
      "step": 3565
    },
    {
      "epoch": 0.8175884575747165,
      "grad_norm": 0.23157624900341034,
      "learning_rate": 6.678303594517715e-05,
      "loss": 0.7592,
      "step": 3570
    },
    {
      "epoch": 0.8187335394480706,
      "grad_norm": 0.26892754435539246,
      "learning_rate": 6.671838634600466e-05,
      "loss": 0.7558,
      "step": 3575
    },
    {
      "epoch": 0.8198786213214245,
      "grad_norm": 0.2596917152404785,
      "learning_rate": 6.665373674683218e-05,
      "loss": 0.7561,
      "step": 3580
    },
    {
      "epoch": 0.8210237031947785,
      "grad_norm": 0.20809470117092133,
      "learning_rate": 6.658908714765969e-05,
      "loss": 0.7588,
      "step": 3585
    },
    {
      "epoch": 0.8221687850681324,
      "grad_norm": 0.19905616343021393,
      "learning_rate": 6.65244375484872e-05,
      "loss": 0.7615,
      "step": 3590
    },
    {
      "epoch": 0.8233138669414863,
      "grad_norm": 0.33176565170288086,
      "learning_rate": 6.645978794931472e-05,
      "loss": 0.7557,
      "step": 3595
    },
    {
      "epoch": 0.8244589488148403,
      "grad_norm": 0.3151164948940277,
      "learning_rate": 6.639513835014223e-05,
      "loss": 0.7541,
      "step": 3600
    },
    {
      "epoch": 0.8256040306881942,
      "grad_norm": 0.21330596506595612,
      "learning_rate": 6.633048875096975e-05,
      "loss": 0.7628,
      "step": 3605
    },
    {
      "epoch": 0.8267491125615481,
      "grad_norm": 0.2487279772758484,
      "learning_rate": 6.626583915179726e-05,
      "loss": 0.7591,
      "step": 3610
    },
    {
      "epoch": 0.8278941944349021,
      "grad_norm": 0.37388402223587036,
      "learning_rate": 6.620118955262477e-05,
      "loss": 0.758,
      "step": 3615
    },
    {
      "epoch": 0.829039276308256,
      "grad_norm": 0.32117441296577454,
      "learning_rate": 6.613653995345229e-05,
      "loss": 0.7577,
      "step": 3620
    },
    {
      "epoch": 0.83018435818161,
      "grad_norm": 0.22614158689975739,
      "learning_rate": 6.60718903542798e-05,
      "loss": 0.7582,
      "step": 3625
    },
    {
      "epoch": 0.8313294400549639,
      "grad_norm": 0.3191037178039551,
      "learning_rate": 6.600724075510733e-05,
      "loss": 0.7567,
      "step": 3630
    },
    {
      "epoch": 0.8324745219283178,
      "grad_norm": 0.22228999435901642,
      "learning_rate": 6.594259115593485e-05,
      "loss": 0.7574,
      "step": 3635
    },
    {
      "epoch": 0.8336196038016718,
      "grad_norm": 0.21925920248031616,
      "learning_rate": 6.587794155676235e-05,
      "loss": 0.7561,
      "step": 3640
    },
    {
      "epoch": 0.8347646856750258,
      "grad_norm": 0.18947115540504456,
      "learning_rate": 6.581329195758986e-05,
      "loss": 0.7517,
      "step": 3645
    },
    {
      "epoch": 0.8359097675483798,
      "grad_norm": 0.25208985805511475,
      "learning_rate": 6.574864235841737e-05,
      "loss": 0.7576,
      "step": 3650
    },
    {
      "epoch": 0.8370548494217337,
      "grad_norm": 0.3698408901691437,
      "learning_rate": 6.568399275924489e-05,
      "loss": 0.7538,
      "step": 3655
    },
    {
      "epoch": 0.8381999312950876,
      "grad_norm": 0.3591291308403015,
      "learning_rate": 6.561934316007242e-05,
      "loss": 0.7572,
      "step": 3660
    },
    {
      "epoch": 0.8393450131684416,
      "grad_norm": 0.268759548664093,
      "learning_rate": 6.555469356089993e-05,
      "loss": 0.7577,
      "step": 3665
    },
    {
      "epoch": 0.8404900950417955,
      "grad_norm": 0.2655838131904602,
      "learning_rate": 6.549004396172744e-05,
      "loss": 0.7597,
      "step": 3670
    },
    {
      "epoch": 0.8416351769151494,
      "grad_norm": 0.29894131422042847,
      "learning_rate": 6.542539436255496e-05,
      "loss": 0.7547,
      "step": 3675
    },
    {
      "epoch": 0.8427802587885034,
      "grad_norm": 0.2238704115152359,
      "learning_rate": 6.536074476338247e-05,
      "loss": 0.757,
      "step": 3680
    },
    {
      "epoch": 0.8439253406618573,
      "grad_norm": 0.14427435398101807,
      "learning_rate": 6.529609516420997e-05,
      "loss": 0.7577,
      "step": 3685
    },
    {
      "epoch": 0.8450704225352113,
      "grad_norm": 0.16628648340702057,
      "learning_rate": 6.52314455650375e-05,
      "loss": 0.7559,
      "step": 3690
    },
    {
      "epoch": 0.8462155044085652,
      "grad_norm": 0.36011502146720886,
      "learning_rate": 6.516679596586502e-05,
      "loss": 0.7559,
      "step": 3695
    },
    {
      "epoch": 0.8473605862819191,
      "grad_norm": 0.3687536418437958,
      "learning_rate": 6.510214636669253e-05,
      "loss": 0.7574,
      "step": 3700
    },
    {
      "epoch": 0.8485056681552731,
      "grad_norm": 0.2624550461769104,
      "learning_rate": 6.503749676752004e-05,
      "loss": 0.7528,
      "step": 3705
    },
    {
      "epoch": 0.849650750028627,
      "grad_norm": 0.30368492007255554,
      "learning_rate": 6.497284716834756e-05,
      "loss": 0.7614,
      "step": 3710
    },
    {
      "epoch": 0.8507958319019809,
      "grad_norm": 0.2152595818042755,
      "learning_rate": 6.490819756917507e-05,
      "loss": 0.7563,
      "step": 3715
    },
    {
      "epoch": 0.851940913775335,
      "grad_norm": 0.1912301778793335,
      "learning_rate": 6.484354797000259e-05,
      "loss": 0.7528,
      "step": 3720
    },
    {
      "epoch": 0.8530859956486889,
      "grad_norm": 0.2414664626121521,
      "learning_rate": 6.47788983708301e-05,
      "loss": 0.7565,
      "step": 3725
    },
    {
      "epoch": 0.8542310775220429,
      "grad_norm": 0.2542330026626587,
      "learning_rate": 6.471424877165762e-05,
      "loss": 0.7581,
      "step": 3730
    },
    {
      "epoch": 0.8553761593953968,
      "grad_norm": 0.21856382489204407,
      "learning_rate": 6.464959917248513e-05,
      "loss": 0.7586,
      "step": 3735
    },
    {
      "epoch": 0.8565212412687507,
      "grad_norm": 0.35254383087158203,
      "learning_rate": 6.458494957331264e-05,
      "loss": 0.7572,
      "step": 3740
    },
    {
      "epoch": 0.8576663231421047,
      "grad_norm": 0.21001994609832764,
      "learning_rate": 6.452029997414016e-05,
      "loss": 0.7616,
      "step": 3745
    },
    {
      "epoch": 0.8588114050154586,
      "grad_norm": 0.4330880045890808,
      "learning_rate": 6.445565037496769e-05,
      "loss": 0.7568,
      "step": 3750
    },
    {
      "epoch": 0.8599564868888125,
      "grad_norm": 0.2603987455368042,
      "learning_rate": 6.43910007757952e-05,
      "loss": 0.7542,
      "step": 3755
    },
    {
      "epoch": 0.8611015687621665,
      "grad_norm": 0.2131146788597107,
      "learning_rate": 6.432635117662271e-05,
      "loss": 0.7556,
      "step": 3760
    },
    {
      "epoch": 0.8622466506355204,
      "grad_norm": 0.296593576669693,
      "learning_rate": 6.426170157745022e-05,
      "loss": 0.7627,
      "step": 3765
    },
    {
      "epoch": 0.8633917325088744,
      "grad_norm": 0.20938166975975037,
      "learning_rate": 6.419705197827773e-05,
      "loss": 0.7571,
      "step": 3770
    },
    {
      "epoch": 0.8645368143822283,
      "grad_norm": 0.31437405943870544,
      "learning_rate": 6.413240237910524e-05,
      "loss": 0.7561,
      "step": 3775
    },
    {
      "epoch": 0.8656818962555822,
      "grad_norm": 0.31622976064682007,
      "learning_rate": 6.406775277993277e-05,
      "loss": 0.7546,
      "step": 3780
    },
    {
      "epoch": 0.8668269781289362,
      "grad_norm": 0.24524053931236267,
      "learning_rate": 6.400310318076029e-05,
      "loss": 0.7599,
      "step": 3785
    },
    {
      "epoch": 0.8679720600022902,
      "grad_norm": 0.17051318287849426,
      "learning_rate": 6.39384535815878e-05,
      "loss": 0.7583,
      "step": 3790
    },
    {
      "epoch": 0.8691171418756442,
      "grad_norm": 0.1901961863040924,
      "learning_rate": 6.387380398241531e-05,
      "loss": 0.7569,
      "step": 3795
    },
    {
      "epoch": 0.8702622237489981,
      "grad_norm": 0.19765377044677734,
      "learning_rate": 6.380915438324283e-05,
      "loss": 0.7561,
      "step": 3800
    },
    {
      "epoch": 0.871407305622352,
      "grad_norm": 0.2318921536207199,
      "learning_rate": 6.374450478407034e-05,
      "loss": 0.7574,
      "step": 3805
    },
    {
      "epoch": 0.872552387495706,
      "grad_norm": 0.23888637125492096,
      "learning_rate": 6.367985518489786e-05,
      "loss": 0.7559,
      "step": 3810
    },
    {
      "epoch": 0.8736974693690599,
      "grad_norm": 0.2582821547985077,
      "learning_rate": 6.361520558572537e-05,
      "loss": 0.759,
      "step": 3815
    },
    {
      "epoch": 0.8748425512424138,
      "grad_norm": 0.2514699697494507,
      "learning_rate": 6.355055598655289e-05,
      "loss": 0.7546,
      "step": 3820
    },
    {
      "epoch": 0.8759876331157678,
      "grad_norm": 0.31897538900375366,
      "learning_rate": 6.34859063873804e-05,
      "loss": 0.758,
      "step": 3825
    },
    {
      "epoch": 0.8771327149891217,
      "grad_norm": 0.18458561599254608,
      "learning_rate": 6.342125678820791e-05,
      "loss": 0.7569,
      "step": 3830
    },
    {
      "epoch": 0.8782777968624756,
      "grad_norm": 0.30276724696159363,
      "learning_rate": 6.335660718903543e-05,
      "loss": 0.7588,
      "step": 3835
    },
    {
      "epoch": 0.8794228787358296,
      "grad_norm": 0.20559334754943848,
      "learning_rate": 6.329195758986296e-05,
      "loss": 0.7593,
      "step": 3840
    },
    {
      "epoch": 0.8805679606091835,
      "grad_norm": 0.23002290725708008,
      "learning_rate": 6.322730799069047e-05,
      "loss": 0.7565,
      "step": 3845
    },
    {
      "epoch": 0.8817130424825375,
      "grad_norm": 0.21782149374485016,
      "learning_rate": 6.316265839151797e-05,
      "loss": 0.7536,
      "step": 3850
    },
    {
      "epoch": 0.8828581243558914,
      "grad_norm": 0.22648191452026367,
      "learning_rate": 6.309800879234548e-05,
      "loss": 0.756,
      "step": 3855
    },
    {
      "epoch": 0.8840032062292454,
      "grad_norm": 0.25743627548217773,
      "learning_rate": 6.3033359193173e-05,
      "loss": 0.7574,
      "step": 3860
    },
    {
      "epoch": 0.8851482881025994,
      "grad_norm": 0.23849058151245117,
      "learning_rate": 6.296870959400051e-05,
      "loss": 0.7532,
      "step": 3865
    },
    {
      "epoch": 0.8862933699759533,
      "grad_norm": 0.24444012343883514,
      "learning_rate": 6.290405999482804e-05,
      "loss": 0.7552,
      "step": 3870
    },
    {
      "epoch": 0.8874384518493073,
      "grad_norm": 0.21026170253753662,
      "learning_rate": 6.283941039565556e-05,
      "loss": 0.7551,
      "step": 3875
    },
    {
      "epoch": 0.8885835337226612,
      "grad_norm": 0.25571170449256897,
      "learning_rate": 6.277476079648307e-05,
      "loss": 0.754,
      "step": 3880
    },
    {
      "epoch": 0.8897286155960151,
      "grad_norm": 0.24130992591381073,
      "learning_rate": 6.271011119731058e-05,
      "loss": 0.7608,
      "step": 3885
    },
    {
      "epoch": 0.8908736974693691,
      "grad_norm": 0.21595409512519836,
      "learning_rate": 6.26454615981381e-05,
      "loss": 0.7563,
      "step": 3890
    },
    {
      "epoch": 0.892018779342723,
      "grad_norm": 0.2847621738910675,
      "learning_rate": 6.25808119989656e-05,
      "loss": 0.7569,
      "step": 3895
    },
    {
      "epoch": 0.8931638612160769,
      "grad_norm": 0.2793238162994385,
      "learning_rate": 6.251616239979313e-05,
      "loss": 0.7572,
      "step": 3900
    },
    {
      "epoch": 0.8943089430894309,
      "grad_norm": 0.19144894182682037,
      "learning_rate": 6.245151280062064e-05,
      "loss": 0.7541,
      "step": 3905
    },
    {
      "epoch": 0.8954540249627848,
      "grad_norm": 0.18905645608901978,
      "learning_rate": 6.238686320144815e-05,
      "loss": 0.7584,
      "step": 3910
    },
    {
      "epoch": 0.8965991068361387,
      "grad_norm": 0.24144595861434937,
      "learning_rate": 6.232221360227567e-05,
      "loss": 0.7566,
      "step": 3915
    },
    {
      "epoch": 0.8977441887094927,
      "grad_norm": 0.22137823700904846,
      "learning_rate": 6.225756400310318e-05,
      "loss": 0.7578,
      "step": 3920
    },
    {
      "epoch": 0.8988892705828466,
      "grad_norm": 0.247211754322052,
      "learning_rate": 6.21929144039307e-05,
      "loss": 0.7566,
      "step": 3925
    },
    {
      "epoch": 0.9000343524562007,
      "grad_norm": 0.14881117641925812,
      "learning_rate": 6.212826480475823e-05,
      "loss": 0.7558,
      "step": 3930
    },
    {
      "epoch": 0.9011794343295546,
      "grad_norm": 0.2065708339214325,
      "learning_rate": 6.206361520558573e-05,
      "loss": 0.759,
      "step": 3935
    },
    {
      "epoch": 0.9023245162029085,
      "grad_norm": 0.21069486439228058,
      "learning_rate": 6.199896560641324e-05,
      "loss": 0.7569,
      "step": 3940
    },
    {
      "epoch": 0.9034695980762625,
      "grad_norm": 0.21884271502494812,
      "learning_rate": 6.193431600724075e-05,
      "loss": 0.7575,
      "step": 3945
    },
    {
      "epoch": 0.9046146799496164,
      "grad_norm": 0.3551616966724396,
      "learning_rate": 6.186966640806827e-05,
      "loss": 0.7536,
      "step": 3950
    },
    {
      "epoch": 0.9057597618229704,
      "grad_norm": 0.28585392236709595,
      "learning_rate": 6.180501680889578e-05,
      "loss": 0.7561,
      "step": 3955
    },
    {
      "epoch": 0.9069048436963243,
      "grad_norm": 0.21221600472927094,
      "learning_rate": 6.174036720972331e-05,
      "loss": 0.753,
      "step": 3960
    },
    {
      "epoch": 0.9080499255696782,
      "grad_norm": 0.24336685240268707,
      "learning_rate": 6.167571761055082e-05,
      "loss": 0.7576,
      "step": 3965
    },
    {
      "epoch": 0.9091950074430322,
      "grad_norm": 0.22866609692573547,
      "learning_rate": 6.161106801137834e-05,
      "loss": 0.7549,
      "step": 3970
    },
    {
      "epoch": 0.9103400893163861,
      "grad_norm": 0.21172703802585602,
      "learning_rate": 6.154641841220584e-05,
      "loss": 0.757,
      "step": 3975
    },
    {
      "epoch": 0.91148517118974,
      "grad_norm": 0.2627602815628052,
      "learning_rate": 6.148176881303335e-05,
      "loss": 0.7549,
      "step": 3980
    },
    {
      "epoch": 0.912630253063094,
      "grad_norm": 0.3001517355442047,
      "learning_rate": 6.141711921386087e-05,
      "loss": 0.7576,
      "step": 3985
    },
    {
      "epoch": 0.9137753349364479,
      "grad_norm": 0.18586446344852448,
      "learning_rate": 6.13524696146884e-05,
      "loss": 0.7527,
      "step": 3990
    },
    {
      "epoch": 0.9149204168098019,
      "grad_norm": 0.2953189015388489,
      "learning_rate": 6.128782001551591e-05,
      "loss": 0.753,
      "step": 3995
    },
    {
      "epoch": 0.9160654986831559,
      "grad_norm": 0.1657901555299759,
      "learning_rate": 6.122317041634342e-05,
      "loss": 0.7552,
      "step": 4000
    },
    {
      "epoch": 0.9172105805565098,
      "grad_norm": 0.26477518677711487,
      "learning_rate": 6.115852081717094e-05,
      "loss": 0.7542,
      "step": 4005
    },
    {
      "epoch": 0.9183556624298638,
      "grad_norm": 0.1784099042415619,
      "learning_rate": 6.109387121799845e-05,
      "loss": 0.7548,
      "step": 4010
    },
    {
      "epoch": 0.9195007443032177,
      "grad_norm": 0.27934473752975464,
      "learning_rate": 6.1029221618825974e-05,
      "loss": 0.7548,
      "step": 4015
    },
    {
      "epoch": 0.9206458261765716,
      "grad_norm": 0.18960314989089966,
      "learning_rate": 6.0964572019653474e-05,
      "loss": 0.7552,
      "step": 4020
    },
    {
      "epoch": 0.9217909080499256,
      "grad_norm": 0.26606252789497375,
      "learning_rate": 6.0899922420480995e-05,
      "loss": 0.757,
      "step": 4025
    },
    {
      "epoch": 0.9229359899232795,
      "grad_norm": 0.19727392494678497,
      "learning_rate": 6.083527282130851e-05,
      "loss": 0.752,
      "step": 4030
    },
    {
      "epoch": 0.9240810717966335,
      "grad_norm": 0.1997370570898056,
      "learning_rate": 6.0770623222136024e-05,
      "loss": 0.7556,
      "step": 4035
    },
    {
      "epoch": 0.9252261536699874,
      "grad_norm": 0.3227653503417969,
      "learning_rate": 6.070597362296354e-05,
      "loss": 0.756,
      "step": 4040
    },
    {
      "epoch": 0.9263712355433413,
      "grad_norm": 0.21063637733459473,
      "learning_rate": 6.064132402379106e-05,
      "loss": 0.7525,
      "step": 4045
    },
    {
      "epoch": 0.9275163174166953,
      "grad_norm": 0.1530616283416748,
      "learning_rate": 6.057667442461857e-05,
      "loss": 0.7552,
      "step": 4050
    },
    {
      "epoch": 0.9286613992900492,
      "grad_norm": 0.3049672245979309,
      "learning_rate": 6.051202482544609e-05,
      "loss": 0.756,
      "step": 4055
    },
    {
      "epoch": 0.9298064811634031,
      "grad_norm": 0.20015349984169006,
      "learning_rate": 6.0447375226273595e-05,
      "loss": 0.7549,
      "step": 4060
    },
    {
      "epoch": 0.9309515630367571,
      "grad_norm": 0.2715403139591217,
      "learning_rate": 6.038272562710111e-05,
      "loss": 0.7544,
      "step": 4065
    },
    {
      "epoch": 0.9320966449101111,
      "grad_norm": 0.200375497341156,
      "learning_rate": 6.031807602792863e-05,
      "loss": 0.7525,
      "step": 4070
    },
    {
      "epoch": 0.9332417267834651,
      "grad_norm": 0.2582535445690155,
      "learning_rate": 6.0253426428756144e-05,
      "loss": 0.7506,
      "step": 4075
    },
    {
      "epoch": 0.934386808656819,
      "grad_norm": 0.18540117144584656,
      "learning_rate": 6.018877682958366e-05,
      "loss": 0.7555,
      "step": 4080
    },
    {
      "epoch": 0.9355318905301729,
      "grad_norm": 0.24330022931098938,
      "learning_rate": 6.012412723041117e-05,
      "loss": 0.7551,
      "step": 4085
    },
    {
      "epoch": 0.9366769724035269,
      "grad_norm": 0.32542139291763306,
      "learning_rate": 6.0059477631238694e-05,
      "loss": 0.7552,
      "step": 4090
    },
    {
      "epoch": 0.9378220542768808,
      "grad_norm": 0.24217607080936432,
      "learning_rate": 5.999482803206621e-05,
      "loss": 0.7557,
      "step": 4095
    },
    {
      "epoch": 0.9389671361502347,
      "grad_norm": 0.2282743602991104,
      "learning_rate": 5.993017843289372e-05,
      "loss": 0.7571,
      "step": 4100
    },
    {
      "epoch": 0.9401122180235887,
      "grad_norm": 0.18588009476661682,
      "learning_rate": 5.986552883372123e-05,
      "loss": 0.7529,
      "step": 4105
    },
    {
      "epoch": 0.9412572998969426,
      "grad_norm": 0.24249807000160217,
      "learning_rate": 5.9800879234548744e-05,
      "loss": 0.756,
      "step": 4110
    },
    {
      "epoch": 0.9424023817702966,
      "grad_norm": 0.20471017062664032,
      "learning_rate": 5.9736229635376265e-05,
      "loss": 0.7577,
      "step": 4115
    },
    {
      "epoch": 0.9435474636436505,
      "grad_norm": 0.2088862508535385,
      "learning_rate": 5.967158003620378e-05,
      "loss": 0.75,
      "step": 4120
    },
    {
      "epoch": 0.9446925455170044,
      "grad_norm": 0.16393999755382538,
      "learning_rate": 5.960693043703129e-05,
      "loss": 0.7508,
      "step": 4125
    },
    {
      "epoch": 0.9458376273903584,
      "grad_norm": 0.1825890988111496,
      "learning_rate": 5.954228083785881e-05,
      "loss": 0.7525,
      "step": 4130
    },
    {
      "epoch": 0.9469827092637123,
      "grad_norm": 0.14733733236789703,
      "learning_rate": 5.947763123868633e-05,
      "loss": 0.7552,
      "step": 4135
    },
    {
      "epoch": 0.9481277911370662,
      "grad_norm": 0.34639453887939453,
      "learning_rate": 5.941298163951384e-05,
      "loss": 0.752,
      "step": 4140
    },
    {
      "epoch": 0.9492728730104203,
      "grad_norm": 0.35142093896865845,
      "learning_rate": 5.934833204034135e-05,
      "loss": 0.7563,
      "step": 4145
    },
    {
      "epoch": 0.9504179548837742,
      "grad_norm": 0.20236782729625702,
      "learning_rate": 5.9283682441168864e-05,
      "loss": 0.7542,
      "step": 4150
    },
    {
      "epoch": 0.9515630367571282,
      "grad_norm": 0.2146655023097992,
      "learning_rate": 5.921903284199638e-05,
      "loss": 0.7539,
      "step": 4155
    },
    {
      "epoch": 0.9527081186304821,
      "grad_norm": 0.20262286067008972,
      "learning_rate": 5.91543832428239e-05,
      "loss": 0.7487,
      "step": 4160
    },
    {
      "epoch": 0.953853200503836,
      "grad_norm": 0.22617624700069427,
      "learning_rate": 5.9089733643651414e-05,
      "loss": 0.7551,
      "step": 4165
    },
    {
      "epoch": 0.95499828237719,
      "grad_norm": 0.18724945187568665,
      "learning_rate": 5.902508404447893e-05,
      "loss": 0.7558,
      "step": 4170
    },
    {
      "epoch": 0.9561433642505439,
      "grad_norm": 0.22609928250312805,
      "learning_rate": 5.896043444530644e-05,
      "loss": 0.7541,
      "step": 4175
    },
    {
      "epoch": 0.9572884461238979,
      "grad_norm": 0.28732478618621826,
      "learning_rate": 5.889578484613396e-05,
      "loss": 0.7535,
      "step": 4180
    },
    {
      "epoch": 0.9584335279972518,
      "grad_norm": 0.18549516797065735,
      "learning_rate": 5.8831135246961464e-05,
      "loss": 0.751,
      "step": 4185
    },
    {
      "epoch": 0.9595786098706057,
      "grad_norm": 0.18325352668762207,
      "learning_rate": 5.8766485647788985e-05,
      "loss": 0.754,
      "step": 4190
    },
    {
      "epoch": 0.9607236917439597,
      "grad_norm": 0.23209375143051147,
      "learning_rate": 5.87018360486165e-05,
      "loss": 0.7526,
      "step": 4195
    },
    {
      "epoch": 0.9618687736173136,
      "grad_norm": 0.312589168548584,
      "learning_rate": 5.863718644944401e-05,
      "loss": 0.7494,
      "step": 4200
    },
    {
      "epoch": 0.9630138554906675,
      "grad_norm": 0.20656633377075195,
      "learning_rate": 5.8572536850271534e-05,
      "loss": 0.7549,
      "step": 4205
    },
    {
      "epoch": 0.9641589373640215,
      "grad_norm": 0.16602040827274323,
      "learning_rate": 5.850788725109905e-05,
      "loss": 0.7553,
      "step": 4210
    },
    {
      "epoch": 0.9653040192373755,
      "grad_norm": 0.32111111283302307,
      "learning_rate": 5.844323765192656e-05,
      "loss": 0.7518,
      "step": 4215
    },
    {
      "epoch": 0.9664491011107295,
      "grad_norm": 0.3452571630477905,
      "learning_rate": 5.837858805275408e-05,
      "loss": 0.753,
      "step": 4220
    },
    {
      "epoch": 0.9675941829840834,
      "grad_norm": 0.18496818840503693,
      "learning_rate": 5.83139384535816e-05,
      "loss": 0.7565,
      "step": 4225
    },
    {
      "epoch": 0.9687392648574373,
      "grad_norm": 0.3221565783023834,
      "learning_rate": 5.82492888544091e-05,
      "loss": 0.7534,
      "step": 4230
    },
    {
      "epoch": 0.9698843467307913,
      "grad_norm": 0.24431121349334717,
      "learning_rate": 5.818463925523662e-05,
      "loss": 0.7519,
      "step": 4235
    },
    {
      "epoch": 0.9710294286041452,
      "grad_norm": 0.21201278269290924,
      "learning_rate": 5.8119989656064134e-05,
      "loss": 0.7536,
      "step": 4240
    },
    {
      "epoch": 0.9721745104774991,
      "grad_norm": 0.1579071283340454,
      "learning_rate": 5.805534005689165e-05,
      "loss": 0.755,
      "step": 4245
    },
    {
      "epoch": 0.9733195923508531,
      "grad_norm": 0.3869153559207916,
      "learning_rate": 5.799069045771917e-05,
      "loss": 0.7513,
      "step": 4250
    },
    {
      "epoch": 0.974464674224207,
      "grad_norm": 0.175177663564682,
      "learning_rate": 5.792604085854668e-05,
      "loss": 0.7513,
      "step": 4255
    },
    {
      "epoch": 0.975609756097561,
      "grad_norm": 0.2530837655067444,
      "learning_rate": 5.78613912593742e-05,
      "loss": 0.7546,
      "step": 4260
    },
    {
      "epoch": 0.9767548379709149,
      "grad_norm": 0.28631719946861267,
      "learning_rate": 5.779674166020171e-05,
      "loss": 0.7511,
      "step": 4265
    },
    {
      "epoch": 0.9778999198442688,
      "grad_norm": 0.1455335021018982,
      "learning_rate": 5.773209206102922e-05,
      "loss": 0.7532,
      "step": 4270
    },
    {
      "epoch": 0.9790450017176228,
      "grad_norm": 0.21537651121616364,
      "learning_rate": 5.766744246185673e-05,
      "loss": 0.7511,
      "step": 4275
    },
    {
      "epoch": 0.9801900835909767,
      "grad_norm": 0.22129929065704346,
      "learning_rate": 5.7602792862684254e-05,
      "loss": 0.7567,
      "step": 4280
    },
    {
      "epoch": 0.9813351654643307,
      "grad_norm": 0.33510035276412964,
      "learning_rate": 5.753814326351177e-05,
      "loss": 0.7509,
      "step": 4285
    },
    {
      "epoch": 0.9824802473376847,
      "grad_norm": 0.31159457564353943,
      "learning_rate": 5.747349366433928e-05,
      "loss": 0.7536,
      "step": 4290
    },
    {
      "epoch": 0.9836253292110386,
      "grad_norm": 0.225836381316185,
      "learning_rate": 5.74088440651668e-05,
      "loss": 0.7546,
      "step": 4295
    },
    {
      "epoch": 0.9847704110843926,
      "grad_norm": 0.18996797502040863,
      "learning_rate": 5.734419446599432e-05,
      "loss": 0.7512,
      "step": 4300
    },
    {
      "epoch": 0.9859154929577465,
      "grad_norm": 0.19356811046600342,
      "learning_rate": 5.727954486682183e-05,
      "loss": 0.7511,
      "step": 4305
    },
    {
      "epoch": 0.9870605748311004,
      "grad_norm": 0.24983671307563782,
      "learning_rate": 5.7214895267649346e-05,
      "loss": 0.7552,
      "step": 4310
    },
    {
      "epoch": 0.9882056567044544,
      "grad_norm": 0.2550750970840454,
      "learning_rate": 5.7150245668476854e-05,
      "loss": 0.7495,
      "step": 4315
    },
    {
      "epoch": 0.9893507385778083,
      "grad_norm": 0.20238997042179108,
      "learning_rate": 5.708559606930437e-05,
      "loss": 0.7531,
      "step": 4320
    },
    {
      "epoch": 0.9904958204511622,
      "grad_norm": 0.23092573881149292,
      "learning_rate": 5.702094647013189e-05,
      "loss": 0.7508,
      "step": 4325
    },
    {
      "epoch": 0.9916409023245162,
      "grad_norm": 0.245700404047966,
      "learning_rate": 5.69562968709594e-05,
      "loss": 0.7517,
      "step": 4330
    },
    {
      "epoch": 0.9927859841978701,
      "grad_norm": 0.2939732074737549,
      "learning_rate": 5.689164727178692e-05,
      "loss": 0.7532,
      "step": 4335
    },
    {
      "epoch": 0.9939310660712241,
      "grad_norm": 0.2650054693222046,
      "learning_rate": 5.682699767261443e-05,
      "loss": 0.7556,
      "step": 4340
    },
    {
      "epoch": 0.995076147944578,
      "grad_norm": 0.18158090114593506,
      "learning_rate": 5.676234807344195e-05,
      "loss": 0.7511,
      "step": 4345
    },
    {
      "epoch": 0.9962212298179319,
      "grad_norm": 0.1984746754169464,
      "learning_rate": 5.669769847426947e-05,
      "loss": 0.755,
      "step": 4350
    },
    {
      "epoch": 0.997366311691286,
      "grad_norm": 0.2304326891899109,
      "learning_rate": 5.6633048875096974e-05,
      "loss": 0.7536,
      "step": 4355
    },
    {
      "epoch": 0.9985113935646399,
      "grad_norm": 0.20437856018543243,
      "learning_rate": 5.656839927592449e-05,
      "loss": 0.7495,
      "step": 4360
    },
    {
      "epoch": 0.9996564754379939,
      "grad_norm": 0.2961784899234772,
      "learning_rate": 5.6503749676752e-05,
      "loss": 0.7505,
      "step": 4365
    },
    {
      "epoch": 1.0006870491240123,
      "grad_norm": 0.27757421135902405,
      "learning_rate": 5.6439100077579524e-05,
      "loss": 0.7592,
      "step": 4370
    },
    {
      "epoch": 1.0018321309973663,
      "grad_norm": 0.17433343827724457,
      "learning_rate": 5.637445047840704e-05,
      "loss": 0.75,
      "step": 4375
    },
    {
      "epoch": 1.0029772128707202,
      "grad_norm": 0.22225502133369446,
      "learning_rate": 5.630980087923455e-05,
      "loss": 0.7547,
      "step": 4380
    },
    {
      "epoch": 1.0041222947440742,
      "grad_norm": 0.21817201375961304,
      "learning_rate": 5.6245151280062066e-05,
      "loss": 0.7504,
      "step": 4385
    },
    {
      "epoch": 1.005267376617428,
      "grad_norm": 0.22865532338619232,
      "learning_rate": 5.618050168088959e-05,
      "loss": 0.7488,
      "step": 4390
    },
    {
      "epoch": 1.006412458490782,
      "grad_norm": 0.24642197787761688,
      "learning_rate": 5.611585208171709e-05,
      "loss": 0.7553,
      "step": 4395
    },
    {
      "epoch": 1.0075575403641361,
      "grad_norm": 0.20586034655570984,
      "learning_rate": 5.605120248254461e-05,
      "loss": 0.7516,
      "step": 4400
    },
    {
      "epoch": 1.00870262223749,
      "grad_norm": 0.20645256340503693,
      "learning_rate": 5.598655288337212e-05,
      "loss": 0.7481,
      "step": 4405
    },
    {
      "epoch": 1.009847704110844,
      "grad_norm": 0.227728009223938,
      "learning_rate": 5.592190328419964e-05,
      "loss": 0.7531,
      "step": 4410
    },
    {
      "epoch": 1.0109927859841978,
      "grad_norm": 0.19497741758823395,
      "learning_rate": 5.585725368502716e-05,
      "loss": 0.7534,
      "step": 4415
    },
    {
      "epoch": 1.0121378678575519,
      "grad_norm": 0.24544782936573029,
      "learning_rate": 5.579260408585467e-05,
      "loss": 0.7522,
      "step": 4420
    },
    {
      "epoch": 1.0132829497309057,
      "grad_norm": 0.21906015276908875,
      "learning_rate": 5.572795448668219e-05,
      "loss": 0.7531,
      "step": 4425
    },
    {
      "epoch": 1.0144280316042598,
      "grad_norm": 0.26968297362327576,
      "learning_rate": 5.56633048875097e-05,
      "loss": 0.7501,
      "step": 4430
    },
    {
      "epoch": 1.0155731134776136,
      "grad_norm": 0.16189736127853394,
      "learning_rate": 5.559865528833722e-05,
      "loss": 0.754,
      "step": 4435
    },
    {
      "epoch": 1.0167181953509676,
      "grad_norm": 0.17086242139339447,
      "learning_rate": 5.553400568916472e-05,
      "loss": 0.7526,
      "step": 4440
    },
    {
      "epoch": 1.0178632772243215,
      "grad_norm": 0.22381778061389923,
      "learning_rate": 5.5469356089992244e-05,
      "loss": 0.7541,
      "step": 4445
    },
    {
      "epoch": 1.0190083590976755,
      "grad_norm": 0.20228935778141022,
      "learning_rate": 5.540470649081976e-05,
      "loss": 0.7502,
      "step": 4450
    },
    {
      "epoch": 1.0201534409710293,
      "grad_norm": 0.25878095626831055,
      "learning_rate": 5.534005689164727e-05,
      "loss": 0.7493,
      "step": 4455
    },
    {
      "epoch": 1.0212985228443834,
      "grad_norm": 0.22310547530651093,
      "learning_rate": 5.527540729247479e-05,
      "loss": 0.755,
      "step": 4460
    },
    {
      "epoch": 1.0224436047177372,
      "grad_norm": 0.2674236297607422,
      "learning_rate": 5.521075769330231e-05,
      "loss": 0.7532,
      "step": 4465
    },
    {
      "epoch": 1.0235886865910913,
      "grad_norm": 0.28843796253204346,
      "learning_rate": 5.514610809412982e-05,
      "loss": 0.7561,
      "step": 4470
    },
    {
      "epoch": 1.0247337684644453,
      "grad_norm": 0.3136085271835327,
      "learning_rate": 5.5081458494957336e-05,
      "loss": 0.752,
      "step": 4475
    },
    {
      "epoch": 1.0258788503377991,
      "grad_norm": 0.2548903822898865,
      "learning_rate": 5.501680889578484e-05,
      "loss": 0.7495,
      "step": 4480
    },
    {
      "epoch": 1.0270239322111532,
      "grad_norm": 0.31399279832839966,
      "learning_rate": 5.495215929661236e-05,
      "loss": 0.754,
      "step": 4485
    },
    {
      "epoch": 1.028169014084507,
      "grad_norm": 0.2506180703639984,
      "learning_rate": 5.488750969743988e-05,
      "loss": 0.7486,
      "step": 4490
    },
    {
      "epoch": 1.029314095957861,
      "grad_norm": 0.21598592400550842,
      "learning_rate": 5.482286009826739e-05,
      "loss": 0.7464,
      "step": 4495
    },
    {
      "epoch": 1.0304591778312149,
      "grad_norm": 0.19644518196582794,
      "learning_rate": 5.475821049909491e-05,
      "loss": 0.7521,
      "step": 4500
    },
    {
      "epoch": 1.031604259704569,
      "grad_norm": 0.22480544447898865,
      "learning_rate": 5.469356089992243e-05,
      "loss": 0.7579,
      "step": 4505
    },
    {
      "epoch": 1.0327493415779228,
      "grad_norm": 0.2105066180229187,
      "learning_rate": 5.462891130074994e-05,
      "loss": 0.7517,
      "step": 4510
    },
    {
      "epoch": 1.0338944234512768,
      "grad_norm": 0.4357909858226776,
      "learning_rate": 5.4564261701577456e-05,
      "loss": 0.75,
      "step": 4515
    },
    {
      "epoch": 1.0350395053246306,
      "grad_norm": 0.3007284700870514,
      "learning_rate": 5.4499612102404964e-05,
      "loss": 0.7505,
      "step": 4520
    },
    {
      "epoch": 1.0361845871979847,
      "grad_norm": 0.13870058953762054,
      "learning_rate": 5.443496250323248e-05,
      "loss": 0.75,
      "step": 4525
    },
    {
      "epoch": 1.0373296690713385,
      "grad_norm": 0.2073334902524948,
      "learning_rate": 5.437031290405999e-05,
      "loss": 0.7487,
      "step": 4530
    },
    {
      "epoch": 1.0384747509446925,
      "grad_norm": 0.21007287502288818,
      "learning_rate": 5.430566330488751e-05,
      "loss": 0.7528,
      "step": 4535
    },
    {
      "epoch": 1.0396198328180466,
      "grad_norm": 0.24089570343494415,
      "learning_rate": 5.424101370571503e-05,
      "loss": 0.752,
      "step": 4540
    },
    {
      "epoch": 1.0407649146914004,
      "grad_norm": 0.21664293110370636,
      "learning_rate": 5.417636410654254e-05,
      "loss": 0.7502,
      "step": 4545
    },
    {
      "epoch": 1.0419099965647545,
      "grad_norm": 0.1680118888616562,
      "learning_rate": 5.4111714507370056e-05,
      "loss": 0.7476,
      "step": 4550
    },
    {
      "epoch": 1.0430550784381083,
      "grad_norm": 0.19197143614292145,
      "learning_rate": 5.404706490819758e-05,
      "loss": 0.7527,
      "step": 4555
    },
    {
      "epoch": 1.0442001603114623,
      "grad_norm": 0.15469254553318024,
      "learning_rate": 5.398241530902509e-05,
      "loss": 0.7508,
      "step": 4560
    },
    {
      "epoch": 1.0453452421848162,
      "grad_norm": 0.16881108283996582,
      "learning_rate": 5.39177657098526e-05,
      "loss": 0.7534,
      "step": 4565
    },
    {
      "epoch": 1.0464903240581702,
      "grad_norm": 0.28262412548065186,
      "learning_rate": 5.385311611068011e-05,
      "loss": 0.7519,
      "step": 4570
    },
    {
      "epoch": 1.047635405931524,
      "grad_norm": 0.2931879460811615,
      "learning_rate": 5.378846651150763e-05,
      "loss": 0.7536,
      "step": 4575
    },
    {
      "epoch": 1.048780487804878,
      "grad_norm": 0.30282384157180786,
      "learning_rate": 5.372381691233515e-05,
      "loss": 0.7514,
      "step": 4580
    },
    {
      "epoch": 1.049925569678232,
      "grad_norm": 0.1762247383594513,
      "learning_rate": 5.365916731316266e-05,
      "loss": 0.7526,
      "step": 4585
    },
    {
      "epoch": 1.051070651551586,
      "grad_norm": 0.2082643061876297,
      "learning_rate": 5.3594517713990177e-05,
      "loss": 0.7507,
      "step": 4590
    },
    {
      "epoch": 1.0522157334249398,
      "grad_norm": 0.26037630438804626,
      "learning_rate": 5.352986811481769e-05,
      "loss": 0.7518,
      "step": 4595
    },
    {
      "epoch": 1.0533608152982938,
      "grad_norm": 0.1809810996055603,
      "learning_rate": 5.346521851564521e-05,
      "loss": 0.754,
      "step": 4600
    },
    {
      "epoch": 1.0545058971716477,
      "grad_norm": 0.2725500762462616,
      "learning_rate": 5.340056891647271e-05,
      "loss": 0.7515,
      "step": 4605
    },
    {
      "epoch": 1.0556509790450017,
      "grad_norm": 0.28422823548316956,
      "learning_rate": 5.3335919317300233e-05,
      "loss": 0.7485,
      "step": 4610
    },
    {
      "epoch": 1.0567960609183558,
      "grad_norm": 0.19163936376571655,
      "learning_rate": 5.327126971812775e-05,
      "loss": 0.7507,
      "step": 4615
    },
    {
      "epoch": 1.0579411427917096,
      "grad_norm": 0.19279161095619202,
      "learning_rate": 5.320662011895526e-05,
      "loss": 0.75,
      "step": 4620
    },
    {
      "epoch": 1.0590862246650636,
      "grad_norm": 0.26162973046302795,
      "learning_rate": 5.314197051978278e-05,
      "loss": 0.7525,
      "step": 4625
    },
    {
      "epoch": 1.0602313065384175,
      "grad_norm": 0.18127937614917755,
      "learning_rate": 5.30773209206103e-05,
      "loss": 0.7504,
      "step": 4630
    },
    {
      "epoch": 1.0613763884117715,
      "grad_norm": 0.16717013716697693,
      "learning_rate": 5.301267132143781e-05,
      "loss": 0.7487,
      "step": 4635
    },
    {
      "epoch": 1.0625214702851253,
      "grad_norm": 0.22202427685260773,
      "learning_rate": 5.2948021722265326e-05,
      "loss": 0.7526,
      "step": 4640
    },
    {
      "epoch": 1.0636665521584794,
      "grad_norm": 0.17828546464443207,
      "learning_rate": 5.2883372123092847e-05,
      "loss": 0.75,
      "step": 4645
    },
    {
      "epoch": 1.0648116340318332,
      "grad_norm": 0.20090904831886292,
      "learning_rate": 5.281872252392035e-05,
      "loss": 0.748,
      "step": 4650
    },
    {
      "epoch": 1.0659567159051873,
      "grad_norm": 0.2585245370864868,
      "learning_rate": 5.275407292474787e-05,
      "loss": 0.7529,
      "step": 4655
    },
    {
      "epoch": 1.067101797778541,
      "grad_norm": 0.21099530160427094,
      "learning_rate": 5.268942332557538e-05,
      "loss": 0.7531,
      "step": 4660
    },
    {
      "epoch": 1.0682468796518951,
      "grad_norm": 0.2740081548690796,
      "learning_rate": 5.26247737264029e-05,
      "loss": 0.7507,
      "step": 4665
    },
    {
      "epoch": 1.069391961525249,
      "grad_norm": 0.2878158688545227,
      "learning_rate": 5.256012412723042e-05,
      "loss": 0.7513,
      "step": 4670
    },
    {
      "epoch": 1.070537043398603,
      "grad_norm": 0.1615872085094452,
      "learning_rate": 5.249547452805793e-05,
      "loss": 0.7505,
      "step": 4675
    },
    {
      "epoch": 1.0716821252719568,
      "grad_norm": 0.18126648664474487,
      "learning_rate": 5.2430824928885446e-05,
      "loss": 0.7493,
      "step": 4680
    },
    {
      "epoch": 1.0728272071453109,
      "grad_norm": 0.1734725534915924,
      "learning_rate": 5.236617532971296e-05,
      "loss": 0.7496,
      "step": 4685
    },
    {
      "epoch": 1.073972289018665,
      "grad_norm": 0.32179296016693115,
      "learning_rate": 5.230152573054047e-05,
      "loss": 0.7504,
      "step": 4690
    },
    {
      "epoch": 1.0751173708920188,
      "grad_norm": 0.37920457124710083,
      "learning_rate": 5.223687613136798e-05,
      "loss": 0.7471,
      "step": 4695
    },
    {
      "epoch": 1.0762624527653728,
      "grad_norm": 0.36169323325157166,
      "learning_rate": 5.21722265321955e-05,
      "loss": 0.7507,
      "step": 4700
    },
    {
      "epoch": 1.0774075346387266,
      "grad_norm": 0.22468143701553345,
      "learning_rate": 5.210757693302302e-05,
      "loss": 0.7534,
      "step": 4705
    },
    {
      "epoch": 1.0785526165120807,
      "grad_norm": 0.20241256058216095,
      "learning_rate": 5.204292733385053e-05,
      "loss": 0.7542,
      "step": 4710
    },
    {
      "epoch": 1.0796976983854345,
      "grad_norm": 0.24470794200897217,
      "learning_rate": 5.197827773467805e-05,
      "loss": 0.7513,
      "step": 4715
    },
    {
      "epoch": 1.0808427802587885,
      "grad_norm": 0.22353743016719818,
      "learning_rate": 5.1913628135505567e-05,
      "loss": 0.7493,
      "step": 4720
    },
    {
      "epoch": 1.0819878621321424,
      "grad_norm": 0.1666325479745865,
      "learning_rate": 5.184897853633308e-05,
      "loss": 0.7505,
      "step": 4725
    },
    {
      "epoch": 1.0831329440054964,
      "grad_norm": 0.23964060842990875,
      "learning_rate": 5.178432893716059e-05,
      "loss": 0.7515,
      "step": 4730
    },
    {
      "epoch": 1.0842780258788502,
      "grad_norm": 0.2576775550842285,
      "learning_rate": 5.17196793379881e-05,
      "loss": 0.753,
      "step": 4735
    },
    {
      "epoch": 1.0854231077522043,
      "grad_norm": 0.2827492952346802,
      "learning_rate": 5.165502973881562e-05,
      "loss": 0.7527,
      "step": 4740
    },
    {
      "epoch": 1.0865681896255581,
      "grad_norm": 0.2897157371044159,
      "learning_rate": 5.159038013964314e-05,
      "loss": 0.7508,
      "step": 4745
    },
    {
      "epoch": 1.0877132714989122,
      "grad_norm": 0.28682994842529297,
      "learning_rate": 5.152573054047065e-05,
      "loss": 0.7525,
      "step": 4750
    },
    {
      "epoch": 1.0888583533722662,
      "grad_norm": 0.14168643951416016,
      "learning_rate": 5.1461080941298166e-05,
      "loss": 0.7547,
      "step": 4755
    },
    {
      "epoch": 1.09000343524562,
      "grad_norm": 0.2121264487504959,
      "learning_rate": 5.139643134212569e-05,
      "loss": 0.7494,
      "step": 4760
    },
    {
      "epoch": 1.091148517118974,
      "grad_norm": 0.2246008962392807,
      "learning_rate": 5.13317817429532e-05,
      "loss": 0.7512,
      "step": 4765
    },
    {
      "epoch": 1.092293598992328,
      "grad_norm": 0.18789461255073547,
      "learning_rate": 5.1267132143780716e-05,
      "loss": 0.7499,
      "step": 4770
    },
    {
      "epoch": 1.093438680865682,
      "grad_norm": 0.2336086481809616,
      "learning_rate": 5.120248254460822e-05,
      "loss": 0.7478,
      "step": 4775
    },
    {
      "epoch": 1.0945837627390358,
      "grad_norm": 0.23505932092666626,
      "learning_rate": 5.113783294543574e-05,
      "loss": 0.7483,
      "step": 4780
    },
    {
      "epoch": 1.0957288446123898,
      "grad_norm": 0.24164508283138275,
      "learning_rate": 5.107318334626325e-05,
      "loss": 0.7524,
      "step": 4785
    },
    {
      "epoch": 1.0968739264857437,
      "grad_norm": 0.22491183876991272,
      "learning_rate": 5.100853374709077e-05,
      "loss": 0.7502,
      "step": 4790
    },
    {
      "epoch": 1.0980190083590977,
      "grad_norm": 0.17309866845607758,
      "learning_rate": 5.094388414791829e-05,
      "loss": 0.751,
      "step": 4795
    },
    {
      "epoch": 1.0991640902324515,
      "grad_norm": 0.20419526100158691,
      "learning_rate": 5.08792345487458e-05,
      "loss": 0.7462,
      "step": 4800
    },
    {
      "epoch": 1.1003091721058056,
      "grad_norm": 0.22812721133232117,
      "learning_rate": 5.0814584949573315e-05,
      "loss": 0.7499,
      "step": 4805
    },
    {
      "epoch": 1.1014542539791594,
      "grad_norm": 0.2595369815826416,
      "learning_rate": 5.0749935350400836e-05,
      "loss": 0.7511,
      "step": 4810
    },
    {
      "epoch": 1.1025993358525135,
      "grad_norm": 0.2934839427471161,
      "learning_rate": 5.068528575122834e-05,
      "loss": 0.7512,
      "step": 4815
    },
    {
      "epoch": 1.1037444177258675,
      "grad_norm": 0.2313292771577835,
      "learning_rate": 5.062063615205586e-05,
      "loss": 0.7491,
      "step": 4820
    },
    {
      "epoch": 1.1048894995992213,
      "grad_norm": 0.2503287196159363,
      "learning_rate": 5.055598655288337e-05,
      "loss": 0.7503,
      "step": 4825
    },
    {
      "epoch": 1.1060345814725754,
      "grad_norm": 0.16885490715503693,
      "learning_rate": 5.0491336953710886e-05,
      "loss": 0.7547,
      "step": 4830
    },
    {
      "epoch": 1.1071796633459292,
      "grad_norm": 0.23992787301540375,
      "learning_rate": 5.042668735453841e-05,
      "loss": 0.7535,
      "step": 4835
    },
    {
      "epoch": 1.1083247452192833,
      "grad_norm": 0.1906599998474121,
      "learning_rate": 5.036203775536592e-05,
      "loss": 0.7516,
      "step": 4840
    },
    {
      "epoch": 1.109469827092637,
      "grad_norm": 0.15855015814304352,
      "learning_rate": 5.0297388156193436e-05,
      "loss": 0.7478,
      "step": 4845
    },
    {
      "epoch": 1.1106149089659911,
      "grad_norm": 0.1988474577665329,
      "learning_rate": 5.023273855702095e-05,
      "loss": 0.7506,
      "step": 4850
    },
    {
      "epoch": 1.111759990839345,
      "grad_norm": 0.1842668205499649,
      "learning_rate": 5.016808895784847e-05,
      "loss": 0.7515,
      "step": 4855
    },
    {
      "epoch": 1.112905072712699,
      "grad_norm": 0.18278862535953522,
      "learning_rate": 5.010343935867597e-05,
      "loss": 0.7539,
      "step": 4860
    },
    {
      "epoch": 1.1140501545860528,
      "grad_norm": 0.18607094883918762,
      "learning_rate": 5.003878975950349e-05,
      "loss": 0.7488,
      "step": 4865
    },
    {
      "epoch": 1.1151952364594069,
      "grad_norm": 0.21544776856899261,
      "learning_rate": 4.997414016033101e-05,
      "loss": 0.7511,
      "step": 4870
    },
    {
      "epoch": 1.1163403183327607,
      "grad_norm": 0.21455517411231995,
      "learning_rate": 4.990949056115852e-05,
      "loss": 0.7526,
      "step": 4875
    },
    {
      "epoch": 1.1174854002061148,
      "grad_norm": 0.2741033434867859,
      "learning_rate": 4.984484096198604e-05,
      "loss": 0.7451,
      "step": 4880
    },
    {
      "epoch": 1.1186304820794686,
      "grad_norm": 0.22744566202163696,
      "learning_rate": 4.9780191362813556e-05,
      "loss": 0.7521,
      "step": 4885
    },
    {
      "epoch": 1.1197755639528226,
      "grad_norm": 0.18673773109912872,
      "learning_rate": 4.9715541763641064e-05,
      "loss": 0.7506,
      "step": 4890
    },
    {
      "epoch": 1.1209206458261765,
      "grad_norm": 0.19755394756793976,
      "learning_rate": 4.9650892164468585e-05,
      "loss": 0.7477,
      "step": 4895
    },
    {
      "epoch": 1.1220657276995305,
      "grad_norm": 0.21924807131290436,
      "learning_rate": 4.95862425652961e-05,
      "loss": 0.7469,
      "step": 4900
    },
    {
      "epoch": 1.1232108095728845,
      "grad_norm": 0.17516666650772095,
      "learning_rate": 4.952159296612361e-05,
      "loss": 0.7499,
      "step": 4905
    },
    {
      "epoch": 1.1243558914462384,
      "grad_norm": 0.18603424727916718,
      "learning_rate": 4.945694336695113e-05,
      "loss": 0.7494,
      "step": 4910
    },
    {
      "epoch": 1.1255009733195924,
      "grad_norm": 0.19065344333648682,
      "learning_rate": 4.939229376777864e-05,
      "loss": 0.748,
      "step": 4915
    },
    {
      "epoch": 1.1266460551929462,
      "grad_norm": 0.2856319546699524,
      "learning_rate": 4.9327644168606156e-05,
      "loss": 0.7501,
      "step": 4920
    },
    {
      "epoch": 1.1277911370663003,
      "grad_norm": 0.21050098538398743,
      "learning_rate": 4.926299456943368e-05,
      "loss": 0.7499,
      "step": 4925
    },
    {
      "epoch": 1.1289362189396541,
      "grad_norm": 0.19090114533901215,
      "learning_rate": 4.9198344970261184e-05,
      "loss": 0.7546,
      "step": 4930
    },
    {
      "epoch": 1.1300813008130082,
      "grad_norm": 0.2405533790588379,
      "learning_rate": 4.91336953710887e-05,
      "loss": 0.7514,
      "step": 4935
    },
    {
      "epoch": 1.131226382686362,
      "grad_norm": 0.18367047607898712,
      "learning_rate": 4.906904577191622e-05,
      "loss": 0.7486,
      "step": 4940
    },
    {
      "epoch": 1.132371464559716,
      "grad_norm": 0.18685941398143768,
      "learning_rate": 4.9004396172743734e-05,
      "loss": 0.7513,
      "step": 4945
    },
    {
      "epoch": 1.1335165464330699,
      "grad_norm": 0.18659988045692444,
      "learning_rate": 4.893974657357124e-05,
      "loss": 0.7511,
      "step": 4950
    },
    {
      "epoch": 1.134661628306424,
      "grad_norm": 0.22466859221458435,
      "learning_rate": 4.887509697439876e-05,
      "loss": 0.7478,
      "step": 4955
    },
    {
      "epoch": 1.1358067101797777,
      "grad_norm": 0.2367299497127533,
      "learning_rate": 4.8810447375226276e-05,
      "loss": 0.7518,
      "step": 4960
    },
    {
      "epoch": 1.1369517920531318,
      "grad_norm": 0.17893458902835846,
      "learning_rate": 4.874579777605379e-05,
      "loss": 0.7492,
      "step": 4965
    },
    {
      "epoch": 1.1380968739264858,
      "grad_norm": 0.22315214574337006,
      "learning_rate": 4.868114817688131e-05,
      "loss": 0.746,
      "step": 4970
    },
    {
      "epoch": 1.1392419557998397,
      "grad_norm": 0.24147474765777588,
      "learning_rate": 4.861649857770882e-05,
      "loss": 0.7509,
      "step": 4975
    },
    {
      "epoch": 1.1403870376731937,
      "grad_norm": 0.17804847657680511,
      "learning_rate": 4.855184897853633e-05,
      "loss": 0.7546,
      "step": 4980
    },
    {
      "epoch": 1.1415321195465475,
      "grad_norm": 0.17471164464950562,
      "learning_rate": 4.8487199379363854e-05,
      "loss": 0.7467,
      "step": 4985
    },
    {
      "epoch": 1.1426772014199016,
      "grad_norm": 0.21423661708831787,
      "learning_rate": 4.842254978019137e-05,
      "loss": 0.7511,
      "step": 4990
    },
    {
      "epoch": 1.1438222832932554,
      "grad_norm": 0.21214281022548676,
      "learning_rate": 4.8357900181018876e-05,
      "loss": 0.7453,
      "step": 4995
    },
    {
      "epoch": 1.1449673651666095,
      "grad_norm": 0.1819201558828354,
      "learning_rate": 4.82932505818464e-05,
      "loss": 0.7513,
      "step": 5000
    },
    {
      "epoch": 1.1461124470399633,
      "grad_norm": 0.17857210338115692,
      "learning_rate": 4.822860098267391e-05,
      "loss": 0.7504,
      "step": 5005
    },
    {
      "epoch": 1.1472575289133173,
      "grad_norm": 0.22314776480197906,
      "learning_rate": 4.8163951383501425e-05,
      "loss": 0.7535,
      "step": 5010
    },
    {
      "epoch": 1.1484026107866712,
      "grad_norm": 0.16417430341243744,
      "learning_rate": 4.809930178432894e-05,
      "loss": 0.7516,
      "step": 5015
    },
    {
      "epoch": 1.1495476926600252,
      "grad_norm": 0.16635441780090332,
      "learning_rate": 4.8034652185156454e-05,
      "loss": 0.7499,
      "step": 5020
    },
    {
      "epoch": 1.150692774533379,
      "grad_norm": 0.254543662071228,
      "learning_rate": 4.797000258598397e-05,
      "loss": 0.75,
      "step": 5025
    },
    {
      "epoch": 1.151837856406733,
      "grad_norm": 0.17547422647476196,
      "learning_rate": 4.790535298681149e-05,
      "loss": 0.752,
      "step": 5030
    },
    {
      "epoch": 1.1529829382800871,
      "grad_norm": 0.18553274869918823,
      "learning_rate": 4.7840703387638996e-05,
      "loss": 0.7505,
      "step": 5035
    },
    {
      "epoch": 1.154128020153441,
      "grad_norm": 0.24924322962760925,
      "learning_rate": 4.777605378846651e-05,
      "loss": 0.7488,
      "step": 5040
    },
    {
      "epoch": 1.155273102026795,
      "grad_norm": 0.3234774172306061,
      "learning_rate": 4.771140418929403e-05,
      "loss": 0.7501,
      "step": 5045
    },
    {
      "epoch": 1.1564181839001488,
      "grad_norm": 0.2411537766456604,
      "learning_rate": 4.7646754590121546e-05,
      "loss": 0.7476,
      "step": 5050
    },
    {
      "epoch": 1.1575632657735029,
      "grad_norm": 0.21320085227489471,
      "learning_rate": 4.758210499094905e-05,
      "loss": 0.7512,
      "step": 5055
    },
    {
      "epoch": 1.1587083476468567,
      "grad_norm": 0.20281052589416504,
      "learning_rate": 4.7517455391776574e-05,
      "loss": 0.7498,
      "step": 5060
    },
    {
      "epoch": 1.1598534295202108,
      "grad_norm": 0.1915905475616455,
      "learning_rate": 4.745280579260409e-05,
      "loss": 0.7465,
      "step": 5065
    },
    {
      "epoch": 1.1609985113935646,
      "grad_norm": 0.17410551011562347,
      "learning_rate": 4.73881561934316e-05,
      "loss": 0.7512,
      "step": 5070
    },
    {
      "epoch": 1.1621435932669186,
      "grad_norm": 0.16328193247318268,
      "learning_rate": 4.7323506594259124e-05,
      "loss": 0.7527,
      "step": 5075
    },
    {
      "epoch": 1.1632886751402725,
      "grad_norm": 0.23142258822917938,
      "learning_rate": 4.725885699508663e-05,
      "loss": 0.7484,
      "step": 5080
    },
    {
      "epoch": 1.1644337570136265,
      "grad_norm": 0.3409397006034851,
      "learning_rate": 4.7194207395914145e-05,
      "loss": 0.7464,
      "step": 5085
    },
    {
      "epoch": 1.1655788388869803,
      "grad_norm": 0.29608339071273804,
      "learning_rate": 4.7129557796741666e-05,
      "loss": 0.7493,
      "step": 5090
    },
    {
      "epoch": 1.1667239207603344,
      "grad_norm": 0.25607064366340637,
      "learning_rate": 4.706490819756918e-05,
      "loss": 0.7495,
      "step": 5095
    },
    {
      "epoch": 1.1678690026336884,
      "grad_norm": 0.21050728857517242,
      "learning_rate": 4.700025859839669e-05,
      "loss": 0.7517,
      "step": 5100
    },
    {
      "epoch": 1.1690140845070423,
      "grad_norm": 0.2006913423538208,
      "learning_rate": 4.693560899922421e-05,
      "loss": 0.7498,
      "step": 5105
    },
    {
      "epoch": 1.170159166380396,
      "grad_norm": 0.1534210741519928,
      "learning_rate": 4.687095940005172e-05,
      "loss": 0.749,
      "step": 5110
    },
    {
      "epoch": 1.1713042482537501,
      "grad_norm": 0.21326866745948792,
      "learning_rate": 4.680630980087924e-05,
      "loss": 0.753,
      "step": 5115
    },
    {
      "epoch": 1.1724493301271042,
      "grad_norm": 0.21120962500572205,
      "learning_rate": 4.674166020170675e-05,
      "loss": 0.7518,
      "step": 5120
    },
    {
      "epoch": 1.173594412000458,
      "grad_norm": 0.1823272407054901,
      "learning_rate": 4.6677010602534266e-05,
      "loss": 0.7484,
      "step": 5125
    },
    {
      "epoch": 1.174739493873812,
      "grad_norm": 0.19098564982414246,
      "learning_rate": 4.661236100336178e-05,
      "loss": 0.7551,
      "step": 5130
    },
    {
      "epoch": 1.1758845757471659,
      "grad_norm": 0.17055176198482513,
      "learning_rate": 4.65477114041893e-05,
      "loss": 0.7512,
      "step": 5135
    },
    {
      "epoch": 1.17702965762052,
      "grad_norm": 0.23654793202877045,
      "learning_rate": 4.648306180501681e-05,
      "loss": 0.7471,
      "step": 5140
    },
    {
      "epoch": 1.1781747394938737,
      "grad_norm": 0.18556205928325653,
      "learning_rate": 4.641841220584432e-05,
      "loss": 0.7527,
      "step": 5145
    },
    {
      "epoch": 1.1793198213672278,
      "grad_norm": 0.2395789623260498,
      "learning_rate": 4.6353762606671844e-05,
      "loss": 0.7502,
      "step": 5150
    },
    {
      "epoch": 1.1804649032405816,
      "grad_norm": 0.23787091672420502,
      "learning_rate": 4.628911300749936e-05,
      "loss": 0.7485,
      "step": 5155
    },
    {
      "epoch": 1.1816099851139357,
      "grad_norm": 0.1994669884443283,
      "learning_rate": 4.6224463408326865e-05,
      "loss": 0.7516,
      "step": 5160
    },
    {
      "epoch": 1.1827550669872897,
      "grad_norm": 0.218184232711792,
      "learning_rate": 4.6159813809154386e-05,
      "loss": 0.7542,
      "step": 5165
    },
    {
      "epoch": 1.1839001488606435,
      "grad_norm": 0.16488322615623474,
      "learning_rate": 4.60951642099819e-05,
      "loss": 0.7472,
      "step": 5170
    },
    {
      "epoch": 1.1850452307339974,
      "grad_norm": 0.2295355200767517,
      "learning_rate": 4.6030514610809415e-05,
      "loss": 0.7485,
      "step": 5175
    },
    {
      "epoch": 1.1861903126073514,
      "grad_norm": 0.19770926237106323,
      "learning_rate": 4.5965865011636936e-05,
      "loss": 0.7515,
      "step": 5180
    },
    {
      "epoch": 1.1873353944807055,
      "grad_norm": 0.20445016026496887,
      "learning_rate": 4.590121541246444e-05,
      "loss": 0.7492,
      "step": 5185
    },
    {
      "epoch": 1.1884804763540593,
      "grad_norm": 0.23801274597644806,
      "learning_rate": 4.583656581329196e-05,
      "loss": 0.7482,
      "step": 5190
    },
    {
      "epoch": 1.1896255582274133,
      "grad_norm": 0.22532589733600616,
      "learning_rate": 4.577191621411948e-05,
      "loss": 0.7491,
      "step": 5195
    },
    {
      "epoch": 1.1907706401007672,
      "grad_norm": 0.2098410725593567,
      "learning_rate": 4.570726661494699e-05,
      "loss": 0.7498,
      "step": 5200
    },
    {
      "epoch": 1.1919157219741212,
      "grad_norm": 0.17258141934871674,
      "learning_rate": 4.56426170157745e-05,
      "loss": 0.7464,
      "step": 5205
    },
    {
      "epoch": 1.193060803847475,
      "grad_norm": 0.23425732553005219,
      "learning_rate": 4.557796741660202e-05,
      "loss": 0.7486,
      "step": 5210
    },
    {
      "epoch": 1.194205885720829,
      "grad_norm": 0.32258808612823486,
      "learning_rate": 4.5513317817429535e-05,
      "loss": 0.7517,
      "step": 5215
    },
    {
      "epoch": 1.195350967594183,
      "grad_norm": 0.29464104771614075,
      "learning_rate": 4.544866821825705e-05,
      "loss": 0.7469,
      "step": 5220
    },
    {
      "epoch": 1.196496049467537,
      "grad_norm": 0.2603786587715149,
      "learning_rate": 4.5384018619084564e-05,
      "loss": 0.7532,
      "step": 5225
    },
    {
      "epoch": 1.1976411313408908,
      "grad_norm": 0.24306389689445496,
      "learning_rate": 4.531936901991208e-05,
      "loss": 0.7481,
      "step": 5230
    },
    {
      "epoch": 1.1987862132142448,
      "grad_norm": 0.2622597813606262,
      "learning_rate": 4.525471942073959e-05,
      "loss": 0.748,
      "step": 5235
    },
    {
      "epoch": 1.1999312950875987,
      "grad_norm": 0.19283488392829895,
      "learning_rate": 4.519006982156711e-05,
      "loss": 0.7476,
      "step": 5240
    },
    {
      "epoch": 1.2010763769609527,
      "grad_norm": 0.22162917256355286,
      "learning_rate": 4.512542022239462e-05,
      "loss": 0.7489,
      "step": 5245
    },
    {
      "epoch": 1.2022214588343068,
      "grad_norm": 0.2552875280380249,
      "learning_rate": 4.5060770623222135e-05,
      "loss": 0.7471,
      "step": 5250
    },
    {
      "epoch": 1.2033665407076606,
      "grad_norm": 0.19390220940113068,
      "learning_rate": 4.4996121024049656e-05,
      "loss": 0.7454,
      "step": 5255
    },
    {
      "epoch": 1.2045116225810146,
      "grad_norm": 0.19998888671398163,
      "learning_rate": 4.493147142487717e-05,
      "loss": 0.7462,
      "step": 5260
    },
    {
      "epoch": 1.2056567044543685,
      "grad_norm": 0.14607127010822296,
      "learning_rate": 4.486682182570468e-05,
      "loss": 0.7489,
      "step": 5265
    },
    {
      "epoch": 1.2068017863277225,
      "grad_norm": 0.17341439425945282,
      "learning_rate": 4.48021722265322e-05,
      "loss": 0.7495,
      "step": 5270
    },
    {
      "epoch": 1.2079468682010763,
      "grad_norm": 0.1995754837989807,
      "learning_rate": 4.473752262735971e-05,
      "loss": 0.7482,
      "step": 5275
    },
    {
      "epoch": 1.2090919500744304,
      "grad_norm": 0.24299126863479614,
      "learning_rate": 4.467287302818723e-05,
      "loss": 0.751,
      "step": 5280
    },
    {
      "epoch": 1.2102370319477842,
      "grad_norm": 0.24882306158542633,
      "learning_rate": 4.460822342901475e-05,
      "loss": 0.7521,
      "step": 5285
    },
    {
      "epoch": 1.2113821138211383,
      "grad_norm": 0.2369246929883957,
      "learning_rate": 4.4543573829842255e-05,
      "loss": 0.749,
      "step": 5290
    },
    {
      "epoch": 1.212527195694492,
      "grad_norm": 0.16573785245418549,
      "learning_rate": 4.447892423066977e-05,
      "loss": 0.7513,
      "step": 5295
    },
    {
      "epoch": 1.2136722775678461,
      "grad_norm": 0.25582486391067505,
      "learning_rate": 4.441427463149729e-05,
      "loss": 0.7494,
      "step": 5300
    },
    {
      "epoch": 1.2148173594412,
      "grad_norm": 0.2501291036605835,
      "learning_rate": 4.4349625032324805e-05,
      "loss": 0.7454,
      "step": 5305
    },
    {
      "epoch": 1.215962441314554,
      "grad_norm": 0.25190266966819763,
      "learning_rate": 4.428497543315231e-05,
      "loss": 0.7508,
      "step": 5310
    },
    {
      "epoch": 1.217107523187908,
      "grad_norm": 0.2685193121433258,
      "learning_rate": 4.422032583397983e-05,
      "loss": 0.7515,
      "step": 5315
    },
    {
      "epoch": 1.2182526050612619,
      "grad_norm": 0.21009624004364014,
      "learning_rate": 4.415567623480735e-05,
      "loss": 0.7524,
      "step": 5320
    },
    {
      "epoch": 1.219397686934616,
      "grad_norm": 0.17537476122379303,
      "learning_rate": 4.409102663563486e-05,
      "loss": 0.7456,
      "step": 5325
    },
    {
      "epoch": 1.2205427688079697,
      "grad_norm": 0.21760603785514832,
      "learning_rate": 4.4026377036462376e-05,
      "loss": 0.7504,
      "step": 5330
    },
    {
      "epoch": 1.2216878506813238,
      "grad_norm": 0.2105751931667328,
      "learning_rate": 4.396172743728989e-05,
      "loss": 0.7458,
      "step": 5335
    },
    {
      "epoch": 1.2228329325546776,
      "grad_norm": 0.3550606667995453,
      "learning_rate": 4.3897077838117404e-05,
      "loss": 0.7498,
      "step": 5340
    },
    {
      "epoch": 1.2239780144280317,
      "grad_norm": 0.22760871052742004,
      "learning_rate": 4.3832428238944925e-05,
      "loss": 0.7445,
      "step": 5345
    },
    {
      "epoch": 1.2251230963013855,
      "grad_norm": 0.2754494249820709,
      "learning_rate": 4.376777863977243e-05,
      "loss": 0.7484,
      "step": 5350
    },
    {
      "epoch": 1.2262681781747395,
      "grad_norm": 0.3162951171398163,
      "learning_rate": 4.370312904059995e-05,
      "loss": 0.7461,
      "step": 5355
    },
    {
      "epoch": 1.2274132600480934,
      "grad_norm": 0.30421894788742065,
      "learning_rate": 4.363847944142747e-05,
      "loss": 0.7468,
      "step": 5360
    },
    {
      "epoch": 1.2285583419214474,
      "grad_norm": 0.22896705567836761,
      "learning_rate": 4.357382984225498e-05,
      "loss": 0.7485,
      "step": 5365
    },
    {
      "epoch": 1.2297034237948012,
      "grad_norm": 0.2728241980075836,
      "learning_rate": 4.350918024308249e-05,
      "loss": 0.7482,
      "step": 5370
    },
    {
      "epoch": 1.2308485056681553,
      "grad_norm": 0.18981365859508514,
      "learning_rate": 4.344453064391001e-05,
      "loss": 0.7493,
      "step": 5375
    },
    {
      "epoch": 1.2319935875415093,
      "grad_norm": 0.24496853351593018,
      "learning_rate": 4.3379881044737525e-05,
      "loss": 0.7466,
      "step": 5380
    },
    {
      "epoch": 1.2331386694148632,
      "grad_norm": 0.20402029156684875,
      "learning_rate": 4.331523144556504e-05,
      "loss": 0.7522,
      "step": 5385
    },
    {
      "epoch": 1.234283751288217,
      "grad_norm": 0.21166785061359406,
      "learning_rate": 4.325058184639256e-05,
      "loss": 0.7484,
      "step": 5390
    },
    {
      "epoch": 1.235428833161571,
      "grad_norm": 0.19772662222385406,
      "learning_rate": 4.318593224722007e-05,
      "loss": 0.7492,
      "step": 5395
    },
    {
      "epoch": 1.236573915034925,
      "grad_norm": 0.18771900236606598,
      "learning_rate": 4.312128264804758e-05,
      "loss": 0.7473,
      "step": 5400
    },
    {
      "epoch": 1.237718996908279,
      "grad_norm": 0.19199180603027344,
      "learning_rate": 4.30566330488751e-05,
      "loss": 0.7474,
      "step": 5405
    },
    {
      "epoch": 1.238864078781633,
      "grad_norm": 0.2975488603115082,
      "learning_rate": 4.299198344970262e-05,
      "loss": 0.7519,
      "step": 5410
    },
    {
      "epoch": 1.2400091606549868,
      "grad_norm": 0.28295421600341797,
      "learning_rate": 4.2927333850530124e-05,
      "loss": 0.7485,
      "step": 5415
    },
    {
      "epoch": 1.2411542425283408,
      "grad_norm": 0.18072207272052765,
      "learning_rate": 4.2862684251357645e-05,
      "loss": 0.7446,
      "step": 5420
    },
    {
      "epoch": 1.2422993244016947,
      "grad_norm": 0.17685651779174805,
      "learning_rate": 4.279803465218516e-05,
      "loss": 0.7469,
      "step": 5425
    },
    {
      "epoch": 1.2434444062750487,
      "grad_norm": 0.16913194954395294,
      "learning_rate": 4.2733385053012674e-05,
      "loss": 0.7466,
      "step": 5430
    },
    {
      "epoch": 1.2445894881484025,
      "grad_norm": 0.2549387514591217,
      "learning_rate": 4.266873545384019e-05,
      "loss": 0.7487,
      "step": 5435
    },
    {
      "epoch": 1.2457345700217566,
      "grad_norm": 0.18447519838809967,
      "learning_rate": 4.26040858546677e-05,
      "loss": 0.7464,
      "step": 5440
    },
    {
      "epoch": 1.2468796518951104,
      "grad_norm": 0.28448691964149475,
      "learning_rate": 4.2539436255495217e-05,
      "loss": 0.7527,
      "step": 5445
    },
    {
      "epoch": 1.2480247337684645,
      "grad_norm": 0.15298114717006683,
      "learning_rate": 4.247478665632274e-05,
      "loss": 0.7492,
      "step": 5450
    },
    {
      "epoch": 1.2491698156418183,
      "grad_norm": 0.24181102216243744,
      "learning_rate": 4.2410137057150245e-05,
      "loss": 0.7444,
      "step": 5455
    },
    {
      "epoch": 1.2503148975151723,
      "grad_norm": 0.2764556109905243,
      "learning_rate": 4.234548745797776e-05,
      "loss": 0.7461,
      "step": 5460
    },
    {
      "epoch": 1.2514599793885264,
      "grad_norm": 0.23078586161136627,
      "learning_rate": 4.228083785880528e-05,
      "loss": 0.7471,
      "step": 5465
    },
    {
      "epoch": 1.2526050612618802,
      "grad_norm": 0.18252263963222504,
      "learning_rate": 4.2216188259632794e-05,
      "loss": 0.7476,
      "step": 5470
    },
    {
      "epoch": 1.2537501431352343,
      "grad_norm": 0.23740258812904358,
      "learning_rate": 4.21515386604603e-05,
      "loss": 0.7491,
      "step": 5475
    },
    {
      "epoch": 1.254895225008588,
      "grad_norm": 0.22824452817440033,
      "learning_rate": 4.208688906128782e-05,
      "loss": 0.7475,
      "step": 5480
    },
    {
      "epoch": 1.2560403068819421,
      "grad_norm": 0.3317946195602417,
      "learning_rate": 4.202223946211534e-05,
      "loss": 0.7441,
      "step": 5485
    },
    {
      "epoch": 1.257185388755296,
      "grad_norm": 0.19173449277877808,
      "learning_rate": 4.195758986294285e-05,
      "loss": 0.7484,
      "step": 5490
    },
    {
      "epoch": 1.25833047062865,
      "grad_norm": 0.2722282111644745,
      "learning_rate": 4.189294026377037e-05,
      "loss": 0.7473,
      "step": 5495
    },
    {
      "epoch": 1.2594755525020038,
      "grad_norm": 0.21009880304336548,
      "learning_rate": 4.182829066459788e-05,
      "loss": 0.7481,
      "step": 5500
    },
    {
      "epoch": 1.2606206343753579,
      "grad_norm": 0.25598183274269104,
      "learning_rate": 4.1763641065425394e-05,
      "loss": 0.7466,
      "step": 5505
    },
    {
      "epoch": 1.261765716248712,
      "grad_norm": 0.17408539354801178,
      "learning_rate": 4.1698991466252915e-05,
      "loss": 0.7497,
      "step": 5510
    },
    {
      "epoch": 1.2629107981220657,
      "grad_norm": 0.1897374391555786,
      "learning_rate": 4.163434186708043e-05,
      "loss": 0.747,
      "step": 5515
    },
    {
      "epoch": 1.2640558799954196,
      "grad_norm": 0.18324466049671173,
      "learning_rate": 4.1569692267907937e-05,
      "loss": 0.7476,
      "step": 5520
    },
    {
      "epoch": 1.2652009618687736,
      "grad_norm": 0.21715322136878967,
      "learning_rate": 4.150504266873546e-05,
      "loss": 0.747,
      "step": 5525
    },
    {
      "epoch": 1.2663460437421277,
      "grad_norm": 0.21431036293506622,
      "learning_rate": 4.144039306956297e-05,
      "loss": 0.7496,
      "step": 5530
    },
    {
      "epoch": 1.2674911256154815,
      "grad_norm": 0.26471084356307983,
      "learning_rate": 4.1375743470390486e-05,
      "loss": 0.7476,
      "step": 5535
    },
    {
      "epoch": 1.2686362074888353,
      "grad_norm": 0.14690008759498596,
      "learning_rate": 4.1311093871218e-05,
      "loss": 0.7527,
      "step": 5540
    },
    {
      "epoch": 1.2697812893621894,
      "grad_norm": 0.15945012867450714,
      "learning_rate": 4.1246444272045514e-05,
      "loss": 0.7463,
      "step": 5545
    },
    {
      "epoch": 1.2709263712355434,
      "grad_norm": 0.22661566734313965,
      "learning_rate": 4.118179467287303e-05,
      "loss": 0.7481,
      "step": 5550
    },
    {
      "epoch": 1.2720714531088972,
      "grad_norm": 0.19860315322875977,
      "learning_rate": 4.111714507370055e-05,
      "loss": 0.747,
      "step": 5555
    },
    {
      "epoch": 1.2732165349822513,
      "grad_norm": 0.23950666189193726,
      "learning_rate": 4.105249547452806e-05,
      "loss": 0.7447,
      "step": 5560
    },
    {
      "epoch": 1.2743616168556051,
      "grad_norm": 0.2150084674358368,
      "learning_rate": 4.098784587535557e-05,
      "loss": 0.7505,
      "step": 5565
    },
    {
      "epoch": 1.2755066987289592,
      "grad_norm": 0.24051061272621155,
      "learning_rate": 4.092319627618309e-05,
      "loss": 0.7494,
      "step": 5570
    },
    {
      "epoch": 1.276651780602313,
      "grad_norm": 0.19743932783603668,
      "learning_rate": 4.0858546677010607e-05,
      "loss": 0.7519,
      "step": 5575
    },
    {
      "epoch": 1.277796862475667,
      "grad_norm": 0.17389178276062012,
      "learning_rate": 4.0793897077838114e-05,
      "loss": 0.7521,
      "step": 5580
    },
    {
      "epoch": 1.2789419443490209,
      "grad_norm": 0.19800211489200592,
      "learning_rate": 4.0729247478665635e-05,
      "loss": 0.7524,
      "step": 5585
    },
    {
      "epoch": 1.280087026222375,
      "grad_norm": 0.21809372305870056,
      "learning_rate": 4.066459787949315e-05,
      "loss": 0.7489,
      "step": 5590
    },
    {
      "epoch": 1.281232108095729,
      "grad_norm": 0.1831665188074112,
      "learning_rate": 4.0599948280320663e-05,
      "loss": 0.7453,
      "step": 5595
    },
    {
      "epoch": 1.2823771899690828,
      "grad_norm": 0.1886046826839447,
      "learning_rate": 4.053529868114818e-05,
      "loss": 0.7483,
      "step": 5600
    },
    {
      "epoch": 1.2835222718424366,
      "grad_norm": 0.1644008457660675,
      "learning_rate": 4.047064908197569e-05,
      "loss": 0.7474,
      "step": 5605
    },
    {
      "epoch": 1.2846673537157907,
      "grad_norm": 0.15591102838516235,
      "learning_rate": 4.0405999482803206e-05,
      "loss": 0.7478,
      "step": 5610
    },
    {
      "epoch": 1.2858124355891447,
      "grad_norm": 0.24285583198070526,
      "learning_rate": 4.034134988363073e-05,
      "loss": 0.7502,
      "step": 5615
    },
    {
      "epoch": 1.2869575174624985,
      "grad_norm": 0.19036328792572021,
      "learning_rate": 4.027670028445824e-05,
      "loss": 0.7475,
      "step": 5620
    },
    {
      "epoch": 1.2881025993358526,
      "grad_norm": 0.1598913073539734,
      "learning_rate": 4.021205068528575e-05,
      "loss": 0.7462,
      "step": 5625
    },
    {
      "epoch": 1.2892476812092064,
      "grad_norm": 0.3020450472831726,
      "learning_rate": 4.014740108611327e-05,
      "loss": 0.7477,
      "step": 5630
    },
    {
      "epoch": 1.2903927630825605,
      "grad_norm": 0.20970624685287476,
      "learning_rate": 4.0082751486940784e-05,
      "loss": 0.7497,
      "step": 5635
    },
    {
      "epoch": 1.2915378449559143,
      "grad_norm": 0.24046367406845093,
      "learning_rate": 4.00181018877683e-05,
      "loss": 0.7441,
      "step": 5640
    },
    {
      "epoch": 1.2926829268292683,
      "grad_norm": 0.22961650788784027,
      "learning_rate": 3.995345228859581e-05,
      "loss": 0.7448,
      "step": 5645
    },
    {
      "epoch": 1.2938280087026222,
      "grad_norm": 0.1839102804660797,
      "learning_rate": 3.988880268942333e-05,
      "loss": 0.7464,
      "step": 5650
    },
    {
      "epoch": 1.2949730905759762,
      "grad_norm": 0.1463252604007721,
      "learning_rate": 3.982415309025084e-05,
      "loss": 0.7461,
      "step": 5655
    },
    {
      "epoch": 1.2961181724493303,
      "grad_norm": 0.22954629361629486,
      "learning_rate": 3.975950349107836e-05,
      "loss": 0.7458,
      "step": 5660
    },
    {
      "epoch": 1.297263254322684,
      "grad_norm": 0.17761455476284027,
      "learning_rate": 3.969485389190587e-05,
      "loss": 0.7475,
      "step": 5665
    },
    {
      "epoch": 1.298408336196038,
      "grad_norm": 0.21988537907600403,
      "learning_rate": 3.9630204292733384e-05,
      "loss": 0.7476,
      "step": 5670
    },
    {
      "epoch": 1.299553418069392,
      "grad_norm": 0.2295353263616562,
      "learning_rate": 3.9565554693560905e-05,
      "loss": 0.7453,
      "step": 5675
    },
    {
      "epoch": 1.300698499942746,
      "grad_norm": 0.1871115267276764,
      "learning_rate": 3.950090509438842e-05,
      "loss": 0.7477,
      "step": 5680
    },
    {
      "epoch": 1.3018435818160998,
      "grad_norm": 0.2102229744195938,
      "learning_rate": 3.9436255495215926e-05,
      "loss": 0.7469,
      "step": 5685
    },
    {
      "epoch": 1.3029886636894539,
      "grad_norm": 0.2002745121717453,
      "learning_rate": 3.937160589604345e-05,
      "loss": 0.751,
      "step": 5690
    },
    {
      "epoch": 1.3041337455628077,
      "grad_norm": 0.23087018728256226,
      "learning_rate": 3.930695629687096e-05,
      "loss": 0.7454,
      "step": 5695
    },
    {
      "epoch": 1.3052788274361617,
      "grad_norm": 0.23787514865398407,
      "learning_rate": 3.9242306697698476e-05,
      "loss": 0.7476,
      "step": 5700
    },
    {
      "epoch": 1.3064239093095156,
      "grad_norm": 0.17557412385940552,
      "learning_rate": 3.917765709852599e-05,
      "loss": 0.7471,
      "step": 5705
    },
    {
      "epoch": 1.3075689911828696,
      "grad_norm": 0.1856442093849182,
      "learning_rate": 3.9113007499353504e-05,
      "loss": 0.748,
      "step": 5710
    },
    {
      "epoch": 1.3087140730562234,
      "grad_norm": 0.242033913731575,
      "learning_rate": 3.904835790018102e-05,
      "loss": 0.7502,
      "step": 5715
    },
    {
      "epoch": 1.3098591549295775,
      "grad_norm": 0.19342093169689178,
      "learning_rate": 3.898370830100854e-05,
      "loss": 0.7477,
      "step": 5720
    },
    {
      "epoch": 1.3110042368029315,
      "grad_norm": 0.2315911203622818,
      "learning_rate": 3.8919058701836053e-05,
      "loss": 0.746,
      "step": 5725
    },
    {
      "epoch": 1.3121493186762854,
      "grad_norm": 0.23462465405464172,
      "learning_rate": 3.885440910266356e-05,
      "loss": 0.7466,
      "step": 5730
    },
    {
      "epoch": 1.3132944005496392,
      "grad_norm": 0.20711112022399902,
      "learning_rate": 3.878975950349108e-05,
      "loss": 0.7495,
      "step": 5735
    },
    {
      "epoch": 1.3144394824229932,
      "grad_norm": 0.21462731063365936,
      "learning_rate": 3.8725109904318596e-05,
      "loss": 0.7484,
      "step": 5740
    },
    {
      "epoch": 1.3155845642963473,
      "grad_norm": 0.1888807862997055,
      "learning_rate": 3.866046030514611e-05,
      "loss": 0.7455,
      "step": 5745
    },
    {
      "epoch": 1.3167296461697011,
      "grad_norm": 0.19394944608211517,
      "learning_rate": 3.8595810705973625e-05,
      "loss": 0.749,
      "step": 5750
    },
    {
      "epoch": 1.317874728043055,
      "grad_norm": 0.1983867883682251,
      "learning_rate": 3.853116110680114e-05,
      "loss": 0.7463,
      "step": 5755
    },
    {
      "epoch": 1.319019809916409,
      "grad_norm": 0.1493363082408905,
      "learning_rate": 3.846651150762865e-05,
      "loss": 0.7472,
      "step": 5760
    },
    {
      "epoch": 1.320164891789763,
      "grad_norm": 0.1743180751800537,
      "learning_rate": 3.8401861908456174e-05,
      "loss": 0.7462,
      "step": 5765
    },
    {
      "epoch": 1.3213099736631169,
      "grad_norm": 0.2253560572862625,
      "learning_rate": 3.833721230928368e-05,
      "loss": 0.7438,
      "step": 5770
    },
    {
      "epoch": 1.322455055536471,
      "grad_norm": 0.19323718547821045,
      "learning_rate": 3.8272562710111196e-05,
      "loss": 0.7486,
      "step": 5775
    },
    {
      "epoch": 1.3236001374098247,
      "grad_norm": 0.1969655156135559,
      "learning_rate": 3.820791311093872e-05,
      "loss": 0.7467,
      "step": 5780
    },
    {
      "epoch": 1.3247452192831788,
      "grad_norm": 0.18611446022987366,
      "learning_rate": 3.814326351176623e-05,
      "loss": 0.7459,
      "step": 5785
    },
    {
      "epoch": 1.3258903011565326,
      "grad_norm": 0.25435754656791687,
      "learning_rate": 3.807861391259374e-05,
      "loss": 0.7469,
      "step": 5790
    },
    {
      "epoch": 1.3270353830298867,
      "grad_norm": 0.2215898036956787,
      "learning_rate": 3.801396431342126e-05,
      "loss": 0.7477,
      "step": 5795
    },
    {
      "epoch": 1.3281804649032405,
      "grad_norm": 0.16783438622951508,
      "learning_rate": 3.7949314714248774e-05,
      "loss": 0.7452,
      "step": 5800
    },
    {
      "epoch": 1.3293255467765945,
      "grad_norm": 0.2280804067850113,
      "learning_rate": 3.788466511507629e-05,
      "loss": 0.7459,
      "step": 5805
    },
    {
      "epoch": 1.3304706286499486,
      "grad_norm": 0.1959756463766098,
      "learning_rate": 3.78200155159038e-05,
      "loss": 0.7471,
      "step": 5810
    },
    {
      "epoch": 1.3316157105233024,
      "grad_norm": 0.28425371646881104,
      "learning_rate": 3.7755365916731316e-05,
      "loss": 0.7485,
      "step": 5815
    },
    {
      "epoch": 1.3327607923966562,
      "grad_norm": 0.1959027647972107,
      "learning_rate": 3.769071631755883e-05,
      "loss": 0.7489,
      "step": 5820
    },
    {
      "epoch": 1.3339058742700103,
      "grad_norm": 0.2708321809768677,
      "learning_rate": 3.762606671838635e-05,
      "loss": 0.7471,
      "step": 5825
    },
    {
      "epoch": 1.3350509561433643,
      "grad_norm": 0.19583362340927124,
      "learning_rate": 3.7561417119213866e-05,
      "loss": 0.7453,
      "step": 5830
    },
    {
      "epoch": 1.3361960380167182,
      "grad_norm": 0.14683350920677185,
      "learning_rate": 3.749676752004137e-05,
      "loss": 0.7472,
      "step": 5835
    },
    {
      "epoch": 1.3373411198900722,
      "grad_norm": 0.24646610021591187,
      "learning_rate": 3.7432117920868894e-05,
      "loss": 0.7475,
      "step": 5840
    },
    {
      "epoch": 1.338486201763426,
      "grad_norm": 0.15096144378185272,
      "learning_rate": 3.736746832169641e-05,
      "loss": 0.7479,
      "step": 5845
    },
    {
      "epoch": 1.33963128363678,
      "grad_norm": 0.14911700785160065,
      "learning_rate": 3.730281872252392e-05,
      "loss": 0.7497,
      "step": 5850
    },
    {
      "epoch": 1.340776365510134,
      "grad_norm": 0.2568439245223999,
      "learning_rate": 3.723816912335144e-05,
      "loss": 0.7464,
      "step": 5855
    },
    {
      "epoch": 1.341921447383488,
      "grad_norm": 0.25193852186203003,
      "learning_rate": 3.717351952417895e-05,
      "loss": 0.7458,
      "step": 5860
    },
    {
      "epoch": 1.3430665292568418,
      "grad_norm": 0.1714603453874588,
      "learning_rate": 3.7108869925006465e-05,
      "loss": 0.747,
      "step": 5865
    },
    {
      "epoch": 1.3442116111301958,
      "grad_norm": 0.2510092854499817,
      "learning_rate": 3.7044220325833986e-05,
      "loss": 0.747,
      "step": 5870
    },
    {
      "epoch": 1.3453566930035499,
      "grad_norm": 0.20950113236904144,
      "learning_rate": 3.6979570726661494e-05,
      "loss": 0.7479,
      "step": 5875
    },
    {
      "epoch": 1.3465017748769037,
      "grad_norm": 0.17263996601104736,
      "learning_rate": 3.691492112748901e-05,
      "loss": 0.7462,
      "step": 5880
    },
    {
      "epoch": 1.3476468567502575,
      "grad_norm": 0.25733307003974915,
      "learning_rate": 3.685027152831653e-05,
      "loss": 0.7486,
      "step": 5885
    },
    {
      "epoch": 1.3487919386236116,
      "grad_norm": 0.23318397998809814,
      "learning_rate": 3.678562192914404e-05,
      "loss": 0.7503,
      "step": 5890
    },
    {
      "epoch": 1.3499370204969656,
      "grad_norm": 0.19126096367835999,
      "learning_rate": 3.672097232997155e-05,
      "loss": 0.7425,
      "step": 5895
    },
    {
      "epoch": 1.3510821023703194,
      "grad_norm": 0.2814708650112152,
      "learning_rate": 3.665632273079907e-05,
      "loss": 0.7476,
      "step": 5900
    },
    {
      "epoch": 1.3522271842436735,
      "grad_norm": 0.23234932124614716,
      "learning_rate": 3.6591673131626586e-05,
      "loss": 0.7485,
      "step": 5905
    },
    {
      "epoch": 1.3533722661170273,
      "grad_norm": 0.34967902302742004,
      "learning_rate": 3.65270235324541e-05,
      "loss": 0.7484,
      "step": 5910
    },
    {
      "epoch": 1.3545173479903814,
      "grad_norm": 0.17776401340961456,
      "learning_rate": 3.6462373933281614e-05,
      "loss": 0.7476,
      "step": 5915
    },
    {
      "epoch": 1.3556624298637352,
      "grad_norm": 0.20090113580226898,
      "learning_rate": 3.639772433410913e-05,
      "loss": 0.7461,
      "step": 5920
    },
    {
      "epoch": 1.3568075117370892,
      "grad_norm": 0.27608180046081543,
      "learning_rate": 3.633307473493664e-05,
      "loss": 0.7471,
      "step": 5925
    },
    {
      "epoch": 1.357952593610443,
      "grad_norm": 0.18629661202430725,
      "learning_rate": 3.6268425135764164e-05,
      "loss": 0.7479,
      "step": 5930
    },
    {
      "epoch": 1.3590976754837971,
      "grad_norm": 0.20956484973430634,
      "learning_rate": 3.620377553659168e-05,
      "loss": 0.7487,
      "step": 5935
    },
    {
      "epoch": 1.3602427573571512,
      "grad_norm": 0.2737008035182953,
      "learning_rate": 3.6139125937419185e-05,
      "loss": 0.7436,
      "step": 5940
    },
    {
      "epoch": 1.361387839230505,
      "grad_norm": 0.16381745040416718,
      "learning_rate": 3.6074476338246706e-05,
      "loss": 0.7469,
      "step": 5945
    },
    {
      "epoch": 1.3625329211038588,
      "grad_norm": 0.24451714754104614,
      "learning_rate": 3.600982673907422e-05,
      "loss": 0.748,
      "step": 5950
    },
    {
      "epoch": 1.3636780029772129,
      "grad_norm": 0.34765592217445374,
      "learning_rate": 3.5945177139901735e-05,
      "loss": 0.7451,
      "step": 5955
    },
    {
      "epoch": 1.364823084850567,
      "grad_norm": 0.1471840888261795,
      "learning_rate": 3.588052754072925e-05,
      "loss": 0.7486,
      "step": 5960
    },
    {
      "epoch": 1.3659681667239207,
      "grad_norm": 0.23996858298778534,
      "learning_rate": 3.581587794155676e-05,
      "loss": 0.7492,
      "step": 5965
    },
    {
      "epoch": 1.3671132485972748,
      "grad_norm": 0.22462098300457,
      "learning_rate": 3.575122834238428e-05,
      "loss": 0.7446,
      "step": 5970
    },
    {
      "epoch": 1.3682583304706286,
      "grad_norm": 0.22103479504585266,
      "learning_rate": 3.56865787432118e-05,
      "loss": 0.7508,
      "step": 5975
    },
    {
      "epoch": 1.3694034123439827,
      "grad_norm": 0.22594700753688812,
      "learning_rate": 3.5621929144039306e-05,
      "loss": 0.7501,
      "step": 5980
    },
    {
      "epoch": 1.3705484942173365,
      "grad_norm": 0.15880168974399567,
      "learning_rate": 3.555727954486682e-05,
      "loss": 0.7407,
      "step": 5985
    },
    {
      "epoch": 1.3716935760906905,
      "grad_norm": 0.18890678882598877,
      "learning_rate": 3.549262994569434e-05,
      "loss": 0.7498,
      "step": 5990
    },
    {
      "epoch": 1.3728386579640444,
      "grad_norm": 0.2300521731376648,
      "learning_rate": 3.5427980346521855e-05,
      "loss": 0.7429,
      "step": 5995
    },
    {
      "epoch": 1.3739837398373984,
      "grad_norm": 0.2027832716703415,
      "learning_rate": 3.536333074734936e-05,
      "loss": 0.7482,
      "step": 6000
    },
    {
      "epoch": 1.3751288217107522,
      "grad_norm": 0.1609947830438614,
      "learning_rate": 3.5298681148176884e-05,
      "loss": 0.7469,
      "step": 6005
    },
    {
      "epoch": 1.3762739035841063,
      "grad_norm": 0.1555599868297577,
      "learning_rate": 3.52340315490044e-05,
      "loss": 0.7448,
      "step": 6010
    },
    {
      "epoch": 1.37741898545746,
      "grad_norm": 0.19723734259605408,
      "learning_rate": 3.516938194983191e-05,
      "loss": 0.7445,
      "step": 6015
    },
    {
      "epoch": 1.3785640673308142,
      "grad_norm": 0.23402778804302216,
      "learning_rate": 3.5104732350659426e-05,
      "loss": 0.7437,
      "step": 6020
    },
    {
      "epoch": 1.3797091492041682,
      "grad_norm": 0.2760639488697052,
      "learning_rate": 3.504008275148694e-05,
      "loss": 0.7429,
      "step": 6025
    },
    {
      "epoch": 1.380854231077522,
      "grad_norm": 0.18020078539848328,
      "learning_rate": 3.4975433152314455e-05,
      "loss": 0.7464,
      "step": 6030
    },
    {
      "epoch": 1.3819993129508759,
      "grad_norm": 0.19109944999217987,
      "learning_rate": 3.4910783553141976e-05,
      "loss": 0.7438,
      "step": 6035
    },
    {
      "epoch": 1.38314439482423,
      "grad_norm": 0.2158491015434265,
      "learning_rate": 3.484613395396949e-05,
      "loss": 0.7462,
      "step": 6040
    },
    {
      "epoch": 1.384289476697584,
      "grad_norm": 0.2594541311264038,
      "learning_rate": 3.4781484354797e-05,
      "loss": 0.7483,
      "step": 6045
    },
    {
      "epoch": 1.3854345585709378,
      "grad_norm": 0.1272057443857193,
      "learning_rate": 3.471683475562452e-05,
      "loss": 0.7449,
      "step": 6050
    },
    {
      "epoch": 1.3865796404442918,
      "grad_norm": 0.1897531896829605,
      "learning_rate": 3.465218515645203e-05,
      "loss": 0.7491,
      "step": 6055
    },
    {
      "epoch": 1.3877247223176457,
      "grad_norm": 0.1548863649368286,
      "learning_rate": 3.458753555727955e-05,
      "loss": 0.7439,
      "step": 6060
    },
    {
      "epoch": 1.3888698041909997,
      "grad_norm": 0.16846147179603577,
      "learning_rate": 3.452288595810706e-05,
      "loss": 0.7437,
      "step": 6065
    },
    {
      "epoch": 1.3900148860643535,
      "grad_norm": 0.1798481047153473,
      "learning_rate": 3.4458236358934575e-05,
      "loss": 0.7464,
      "step": 6070
    },
    {
      "epoch": 1.3911599679377076,
      "grad_norm": 0.31617963314056396,
      "learning_rate": 3.439358675976209e-05,
      "loss": 0.7442,
      "step": 6075
    },
    {
      "epoch": 1.3923050498110614,
      "grad_norm": 0.17033277451992035,
      "learning_rate": 3.432893716058961e-05,
      "loss": 0.7494,
      "step": 6080
    },
    {
      "epoch": 1.3934501316844154,
      "grad_norm": 0.30593806505203247,
      "learning_rate": 3.426428756141712e-05,
      "loss": 0.744,
      "step": 6085
    },
    {
      "epoch": 1.3945952135577695,
      "grad_norm": 0.18552587926387787,
      "learning_rate": 3.419963796224463e-05,
      "loss": 0.7489,
      "step": 6090
    },
    {
      "epoch": 1.3957402954311233,
      "grad_norm": 0.15603308379650116,
      "learning_rate": 3.413498836307215e-05,
      "loss": 0.7412,
      "step": 6095
    },
    {
      "epoch": 1.3968853773044771,
      "grad_norm": 0.20843032002449036,
      "learning_rate": 3.407033876389967e-05,
      "loss": 0.7417,
      "step": 6100
    },
    {
      "epoch": 1.3980304591778312,
      "grad_norm": 0.20933730900287628,
      "learning_rate": 3.4005689164727175e-05,
      "loss": 0.7426,
      "step": 6105
    },
    {
      "epoch": 1.3991755410511852,
      "grad_norm": 0.17963987588882446,
      "learning_rate": 3.3941039565554696e-05,
      "loss": 0.747,
      "step": 6110
    },
    {
      "epoch": 1.400320622924539,
      "grad_norm": 0.22531652450561523,
      "learning_rate": 3.387638996638221e-05,
      "loss": 0.7469,
      "step": 6115
    },
    {
      "epoch": 1.4014657047978931,
      "grad_norm": 0.2187807708978653,
      "learning_rate": 3.3811740367209724e-05,
      "loss": 0.7439,
      "step": 6120
    },
    {
      "epoch": 1.402610786671247,
      "grad_norm": 0.21380850672721863,
      "learning_rate": 3.374709076803724e-05,
      "loss": 0.7443,
      "step": 6125
    },
    {
      "epoch": 1.403755868544601,
      "grad_norm": 0.22001244127750397,
      "learning_rate": 3.368244116886475e-05,
      "loss": 0.7509,
      "step": 6130
    },
    {
      "epoch": 1.4049009504179548,
      "grad_norm": 0.17430852353572845,
      "learning_rate": 3.361779156969227e-05,
      "loss": 0.7496,
      "step": 6135
    },
    {
      "epoch": 1.4060460322913089,
      "grad_norm": 0.2584846317768097,
      "learning_rate": 3.355314197051979e-05,
      "loss": 0.7483,
      "step": 6140
    },
    {
      "epoch": 1.4071911141646627,
      "grad_norm": 0.15371204912662506,
      "learning_rate": 3.34884923713473e-05,
      "loss": 0.7454,
      "step": 6145
    },
    {
      "epoch": 1.4083361960380167,
      "grad_norm": 0.24281756579875946,
      "learning_rate": 3.342384277217481e-05,
      "loss": 0.7459,
      "step": 6150
    },
    {
      "epoch": 1.4094812779113708,
      "grad_norm": 0.18498913943767548,
      "learning_rate": 3.335919317300233e-05,
      "loss": 0.7458,
      "step": 6155
    },
    {
      "epoch": 1.4106263597847246,
      "grad_norm": 0.19410498440265656,
      "learning_rate": 3.3294543573829845e-05,
      "loss": 0.7465,
      "step": 6160
    },
    {
      "epoch": 1.4117714416580784,
      "grad_norm": 0.1695197969675064,
      "learning_rate": 3.322989397465736e-05,
      "loss": 0.7485,
      "step": 6165
    },
    {
      "epoch": 1.4129165235314325,
      "grad_norm": 0.1701662689447403,
      "learning_rate": 3.316524437548487e-05,
      "loss": 0.7511,
      "step": 6170
    },
    {
      "epoch": 1.4140616054047865,
      "grad_norm": 0.1659391224384308,
      "learning_rate": 3.310059477631239e-05,
      "loss": 0.7421,
      "step": 6175
    },
    {
      "epoch": 1.4152066872781404,
      "grad_norm": 0.17819622159004211,
      "learning_rate": 3.30359451771399e-05,
      "loss": 0.7465,
      "step": 6180
    },
    {
      "epoch": 1.4163517691514944,
      "grad_norm": 0.2059786170721054,
      "learning_rate": 3.297129557796742e-05,
      "loss": 0.7437,
      "step": 6185
    },
    {
      "epoch": 1.4174968510248482,
      "grad_norm": 0.24188071489334106,
      "learning_rate": 3.290664597879493e-05,
      "loss": 0.7426,
      "step": 6190
    },
    {
      "epoch": 1.4186419328982023,
      "grad_norm": 0.2378331571817398,
      "learning_rate": 3.2841996379622444e-05,
      "loss": 0.7468,
      "step": 6195
    },
    {
      "epoch": 1.419787014771556,
      "grad_norm": 0.2121928632259369,
      "learning_rate": 3.2777346780449965e-05,
      "loss": 0.7437,
      "step": 6200
    },
    {
      "epoch": 1.4209320966449102,
      "grad_norm": 0.18138332664966583,
      "learning_rate": 3.271269718127748e-05,
      "loss": 0.7473,
      "step": 6205
    },
    {
      "epoch": 1.422077178518264,
      "grad_norm": 0.18859916925430298,
      "learning_rate": 3.264804758210499e-05,
      "loss": 0.7499,
      "step": 6210
    },
    {
      "epoch": 1.423222260391618,
      "grad_norm": 0.250404417514801,
      "learning_rate": 3.258339798293251e-05,
      "loss": 0.7473,
      "step": 6215
    },
    {
      "epoch": 1.424367342264972,
      "grad_norm": 0.26826900243759155,
      "learning_rate": 3.251874838376002e-05,
      "loss": 0.7442,
      "step": 6220
    },
    {
      "epoch": 1.425512424138326,
      "grad_norm": 0.20250312983989716,
      "learning_rate": 3.2454098784587536e-05,
      "loss": 0.7453,
      "step": 6225
    },
    {
      "epoch": 1.4266575060116797,
      "grad_norm": 0.18097352981567383,
      "learning_rate": 3.238944918541505e-05,
      "loss": 0.7525,
      "step": 6230
    },
    {
      "epoch": 1.4278025878850338,
      "grad_norm": 0.19572164118289948,
      "learning_rate": 3.2324799586242565e-05,
      "loss": 0.748,
      "step": 6235
    },
    {
      "epoch": 1.4289476697583878,
      "grad_norm": 0.2311696708202362,
      "learning_rate": 3.226014998707008e-05,
      "loss": 0.7479,
      "step": 6240
    },
    {
      "epoch": 1.4300927516317417,
      "grad_norm": 0.22108034789562225,
      "learning_rate": 3.21955003878976e-05,
      "loss": 0.7435,
      "step": 6245
    },
    {
      "epoch": 1.4312378335050955,
      "grad_norm": 0.19984404742717743,
      "learning_rate": 3.213085078872511e-05,
      "loss": 0.7446,
      "step": 6250
    },
    {
      "epoch": 1.4323829153784495,
      "grad_norm": 0.17798064649105072,
      "learning_rate": 3.206620118955262e-05,
      "loss": 0.7499,
      "step": 6255
    },
    {
      "epoch": 1.4335279972518036,
      "grad_norm": 0.19513164460659027,
      "learning_rate": 3.200155159038014e-05,
      "loss": 0.7478,
      "step": 6260
    },
    {
      "epoch": 1.4346730791251574,
      "grad_norm": 0.21562044322490692,
      "learning_rate": 3.193690199120766e-05,
      "loss": 0.7471,
      "step": 6265
    },
    {
      "epoch": 1.4358181609985115,
      "grad_norm": 0.16316361725330353,
      "learning_rate": 3.187225239203517e-05,
      "loss": 0.7473,
      "step": 6270
    },
    {
      "epoch": 1.4369632428718653,
      "grad_norm": 0.17712213099002838,
      "learning_rate": 3.1807602792862685e-05,
      "loss": 0.7458,
      "step": 6275
    },
    {
      "epoch": 1.4381083247452193,
      "grad_norm": 0.18155217170715332,
      "learning_rate": 3.17429531936902e-05,
      "loss": 0.7513,
      "step": 6280
    },
    {
      "epoch": 1.4392534066185732,
      "grad_norm": 0.18960131704807281,
      "learning_rate": 3.1678303594517714e-05,
      "loss": 0.7469,
      "step": 6285
    },
    {
      "epoch": 1.4403984884919272,
      "grad_norm": 0.15972264111042023,
      "learning_rate": 3.1613653995345235e-05,
      "loss": 0.7422,
      "step": 6290
    },
    {
      "epoch": 1.441543570365281,
      "grad_norm": 0.23653262853622437,
      "learning_rate": 3.154900439617274e-05,
      "loss": 0.7424,
      "step": 6295
    },
    {
      "epoch": 1.442688652238635,
      "grad_norm": 0.22679731249809265,
      "learning_rate": 3.1484354797000257e-05,
      "loss": 0.7452,
      "step": 6300
    },
    {
      "epoch": 1.4438337341119891,
      "grad_norm": 0.25342831015586853,
      "learning_rate": 3.141970519782778e-05,
      "loss": 0.7401,
      "step": 6305
    },
    {
      "epoch": 1.444978815985343,
      "grad_norm": 0.21492965519428253,
      "learning_rate": 3.135505559865529e-05,
      "loss": 0.7443,
      "step": 6310
    },
    {
      "epoch": 1.4461238978586968,
      "grad_norm": 0.19540521502494812,
      "learning_rate": 3.12904059994828e-05,
      "loss": 0.7442,
      "step": 6315
    },
    {
      "epoch": 1.4472689797320508,
      "grad_norm": 0.28760331869125366,
      "learning_rate": 3.122575640031032e-05,
      "loss": 0.7408,
      "step": 6320
    },
    {
      "epoch": 1.4484140616054049,
      "grad_norm": 0.1781582087278366,
      "learning_rate": 3.1161106801137834e-05,
      "loss": 0.7459,
      "step": 6325
    },
    {
      "epoch": 1.4495591434787587,
      "grad_norm": 0.21897757053375244,
      "learning_rate": 3.109645720196535e-05,
      "loss": 0.7438,
      "step": 6330
    },
    {
      "epoch": 1.4507042253521127,
      "grad_norm": 0.3101745843887329,
      "learning_rate": 3.103180760279286e-05,
      "loss": 0.7477,
      "step": 6335
    },
    {
      "epoch": 1.4518493072254666,
      "grad_norm": 0.18157465755939484,
      "learning_rate": 3.096715800362038e-05,
      "loss": 0.7497,
      "step": 6340
    },
    {
      "epoch": 1.4529943890988206,
      "grad_norm": 0.2329101860523224,
      "learning_rate": 3.090250840444789e-05,
      "loss": 0.7449,
      "step": 6345
    },
    {
      "epoch": 1.4541394709721744,
      "grad_norm": 0.17827336490154266,
      "learning_rate": 3.083785880527541e-05,
      "loss": 0.745,
      "step": 6350
    },
    {
      "epoch": 1.4552845528455285,
      "grad_norm": 0.22005167603492737,
      "learning_rate": 3.077320920610292e-05,
      "loss": 0.7458,
      "step": 6355
    },
    {
      "epoch": 1.4564296347188823,
      "grad_norm": 0.17929284274578094,
      "learning_rate": 3.0708559606930434e-05,
      "loss": 0.7451,
      "step": 6360
    },
    {
      "epoch": 1.4575747165922364,
      "grad_norm": 0.20535363256931305,
      "learning_rate": 3.0643910007757955e-05,
      "loss": 0.7398,
      "step": 6365
    },
    {
      "epoch": 1.4587197984655904,
      "grad_norm": 0.19280757009983063,
      "learning_rate": 3.057926040858547e-05,
      "loss": 0.7476,
      "step": 6370
    },
    {
      "epoch": 1.4598648803389442,
      "grad_norm": 0.19142794609069824,
      "learning_rate": 3.0514610809412987e-05,
      "loss": 0.7427,
      "step": 6375
    },
    {
      "epoch": 1.461009962212298,
      "grad_norm": 0.2704218626022339,
      "learning_rate": 3.0449961210240498e-05,
      "loss": 0.7455,
      "step": 6380
    },
    {
      "epoch": 1.4621550440856521,
      "grad_norm": 0.21759334206581116,
      "learning_rate": 3.0385311611068012e-05,
      "loss": 0.7429,
      "step": 6385
    },
    {
      "epoch": 1.4633001259590062,
      "grad_norm": 0.1478096842765808,
      "learning_rate": 3.032066201189553e-05,
      "loss": 0.7422,
      "step": 6390
    },
    {
      "epoch": 1.46444520783236,
      "grad_norm": 0.19636553525924683,
      "learning_rate": 3.0256012412723044e-05,
      "loss": 0.7478,
      "step": 6395
    },
    {
      "epoch": 1.465590289705714,
      "grad_norm": 0.1897573620080948,
      "learning_rate": 3.0191362813550554e-05,
      "loss": 0.7432,
      "step": 6400
    },
    {
      "epoch": 1.4667353715790679,
      "grad_norm": 0.17960022389888763,
      "learning_rate": 3.0126713214378072e-05,
      "loss": 0.746,
      "step": 6405
    },
    {
      "epoch": 1.467880453452422,
      "grad_norm": 0.18701818585395813,
      "learning_rate": 3.0062063615205586e-05,
      "loss": 0.7419,
      "step": 6410
    },
    {
      "epoch": 1.4690255353257757,
      "grad_norm": 0.1686282753944397,
      "learning_rate": 2.9997414016033104e-05,
      "loss": 0.743,
      "step": 6415
    },
    {
      "epoch": 1.4701706171991298,
      "grad_norm": 0.2564958333969116,
      "learning_rate": 2.9932764416860615e-05,
      "loss": 0.7475,
      "step": 6420
    },
    {
      "epoch": 1.4713156990724836,
      "grad_norm": 0.25114843249320984,
      "learning_rate": 2.9868114817688132e-05,
      "loss": 0.7476,
      "step": 6425
    },
    {
      "epoch": 1.4724607809458377,
      "grad_norm": 0.16929112374782562,
      "learning_rate": 2.9803465218515647e-05,
      "loss": 0.7429,
      "step": 6430
    },
    {
      "epoch": 1.4736058628191917,
      "grad_norm": 0.25373727083206177,
      "learning_rate": 2.9738815619343164e-05,
      "loss": 0.7429,
      "step": 6435
    },
    {
      "epoch": 1.4747509446925455,
      "grad_norm": 0.15113116800785065,
      "learning_rate": 2.9674166020170675e-05,
      "loss": 0.7416,
      "step": 6440
    },
    {
      "epoch": 1.4758960265658994,
      "grad_norm": 0.2088536024093628,
      "learning_rate": 2.960951642099819e-05,
      "loss": 0.7457,
      "step": 6445
    },
    {
      "epoch": 1.4770411084392534,
      "grad_norm": 0.16958044469356537,
      "learning_rate": 2.9544866821825707e-05,
      "loss": 0.7446,
      "step": 6450
    },
    {
      "epoch": 1.4781861903126075,
      "grad_norm": 0.2599340081214905,
      "learning_rate": 2.948021722265322e-05,
      "loss": 0.7422,
      "step": 6455
    },
    {
      "epoch": 1.4793312721859613,
      "grad_norm": 0.1796865016222,
      "learning_rate": 2.9415567623480732e-05,
      "loss": 0.7466,
      "step": 6460
    },
    {
      "epoch": 1.480476354059315,
      "grad_norm": 0.265708863735199,
      "learning_rate": 2.935091802430825e-05,
      "loss": 0.7434,
      "step": 6465
    },
    {
      "epoch": 1.4816214359326692,
      "grad_norm": 0.2140311598777771,
      "learning_rate": 2.9286268425135767e-05,
      "loss": 0.7485,
      "step": 6470
    },
    {
      "epoch": 1.4827665178060232,
      "grad_norm": 0.1847228854894638,
      "learning_rate": 2.922161882596328e-05,
      "loss": 0.743,
      "step": 6475
    },
    {
      "epoch": 1.483911599679377,
      "grad_norm": 0.24625612795352936,
      "learning_rate": 2.91569692267908e-05,
      "loss": 0.7481,
      "step": 6480
    },
    {
      "epoch": 1.485056681552731,
      "grad_norm": 0.199641615152359,
      "learning_rate": 2.909231962761831e-05,
      "loss": 0.7463,
      "step": 6485
    },
    {
      "epoch": 1.486201763426085,
      "grad_norm": 0.20555761456489563,
      "learning_rate": 2.9027670028445824e-05,
      "loss": 0.7499,
      "step": 6490
    },
    {
      "epoch": 1.487346845299439,
      "grad_norm": 0.1842489093542099,
      "learning_rate": 2.896302042927334e-05,
      "loss": 0.7497,
      "step": 6495
    },
    {
      "epoch": 1.4884919271727928,
      "grad_norm": 0.24474182724952698,
      "learning_rate": 2.8898370830100856e-05,
      "loss": 0.7429,
      "step": 6500
    },
    {
      "epoch": 1.4896370090461468,
      "grad_norm": 0.2305942177772522,
      "learning_rate": 2.8833721230928367e-05,
      "loss": 0.7456,
      "step": 6505
    },
    {
      "epoch": 1.4907820909195006,
      "grad_norm": 0.19151118397712708,
      "learning_rate": 2.8769071631755884e-05,
      "loss": 0.7469,
      "step": 6510
    },
    {
      "epoch": 1.4919271727928547,
      "grad_norm": 0.17481862008571625,
      "learning_rate": 2.87044220325834e-05,
      "loss": 0.7435,
      "step": 6515
    },
    {
      "epoch": 1.4930722546662087,
      "grad_norm": 0.1783713400363922,
      "learning_rate": 2.8639772433410916e-05,
      "loss": 0.7466,
      "step": 6520
    },
    {
      "epoch": 1.4942173365395626,
      "grad_norm": 0.18017755448818207,
      "learning_rate": 2.8575122834238427e-05,
      "loss": 0.7424,
      "step": 6525
    },
    {
      "epoch": 1.4953624184129164,
      "grad_norm": 0.25057896971702576,
      "learning_rate": 2.8510473235065945e-05,
      "loss": 0.7395,
      "step": 6530
    },
    {
      "epoch": 1.4965075002862704,
      "grad_norm": 0.25676342844963074,
      "learning_rate": 2.844582363589346e-05,
      "loss": 0.7402,
      "step": 6535
    },
    {
      "epoch": 1.4976525821596245,
      "grad_norm": 0.17843708395957947,
      "learning_rate": 2.8381174036720976e-05,
      "loss": 0.7485,
      "step": 6540
    },
    {
      "epoch": 1.4987976640329783,
      "grad_norm": 0.19747896492481232,
      "learning_rate": 2.8316524437548487e-05,
      "loss": 0.7466,
      "step": 6545
    },
    {
      "epoch": 1.4999427459063324,
      "grad_norm": 0.24598313868045807,
      "learning_rate": 2.8251874838376e-05,
      "loss": 0.7434,
      "step": 6550
    },
    {
      "epoch": 1.5010878277796862,
      "grad_norm": 0.1990305483341217,
      "learning_rate": 2.818722523920352e-05,
      "loss": 0.7455,
      "step": 6555
    },
    {
      "epoch": 1.5022329096530402,
      "grad_norm": 0.20389562845230103,
      "learning_rate": 2.8122575640031033e-05,
      "loss": 0.7472,
      "step": 6560
    },
    {
      "epoch": 1.5033779915263943,
      "grad_norm": 0.18377013504505157,
      "learning_rate": 2.8057926040858544e-05,
      "loss": 0.7445,
      "step": 6565
    },
    {
      "epoch": 1.5045230733997481,
      "grad_norm": 0.17699111998081207,
      "learning_rate": 2.799327644168606e-05,
      "loss": 0.7433,
      "step": 6570
    },
    {
      "epoch": 1.505668155273102,
      "grad_norm": 0.1928340494632721,
      "learning_rate": 2.792862684251358e-05,
      "loss": 0.7442,
      "step": 6575
    },
    {
      "epoch": 1.506813237146456,
      "grad_norm": 0.1634168028831482,
      "learning_rate": 2.7863977243341093e-05,
      "loss": 0.7422,
      "step": 6580
    },
    {
      "epoch": 1.50795831901981,
      "grad_norm": 0.1806732714176178,
      "learning_rate": 2.779932764416861e-05,
      "loss": 0.7436,
      "step": 6585
    },
    {
      "epoch": 1.5091034008931639,
      "grad_norm": 0.17415916919708252,
      "learning_rate": 2.7734678044996122e-05,
      "loss": 0.7457,
      "step": 6590
    },
    {
      "epoch": 1.5102484827665177,
      "grad_norm": 0.15790635347366333,
      "learning_rate": 2.7670028445823636e-05,
      "loss": 0.7445,
      "step": 6595
    },
    {
      "epoch": 1.5113935646398717,
      "grad_norm": 0.26694029569625854,
      "learning_rate": 2.7605378846651154e-05,
      "loss": 0.7469,
      "step": 6600
    },
    {
      "epoch": 1.5125386465132258,
      "grad_norm": 0.2555330991744995,
      "learning_rate": 2.7540729247478668e-05,
      "loss": 0.7449,
      "step": 6605
    },
    {
      "epoch": 1.5136837283865796,
      "grad_norm": 0.1820402443408966,
      "learning_rate": 2.747607964830618e-05,
      "loss": 0.7433,
      "step": 6610
    },
    {
      "epoch": 1.5148288102599334,
      "grad_norm": 0.17936120927333832,
      "learning_rate": 2.7411430049133696e-05,
      "loss": 0.7464,
      "step": 6615
    },
    {
      "epoch": 1.5159738921332875,
      "grad_norm": 0.17956532537937164,
      "learning_rate": 2.7346780449961214e-05,
      "loss": 0.7438,
      "step": 6620
    },
    {
      "epoch": 1.5171189740066415,
      "grad_norm": 0.23218995332717896,
      "learning_rate": 2.7282130850788728e-05,
      "loss": 0.7451,
      "step": 6625
    },
    {
      "epoch": 1.5182640558799956,
      "grad_norm": 0.21373704075813293,
      "learning_rate": 2.721748125161624e-05,
      "loss": 0.7436,
      "step": 6630
    },
    {
      "epoch": 1.5194091377533494,
      "grad_norm": 0.17974428832530975,
      "learning_rate": 2.7152831652443757e-05,
      "loss": 0.7414,
      "step": 6635
    },
    {
      "epoch": 1.5205542196267032,
      "grad_norm": 0.16060051321983337,
      "learning_rate": 2.708818205327127e-05,
      "loss": 0.7446,
      "step": 6640
    },
    {
      "epoch": 1.5216993015000573,
      "grad_norm": 0.21203061938285828,
      "learning_rate": 2.702353245409879e-05,
      "loss": 0.7418,
      "step": 6645
    },
    {
      "epoch": 1.5228443833734113,
      "grad_norm": 0.16728591918945312,
      "learning_rate": 2.69588828549263e-05,
      "loss": 0.7469,
      "step": 6650
    },
    {
      "epoch": 1.5239894652467652,
      "grad_norm": 0.1729734241962433,
      "learning_rate": 2.6894233255753814e-05,
      "loss": 0.7429,
      "step": 6655
    },
    {
      "epoch": 1.525134547120119,
      "grad_norm": 0.1781495064496994,
      "learning_rate": 2.682958365658133e-05,
      "loss": 0.7416,
      "step": 6660
    },
    {
      "epoch": 1.526279628993473,
      "grad_norm": 0.17667962610721588,
      "learning_rate": 2.6764934057408845e-05,
      "loss": 0.743,
      "step": 6665
    },
    {
      "epoch": 1.527424710866827,
      "grad_norm": 0.20562386512756348,
      "learning_rate": 2.6700284458236356e-05,
      "loss": 0.7439,
      "step": 6670
    },
    {
      "epoch": 1.528569792740181,
      "grad_norm": 0.22437416017055511,
      "learning_rate": 2.6635634859063874e-05,
      "loss": 0.7436,
      "step": 6675
    },
    {
      "epoch": 1.5297148746135347,
      "grad_norm": 0.20179647207260132,
      "learning_rate": 2.657098525989139e-05,
      "loss": 0.7446,
      "step": 6680
    },
    {
      "epoch": 1.5308599564868888,
      "grad_norm": 0.2053723931312561,
      "learning_rate": 2.6506335660718906e-05,
      "loss": 0.7424,
      "step": 6685
    },
    {
      "epoch": 1.5320050383602428,
      "grad_norm": 0.2075263410806656,
      "learning_rate": 2.6441686061546423e-05,
      "loss": 0.7439,
      "step": 6690
    },
    {
      "epoch": 1.5331501202335966,
      "grad_norm": 0.2276289463043213,
      "learning_rate": 2.6377036462373934e-05,
      "loss": 0.7428,
      "step": 6695
    },
    {
      "epoch": 1.5342952021069507,
      "grad_norm": 0.2265288233757019,
      "learning_rate": 2.631238686320145e-05,
      "loss": 0.7437,
      "step": 6700
    },
    {
      "epoch": 1.5354402839803045,
      "grad_norm": 0.17774851620197296,
      "learning_rate": 2.6247737264028966e-05,
      "loss": 0.7469,
      "step": 6705
    },
    {
      "epoch": 1.5365853658536586,
      "grad_norm": 0.1712876409292221,
      "learning_rate": 2.618308766485648e-05,
      "loss": 0.7398,
      "step": 6710
    },
    {
      "epoch": 1.5377304477270126,
      "grad_norm": 0.16502240300178528,
      "learning_rate": 2.611843806568399e-05,
      "loss": 0.7486,
      "step": 6715
    },
    {
      "epoch": 1.5388755296003664,
      "grad_norm": 0.17795094847679138,
      "learning_rate": 2.605378846651151e-05,
      "loss": 0.7474,
      "step": 6720
    },
    {
      "epoch": 1.5400206114737203,
      "grad_norm": 0.16823707520961761,
      "learning_rate": 2.5989138867339026e-05,
      "loss": 0.7411,
      "step": 6725
    },
    {
      "epoch": 1.5411656933470743,
      "grad_norm": 0.16455113887786865,
      "learning_rate": 2.592448926816654e-05,
      "loss": 0.7409,
      "step": 6730
    },
    {
      "epoch": 1.5423107752204284,
      "grad_norm": 0.1896887868642807,
      "learning_rate": 2.585983966899405e-05,
      "loss": 0.7445,
      "step": 6735
    },
    {
      "epoch": 1.5434558570937822,
      "grad_norm": 0.16402652859687805,
      "learning_rate": 2.579519006982157e-05,
      "loss": 0.748,
      "step": 6740
    },
    {
      "epoch": 1.544600938967136,
      "grad_norm": 0.17740361392498016,
      "learning_rate": 2.5730540470649083e-05,
      "loss": 0.7454,
      "step": 6745
    },
    {
      "epoch": 1.54574602084049,
      "grad_norm": 0.16312089562416077,
      "learning_rate": 2.56658908714766e-05,
      "loss": 0.7451,
      "step": 6750
    },
    {
      "epoch": 1.5468911027138441,
      "grad_norm": 0.15862400829792023,
      "learning_rate": 2.560124127230411e-05,
      "loss": 0.7472,
      "step": 6755
    },
    {
      "epoch": 1.548036184587198,
      "grad_norm": 0.23264558613300323,
      "learning_rate": 2.5536591673131626e-05,
      "loss": 0.7446,
      "step": 6760
    },
    {
      "epoch": 1.5491812664605518,
      "grad_norm": 0.20183593034744263,
      "learning_rate": 2.5471942073959143e-05,
      "loss": 0.7426,
      "step": 6765
    },
    {
      "epoch": 1.5503263483339058,
      "grad_norm": 0.16477009654045105,
      "learning_rate": 2.5407292474786658e-05,
      "loss": 0.7452,
      "step": 6770
    },
    {
      "epoch": 1.5514714302072599,
      "grad_norm": 0.14841727912425995,
      "learning_rate": 2.534264287561417e-05,
      "loss": 0.7439,
      "step": 6775
    },
    {
      "epoch": 1.552616512080614,
      "grad_norm": 0.16832435131072998,
      "learning_rate": 2.5277993276441686e-05,
      "loss": 0.7428,
      "step": 6780
    },
    {
      "epoch": 1.5537615939539677,
      "grad_norm": 0.16100366413593292,
      "learning_rate": 2.5213343677269204e-05,
      "loss": 0.7461,
      "step": 6785
    },
    {
      "epoch": 1.5549066758273216,
      "grad_norm": 0.16844795644283295,
      "learning_rate": 2.5148694078096718e-05,
      "loss": 0.7407,
      "step": 6790
    },
    {
      "epoch": 1.5560517577006756,
      "grad_norm": 0.2388097047805786,
      "learning_rate": 2.5084044478924235e-05,
      "loss": 0.7446,
      "step": 6795
    },
    {
      "epoch": 1.5571968395740297,
      "grad_norm": 0.2388181835412979,
      "learning_rate": 2.5019394879751746e-05,
      "loss": 0.7424,
      "step": 6800
    },
    {
      "epoch": 1.5583419214473835,
      "grad_norm": 0.1726500540971756,
      "learning_rate": 2.495474528057926e-05,
      "loss": 0.7445,
      "step": 6805
    },
    {
      "epoch": 1.5594870033207373,
      "grad_norm": 0.19557859003543854,
      "learning_rate": 2.4890095681406778e-05,
      "loss": 0.7411,
      "step": 6810
    },
    {
      "epoch": 1.5606320851940914,
      "grad_norm": 0.14564669132232666,
      "learning_rate": 2.4825446082234292e-05,
      "loss": 0.7441,
      "step": 6815
    },
    {
      "epoch": 1.5617771670674454,
      "grad_norm": 0.14897030591964722,
      "learning_rate": 2.4760796483061807e-05,
      "loss": 0.7449,
      "step": 6820
    },
    {
      "epoch": 1.5629222489407992,
      "grad_norm": 0.17548809945583344,
      "learning_rate": 2.469614688388932e-05,
      "loss": 0.7452,
      "step": 6825
    },
    {
      "epoch": 1.564067330814153,
      "grad_norm": 0.21593323349952698,
      "learning_rate": 2.463149728471684e-05,
      "loss": 0.7428,
      "step": 6830
    },
    {
      "epoch": 1.565212412687507,
      "grad_norm": 0.1762513667345047,
      "learning_rate": 2.456684768554435e-05,
      "loss": 0.7441,
      "step": 6835
    },
    {
      "epoch": 1.5663574945608612,
      "grad_norm": 0.2051253616809845,
      "learning_rate": 2.4502198086371867e-05,
      "loss": 0.743,
      "step": 6840
    },
    {
      "epoch": 1.5675025764342152,
      "grad_norm": 0.15214301645755768,
      "learning_rate": 2.443754848719938e-05,
      "loss": 0.7448,
      "step": 6845
    },
    {
      "epoch": 1.568647658307569,
      "grad_norm": 0.1648416966199875,
      "learning_rate": 2.4372898888026895e-05,
      "loss": 0.7378,
      "step": 6850
    },
    {
      "epoch": 1.5697927401809229,
      "grad_norm": 0.1519552320241928,
      "learning_rate": 2.430824928885441e-05,
      "loss": 0.7445,
      "step": 6855
    },
    {
      "epoch": 1.570937822054277,
      "grad_norm": 0.18640117347240448,
      "learning_rate": 2.4243599689681927e-05,
      "loss": 0.7416,
      "step": 6860
    },
    {
      "epoch": 1.572082903927631,
      "grad_norm": 0.19299635291099548,
      "learning_rate": 2.4178950090509438e-05,
      "loss": 0.7456,
      "step": 6865
    },
    {
      "epoch": 1.5732279858009848,
      "grad_norm": 0.19505447149276733,
      "learning_rate": 2.4114300491336955e-05,
      "loss": 0.7468,
      "step": 6870
    },
    {
      "epoch": 1.5743730676743386,
      "grad_norm": 0.17147518694400787,
      "learning_rate": 2.404965089216447e-05,
      "loss": 0.7414,
      "step": 6875
    },
    {
      "epoch": 1.5755181495476926,
      "grad_norm": 0.22106440365314484,
      "learning_rate": 2.3985001292991984e-05,
      "loss": 0.7444,
      "step": 6880
    },
    {
      "epoch": 1.5766632314210467,
      "grad_norm": 0.27256956696510315,
      "learning_rate": 2.3920351693819498e-05,
      "loss": 0.7448,
      "step": 6885
    },
    {
      "epoch": 1.5778083132944005,
      "grad_norm": 0.23254042863845825,
      "learning_rate": 2.3855702094647016e-05,
      "loss": 0.7449,
      "step": 6890
    },
    {
      "epoch": 1.5789533951677543,
      "grad_norm": 0.17433930933475494,
      "learning_rate": 2.3791052495474527e-05,
      "loss": 0.7433,
      "step": 6895
    },
    {
      "epoch": 1.5800984770411084,
      "grad_norm": 0.21645629405975342,
      "learning_rate": 2.3726402896302044e-05,
      "loss": 0.7444,
      "step": 6900
    },
    {
      "epoch": 1.5812435589144624,
      "grad_norm": 0.2150690108537674,
      "learning_rate": 2.3661753297129562e-05,
      "loss": 0.7435,
      "step": 6905
    },
    {
      "epoch": 1.5823886407878165,
      "grad_norm": 0.3046257793903351,
      "learning_rate": 2.3597103697957073e-05,
      "loss": 0.7452,
      "step": 6910
    },
    {
      "epoch": 1.5835337226611703,
      "grad_norm": 0.18164703249931335,
      "learning_rate": 2.353245409878459e-05,
      "loss": 0.7457,
      "step": 6915
    },
    {
      "epoch": 1.5846788045345241,
      "grad_norm": 0.16916827857494354,
      "learning_rate": 2.3467804499612104e-05,
      "loss": 0.7408,
      "step": 6920
    },
    {
      "epoch": 1.5858238864078782,
      "grad_norm": 0.17933885753154755,
      "learning_rate": 2.340315490043962e-05,
      "loss": 0.7411,
      "step": 6925
    },
    {
      "epoch": 1.5869689682812322,
      "grad_norm": 0.22185829281806946,
      "learning_rate": 2.3338505301267133e-05,
      "loss": 0.7463,
      "step": 6930
    },
    {
      "epoch": 1.588114050154586,
      "grad_norm": 0.21240991353988647,
      "learning_rate": 2.327385570209465e-05,
      "loss": 0.7432,
      "step": 6935
    },
    {
      "epoch": 1.58925913202794,
      "grad_norm": 0.1964518278837204,
      "learning_rate": 2.320920610292216e-05,
      "loss": 0.7425,
      "step": 6940
    },
    {
      "epoch": 1.590404213901294,
      "grad_norm": 0.19527554512023926,
      "learning_rate": 2.314455650374968e-05,
      "loss": 0.7439,
      "step": 6945
    },
    {
      "epoch": 1.591549295774648,
      "grad_norm": 0.16673538088798523,
      "learning_rate": 2.3079906904577193e-05,
      "loss": 0.7429,
      "step": 6950
    },
    {
      "epoch": 1.5926943776480018,
      "grad_norm": 0.2501964271068573,
      "learning_rate": 2.3015257305404707e-05,
      "loss": 0.7456,
      "step": 6955
    },
    {
      "epoch": 1.5938394595213556,
      "grad_norm": 0.21052591502666473,
      "learning_rate": 2.295060770623222e-05,
      "loss": 0.7451,
      "step": 6960
    },
    {
      "epoch": 1.5949845413947097,
      "grad_norm": 0.1688924878835678,
      "learning_rate": 2.288595810705974e-05,
      "loss": 0.7431,
      "step": 6965
    },
    {
      "epoch": 1.5961296232680637,
      "grad_norm": 0.2228127270936966,
      "learning_rate": 2.282130850788725e-05,
      "loss": 0.7368,
      "step": 6970
    },
    {
      "epoch": 1.5972747051414176,
      "grad_norm": 0.16818226873874664,
      "learning_rate": 2.2756658908714768e-05,
      "loss": 0.7402,
      "step": 6975
    },
    {
      "epoch": 1.5984197870147716,
      "grad_norm": 0.22381144762039185,
      "learning_rate": 2.2692009309542282e-05,
      "loss": 0.7424,
      "step": 6980
    },
    {
      "epoch": 1.5995648688881254,
      "grad_norm": 0.15830406546592712,
      "learning_rate": 2.2627359710369796e-05,
      "loss": 0.7477,
      "step": 6985
    },
    {
      "epoch": 1.6007099507614795,
      "grad_norm": 0.2143416404724121,
      "learning_rate": 2.256271011119731e-05,
      "loss": 0.7439,
      "step": 6990
    },
    {
      "epoch": 1.6018550326348335,
      "grad_norm": 0.20761322975158691,
      "learning_rate": 2.2498060512024828e-05,
      "loss": 0.7463,
      "step": 6995
    },
    {
      "epoch": 1.6030001145081874,
      "grad_norm": 0.23132385313510895,
      "learning_rate": 2.243341091285234e-05,
      "loss": 0.7459,
      "step": 7000
    },
    {
      "epoch": 1.6041451963815412,
      "grad_norm": 0.1932103931903839,
      "learning_rate": 2.2368761313679856e-05,
      "loss": 0.743,
      "step": 7005
    },
    {
      "epoch": 1.6052902782548952,
      "grad_norm": 0.18202821910381317,
      "learning_rate": 2.2304111714507374e-05,
      "loss": 0.742,
      "step": 7010
    },
    {
      "epoch": 1.6064353601282493,
      "grad_norm": 0.17258566617965698,
      "learning_rate": 2.2239462115334885e-05,
      "loss": 0.7414,
      "step": 7015
    },
    {
      "epoch": 1.607580442001603,
      "grad_norm": 0.223942369222641,
      "learning_rate": 2.2174812516162402e-05,
      "loss": 0.7452,
      "step": 7020
    },
    {
      "epoch": 1.608725523874957,
      "grad_norm": 0.15610426664352417,
      "learning_rate": 2.2110162916989917e-05,
      "loss": 0.744,
      "step": 7025
    },
    {
      "epoch": 1.609870605748311,
      "grad_norm": 0.1526680737733841,
      "learning_rate": 2.204551331781743e-05,
      "loss": 0.7444,
      "step": 7030
    },
    {
      "epoch": 1.611015687621665,
      "grad_norm": 0.15396279096603394,
      "learning_rate": 2.1980863718644945e-05,
      "loss": 0.738,
      "step": 7035
    },
    {
      "epoch": 1.6121607694950189,
      "grad_norm": 0.17607542872428894,
      "learning_rate": 2.1916214119472463e-05,
      "loss": 0.7445,
      "step": 7040
    },
    {
      "epoch": 1.6133058513683727,
      "grad_norm": 0.270062118768692,
      "learning_rate": 2.1851564520299974e-05,
      "loss": 0.7479,
      "step": 7045
    },
    {
      "epoch": 1.6144509332417267,
      "grad_norm": 0.18403637409210205,
      "learning_rate": 2.178691492112749e-05,
      "loss": 0.7425,
      "step": 7050
    },
    {
      "epoch": 1.6155960151150808,
      "grad_norm": 0.24163562059402466,
      "learning_rate": 2.1722265321955005e-05,
      "loss": 0.7418,
      "step": 7055
    },
    {
      "epoch": 1.6167410969884348,
      "grad_norm": 0.16625964641571045,
      "learning_rate": 2.165761572278252e-05,
      "loss": 0.7427,
      "step": 7060
    },
    {
      "epoch": 1.6178861788617886,
      "grad_norm": 0.19170767068862915,
      "learning_rate": 2.1592966123610034e-05,
      "loss": 0.7423,
      "step": 7065
    },
    {
      "epoch": 1.6190312607351425,
      "grad_norm": 0.1749299019575119,
      "learning_rate": 2.152831652443755e-05,
      "loss": 0.7442,
      "step": 7070
    },
    {
      "epoch": 1.6201763426084965,
      "grad_norm": 0.28893375396728516,
      "learning_rate": 2.1463666925265062e-05,
      "loss": 0.7441,
      "step": 7075
    },
    {
      "epoch": 1.6213214244818506,
      "grad_norm": 0.15919381380081177,
      "learning_rate": 2.139901732609258e-05,
      "loss": 0.7454,
      "step": 7080
    },
    {
      "epoch": 1.6224665063552044,
      "grad_norm": 0.18021824955940247,
      "learning_rate": 2.1334367726920094e-05,
      "loss": 0.7467,
      "step": 7085
    },
    {
      "epoch": 1.6236115882285582,
      "grad_norm": 0.21050041913986206,
      "learning_rate": 2.1269718127747608e-05,
      "loss": 0.7425,
      "step": 7090
    },
    {
      "epoch": 1.6247566701019123,
      "grad_norm": 0.1786871701478958,
      "learning_rate": 2.1205068528575122e-05,
      "loss": 0.7471,
      "step": 7095
    },
    {
      "epoch": 1.6259017519752663,
      "grad_norm": 0.18213413655757904,
      "learning_rate": 2.114041892940264e-05,
      "loss": 0.7435,
      "step": 7100
    },
    {
      "epoch": 1.6270468338486201,
      "grad_norm": 0.18353727459907532,
      "learning_rate": 2.107576933023015e-05,
      "loss": 0.7407,
      "step": 7105
    },
    {
      "epoch": 1.628191915721974,
      "grad_norm": 0.1717100739479065,
      "learning_rate": 2.101111973105767e-05,
      "loss": 0.7417,
      "step": 7110
    },
    {
      "epoch": 1.629336997595328,
      "grad_norm": 0.1539594978094101,
      "learning_rate": 2.0946470131885186e-05,
      "loss": 0.744,
      "step": 7115
    },
    {
      "epoch": 1.630482079468682,
      "grad_norm": 0.1720319539308548,
      "learning_rate": 2.0881820532712697e-05,
      "loss": 0.7472,
      "step": 7120
    },
    {
      "epoch": 1.6316271613420361,
      "grad_norm": 0.15920764207839966,
      "learning_rate": 2.0817170933540215e-05,
      "loss": 0.7439,
      "step": 7125
    },
    {
      "epoch": 1.63277224321539,
      "grad_norm": 0.2293880581855774,
      "learning_rate": 2.075252133436773e-05,
      "loss": 0.743,
      "step": 7130
    },
    {
      "epoch": 1.6339173250887438,
      "grad_norm": 0.16954880952835083,
      "learning_rate": 2.0687871735195243e-05,
      "loss": 0.7437,
      "step": 7135
    },
    {
      "epoch": 1.6350624069620978,
      "grad_norm": 0.2514009475708008,
      "learning_rate": 2.0623222136022757e-05,
      "loss": 0.7427,
      "step": 7140
    },
    {
      "epoch": 1.6362074888354519,
      "grad_norm": 0.1630130112171173,
      "learning_rate": 2.0558572536850275e-05,
      "loss": 0.7428,
      "step": 7145
    },
    {
      "epoch": 1.6373525707088057,
      "grad_norm": 0.2145601212978363,
      "learning_rate": 2.0493922937677786e-05,
      "loss": 0.7485,
      "step": 7150
    },
    {
      "epoch": 1.6384976525821595,
      "grad_norm": 0.2102821171283722,
      "learning_rate": 2.0429273338505303e-05,
      "loss": 0.7454,
      "step": 7155
    },
    {
      "epoch": 1.6396427344555136,
      "grad_norm": 0.24257567524909973,
      "learning_rate": 2.0364623739332818e-05,
      "loss": 0.7436,
      "step": 7160
    },
    {
      "epoch": 1.6407878163288676,
      "grad_norm": 0.198568657040596,
      "learning_rate": 2.0299974140160332e-05,
      "loss": 0.7436,
      "step": 7165
    },
    {
      "epoch": 1.6419328982022214,
      "grad_norm": 0.1587372124195099,
      "learning_rate": 2.0235324540987846e-05,
      "loss": 0.7453,
      "step": 7170
    },
    {
      "epoch": 1.6430779800755753,
      "grad_norm": 0.1822303682565689,
      "learning_rate": 2.0170674941815364e-05,
      "loss": 0.7407,
      "step": 7175
    },
    {
      "epoch": 1.6442230619489293,
      "grad_norm": 0.19832190871238708,
      "learning_rate": 2.0106025342642874e-05,
      "loss": 0.7437,
      "step": 7180
    },
    {
      "epoch": 1.6453681438222834,
      "grad_norm": 0.14971603453159332,
      "learning_rate": 2.0041375743470392e-05,
      "loss": 0.742,
      "step": 7185
    },
    {
      "epoch": 1.6465132256956372,
      "grad_norm": 0.22691278159618378,
      "learning_rate": 1.9976726144297906e-05,
      "loss": 0.7446,
      "step": 7190
    },
    {
      "epoch": 1.6476583075689912,
      "grad_norm": 0.2300955355167389,
      "learning_rate": 1.991207654512542e-05,
      "loss": 0.7449,
      "step": 7195
    },
    {
      "epoch": 1.648803389442345,
      "grad_norm": 0.22187408804893494,
      "learning_rate": 1.9847426945952935e-05,
      "loss": 0.7428,
      "step": 7200
    },
    {
      "epoch": 1.649948471315699,
      "grad_norm": 0.18663349747657776,
      "learning_rate": 1.9782777346780452e-05,
      "loss": 0.7431,
      "step": 7205
    },
    {
      "epoch": 1.6510935531890532,
      "grad_norm": 0.212881401181221,
      "learning_rate": 1.9718127747607963e-05,
      "loss": 0.7414,
      "step": 7210
    },
    {
      "epoch": 1.652238635062407,
      "grad_norm": 0.18411917984485626,
      "learning_rate": 1.965347814843548e-05,
      "loss": 0.738,
      "step": 7215
    },
    {
      "epoch": 1.6533837169357608,
      "grad_norm": 0.15346644818782806,
      "learning_rate": 1.9588828549262995e-05,
      "loss": 0.7422,
      "step": 7220
    },
    {
      "epoch": 1.6545287988091149,
      "grad_norm": 0.15337932109832764,
      "learning_rate": 1.952417895009051e-05,
      "loss": 0.7421,
      "step": 7225
    },
    {
      "epoch": 1.655673880682469,
      "grad_norm": 0.24280506372451782,
      "learning_rate": 1.9459529350918027e-05,
      "loss": 0.7395,
      "step": 7230
    },
    {
      "epoch": 1.6568189625558227,
      "grad_norm": 0.18352577090263367,
      "learning_rate": 1.939487975174554e-05,
      "loss": 0.7444,
      "step": 7235
    },
    {
      "epoch": 1.6579640444291766,
      "grad_norm": 0.1767406165599823,
      "learning_rate": 1.9330230152573055e-05,
      "loss": 0.7427,
      "step": 7240
    },
    {
      "epoch": 1.6591091263025306,
      "grad_norm": 0.20321263372898102,
      "learning_rate": 1.926558055340057e-05,
      "loss": 0.7432,
      "step": 7245
    },
    {
      "epoch": 1.6602542081758846,
      "grad_norm": 0.2268437296152115,
      "learning_rate": 1.9200930954228087e-05,
      "loss": 0.739,
      "step": 7250
    },
    {
      "epoch": 1.6613992900492385,
      "grad_norm": 0.25249117612838745,
      "learning_rate": 1.9136281355055598e-05,
      "loss": 0.7461,
      "step": 7255
    },
    {
      "epoch": 1.6625443719225923,
      "grad_norm": 0.24010123312473297,
      "learning_rate": 1.9071631755883115e-05,
      "loss": 0.741,
      "step": 7260
    },
    {
      "epoch": 1.6636894537959463,
      "grad_norm": 0.19712349772453308,
      "learning_rate": 1.900698215671063e-05,
      "loss": 0.738,
      "step": 7265
    },
    {
      "epoch": 1.6648345356693004,
      "grad_norm": 0.22672364115715027,
      "learning_rate": 1.8942332557538144e-05,
      "loss": 0.7448,
      "step": 7270
    },
    {
      "epoch": 1.6659796175426544,
      "grad_norm": 0.24020400643348694,
      "learning_rate": 1.8877682958365658e-05,
      "loss": 0.7407,
      "step": 7275
    },
    {
      "epoch": 1.6671246994160083,
      "grad_norm": 0.2094317078590393,
      "learning_rate": 1.8813033359193176e-05,
      "loss": 0.742,
      "step": 7280
    },
    {
      "epoch": 1.668269781289362,
      "grad_norm": 0.34122583270072937,
      "learning_rate": 1.8748383760020687e-05,
      "loss": 0.744,
      "step": 7285
    },
    {
      "epoch": 1.6694148631627161,
      "grad_norm": 0.29159530997276306,
      "learning_rate": 1.8683734160848204e-05,
      "loss": 0.7414,
      "step": 7290
    },
    {
      "epoch": 1.6705599450360702,
      "grad_norm": 0.2118174433708191,
      "learning_rate": 1.861908456167572e-05,
      "loss": 0.742,
      "step": 7295
    },
    {
      "epoch": 1.671705026909424,
      "grad_norm": 0.1960996836423874,
      "learning_rate": 1.8554434962503233e-05,
      "loss": 0.7412,
      "step": 7300
    },
    {
      "epoch": 1.6728501087827778,
      "grad_norm": 0.20879942178726196,
      "learning_rate": 1.8489785363330747e-05,
      "loss": 0.7422,
      "step": 7305
    },
    {
      "epoch": 1.673995190656132,
      "grad_norm": 0.1483440101146698,
      "learning_rate": 1.8425135764158264e-05,
      "loss": 0.742,
      "step": 7310
    },
    {
      "epoch": 1.675140272529486,
      "grad_norm": 0.15710671246051788,
      "learning_rate": 1.8360486164985775e-05,
      "loss": 0.7458,
      "step": 7315
    },
    {
      "epoch": 1.6762853544028398,
      "grad_norm": 0.15391378104686737,
      "learning_rate": 1.8295836565813293e-05,
      "loss": 0.7462,
      "step": 7320
    },
    {
      "epoch": 1.6774304362761936,
      "grad_norm": 0.15545009076595306,
      "learning_rate": 1.8231186966640807e-05,
      "loss": 0.7435,
      "step": 7325
    },
    {
      "epoch": 1.6785755181495476,
      "grad_norm": 0.16694271564483643,
      "learning_rate": 1.816653736746832e-05,
      "loss": 0.7436,
      "step": 7330
    },
    {
      "epoch": 1.6797206000229017,
      "grad_norm": 0.21401578187942505,
      "learning_rate": 1.810188776829584e-05,
      "loss": 0.7442,
      "step": 7335
    },
    {
      "epoch": 1.6808656818962557,
      "grad_norm": 0.1898529827594757,
      "learning_rate": 1.8037238169123353e-05,
      "loss": 0.743,
      "step": 7340
    },
    {
      "epoch": 1.6820107637696096,
      "grad_norm": 0.22047021985054016,
      "learning_rate": 1.7972588569950867e-05,
      "loss": 0.7419,
      "step": 7345
    },
    {
      "epoch": 1.6831558456429634,
      "grad_norm": 0.1430683583021164,
      "learning_rate": 1.790793897077838e-05,
      "loss": 0.7432,
      "step": 7350
    },
    {
      "epoch": 1.6843009275163174,
      "grad_norm": 0.1772889792919159,
      "learning_rate": 1.78432893716059e-05,
      "loss": 0.7423,
      "step": 7355
    },
    {
      "epoch": 1.6854460093896715,
      "grad_norm": 0.2118036150932312,
      "learning_rate": 1.777863977243341e-05,
      "loss": 0.743,
      "step": 7360
    },
    {
      "epoch": 1.6865910912630253,
      "grad_norm": 0.1891527622938156,
      "learning_rate": 1.7713990173260928e-05,
      "loss": 0.7424,
      "step": 7365
    },
    {
      "epoch": 1.6877361731363791,
      "grad_norm": 0.1738433688879013,
      "learning_rate": 1.7649340574088442e-05,
      "loss": 0.7403,
      "step": 7370
    },
    {
      "epoch": 1.6888812550097332,
      "grad_norm": 0.2036980539560318,
      "learning_rate": 1.7584690974915956e-05,
      "loss": 0.7409,
      "step": 7375
    },
    {
      "epoch": 1.6900263368830872,
      "grad_norm": 0.1397104263305664,
      "learning_rate": 1.752004137574347e-05,
      "loss": 0.7433,
      "step": 7380
    },
    {
      "epoch": 1.691171418756441,
      "grad_norm": 0.1923220157623291,
      "learning_rate": 1.7455391776570988e-05,
      "loss": 0.7432,
      "step": 7385
    },
    {
      "epoch": 1.6923165006297949,
      "grad_norm": 0.17131070792675018,
      "learning_rate": 1.73907421773985e-05,
      "loss": 0.7423,
      "step": 7390
    },
    {
      "epoch": 1.693461582503149,
      "grad_norm": 0.19466909766197205,
      "learning_rate": 1.7326092578226016e-05,
      "loss": 0.743,
      "step": 7395
    },
    {
      "epoch": 1.694606664376503,
      "grad_norm": 0.15853409469127655,
      "learning_rate": 1.726144297905353e-05,
      "loss": 0.7405,
      "step": 7400
    },
    {
      "epoch": 1.695751746249857,
      "grad_norm": 0.1643185168504715,
      "learning_rate": 1.7196793379881045e-05,
      "loss": 0.7466,
      "step": 7405
    },
    {
      "epoch": 1.6968968281232109,
      "grad_norm": 0.20322449505329132,
      "learning_rate": 1.713214378070856e-05,
      "loss": 0.7463,
      "step": 7410
    },
    {
      "epoch": 1.6980419099965647,
      "grad_norm": 0.1877737194299698,
      "learning_rate": 1.7067494181536077e-05,
      "loss": 0.737,
      "step": 7415
    },
    {
      "epoch": 1.6991869918699187,
      "grad_norm": 0.16088910400867462,
      "learning_rate": 1.7002844582363587e-05,
      "loss": 0.7466,
      "step": 7420
    },
    {
      "epoch": 1.7003320737432728,
      "grad_norm": 0.16621771454811096,
      "learning_rate": 1.6938194983191105e-05,
      "loss": 0.7427,
      "step": 7425
    },
    {
      "epoch": 1.7014771556166266,
      "grad_norm": 0.17676587402820587,
      "learning_rate": 1.687354538401862e-05,
      "loss": 0.7461,
      "step": 7430
    },
    {
      "epoch": 1.7026222374899804,
      "grad_norm": 0.22310510277748108,
      "learning_rate": 1.6808895784846133e-05,
      "loss": 0.7416,
      "step": 7435
    },
    {
      "epoch": 1.7037673193633345,
      "grad_norm": 0.1830226629972458,
      "learning_rate": 1.674424618567365e-05,
      "loss": 0.7393,
      "step": 7440
    },
    {
      "epoch": 1.7049124012366885,
      "grad_norm": 0.17793413996696472,
      "learning_rate": 1.6679596586501165e-05,
      "loss": 0.7447,
      "step": 7445
    },
    {
      "epoch": 1.7060574831100423,
      "grad_norm": 0.2182731181383133,
      "learning_rate": 1.661494698732868e-05,
      "loss": 0.7424,
      "step": 7450
    },
    {
      "epoch": 1.7072025649833962,
      "grad_norm": 0.15893946588039398,
      "learning_rate": 1.6550297388156194e-05,
      "loss": 0.7453,
      "step": 7455
    },
    {
      "epoch": 1.7083476468567502,
      "grad_norm": 0.18081578612327576,
      "learning_rate": 1.648564778898371e-05,
      "loss": 0.7473,
      "step": 7460
    },
    {
      "epoch": 1.7094927287301043,
      "grad_norm": 0.18333902955055237,
      "learning_rate": 1.6420998189811222e-05,
      "loss": 0.7386,
      "step": 7465
    },
    {
      "epoch": 1.710637810603458,
      "grad_norm": 0.18723733723163605,
      "learning_rate": 1.635634859063874e-05,
      "loss": 0.7439,
      "step": 7470
    },
    {
      "epoch": 1.7117828924768121,
      "grad_norm": 0.15768438577651978,
      "learning_rate": 1.6291698991466254e-05,
      "loss": 0.741,
      "step": 7475
    },
    {
      "epoch": 1.712927974350166,
      "grad_norm": 0.15292543172836304,
      "learning_rate": 1.6227049392293768e-05,
      "loss": 0.7439,
      "step": 7480
    },
    {
      "epoch": 1.71407305622352,
      "grad_norm": 0.15622615814208984,
      "learning_rate": 1.6162399793121282e-05,
      "loss": 0.7415,
      "step": 7485
    },
    {
      "epoch": 1.715218138096874,
      "grad_norm": 0.2150067538022995,
      "learning_rate": 1.60977501939488e-05,
      "loss": 0.7441,
      "step": 7490
    },
    {
      "epoch": 1.716363219970228,
      "grad_norm": 0.17251062393188477,
      "learning_rate": 1.603310059477631e-05,
      "loss": 0.7394,
      "step": 7495
    },
    {
      "epoch": 1.7175083018435817,
      "grad_norm": 0.1846706122159958,
      "learning_rate": 1.596845099560383e-05,
      "loss": 0.7433,
      "step": 7500
    },
    {
      "epoch": 1.7186533837169358,
      "grad_norm": 0.16406218707561493,
      "learning_rate": 1.5903801396431343e-05,
      "loss": 0.7412,
      "step": 7505
    },
    {
      "epoch": 1.7197984655902898,
      "grad_norm": 0.17380113899707794,
      "learning_rate": 1.5839151797258857e-05,
      "loss": 0.7407,
      "step": 7510
    },
    {
      "epoch": 1.7209435474636436,
      "grad_norm": 0.18964682519435883,
      "learning_rate": 1.577450219808637e-05,
      "loss": 0.7447,
      "step": 7515
    },
    {
      "epoch": 1.7220886293369975,
      "grad_norm": 0.24054868519306183,
      "learning_rate": 1.570985259891389e-05,
      "loss": 0.7437,
      "step": 7520
    },
    {
      "epoch": 1.7232337112103515,
      "grad_norm": 0.206729918718338,
      "learning_rate": 1.56452029997414e-05,
      "loss": 0.7403,
      "step": 7525
    },
    {
      "epoch": 1.7243787930837056,
      "grad_norm": 0.14928388595581055,
      "learning_rate": 1.5580553400568917e-05,
      "loss": 0.7429,
      "step": 7530
    },
    {
      "epoch": 1.7255238749570594,
      "grad_norm": 0.13767731189727783,
      "learning_rate": 1.551590380139643e-05,
      "loss": 0.7423,
      "step": 7535
    },
    {
      "epoch": 1.7266689568304132,
      "grad_norm": 0.21498754620552063,
      "learning_rate": 1.5451254202223946e-05,
      "loss": 0.7395,
      "step": 7540
    },
    {
      "epoch": 1.7278140387037673,
      "grad_norm": 0.19207172095775604,
      "learning_rate": 1.538660460305146e-05,
      "loss": 0.7423,
      "step": 7545
    },
    {
      "epoch": 1.7289591205771213,
      "grad_norm": 0.16116943955421448,
      "learning_rate": 1.5321955003878977e-05,
      "loss": 0.7404,
      "step": 7550
    },
    {
      "epoch": 1.7301042024504754,
      "grad_norm": 0.18703986704349518,
      "learning_rate": 1.5257305404706493e-05,
      "loss": 0.7387,
      "step": 7555
    },
    {
      "epoch": 1.7312492843238292,
      "grad_norm": 0.14413946866989136,
      "learning_rate": 1.5192655805534006e-05,
      "loss": 0.745,
      "step": 7560
    },
    {
      "epoch": 1.732394366197183,
      "grad_norm": 0.18334537744522095,
      "learning_rate": 1.5128006206361522e-05,
      "loss": 0.7456,
      "step": 7565
    },
    {
      "epoch": 1.733539448070537,
      "grad_norm": 0.1862698346376419,
      "learning_rate": 1.5063356607189036e-05,
      "loss": 0.7441,
      "step": 7570
    },
    {
      "epoch": 1.734684529943891,
      "grad_norm": 0.18286314606666565,
      "learning_rate": 1.4998707008016552e-05,
      "loss": 0.743,
      "step": 7575
    },
    {
      "epoch": 1.735829611817245,
      "grad_norm": 0.17219437658786774,
      "learning_rate": 1.4934057408844066e-05,
      "loss": 0.7403,
      "step": 7580
    },
    {
      "epoch": 1.7369746936905988,
      "grad_norm": 0.22169716656208038,
      "learning_rate": 1.4869407809671582e-05,
      "loss": 0.7379,
      "step": 7585
    },
    {
      "epoch": 1.7381197755639528,
      "grad_norm": 0.19314883649349213,
      "learning_rate": 1.4804758210499095e-05,
      "loss": 0.743,
      "step": 7590
    },
    {
      "epoch": 1.7392648574373069,
      "grad_norm": 0.22871583700180054,
      "learning_rate": 1.474010861132661e-05,
      "loss": 0.742,
      "step": 7595
    },
    {
      "epoch": 1.7404099393106607,
      "grad_norm": 0.17754560708999634,
      "learning_rate": 1.4675459012154125e-05,
      "loss": 0.7414,
      "step": 7600
    },
    {
      "epoch": 1.7415550211840145,
      "grad_norm": 0.16177891194820404,
      "learning_rate": 1.461080941298164e-05,
      "loss": 0.7418,
      "step": 7605
    },
    {
      "epoch": 1.7427001030573686,
      "grad_norm": 0.17212548851966858,
      "learning_rate": 1.4546159813809155e-05,
      "loss": 0.7438,
      "step": 7610
    },
    {
      "epoch": 1.7438451849307226,
      "grad_norm": 0.20453935861587524,
      "learning_rate": 1.448151021463667e-05,
      "loss": 0.7383,
      "step": 7615
    },
    {
      "epoch": 1.7449902668040767,
      "grad_norm": 0.17530488967895508,
      "learning_rate": 1.4416860615464183e-05,
      "loss": 0.7422,
      "step": 7620
    },
    {
      "epoch": 1.7461353486774305,
      "grad_norm": 0.15790791809558868,
      "learning_rate": 1.43522110162917e-05,
      "loss": 0.7426,
      "step": 7625
    },
    {
      "epoch": 1.7472804305507843,
      "grad_norm": 0.19352249801158905,
      "learning_rate": 1.4287561417119213e-05,
      "loss": 0.744,
      "step": 7630
    },
    {
      "epoch": 1.7484255124241384,
      "grad_norm": 0.19556356966495514,
      "learning_rate": 1.422291181794673e-05,
      "loss": 0.7412,
      "step": 7635
    },
    {
      "epoch": 1.7495705942974924,
      "grad_norm": 0.1878131479024887,
      "learning_rate": 1.4158262218774244e-05,
      "loss": 0.7456,
      "step": 7640
    },
    {
      "epoch": 1.7507156761708462,
      "grad_norm": 0.14573533833026886,
      "learning_rate": 1.409361261960176e-05,
      "loss": 0.743,
      "step": 7645
    },
    {
      "epoch": 1.7518607580442,
      "grad_norm": 0.15882503986358643,
      "learning_rate": 1.4028963020429272e-05,
      "loss": 0.7424,
      "step": 7650
    },
    {
      "epoch": 1.753005839917554,
      "grad_norm": 0.14983394742012024,
      "learning_rate": 1.396431342125679e-05,
      "loss": 0.745,
      "step": 7655
    },
    {
      "epoch": 1.7541509217909081,
      "grad_norm": 0.13475537300109863,
      "learning_rate": 1.3899663822084306e-05,
      "loss": 0.739,
      "step": 7660
    },
    {
      "epoch": 1.755296003664262,
      "grad_norm": 0.20346559584140778,
      "learning_rate": 1.3835014222911818e-05,
      "loss": 0.7376,
      "step": 7665
    },
    {
      "epoch": 1.7564410855376158,
      "grad_norm": 0.18719197809696198,
      "learning_rate": 1.3770364623739334e-05,
      "loss": 0.7371,
      "step": 7670
    },
    {
      "epoch": 1.7575861674109698,
      "grad_norm": 0.2041907161474228,
      "learning_rate": 1.3705715024566848e-05,
      "loss": 0.745,
      "step": 7675
    },
    {
      "epoch": 1.758731249284324,
      "grad_norm": 0.17985425889492035,
      "learning_rate": 1.3641065425394364e-05,
      "loss": 0.739,
      "step": 7680
    },
    {
      "epoch": 1.7598763311576777,
      "grad_norm": 0.1952625960111618,
      "learning_rate": 1.3576415826221878e-05,
      "loss": 0.7407,
      "step": 7685
    },
    {
      "epoch": 1.7610214130310318,
      "grad_norm": 0.19944436848163605,
      "learning_rate": 1.3511766227049394e-05,
      "loss": 0.7432,
      "step": 7690
    },
    {
      "epoch": 1.7621664949043856,
      "grad_norm": 0.1849888414144516,
      "learning_rate": 1.3447116627876907e-05,
      "loss": 0.7385,
      "step": 7695
    },
    {
      "epoch": 1.7633115767777396,
      "grad_norm": 0.20331071317195892,
      "learning_rate": 1.3382467028704423e-05,
      "loss": 0.7407,
      "step": 7700
    },
    {
      "epoch": 1.7644566586510937,
      "grad_norm": 0.15034550428390503,
      "learning_rate": 1.3317817429531937e-05,
      "loss": 0.742,
      "step": 7705
    },
    {
      "epoch": 1.7656017405244475,
      "grad_norm": 0.19090567529201508,
      "learning_rate": 1.3253167830359453e-05,
      "loss": 0.7434,
      "step": 7710
    },
    {
      "epoch": 1.7667468223978013,
      "grad_norm": 0.19512049853801727,
      "learning_rate": 1.3188518231186967e-05,
      "loss": 0.7418,
      "step": 7715
    },
    {
      "epoch": 1.7678919042711554,
      "grad_norm": 0.15119600296020508,
      "learning_rate": 1.3123868632014483e-05,
      "loss": 0.7441,
      "step": 7720
    },
    {
      "epoch": 1.7690369861445094,
      "grad_norm": 0.16292977333068848,
      "learning_rate": 1.3059219032841995e-05,
      "loss": 0.741,
      "step": 7725
    },
    {
      "epoch": 1.7701820680178633,
      "grad_norm": 0.14958208799362183,
      "learning_rate": 1.2994569433669513e-05,
      "loss": 0.7434,
      "step": 7730
    },
    {
      "epoch": 1.771327149891217,
      "grad_norm": 0.1666942834854126,
      "learning_rate": 1.2929919834497026e-05,
      "loss": 0.7415,
      "step": 7735
    },
    {
      "epoch": 1.7724722317645711,
      "grad_norm": 0.17329080402851105,
      "learning_rate": 1.2865270235324542e-05,
      "loss": 0.7401,
      "step": 7740
    },
    {
      "epoch": 1.7736173136379252,
      "grad_norm": 0.18390893936157227,
      "learning_rate": 1.2800620636152056e-05,
      "loss": 0.7392,
      "step": 7745
    },
    {
      "epoch": 1.774762395511279,
      "grad_norm": 0.20777173340320587,
      "learning_rate": 1.2735971036979572e-05,
      "loss": 0.7461,
      "step": 7750
    },
    {
      "epoch": 1.7759074773846328,
      "grad_norm": 0.16422134637832642,
      "learning_rate": 1.2671321437807084e-05,
      "loss": 0.7416,
      "step": 7755
    },
    {
      "epoch": 1.7770525592579869,
      "grad_norm": 0.20069563388824463,
      "learning_rate": 1.2606671838634602e-05,
      "loss": 0.7422,
      "step": 7760
    },
    {
      "epoch": 1.778197641131341,
      "grad_norm": 0.1575007140636444,
      "learning_rate": 1.2542022239462118e-05,
      "loss": 0.736,
      "step": 7765
    },
    {
      "epoch": 1.779342723004695,
      "grad_norm": 0.1856757402420044,
      "learning_rate": 1.247737264028963e-05,
      "loss": 0.7456,
      "step": 7770
    },
    {
      "epoch": 1.7804878048780488,
      "grad_norm": 0.1524512767791748,
      "learning_rate": 1.2412723041117146e-05,
      "loss": 0.7419,
      "step": 7775
    },
    {
      "epoch": 1.7816328867514026,
      "grad_norm": 0.17118653655052185,
      "learning_rate": 1.234807344194466e-05,
      "loss": 0.7465,
      "step": 7780
    },
    {
      "epoch": 1.7827779686247567,
      "grad_norm": 0.17673975229263306,
      "learning_rate": 1.2283423842772175e-05,
      "loss": 0.7377,
      "step": 7785
    },
    {
      "epoch": 1.7839230504981107,
      "grad_norm": 0.16801375150680542,
      "learning_rate": 1.221877424359969e-05,
      "loss": 0.7417,
      "step": 7790
    },
    {
      "epoch": 1.7850681323714646,
      "grad_norm": 0.21200990676879883,
      "learning_rate": 1.2154124644427205e-05,
      "loss": 0.7421,
      "step": 7795
    },
    {
      "epoch": 1.7862132142448184,
      "grad_norm": 0.1710125207901001,
      "learning_rate": 1.2089475045254719e-05,
      "loss": 0.7392,
      "step": 7800
    },
    {
      "epoch": 1.7873582961181724,
      "grad_norm": 0.17124149203300476,
      "learning_rate": 1.2024825446082235e-05,
      "loss": 0.7419,
      "step": 7805
    },
    {
      "epoch": 1.7885033779915265,
      "grad_norm": 0.1881469190120697,
      "learning_rate": 1.1960175846909749e-05,
      "loss": 0.7431,
      "step": 7810
    },
    {
      "epoch": 1.7896484598648803,
      "grad_norm": 0.138245090842247,
      "learning_rate": 1.1895526247737263e-05,
      "loss": 0.7416,
      "step": 7815
    },
    {
      "epoch": 1.7907935417382341,
      "grad_norm": 0.19728197157382965,
      "learning_rate": 1.1830876648564781e-05,
      "loss": 0.7388,
      "step": 7820
    },
    {
      "epoch": 1.7919386236115882,
      "grad_norm": 0.15021589398384094,
      "learning_rate": 1.1766227049392295e-05,
      "loss": 0.7395,
      "step": 7825
    },
    {
      "epoch": 1.7930837054849422,
      "grad_norm": 0.1715056300163269,
      "learning_rate": 1.170157745021981e-05,
      "loss": 0.7404,
      "step": 7830
    },
    {
      "epoch": 1.7942287873582963,
      "grad_norm": 0.14820553362369537,
      "learning_rate": 1.1636927851047325e-05,
      "loss": 0.745,
      "step": 7835
    },
    {
      "epoch": 1.79537386923165,
      "grad_norm": 0.1573396623134613,
      "learning_rate": 1.157227825187484e-05,
      "loss": 0.7412,
      "step": 7840
    },
    {
      "epoch": 1.796518951105004,
      "grad_norm": 0.17530624568462372,
      "learning_rate": 1.1507628652702354e-05,
      "loss": 0.7416,
      "step": 7845
    },
    {
      "epoch": 1.797664032978358,
      "grad_norm": 0.1957489401102066,
      "learning_rate": 1.144297905352987e-05,
      "loss": 0.7417,
      "step": 7850
    },
    {
      "epoch": 1.798809114851712,
      "grad_norm": 0.14243096113204956,
      "learning_rate": 1.1378329454357384e-05,
      "loss": 0.7407,
      "step": 7855
    },
    {
      "epoch": 1.7999541967250658,
      "grad_norm": 0.17198410630226135,
      "learning_rate": 1.1313679855184898e-05,
      "loss": 0.7396,
      "step": 7860
    },
    {
      "epoch": 1.8010992785984197,
      "grad_norm": 0.170758455991745,
      "learning_rate": 1.1249030256012414e-05,
      "loss": 0.7431,
      "step": 7865
    },
    {
      "epoch": 1.8022443604717737,
      "grad_norm": 0.1600438803434372,
      "learning_rate": 1.1184380656839928e-05,
      "loss": 0.7441,
      "step": 7870
    },
    {
      "epoch": 1.8033894423451278,
      "grad_norm": 0.18139535188674927,
      "learning_rate": 1.1119731057667442e-05,
      "loss": 0.7433,
      "step": 7875
    },
    {
      "epoch": 1.8045345242184816,
      "grad_norm": 0.24969196319580078,
      "learning_rate": 1.1055081458494958e-05,
      "loss": 0.7426,
      "step": 7880
    },
    {
      "epoch": 1.8056796060918354,
      "grad_norm": 0.16466961801052094,
      "learning_rate": 1.0990431859322473e-05,
      "loss": 0.7426,
      "step": 7885
    },
    {
      "epoch": 1.8068246879651895,
      "grad_norm": 0.19945576786994934,
      "learning_rate": 1.0925782260149987e-05,
      "loss": 0.7396,
      "step": 7890
    },
    {
      "epoch": 1.8079697698385435,
      "grad_norm": 0.2034262865781784,
      "learning_rate": 1.0861132660977503e-05,
      "loss": 0.7378,
      "step": 7895
    },
    {
      "epoch": 1.8091148517118976,
      "grad_norm": 0.16968347132205963,
      "learning_rate": 1.0796483061805017e-05,
      "loss": 0.7384,
      "step": 7900
    },
    {
      "epoch": 1.8102599335852514,
      "grad_norm": 0.20821303129196167,
      "learning_rate": 1.0731833462632531e-05,
      "loss": 0.7362,
      "step": 7905
    },
    {
      "epoch": 1.8114050154586052,
      "grad_norm": 0.20150309801101685,
      "learning_rate": 1.0667183863460047e-05,
      "loss": 0.7452,
      "step": 7910
    },
    {
      "epoch": 1.8125500973319593,
      "grad_norm": 0.16308322548866272,
      "learning_rate": 1.0602534264287561e-05,
      "loss": 0.7387,
      "step": 7915
    },
    {
      "epoch": 1.8136951792053133,
      "grad_norm": 0.19306686520576477,
      "learning_rate": 1.0537884665115075e-05,
      "loss": 0.7388,
      "step": 7920
    },
    {
      "epoch": 1.8148402610786671,
      "grad_norm": 0.1482827216386795,
      "learning_rate": 1.0473235065942593e-05,
      "loss": 0.7394,
      "step": 7925
    },
    {
      "epoch": 1.815985342952021,
      "grad_norm": 0.16145211458206177,
      "learning_rate": 1.0408585466770107e-05,
      "loss": 0.7387,
      "step": 7930
    },
    {
      "epoch": 1.817130424825375,
      "grad_norm": 0.15279000997543335,
      "learning_rate": 1.0343935867597622e-05,
      "loss": 0.7397,
      "step": 7935
    },
    {
      "epoch": 1.818275506698729,
      "grad_norm": 0.15904763340950012,
      "learning_rate": 1.0279286268425137e-05,
      "loss": 0.7404,
      "step": 7940
    },
    {
      "epoch": 1.8194205885720829,
      "grad_norm": 0.14108118414878845,
      "learning_rate": 1.0214636669252652e-05,
      "loss": 0.7407,
      "step": 7945
    },
    {
      "epoch": 1.8205656704454367,
      "grad_norm": 0.14389878511428833,
      "learning_rate": 1.0149987070080166e-05,
      "loss": 0.7454,
      "step": 7950
    },
    {
      "epoch": 1.8217107523187908,
      "grad_norm": 0.15871091187000275,
      "learning_rate": 1.0085337470907682e-05,
      "loss": 0.7398,
      "step": 7955
    },
    {
      "epoch": 1.8228558341921448,
      "grad_norm": 0.17688189446926117,
      "learning_rate": 1.0020687871735196e-05,
      "loss": 0.7433,
      "step": 7960
    },
    {
      "epoch": 1.8240009160654986,
      "grad_norm": 0.14944761991500854,
      "learning_rate": 9.95603827256271e-06,
      "loss": 0.7453,
      "step": 7965
    },
    {
      "epoch": 1.8251459979388527,
      "grad_norm": 0.19329968094825745,
      "learning_rate": 9.891388673390226e-06,
      "loss": 0.744,
      "step": 7970
    },
    {
      "epoch": 1.8262910798122065,
      "grad_norm": 0.15224604308605194,
      "learning_rate": 9.82673907421774e-06,
      "loss": 0.7419,
      "step": 7975
    },
    {
      "epoch": 1.8274361616855606,
      "grad_norm": 0.19345805048942566,
      "learning_rate": 9.762089475045255e-06,
      "loss": 0.7453,
      "step": 7980
    },
    {
      "epoch": 1.8285812435589146,
      "grad_norm": 0.17446333169937134,
      "learning_rate": 9.69743987587277e-06,
      "loss": 0.7404,
      "step": 7985
    },
    {
      "epoch": 1.8297263254322684,
      "grad_norm": 0.14764748513698578,
      "learning_rate": 9.632790276700285e-06,
      "loss": 0.7413,
      "step": 7990
    },
    {
      "epoch": 1.8308714073056223,
      "grad_norm": 0.15941238403320312,
      "learning_rate": 9.568140677527799e-06,
      "loss": 0.7429,
      "step": 7995
    },
    {
      "epoch": 1.8320164891789763,
      "grad_norm": 0.1465342789888382,
      "learning_rate": 9.503491078355315e-06,
      "loss": 0.7422,
      "step": 8000
    },
    {
      "epoch": 1.8331615710523304,
      "grad_norm": 0.15987376868724823,
      "learning_rate": 9.438841479182829e-06,
      "loss": 0.7432,
      "step": 8005
    },
    {
      "epoch": 1.8343066529256842,
      "grad_norm": 0.153104767203331,
      "learning_rate": 9.374191880010343e-06,
      "loss": 0.7391,
      "step": 8010
    },
    {
      "epoch": 1.835451734799038,
      "grad_norm": 0.14702925086021423,
      "learning_rate": 9.30954228083786e-06,
      "loss": 0.742,
      "step": 8015
    },
    {
      "epoch": 1.836596816672392,
      "grad_norm": 0.14339160919189453,
      "learning_rate": 9.244892681665373e-06,
      "loss": 0.743,
      "step": 8020
    },
    {
      "epoch": 1.837741898545746,
      "grad_norm": 0.16755425930023193,
      "learning_rate": 9.180243082492888e-06,
      "loss": 0.7434,
      "step": 8025
    },
    {
      "epoch": 1.8388869804191,
      "grad_norm": 0.19335520267486572,
      "learning_rate": 9.115593483320404e-06,
      "loss": 0.7422,
      "step": 8030
    },
    {
      "epoch": 1.8400320622924538,
      "grad_norm": 0.1825624406337738,
      "learning_rate": 9.05094388414792e-06,
      "loss": 0.741,
      "step": 8035
    },
    {
      "epoch": 1.8411771441658078,
      "grad_norm": 0.1814911663532257,
      "learning_rate": 8.986294284975434e-06,
      "loss": 0.7431,
      "step": 8040
    },
    {
      "epoch": 1.8423222260391618,
      "grad_norm": 0.16757063567638397,
      "learning_rate": 8.92164468580295e-06,
      "loss": 0.7367,
      "step": 8045
    },
    {
      "epoch": 1.843467307912516,
      "grad_norm": 0.18699325621128082,
      "learning_rate": 8.856995086630464e-06,
      "loss": 0.7431,
      "step": 8050
    },
    {
      "epoch": 1.8446123897858697,
      "grad_norm": 0.15374694764614105,
      "learning_rate": 8.792345487457978e-06,
      "loss": 0.7463,
      "step": 8055
    },
    {
      "epoch": 1.8457574716592235,
      "grad_norm": 0.17602622509002686,
      "learning_rate": 8.727695888285494e-06,
      "loss": 0.7433,
      "step": 8060
    },
    {
      "epoch": 1.8469025535325776,
      "grad_norm": 0.15638279914855957,
      "learning_rate": 8.663046289113008e-06,
      "loss": 0.7399,
      "step": 8065
    },
    {
      "epoch": 1.8480476354059316,
      "grad_norm": 0.16193757951259613,
      "learning_rate": 8.598396689940522e-06,
      "loss": 0.7422,
      "step": 8070
    },
    {
      "epoch": 1.8491927172792855,
      "grad_norm": 0.17268849909305573,
      "learning_rate": 8.533747090768038e-06,
      "loss": 0.7448,
      "step": 8075
    },
    {
      "epoch": 1.8503377991526393,
      "grad_norm": 0.14754731953144073,
      "learning_rate": 8.469097491595553e-06,
      "loss": 0.7401,
      "step": 8080
    },
    {
      "epoch": 1.8514828810259933,
      "grad_norm": 0.17625221610069275,
      "learning_rate": 8.404447892423067e-06,
      "loss": 0.7385,
      "step": 8085
    },
    {
      "epoch": 1.8526279628993474,
      "grad_norm": 0.14788591861724854,
      "learning_rate": 8.339798293250583e-06,
      "loss": 0.7409,
      "step": 8090
    },
    {
      "epoch": 1.8537730447727012,
      "grad_norm": 0.18604493141174316,
      "learning_rate": 8.275148694078097e-06,
      "loss": 0.7394,
      "step": 8095
    },
    {
      "epoch": 1.854918126646055,
      "grad_norm": 0.16041810810565948,
      "learning_rate": 8.210499094905611e-06,
      "loss": 0.7386,
      "step": 8100
    },
    {
      "epoch": 1.856063208519409,
      "grad_norm": 0.18122714757919312,
      "learning_rate": 8.145849495733127e-06,
      "loss": 0.7393,
      "step": 8105
    },
    {
      "epoch": 1.8572082903927631,
      "grad_norm": 0.16307488083839417,
      "learning_rate": 8.081199896560641e-06,
      "loss": 0.7413,
      "step": 8110
    },
    {
      "epoch": 1.8583533722661172,
      "grad_norm": 0.22801965475082397,
      "learning_rate": 8.016550297388155e-06,
      "loss": 0.7459,
      "step": 8115
    },
    {
      "epoch": 1.859498454139471,
      "grad_norm": 0.18238000571727753,
      "learning_rate": 7.951900698215671e-06,
      "loss": 0.74,
      "step": 8120
    },
    {
      "epoch": 1.8606435360128248,
      "grad_norm": 0.20741623640060425,
      "learning_rate": 7.887251099043186e-06,
      "loss": 0.7382,
      "step": 8125
    },
    {
      "epoch": 1.8617886178861789,
      "grad_norm": 0.22925831377506256,
      "learning_rate": 7.8226014998707e-06,
      "loss": 0.7391,
      "step": 8130
    },
    {
      "epoch": 1.862933699759533,
      "grad_norm": 0.16230937838554382,
      "learning_rate": 7.757951900698216e-06,
      "loss": 0.7397,
      "step": 8135
    },
    {
      "epoch": 1.8640787816328868,
      "grad_norm": 0.17243419587612152,
      "learning_rate": 7.69330230152573e-06,
      "loss": 0.7401,
      "step": 8140
    },
    {
      "epoch": 1.8652238635062406,
      "grad_norm": 0.1455468386411667,
      "learning_rate": 7.628652702353247e-06,
      "loss": 0.7411,
      "step": 8145
    },
    {
      "epoch": 1.8663689453795946,
      "grad_norm": 0.1580912470817566,
      "learning_rate": 7.564003103180761e-06,
      "loss": 0.7408,
      "step": 8150
    },
    {
      "epoch": 1.8675140272529487,
      "grad_norm": 0.18351049721240997,
      "learning_rate": 7.499353504008276e-06,
      "loss": 0.7425,
      "step": 8155
    },
    {
      "epoch": 1.8686591091263025,
      "grad_norm": 0.1617344319820404,
      "learning_rate": 7.434703904835791e-06,
      "loss": 0.7398,
      "step": 8160
    },
    {
      "epoch": 1.8698041909996563,
      "grad_norm": 0.14651767909526825,
      "learning_rate": 7.370054305663305e-06,
      "loss": 0.7403,
      "step": 8165
    },
    {
      "epoch": 1.8709492728730104,
      "grad_norm": 0.15270645916461945,
      "learning_rate": 7.30540470649082e-06,
      "loss": 0.7394,
      "step": 8170
    },
    {
      "epoch": 1.8720943547463644,
      "grad_norm": 0.15083663165569305,
      "learning_rate": 7.240755107318335e-06,
      "loss": 0.7384,
      "step": 8175
    },
    {
      "epoch": 1.8732394366197183,
      "grad_norm": 0.18049649894237518,
      "learning_rate": 7.17610550814585e-06,
      "loss": 0.7441,
      "step": 8180
    },
    {
      "epoch": 1.8743845184930723,
      "grad_norm": 0.18089523911476135,
      "learning_rate": 7.111455908973365e-06,
      "loss": 0.7386,
      "step": 8185
    },
    {
      "epoch": 1.8755296003664261,
      "grad_norm": 0.19794507324695587,
      "learning_rate": 7.04680630980088e-06,
      "loss": 0.7408,
      "step": 8190
    },
    {
      "epoch": 1.8766746822397802,
      "grad_norm": 0.15723812580108643,
      "learning_rate": 6.982156710628395e-06,
      "loss": 0.7404,
      "step": 8195
    },
    {
      "epoch": 1.8778197641131342,
      "grad_norm": 0.14659984409809113,
      "learning_rate": 6.917507111455909e-06,
      "loss": 0.7427,
      "step": 8200
    },
    {
      "epoch": 1.878964845986488,
      "grad_norm": 0.14966751635074615,
      "learning_rate": 6.852857512283424e-06,
      "loss": 0.7358,
      "step": 8205
    },
    {
      "epoch": 1.8801099278598419,
      "grad_norm": 0.1722017079591751,
      "learning_rate": 6.788207913110939e-06,
      "loss": 0.7426,
      "step": 8210
    },
    {
      "epoch": 1.881255009733196,
      "grad_norm": 0.15833428502082825,
      "learning_rate": 6.723558313938453e-06,
      "loss": 0.7438,
      "step": 8215
    },
    {
      "epoch": 1.88240009160655,
      "grad_norm": 0.15574586391448975,
      "learning_rate": 6.6589087147659685e-06,
      "loss": 0.7382,
      "step": 8220
    },
    {
      "epoch": 1.8835451734799038,
      "grad_norm": 0.18555963039398193,
      "learning_rate": 6.5942591155934835e-06,
      "loss": 0.7406,
      "step": 8225
    },
    {
      "epoch": 1.8846902553532576,
      "grad_norm": 0.14833788573741913,
      "learning_rate": 6.529609516420998e-06,
      "loss": 0.7443,
      "step": 8230
    },
    {
      "epoch": 1.8858353372266117,
      "grad_norm": 0.15129245817661285,
      "learning_rate": 6.464959917248513e-06,
      "loss": 0.7388,
      "step": 8235
    },
    {
      "epoch": 1.8869804190999657,
      "grad_norm": 0.16850748658180237,
      "learning_rate": 6.400310318076028e-06,
      "loss": 0.741,
      "step": 8240
    },
    {
      "epoch": 1.8881255009733195,
      "grad_norm": 0.1815343052148819,
      "learning_rate": 6.335660718903542e-06,
      "loss": 0.7415,
      "step": 8245
    },
    {
      "epoch": 1.8892705828466734,
      "grad_norm": 0.17904560267925262,
      "learning_rate": 6.271011119731059e-06,
      "loss": 0.741,
      "step": 8250
    },
    {
      "epoch": 1.8904156647200274,
      "grad_norm": 0.1715519279241562,
      "learning_rate": 6.206361520558573e-06,
      "loss": 0.7427,
      "step": 8255
    },
    {
      "epoch": 1.8915607465933815,
      "grad_norm": 0.16421706974506378,
      "learning_rate": 6.141711921386087e-06,
      "loss": 0.7419,
      "step": 8260
    },
    {
      "epoch": 1.8927058284667355,
      "grad_norm": 0.150858074426651,
      "learning_rate": 6.077062322213602e-06,
      "loss": 0.7424,
      "step": 8265
    },
    {
      "epoch": 1.8938509103400893,
      "grad_norm": 0.19547419250011444,
      "learning_rate": 6.0124127230411174e-06,
      "loss": 0.741,
      "step": 8270
    },
    {
      "epoch": 1.8949959922134432,
      "grad_norm": 0.1885257363319397,
      "learning_rate": 5.947763123868632e-06,
      "loss": 0.7398,
      "step": 8275
    },
    {
      "epoch": 1.8961410740867972,
      "grad_norm": 0.17618173360824585,
      "learning_rate": 5.8831135246961476e-06,
      "loss": 0.7426,
      "step": 8280
    },
    {
      "epoch": 1.8972861559601513,
      "grad_norm": 0.19151267409324646,
      "learning_rate": 5.818463925523663e-06,
      "loss": 0.7419,
      "step": 8285
    },
    {
      "epoch": 1.898431237833505,
      "grad_norm": 0.17066292464733124,
      "learning_rate": 5.753814326351177e-06,
      "loss": 0.7412,
      "step": 8290
    },
    {
      "epoch": 1.899576319706859,
      "grad_norm": 0.17010915279388428,
      "learning_rate": 5.689164727178692e-06,
      "loss": 0.7413,
      "step": 8295
    },
    {
      "epoch": 1.900721401580213,
      "grad_norm": 0.14419379830360413,
      "learning_rate": 5.624515128006207e-06,
      "loss": 0.7415,
      "step": 8300
    },
    {
      "epoch": 1.901866483453567,
      "grad_norm": 0.15456563234329224,
      "learning_rate": 5.559865528833721e-06,
      "loss": 0.7393,
      "step": 8305
    },
    {
      "epoch": 1.9030115653269208,
      "grad_norm": 0.15366721153259277,
      "learning_rate": 5.495215929661236e-06,
      "loss": 0.7393,
      "step": 8310
    },
    {
      "epoch": 1.9041566472002747,
      "grad_norm": 0.13521674275398254,
      "learning_rate": 5.430566330488751e-06,
      "loss": 0.7416,
      "step": 8315
    },
    {
      "epoch": 1.9053017290736287,
      "grad_norm": 0.1475028693675995,
      "learning_rate": 5.3659167313162656e-06,
      "loss": 0.7368,
      "step": 8320
    },
    {
      "epoch": 1.9064468109469828,
      "grad_norm": 0.14698173105716705,
      "learning_rate": 5.301267132143781e-06,
      "loss": 0.7395,
      "step": 8325
    },
    {
      "epoch": 1.9075918928203368,
      "grad_norm": 0.1603127121925354,
      "learning_rate": 5.2366175329712965e-06,
      "loss": 0.7422,
      "step": 8330
    },
    {
      "epoch": 1.9087369746936906,
      "grad_norm": 0.13494108617305756,
      "learning_rate": 5.171967933798811e-06,
      "loss": 0.7399,
      "step": 8335
    },
    {
      "epoch": 1.9098820565670445,
      "grad_norm": 0.1673167645931244,
      "learning_rate": 5.107318334626326e-06,
      "loss": 0.7398,
      "step": 8340
    },
    {
      "epoch": 1.9110271384403985,
      "grad_norm": 0.19458456337451935,
      "learning_rate": 5.042668735453841e-06,
      "loss": 0.7438,
      "step": 8345
    },
    {
      "epoch": 1.9121722203137526,
      "grad_norm": 0.1579926759004593,
      "learning_rate": 4.978019136281355e-06,
      "loss": 0.7418,
      "step": 8350
    },
    {
      "epoch": 1.9133173021871064,
      "grad_norm": 0.18511444330215454,
      "learning_rate": 4.91336953710887e-06,
      "loss": 0.7379,
      "step": 8355
    },
    {
      "epoch": 1.9144623840604602,
      "grad_norm": 0.1794387549161911,
      "learning_rate": 4.848719937936385e-06,
      "loss": 0.743,
      "step": 8360
    },
    {
      "epoch": 1.9156074659338143,
      "grad_norm": 0.14920532703399658,
      "learning_rate": 4.7840703387638995e-06,
      "loss": 0.7411,
      "step": 8365
    },
    {
      "epoch": 1.9167525478071683,
      "grad_norm": 0.14493878185749054,
      "learning_rate": 4.7194207395914145e-06,
      "loss": 0.7393,
      "step": 8370
    },
    {
      "epoch": 1.9178976296805221,
      "grad_norm": 0.17128613591194153,
      "learning_rate": 4.65477114041893e-06,
      "loss": 0.7426,
      "step": 8375
    },
    {
      "epoch": 1.919042711553876,
      "grad_norm": 0.13501162827014923,
      "learning_rate": 4.590121541246444e-06,
      "loss": 0.74,
      "step": 8380
    },
    {
      "epoch": 1.92018779342723,
      "grad_norm": 0.18030951917171478,
      "learning_rate": 4.52547194207396e-06,
      "loss": 0.7379,
      "step": 8385
    },
    {
      "epoch": 1.921332875300584,
      "grad_norm": 0.1932855099439621,
      "learning_rate": 4.460822342901475e-06,
      "loss": 0.7438,
      "step": 8390
    },
    {
      "epoch": 1.922477957173938,
      "grad_norm": 0.1626242697238922,
      "learning_rate": 4.396172743728989e-06,
      "loss": 0.743,
      "step": 8395
    },
    {
      "epoch": 1.923623039047292,
      "grad_norm": 0.1467551589012146,
      "learning_rate": 4.331523144556504e-06,
      "loss": 0.7406,
      "step": 8400
    },
    {
      "epoch": 1.9247681209206458,
      "grad_norm": 0.1600460261106491,
      "learning_rate": 4.266873545384019e-06,
      "loss": 0.7414,
      "step": 8405
    },
    {
      "epoch": 1.9259132027939998,
      "grad_norm": 0.1701911985874176,
      "learning_rate": 4.202223946211533e-06,
      "loss": 0.7399,
      "step": 8410
    },
    {
      "epoch": 1.9270582846673538,
      "grad_norm": 0.15907885134220123,
      "learning_rate": 4.1375743470390484e-06,
      "loss": 0.7355,
      "step": 8415
    },
    {
      "epoch": 1.9282033665407077,
      "grad_norm": 0.14833064377307892,
      "learning_rate": 4.0729247478665635e-06,
      "loss": 0.747,
      "step": 8420
    },
    {
      "epoch": 1.9293484484140615,
      "grad_norm": 0.1576046198606491,
      "learning_rate": 4.008275148694078e-06,
      "loss": 0.7372,
      "step": 8425
    },
    {
      "epoch": 1.9304935302874155,
      "grad_norm": 0.1346118301153183,
      "learning_rate": 3.943625549521593e-06,
      "loss": 0.7375,
      "step": 8430
    },
    {
      "epoch": 1.9316386121607696,
      "grad_norm": 0.16976960003376007,
      "learning_rate": 3.878975950349108e-06,
      "loss": 0.7393,
      "step": 8435
    },
    {
      "epoch": 1.9327836940341234,
      "grad_norm": 0.16566011309623718,
      "learning_rate": 3.8143263511766233e-06,
      "loss": 0.7351,
      "step": 8440
    },
    {
      "epoch": 1.9339287759074772,
      "grad_norm": 0.1495964080095291,
      "learning_rate": 3.749676752004138e-06,
      "loss": 0.7378,
      "step": 8445
    },
    {
      "epoch": 1.9350738577808313,
      "grad_norm": 0.17395488917827606,
      "learning_rate": 3.6850271528316526e-06,
      "loss": 0.7394,
      "step": 8450
    },
    {
      "epoch": 1.9362189396541853,
      "grad_norm": 0.1591861993074417,
      "learning_rate": 3.6203775536591677e-06,
      "loss": 0.7377,
      "step": 8455
    },
    {
      "epoch": 1.9373640215275392,
      "grad_norm": 0.14285223186016083,
      "learning_rate": 3.5557279544866823e-06,
      "loss": 0.7416,
      "step": 8460
    },
    {
      "epoch": 1.9385091034008932,
      "grad_norm": 0.1831120252609253,
      "learning_rate": 3.4910783553141974e-06,
      "loss": 0.7338,
      "step": 8465
    },
    {
      "epoch": 1.939654185274247,
      "grad_norm": 0.1750773787498474,
      "learning_rate": 3.426428756141712e-06,
      "loss": 0.7395,
      "step": 8470
    },
    {
      "epoch": 1.940799267147601,
      "grad_norm": 0.16022038459777832,
      "learning_rate": 3.3617791569692267e-06,
      "loss": 0.7399,
      "step": 8475
    },
    {
      "epoch": 1.9419443490209551,
      "grad_norm": 0.1562938690185547,
      "learning_rate": 3.2971295577967418e-06,
      "loss": 0.7374,
      "step": 8480
    },
    {
      "epoch": 1.943089430894309,
      "grad_norm": 0.1531076729297638,
      "learning_rate": 3.2324799586242564e-06,
      "loss": 0.7366,
      "step": 8485
    },
    {
      "epoch": 1.9442345127676628,
      "grad_norm": 0.13733965158462524,
      "learning_rate": 3.167830359451771e-06,
      "loss": 0.7387,
      "step": 8490
    },
    {
      "epoch": 1.9453795946410168,
      "grad_norm": 0.1592342108488083,
      "learning_rate": 3.1031807602792865e-06,
      "loss": 0.7412,
      "step": 8495
    },
    {
      "epoch": 1.9465246765143709,
      "grad_norm": 0.13735567033290863,
      "learning_rate": 3.038531161106801e-06,
      "loss": 0.7424,
      "step": 8500
    },
    {
      "epoch": 1.9476697583877247,
      "grad_norm": 0.1416022926568985,
      "learning_rate": 2.973881561934316e-06,
      "loss": 0.7371,
      "step": 8505
    },
    {
      "epoch": 1.9488148402610785,
      "grad_norm": 0.15780256688594818,
      "learning_rate": 2.9092319627618313e-06,
      "loss": 0.7413,
      "step": 8510
    },
    {
      "epoch": 1.9499599221344326,
      "grad_norm": 0.15426452457904816,
      "learning_rate": 2.844582363589346e-06,
      "loss": 0.7399,
      "step": 8515
    },
    {
      "epoch": 1.9511050040077866,
      "grad_norm": 0.1280713975429535,
      "learning_rate": 2.7799327644168606e-06,
      "loss": 0.7386,
      "step": 8520
    },
    {
      "epoch": 1.9522500858811405,
      "grad_norm": 0.1521618515253067,
      "learning_rate": 2.7152831652443757e-06,
      "loss": 0.7361,
      "step": 8525
    },
    {
      "epoch": 1.9533951677544943,
      "grad_norm": 0.15051406621932983,
      "learning_rate": 2.6506335660718903e-06,
      "loss": 0.7436,
      "step": 8530
    },
    {
      "epoch": 1.9545402496278483,
      "grad_norm": 0.19529180228710175,
      "learning_rate": 2.5859839668994054e-06,
      "loss": 0.7364,
      "step": 8535
    },
    {
      "epoch": 1.9556853315012024,
      "grad_norm": 0.17485053837299347,
      "learning_rate": 2.5213343677269204e-06,
      "loss": 0.7413,
      "step": 8540
    },
    {
      "epoch": 1.9568304133745564,
      "grad_norm": 0.13643115758895874,
      "learning_rate": 2.456684768554435e-06,
      "loss": 0.7377,
      "step": 8545
    },
    {
      "epoch": 1.9579754952479103,
      "grad_norm": 0.16808709502220154,
      "learning_rate": 2.3920351693819497e-06,
      "loss": 0.7398,
      "step": 8550
    },
    {
      "epoch": 1.959120577121264,
      "grad_norm": 0.14400559663772583,
      "learning_rate": 2.327385570209465e-06,
      "loss": 0.7403,
      "step": 8555
    },
    {
      "epoch": 1.9602656589946181,
      "grad_norm": 0.15056101977825165,
      "learning_rate": 2.26273597103698e-06,
      "loss": 0.7426,
      "step": 8560
    },
    {
      "epoch": 1.9614107408679722,
      "grad_norm": 0.1426333487033844,
      "learning_rate": 2.1980863718644945e-06,
      "loss": 0.7387,
      "step": 8565
    },
    {
      "epoch": 1.962555822741326,
      "grad_norm": 0.14333635568618774,
      "learning_rate": 2.1334367726920096e-06,
      "loss": 0.7378,
      "step": 8570
    },
    {
      "epoch": 1.9637009046146798,
      "grad_norm": 0.15889368951320648,
      "learning_rate": 2.0687871735195242e-06,
      "loss": 0.7417,
      "step": 8575
    },
    {
      "epoch": 1.9648459864880339,
      "grad_norm": 0.1589675098657608,
      "learning_rate": 2.004137574347039e-06,
      "loss": 0.7391,
      "step": 8580
    },
    {
      "epoch": 1.965991068361388,
      "grad_norm": 0.14005248248577118,
      "learning_rate": 1.939487975174554e-06,
      "loss": 0.7368,
      "step": 8585
    },
    {
      "epoch": 1.9671361502347418,
      "grad_norm": 0.12994040548801422,
      "learning_rate": 1.874838376002069e-06,
      "loss": 0.7369,
      "step": 8590
    },
    {
      "epoch": 1.9682812321080956,
      "grad_norm": 0.15902163088321686,
      "learning_rate": 1.8101887768295838e-06,
      "loss": 0.7406,
      "step": 8595
    },
    {
      "epoch": 1.9694263139814496,
      "grad_norm": 0.14345116913318634,
      "learning_rate": 1.7455391776570987e-06,
      "loss": 0.7397,
      "step": 8600
    },
    {
      "epoch": 1.9705713958548037,
      "grad_norm": 0.12906809151172638,
      "learning_rate": 1.6808895784846133e-06,
      "loss": 0.7412,
      "step": 8605
    },
    {
      "epoch": 1.9717164777281577,
      "grad_norm": 0.15647900104522705,
      "learning_rate": 1.6162399793121282e-06,
      "loss": 0.7395,
      "step": 8610
    },
    {
      "epoch": 1.9728615596015115,
      "grad_norm": 0.14507581293582916,
      "learning_rate": 1.5515903801396433e-06,
      "loss": 0.7436,
      "step": 8615
    },
    {
      "epoch": 1.9740066414748654,
      "grad_norm": 0.1292596310377121,
      "learning_rate": 1.486940780967158e-06,
      "loss": 0.7404,
      "step": 8620
    },
    {
      "epoch": 1.9751517233482194,
      "grad_norm": 0.13533343374729156,
      "learning_rate": 1.422291181794673e-06,
      "loss": 0.7395,
      "step": 8625
    },
    {
      "epoch": 1.9762968052215735,
      "grad_norm": 0.14757055044174194,
      "learning_rate": 1.3576415826221878e-06,
      "loss": 0.7355,
      "step": 8630
    },
    {
      "epoch": 1.9774418870949273,
      "grad_norm": 0.15465658903121948,
      "learning_rate": 1.2929919834497027e-06,
      "loss": 0.7399,
      "step": 8635
    },
    {
      "epoch": 1.9785869689682811,
      "grad_norm": 0.16538003087043762,
      "learning_rate": 1.2283423842772175e-06,
      "loss": 0.7384,
      "step": 8640
    },
    {
      "epoch": 1.9797320508416352,
      "grad_norm": 0.14361347258090973,
      "learning_rate": 1.1636927851047324e-06,
      "loss": 0.7421,
      "step": 8645
    },
    {
      "epoch": 1.9808771327149892,
      "grad_norm": 0.13651369512081146,
      "learning_rate": 1.0990431859322473e-06,
      "loss": 0.7371,
      "step": 8650
    },
    {
      "epoch": 1.982022214588343,
      "grad_norm": 0.1409565657377243,
      "learning_rate": 1.0343935867597621e-06,
      "loss": 0.7391,
      "step": 8655
    },
    {
      "epoch": 1.9831672964616969,
      "grad_norm": 0.13974834978580475,
      "learning_rate": 9.69743987587277e-07,
      "loss": 0.7457,
      "step": 8660
    },
    {
      "epoch": 1.984312378335051,
      "grad_norm": 0.14657549560070038,
      "learning_rate": 9.050943884147919e-07,
      "loss": 0.7348,
      "step": 8665
    },
    {
      "epoch": 1.985457460208405,
      "grad_norm": 0.156954824924469,
      "learning_rate": 8.404447892423067e-07,
      "loss": 0.7441,
      "step": 8670
    },
    {
      "epoch": 1.9866025420817588,
      "grad_norm": 0.1401233673095703,
      "learning_rate": 7.757951900698216e-07,
      "loss": 0.7386,
      "step": 8675
    },
    {
      "epoch": 1.9877476239551128,
      "grad_norm": 0.14088112115859985,
      "learning_rate": 7.111455908973365e-07,
      "loss": 0.7359,
      "step": 8680
    },
    {
      "epoch": 1.9888927058284667,
      "grad_norm": 0.14570772647857666,
      "learning_rate": 6.464959917248513e-07,
      "loss": 0.7392,
      "step": 8685
    },
    {
      "epoch": 1.9900377877018207,
      "grad_norm": 0.15030637383460999,
      "learning_rate": 5.818463925523662e-07,
      "loss": 0.744,
      "step": 8690
    },
    {
      "epoch": 1.9911828695751748,
      "grad_norm": 0.13755646347999573,
      "learning_rate": 5.171967933798811e-07,
      "loss": 0.7355,
      "step": 8695
    },
    {
      "epoch": 1.9923279514485286,
      "grad_norm": 0.155239075422287,
      "learning_rate": 4.5254719420739596e-07,
      "loss": 0.7408,
      "step": 8700
    },
    {
      "epoch": 1.9934730333218824,
      "grad_norm": 0.1403430551290512,
      "learning_rate": 3.878975950349108e-07,
      "loss": 0.7427,
      "step": 8705
    },
    {
      "epoch": 1.9946181151952365,
      "grad_norm": 0.14680492877960205,
      "learning_rate": 3.2324799586242567e-07,
      "loss": 0.7434,
      "step": 8710
    },
    {
      "epoch": 1.9957631970685905,
      "grad_norm": 0.1654711365699768,
      "learning_rate": 2.5859839668994053e-07,
      "loss": 0.74,
      "step": 8715
    },
    {
      "epoch": 1.9969082789419443,
      "grad_norm": 0.14024050533771515,
      "learning_rate": 1.939487975174554e-07,
      "loss": 0.7425,
      "step": 8720
    },
    {
      "epoch": 1.9980533608152982,
      "grad_norm": 0.12798629701137543,
      "learning_rate": 1.2929919834497026e-07,
      "loss": 0.7389,
      "step": 8725
    },
    {
      "epoch": 1.9991984426886522,
      "grad_norm": 0.1456129103899002,
      "learning_rate": 6.464959917248513e-08,
      "loss": 0.7414,
      "step": 8730
    }
  ],
  "logging_steps": 5,
  "max_steps": 8734,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}