{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.0631463871386596,
  "eval_steps": 500,
  "global_step": 8000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 15.142339706420898,
      "learning_rate": 5e-06,
      "loss": 12.0133,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.963080406188965,
      "learning_rate": 1e-05,
      "loss": 11.7711,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.329822540283203,
      "learning_rate": 1.5e-05,
      "loss": 11.2171,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.5315165519714355,
      "learning_rate": 2e-05,
      "loss": 10.5225,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.707030296325684,
      "learning_rate": 2.5e-05,
      "loss": 9.8322,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2473812103271484,
      "learning_rate": 3e-05,
      "loss": 9.1367,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.32763409614563,
      "learning_rate": 3.5e-05,
      "loss": 8.5446,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.939070701599121,
      "learning_rate": 4e-05,
      "loss": 8.0765,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5867785215377808,
      "learning_rate": 4.5e-05,
      "loss": 7.7378,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3308677673339844,
      "learning_rate": 5e-05,
      "loss": 7.5431,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8467875123023987,
      "learning_rate": 5.500000000000001e-05,
      "loss": 7.3992,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7966635823249817,
      "learning_rate": 6e-05,
      "loss": 7.2515,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7197663187980652,
      "learning_rate": 6.500000000000001e-05,
      "loss": 7.2282,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7674877047538757,
      "learning_rate": 7e-05,
      "loss": 7.0993,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7982838153839111,
      "learning_rate": 7.500000000000001e-05,
      "loss": 6.9743,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9206609129905701,
      "learning_rate": 8e-05,
      "loss": 6.8044,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8050782680511475,
      "learning_rate": 8.5e-05,
      "loss": 6.6585,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8832933902740479,
      "learning_rate": 9e-05,
      "loss": 6.6112,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1636277437210083,
      "learning_rate": 9.5e-05,
      "loss": 6.3943,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8433125019073486,
      "learning_rate": 0.0001,
      "loss": 6.2563,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6152033805847168,
      "learning_rate": 9.999999961506464e-05,
      "loss": 6.1552,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2387120723724365,
      "learning_rate": 9.999999846025851e-05,
      "loss": 6.0804,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6138665676116943,
      "learning_rate": 9.999999653558166e-05,
      "loss": 5.9959,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.739707350730896,
      "learning_rate": 9.999999384103412e-05,
      "loss": 5.8416,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8446868658065796,
      "learning_rate": 9.999999037661592e-05,
      "loss": 5.6861,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.326461911201477,
      "learning_rate": 9.999998614232713e-05,
      "loss": 5.7214,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2305861711502075,
      "learning_rate": 9.99999811381678e-05,
      "loss": 5.5801,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4233742952346802,
      "learning_rate": 9.9999975364138e-05,
      "loss": 5.55,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1713918447494507,
      "learning_rate": 9.999996882023784e-05,
      "loss": 5.5716,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2854537963867188,
      "learning_rate": 9.99999615064674e-05,
      "loss": 5.4537,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4329421520233154,
      "learning_rate": 9.99999534228268e-05,
      "loss": 5.3801,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4164565801620483,
      "learning_rate": 9.999994456931619e-05,
      "loss": 5.31,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1054998636245728,
      "learning_rate": 9.999993494593566e-05,
      "loss": 5.2541,
      "step": 165
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5174094438552856,
      "learning_rate": 9.999992455268539e-05,
      "loss": 5.1326,
      "step": 170
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1391857862472534,
      "learning_rate": 9.999991338956554e-05,
      "loss": 5.3246,
      "step": 175
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4162404537200928,
      "learning_rate": 9.999990145657626e-05,
      "loss": 5.2027,
      "step": 180
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2971961498260498,
      "learning_rate": 9.999988875371776e-05,
      "loss": 5.1073,
      "step": 185
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4737606048583984,
      "learning_rate": 9.999987528099021e-05,
      "loss": 5.1143,
      "step": 190
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4011461734771729,
      "learning_rate": 9.999986103839384e-05,
      "loss": 5.0899,
      "step": 195
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4744013547897339,
      "learning_rate": 9.999984602592884e-05,
      "loss": 5.0373,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4590356349945068,
      "learning_rate": 9.999983024359549e-05,
      "loss": 5.0199,
      "step": 205
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0719900131225586,
      "learning_rate": 9.9999813691394e-05,
      "loss": 4.902,
      "step": 210
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1954728364944458,
      "learning_rate": 9.999979636932461e-05,
      "loss": 4.9605,
      "step": 215
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5145868062973022,
      "learning_rate": 9.999977827738762e-05,
      "loss": 4.985,
      "step": 220
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3508830070495605,
      "learning_rate": 9.999975941558328e-05,
      "loss": 4.8785,
      "step": 225
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.329408049583435,
      "learning_rate": 9.999973978391191e-05,
      "loss": 4.9955,
      "step": 230
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5139230489730835,
      "learning_rate": 9.99997193823738e-05,
      "loss": 4.7139,
      "step": 235
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6292555332183838,
      "learning_rate": 9.999969821096924e-05,
      "loss": 4.7876,
      "step": 240
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.664516806602478,
      "learning_rate": 9.99996762696986e-05,
      "loss": 4.7561,
      "step": 245
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5122953653335571,
      "learning_rate": 9.999965355856218e-05,
      "loss": 4.734,
      "step": 250
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4214054346084595,
      "learning_rate": 9.999963007756035e-05,
      "loss": 4.7356,
      "step": 255
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.507981061935425,
      "learning_rate": 9.999960582669346e-05,
      "loss": 4.7801,
      "step": 260
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9237830638885498,
      "learning_rate": 9.99995808059619e-05,
      "loss": 4.7508,
      "step": 265
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.620719313621521,
      "learning_rate": 9.999955501536603e-05,
      "loss": 4.524,
      "step": 270
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7300649881362915,
      "learning_rate": 9.999952845490626e-05,
      "loss": 4.6056,
      "step": 275
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.511616587638855,
      "learning_rate": 9.999950112458302e-05,
      "loss": 4.6749,
      "step": 280
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.47576904296875,
      "learning_rate": 9.99994730243967e-05,
      "loss": 4.6024,
      "step": 285
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4870834350585938,
      "learning_rate": 9.999944415434774e-05,
      "loss": 4.5686,
      "step": 290
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7283949851989746,
      "learning_rate": 9.99994145144366e-05,
      "loss": 4.5821,
      "step": 295
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4078936576843262,
      "learning_rate": 9.999938410466371e-05,
      "loss": 4.574,
      "step": 300
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.500351071357727,
      "learning_rate": 9.999935292502958e-05,
      "loss": 4.4652,
      "step": 305
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9204586744308472,
      "learning_rate": 9.999932097553463e-05,
      "loss": 4.5241,
      "step": 310
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5964891910552979,
      "learning_rate": 9.999928825617941e-05,
      "loss": 4.447,
      "step": 315
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.329533576965332,
      "learning_rate": 9.999925476696439e-05,
      "loss": 4.4309,
      "step": 320
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.473573088645935,
      "learning_rate": 9.99992205078901e-05,
      "loss": 4.5307,
      "step": 325
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2743762731552124,
      "learning_rate": 9.999918547895705e-05,
      "loss": 4.4758,
      "step": 330
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3271675109863281,
      "learning_rate": 9.999914968016579e-05,
      "loss": 4.3579,
      "step": 335
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5689301490783691,
      "learning_rate": 9.999911311151689e-05,
      "loss": 4.2856,
      "step": 340
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3750799894332886,
      "learning_rate": 9.999907577301089e-05,
      "loss": 4.3412,
      "step": 345
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3108932971954346,
      "learning_rate": 9.999903766464837e-05,
      "loss": 4.4266,
      "step": 350
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3405834436416626,
      "learning_rate": 9.99989987864299e-05,
      "loss": 4.3378,
      "step": 355
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.525265097618103,
      "learning_rate": 9.999895913835613e-05,
      "loss": 4.3734,
      "step": 360
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4694818258285522,
      "learning_rate": 9.999891872042762e-05,
      "loss": 4.2962,
      "step": 365
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3874412775039673,
      "learning_rate": 9.9998877532645e-05,
      "loss": 4.2287,
      "step": 370
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4830822944641113,
      "learning_rate": 9.999883557500892e-05,
      "loss": 4.3139,
      "step": 375
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.428941249847412,
      "learning_rate": 9.999879284752003e-05,
      "loss": 4.2367,
      "step": 380
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2547497749328613,
      "learning_rate": 9.999874935017897e-05,
      "loss": 4.2918,
      "step": 385
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1645982265472412,
      "learning_rate": 9.999870508298641e-05,
      "loss": 4.2327,
      "step": 390
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2057344913482666,
      "learning_rate": 9.999866004594303e-05,
      "loss": 4.281,
      "step": 395
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.135153889656067,
      "learning_rate": 9.999861423904954e-05,
      "loss": 4.1885,
      "step": 400
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.204508662223816,
      "learning_rate": 9.999856766230665e-05,
      "loss": 4.259,
      "step": 405
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3388700485229492,
      "learning_rate": 9.999852031571506e-05,
      "loss": 4.2688,
      "step": 410
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3300930261611938,
      "learning_rate": 9.999847219927549e-05,
      "loss": 4.1752,
      "step": 415
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1948955059051514,
      "learning_rate": 9.999842331298872e-05,
      "loss": 4.0834,
      "step": 420
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2118010520935059,
      "learning_rate": 9.999837365685544e-05,
      "loss": 4.1875,
      "step": 425
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2062677145004272,
      "learning_rate": 9.999832323087649e-05,
      "loss": 4.2102,
      "step": 430
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1632463932037354,
      "learning_rate": 9.999827203505259e-05,
      "loss": 4.0892,
      "step": 435
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2679600715637207,
      "learning_rate": 9.999822006938455e-05,
      "loss": 4.1411,
      "step": 440
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.247983694076538,
      "learning_rate": 9.999816733387316e-05,
      "loss": 4.0653,
      "step": 445
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0861722230911255,
      "learning_rate": 9.999811382851924e-05,
      "loss": 3.8844,
      "step": 450
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3868489265441895,
      "learning_rate": 9.999805955332362e-05,
      "loss": 4.0288,
      "step": 455
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1216765642166138,
      "learning_rate": 9.999800450828711e-05,
      "loss": 4.0726,
      "step": 460
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1698716878890991,
      "learning_rate": 9.99979486934106e-05,
      "loss": 4.0056,
      "step": 465
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.154384732246399,
      "learning_rate": 9.99978921086949e-05,
      "loss": 3.9945,
      "step": 470
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3986250162124634,
      "learning_rate": 9.999783475414093e-05,
      "loss": 4.0358,
      "step": 475
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1886556148529053,
      "learning_rate": 9.999777662974952e-05,
      "loss": 3.9156,
      "step": 480
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.145485520362854,
      "learning_rate": 9.999771773552162e-05,
      "loss": 3.9793,
      "step": 485
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2868390083312988,
      "learning_rate": 9.999765807145811e-05,
      "loss": 4.0186,
      "step": 490
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1428489685058594,
      "learning_rate": 9.99975976375599e-05,
      "loss": 3.907,
      "step": 495
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1288151741027832,
      "learning_rate": 9.999753643382793e-05,
      "loss": 4.0657,
      "step": 500
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1802102327346802,
      "learning_rate": 9.999747446026315e-05,
      "loss": 3.9334,
      "step": 505
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.244258999824524,
      "learning_rate": 9.999741171686649e-05,
      "loss": 3.8593,
      "step": 510
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0476657152175903,
      "learning_rate": 9.999734820363895e-05,
      "loss": 3.8586,
      "step": 515
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1787341833114624,
      "learning_rate": 9.99972839205815e-05,
      "loss": 3.9123,
      "step": 520
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1357966661453247,
      "learning_rate": 9.999721886769511e-05,
      "loss": 3.9207,
      "step": 525
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.204612374305725,
      "learning_rate": 9.999715304498078e-05,
      "loss": 3.8202,
      "step": 530
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1605384349822998,
      "learning_rate": 9.999708645243955e-05,
      "loss": 3.774,
      "step": 535
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0668116807937622,
      "learning_rate": 9.999701909007242e-05,
      "loss": 3.8881,
      "step": 540
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0532135963439941,
      "learning_rate": 9.999695095788044e-05,
      "loss": 3.8384,
      "step": 545
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0724900960922241,
      "learning_rate": 9.999688205586468e-05,
      "loss": 3.8253,
      "step": 550
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0534138679504395,
      "learning_rate": 9.999681238402615e-05,
      "loss": 3.7836,
      "step": 555
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0635172128677368,
      "learning_rate": 9.999674194236597e-05,
      "loss": 3.891,
      "step": 560
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1336596012115479,
      "learning_rate": 9.999667073088519e-05,
      "loss": 3.8791,
      "step": 565
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.086419701576233,
      "learning_rate": 9.999659874958493e-05,
      "loss": 3.7613,
      "step": 570
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0388224124908447,
      "learning_rate": 9.99965259984663e-05,
      "loss": 3.874,
      "step": 575
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0822333097457886,
      "learning_rate": 9.999645247753039e-05,
      "loss": 3.8302,
      "step": 580
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2413557767868042,
      "learning_rate": 9.999637818677836e-05,
      "loss": 3.8565,
      "step": 585
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3178820610046387,
      "learning_rate": 9.999630312621136e-05,
      "loss": 3.7802,
      "step": 590
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2712346315383911,
      "learning_rate": 9.999622729583051e-05,
      "loss": 3.7677,
      "step": 595
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1075167655944824,
      "learning_rate": 9.999615069563701e-05,
      "loss": 3.8157,
      "step": 600
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9688358902931213,
      "learning_rate": 9.999607332563205e-05,
      "loss": 3.7215,
      "step": 605
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0530669689178467,
      "learning_rate": 9.999599518581676e-05,
      "loss": 3.7789,
      "step": 610
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.048405647277832,
      "learning_rate": 9.999591627619242e-05,
      "loss": 3.7802,
      "step": 615
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0187485218048096,
      "learning_rate": 9.999583659676019e-05,
      "loss": 3.6741,
      "step": 620
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.144195795059204,
      "learning_rate": 9.999575614752133e-05,
      "loss": 3.774,
      "step": 625
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0065631866455078,
      "learning_rate": 9.999567492847706e-05,
      "loss": 3.7138,
      "step": 630
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1354420185089111,
      "learning_rate": 9.999559293962864e-05,
      "loss": 3.7273,
      "step": 635
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.044249415397644,
      "learning_rate": 9.99955101809773e-05,
      "loss": 3.6852,
      "step": 640
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2161166667938232,
      "learning_rate": 9.999542665252437e-05,
      "loss": 3.7468,
      "step": 645
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1312017440795898,
      "learning_rate": 9.999534235427109e-05,
      "loss": 3.631,
      "step": 650
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0620629787445068,
      "learning_rate": 9.999525728621879e-05,
      "loss": 3.6775,
      "step": 655
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0445822477340698,
      "learning_rate": 9.999517144836876e-05,
      "loss": 3.6206,
      "step": 660
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0086705684661865,
      "learning_rate": 9.999508484072233e-05,
      "loss": 3.6288,
      "step": 665
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.975386381149292,
      "learning_rate": 9.999499746328083e-05,
      "loss": 3.6767,
      "step": 670
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.992201566696167,
      "learning_rate": 9.99949093160456e-05,
      "loss": 3.6763,
      "step": 675
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2186998128890991,
      "learning_rate": 9.9994820399018e-05,
      "loss": 3.6571,
      "step": 680
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0757606029510498,
      "learning_rate": 9.999473071219941e-05,
      "loss": 3.5341,
      "step": 685
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0610624551773071,
      "learning_rate": 9.999464025559121e-05,
      "loss": 3.6831,
      "step": 690
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.904817521572113,
      "learning_rate": 9.999454902919478e-05,
      "loss": 3.638,
      "step": 695
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0359525680541992,
      "learning_rate": 9.999445703301154e-05,
      "loss": 3.5854,
      "step": 700
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1337298154830933,
      "learning_rate": 9.999436426704287e-05,
      "loss": 3.659,
      "step": 705
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0919194221496582,
      "learning_rate": 9.999427073129026e-05,
      "loss": 3.5774,
      "step": 710
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8795281648635864,
      "learning_rate": 9.999417642575509e-05,
      "loss": 3.6313,
      "step": 715
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0229532718658447,
      "learning_rate": 9.999408135043886e-05,
      "loss": 3.5507,
      "step": 720
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.104490041732788,
      "learning_rate": 9.999398550534298e-05,
      "loss": 3.5273,
      "step": 725
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9430055618286133,
      "learning_rate": 9.999388889046899e-05,
      "loss": 3.5848,
      "step": 730
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0396498441696167,
      "learning_rate": 9.999379150581833e-05,
      "loss": 3.5229,
      "step": 735
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.97342449426651,
      "learning_rate": 9.999369335139251e-05,
      "loss": 3.5035,
      "step": 740
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9150033593177795,
      "learning_rate": 9.999359442719306e-05,
      "loss": 3.5286,
      "step": 745
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9647395610809326,
      "learning_rate": 9.999349473322149e-05,
      "loss": 3.4627,
      "step": 750
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9407553672790527,
      "learning_rate": 9.999339426947932e-05,
      "loss": 3.5536,
      "step": 755
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9081340432167053,
      "learning_rate": 9.999329303596812e-05,
      "loss": 3.5181,
      "step": 760
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8502491116523743,
      "learning_rate": 9.999319103268945e-05,
      "loss": 3.372,
      "step": 765
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9379911422729492,
      "learning_rate": 9.999308825964485e-05,
      "loss": 3.5191,
      "step": 770
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0037740468978882,
      "learning_rate": 9.999298471683593e-05,
      "loss": 3.603,
      "step": 775
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8996877074241638,
      "learning_rate": 9.999288040426429e-05,
      "loss": 3.4557,
      "step": 780
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9344475269317627,
      "learning_rate": 9.999277532193151e-05,
      "loss": 3.5814,
      "step": 785
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.904925525188446,
      "learning_rate": 9.999266946983923e-05,
      "loss": 3.5173,
      "step": 790
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8760194182395935,
      "learning_rate": 9.999256284798906e-05,
      "loss": 3.4082,
      "step": 795
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9515339732170105,
      "learning_rate": 9.999245545638267e-05,
      "loss": 3.4768,
      "step": 800
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9361962080001831,
      "learning_rate": 9.999234729502167e-05,
      "loss": 3.4689,
      "step": 805
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9242522120475769,
      "learning_rate": 9.999223836390777e-05,
      "loss": 3.4464,
      "step": 810
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9124124050140381,
      "learning_rate": 9.999212866304262e-05,
      "loss": 3.4256,
      "step": 815
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1004537343978882,
      "learning_rate": 9.999201819242794e-05,
      "loss": 3.3863,
      "step": 820
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9508906602859497,
      "learning_rate": 9.99919069520654e-05,
      "loss": 3.4356,
      "step": 825
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9869884252548218,
      "learning_rate": 9.99917949419567e-05,
      "loss": 3.5092,
      "step": 830
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8730179667472839,
      "learning_rate": 9.999168216210362e-05,
      "loss": 3.4543,
      "step": 835
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8838622570037842,
      "learning_rate": 9.999156861250784e-05,
      "loss": 3.4752,
      "step": 840
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8449395895004272,
      "learning_rate": 9.999145429317114e-05,
      "loss": 3.4187,
      "step": 845
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7765811085700989,
      "learning_rate": 9.999133920409527e-05,
      "loss": 3.3923,
      "step": 850
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8263750672340393,
      "learning_rate": 9.999122334528199e-05,
      "loss": 3.4519,
      "step": 855
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8535619974136353,
      "learning_rate": 9.999110671673313e-05,
      "loss": 3.4007,
      "step": 860
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8383920788764954,
      "learning_rate": 9.999098931845044e-05,
      "loss": 3.2882,
      "step": 865
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8224014639854431,
      "learning_rate": 9.999087115043573e-05,
      "loss": 3.4029,
      "step": 870
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7796249389648438,
      "learning_rate": 9.999075221269083e-05,
      "loss": 3.2994,
      "step": 875
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8989777565002441,
      "learning_rate": 9.999063250521759e-05,
      "loss": 3.2909,
      "step": 880
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8340718746185303,
      "learning_rate": 9.999051202801781e-05,
      "loss": 3.4116,
      "step": 885
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8410345315933228,
      "learning_rate": 9.999039078109341e-05,
      "loss": 3.3935,
      "step": 890
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.011084794998169,
      "learning_rate": 9.999026876444617e-05,
      "loss": 3.3324,
      "step": 895
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9302998185157776,
      "learning_rate": 9.999014597807805e-05,
      "loss": 3.2815,
      "step": 900
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8717480301856995,
      "learning_rate": 9.99900224219909e-05,
      "loss": 3.4572,
      "step": 905
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.914622962474823,
      "learning_rate": 9.998989809618662e-05,
      "loss": 3.3905,
      "step": 910
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7772191762924194,
      "learning_rate": 9.998977300066715e-05,
      "loss": 3.3004,
      "step": 915
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8308324217796326,
      "learning_rate": 9.998964713543437e-05,
      "loss": 3.2163,
      "step": 920
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7868050336837769,
      "learning_rate": 9.998952050049026e-05,
      "loss": 3.33,
      "step": 925
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8267785906791687,
      "learning_rate": 9.998939309583677e-05,
      "loss": 3.3585,
      "step": 930
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7986210584640503,
      "learning_rate": 9.998926492147585e-05,
      "loss": 3.3938,
      "step": 935
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8491371273994446,
      "learning_rate": 9.998913597740945e-05,
      "loss": 3.4038,
      "step": 940
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7769982218742371,
      "learning_rate": 9.99890062636396e-05,
      "loss": 3.2976,
      "step": 945
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7924196720123291,
      "learning_rate": 9.998887578016827e-05,
      "loss": 3.3635,
      "step": 950
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7578513026237488,
      "learning_rate": 9.998874452699747e-05,
      "loss": 3.2376,
      "step": 955
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.91054368019104,
      "learning_rate": 9.998861250412922e-05,
      "loss": 3.3373,
      "step": 960
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9622986316680908,
      "learning_rate": 9.998847971156556e-05,
      "loss": 3.3337,
      "step": 965
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7796733379364014,
      "learning_rate": 9.998834614930855e-05,
      "loss": 3.2744,
      "step": 970
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7877930998802185,
      "learning_rate": 9.99882118173602e-05,
      "loss": 3.2744,
      "step": 975
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8464511632919312,
      "learning_rate": 9.998807671572263e-05,
      "loss": 3.2125,
      "step": 980
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9003927111625671,
      "learning_rate": 9.998794084439789e-05,
      "loss": 3.2361,
      "step": 985
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8311697244644165,
      "learning_rate": 9.998780420338809e-05,
      "loss": 3.265,
      "step": 990
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8703957200050354,
      "learning_rate": 9.998766679269531e-05,
      "loss": 3.2385,
      "step": 995
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7914534211158752,
      "learning_rate": 9.998752861232168e-05,
      "loss": 3.2925,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8593668341636658,
      "learning_rate": 9.998738966226933e-05,
      "loss": 3.3447,
      "step": 1005
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7278022170066833,
      "learning_rate": 9.998724994254041e-05,
      "loss": 3.2174,
      "step": 1010
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7596559524536133,
      "learning_rate": 9.998710945313705e-05,
      "loss": 3.1526,
      "step": 1015
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8033633232116699,
      "learning_rate": 9.998696819406141e-05,
      "loss": 3.2767,
      "step": 1020
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.859947681427002,
      "learning_rate": 9.99868261653157e-05,
      "loss": 3.2191,
      "step": 1025
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7699180245399475,
      "learning_rate": 9.998668336690206e-05,
      "loss": 3.4164,
      "step": 1030
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8244397640228271,
      "learning_rate": 9.998653979882271e-05,
      "loss": 3.3604,
      "step": 1035
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7424900531768799,
      "learning_rate": 9.998639546107989e-05,
      "loss": 3.3014,
      "step": 1040
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7951467037200928,
      "learning_rate": 9.998625035367577e-05,
      "loss": 3.2752,
      "step": 1045
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7850064039230347,
      "learning_rate": 9.998610447661261e-05,
      "loss": 3.2397,
      "step": 1050
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7577301263809204,
      "learning_rate": 9.998595782989266e-05,
      "loss": 3.3126,
      "step": 1055
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7764389514923096,
      "learning_rate": 9.998581041351818e-05,
      "loss": 3.1815,
      "step": 1060
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8500764966011047,
      "learning_rate": 9.998566222749142e-05,
      "loss": 3.1863,
      "step": 1065
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8479288220405579,
      "learning_rate": 9.998551327181468e-05,
      "loss": 3.3124,
      "step": 1070
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8878303170204163,
      "learning_rate": 9.998536354649026e-05,
      "loss": 3.3759,
      "step": 1075
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7699640989303589,
      "learning_rate": 9.998521305152045e-05,
      "loss": 3.1993,
      "step": 1080
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8837441205978394,
      "learning_rate": 9.998506178690755e-05,
      "loss": 3.1428,
      "step": 1085
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6983273029327393,
      "learning_rate": 9.998490975265393e-05,
      "loss": 3.1786,
      "step": 1090
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7437112331390381,
      "learning_rate": 9.99847569487619e-05,
      "loss": 3.1546,
      "step": 1095
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9101467132568359,
      "learning_rate": 9.998460337523383e-05,
      "loss": 3.1859,
      "step": 1100
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8673747777938843,
      "learning_rate": 9.998444903207207e-05,
      "loss": 3.1969,
      "step": 1105
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9224711656570435,
      "learning_rate": 9.998429391927902e-05,
      "loss": 3.1633,
      "step": 1110
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6755993962287903,
      "learning_rate": 9.998413803685705e-05,
      "loss": 3.1122,
      "step": 1115
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7709223628044128,
      "learning_rate": 9.998398138480855e-05,
      "loss": 3.2095,
      "step": 1120
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7947278618812561,
      "learning_rate": 9.998382396313597e-05,
      "loss": 3.2037,
      "step": 1125
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7263473868370056,
      "learning_rate": 9.998366577184168e-05,
      "loss": 3.1079,
      "step": 1130
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8350757956504822,
      "learning_rate": 9.998350681092817e-05,
      "loss": 3.1587,
      "step": 1135
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7083161473274231,
      "learning_rate": 9.998334708039785e-05,
      "loss": 3.2196,
      "step": 1140
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7883744835853577,
      "learning_rate": 9.998318658025318e-05,
      "loss": 3.1765,
      "step": 1145
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7652791142463684,
      "learning_rate": 9.998302531049667e-05,
      "loss": 3.1145,
      "step": 1150
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.769584059715271,
      "learning_rate": 9.998286327113076e-05,
      "loss": 3.1976,
      "step": 1155
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.718218982219696,
      "learning_rate": 9.998270046215797e-05,
      "loss": 3.1375,
      "step": 1160
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7185037732124329,
      "learning_rate": 9.998253688358078e-05,
      "loss": 3.2753,
      "step": 1165
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.791697084903717,
      "learning_rate": 9.998237253540174e-05,
      "loss": 3.21,
      "step": 1170
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7872958779335022,
      "learning_rate": 9.998220741762338e-05,
      "loss": 3.2575,
      "step": 1175
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7162549495697021,
      "learning_rate": 9.998204153024821e-05,
      "loss": 3.2133,
      "step": 1180
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8020044565200806,
      "learning_rate": 9.998187487327882e-05,
      "loss": 3.1511,
      "step": 1185
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7478770613670349,
      "learning_rate": 9.998170744671775e-05,
      "loss": 3.2126,
      "step": 1190
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.906419038772583,
      "learning_rate": 9.998153925056758e-05,
      "loss": 3.076,
      "step": 1195
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7775239944458008,
      "learning_rate": 9.998137028483092e-05,
      "loss": 3.1069,
      "step": 1200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7984919548034668,
      "learning_rate": 9.998120054951035e-05,
      "loss": 3.1638,
      "step": 1205
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7629850506782532,
      "learning_rate": 9.99810300446085e-05,
      "loss": 3.1577,
      "step": 1210
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7063381671905518,
      "learning_rate": 9.9980858770128e-05,
      "loss": 3.1524,
      "step": 1215
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.697745680809021,
      "learning_rate": 9.998068672607146e-05,
      "loss": 3.0961,
      "step": 1220
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7392011284828186,
      "learning_rate": 9.998051391244155e-05,
      "loss": 3.0844,
      "step": 1225
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7307007312774658,
      "learning_rate": 9.998034032924092e-05,
      "loss": 3.1287,
      "step": 1230
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.778901219367981,
      "learning_rate": 9.998016597647225e-05,
      "loss": 3.0723,
      "step": 1235
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7319424748420715,
      "learning_rate": 9.997999085413823e-05,
      "loss": 3.026,
      "step": 1240
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7516511678695679,
      "learning_rate": 9.997981496224153e-05,
      "loss": 3.1086,
      "step": 1245
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7049089074134827,
      "learning_rate": 9.99796383007849e-05,
      "loss": 3.1041,
      "step": 1250
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7009226679801941,
      "learning_rate": 9.997946086977103e-05,
      "loss": 3.0769,
      "step": 1255
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6842823624610901,
      "learning_rate": 9.997928266920266e-05,
      "loss": 2.9793,
      "step": 1260
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6773582696914673,
      "learning_rate": 9.997910369908254e-05,
      "loss": 3.1953,
      "step": 1265
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7307072877883911,
      "learning_rate": 9.99789239594134e-05,
      "loss": 3.1331,
      "step": 1270
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.772514820098877,
      "learning_rate": 9.997874345019805e-05,
      "loss": 3.0518,
      "step": 1275
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7450544834136963,
      "learning_rate": 9.997856217143921e-05,
      "loss": 3.1585,
      "step": 1280
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7175887823104858,
      "learning_rate": 9.997838012313974e-05,
      "loss": 3.1429,
      "step": 1285
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.708309531211853,
      "learning_rate": 9.99781973053024e-05,
      "loss": 3.1191,
      "step": 1290
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6933179497718811,
      "learning_rate": 9.997801371793001e-05,
      "loss": 3.1186,
      "step": 1295
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7419211268424988,
      "learning_rate": 9.997782936102542e-05,
      "loss": 3.0695,
      "step": 1300
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6648455858230591,
      "learning_rate": 9.997764423459144e-05,
      "loss": 3.1001,
      "step": 1305
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7579753994941711,
      "learning_rate": 9.997745833863092e-05,
      "loss": 3.0819,
      "step": 1310
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6626492738723755,
      "learning_rate": 9.997727167314676e-05,
      "loss": 3.0989,
      "step": 1315
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7359100580215454,
      "learning_rate": 9.997708423814178e-05,
      "loss": 3.1212,
      "step": 1320
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6784576773643494,
      "learning_rate": 9.99768960336189e-05,
      "loss": 3.232,
      "step": 1325
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.771381676197052,
      "learning_rate": 9.997670705958102e-05,
      "loss": 3.0066,
      "step": 1330
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6609118580818176,
      "learning_rate": 9.997651731603104e-05,
      "loss": 3.0576,
      "step": 1335
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7440353631973267,
      "learning_rate": 9.997632680297188e-05,
      "loss": 3.0011,
      "step": 1340
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7288987040519714,
      "learning_rate": 9.997613552040648e-05,
      "loss": 3.0711,
      "step": 1345
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7144935131072998,
      "learning_rate": 9.997594346833779e-05,
      "loss": 3.1468,
      "step": 1350
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7377967238426208,
      "learning_rate": 9.997575064676875e-05,
      "loss": 2.9931,
      "step": 1355
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7527424693107605,
      "learning_rate": 9.997555705570232e-05,
      "loss": 3.0382,
      "step": 1360
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7120419144630432,
      "learning_rate": 9.997536269514152e-05,
      "loss": 3.0208,
      "step": 1365
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7157734632492065,
      "learning_rate": 9.99751675650893e-05,
      "loss": 3.0208,
      "step": 1370
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.734337568283081,
      "learning_rate": 9.99749716655487e-05,
      "loss": 3.0317,
      "step": 1375
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6594445109367371,
      "learning_rate": 9.997477499652273e-05,
      "loss": 3.1288,
      "step": 1380
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6981396675109863,
      "learning_rate": 9.99745775580144e-05,
      "loss": 3.0433,
      "step": 1385
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7160163521766663,
      "learning_rate": 9.997437935002676e-05,
      "loss": 3.135,
      "step": 1390
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6922858953475952,
      "learning_rate": 9.997418037256285e-05,
      "loss": 3.0693,
      "step": 1395
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.659522294998169,
      "learning_rate": 9.997398062562575e-05,
      "loss": 3.0013,
      "step": 1400
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6595268845558167,
      "learning_rate": 9.997378010921856e-05,
      "loss": 2.9828,
      "step": 1405
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7134171724319458,
      "learning_rate": 9.99735788233443e-05,
      "loss": 3.0607,
      "step": 1410
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7301485538482666,
      "learning_rate": 9.997337676800613e-05,
      "loss": 2.9043,
      "step": 1415
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7607885599136353,
      "learning_rate": 9.997317394320713e-05,
      "loss": 3.0819,
      "step": 1420
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6910939812660217,
      "learning_rate": 9.997297034895044e-05,
      "loss": 3.0466,
      "step": 1425
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6826831698417664,
      "learning_rate": 9.99727659852392e-05,
      "loss": 3.0934,
      "step": 1430
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7005428075790405,
      "learning_rate": 9.997256085207652e-05,
      "loss": 3.0576,
      "step": 1435
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.746114194393158,
      "learning_rate": 9.997235494946557e-05,
      "loss": 3.0963,
      "step": 1440
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7530882358551025,
      "learning_rate": 9.997214827740955e-05,
      "loss": 3.0065,
      "step": 1445
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.687054455280304,
      "learning_rate": 9.997194083591164e-05,
      "loss": 2.9684,
      "step": 1450
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6739320158958435,
      "learning_rate": 9.9971732624975e-05,
      "loss": 2.946,
      "step": 1455
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7423838376998901,
      "learning_rate": 9.997152364460287e-05,
      "loss": 3.0657,
      "step": 1460
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6832091808319092,
      "learning_rate": 9.997131389479844e-05,
      "loss": 3.061,
      "step": 1465
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7025681734085083,
      "learning_rate": 9.997110337556494e-05,
      "loss": 3.0702,
      "step": 1470
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6948280334472656,
      "learning_rate": 9.997089208690564e-05,
      "loss": 2.8701,
      "step": 1475
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.762286365032196,
      "learning_rate": 9.997068002882377e-05,
      "loss": 2.9458,
      "step": 1480
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7215501070022583,
      "learning_rate": 9.997046720132261e-05,
      "loss": 3.002,
      "step": 1485
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7120538949966431,
      "learning_rate": 9.997025360440541e-05,
      "loss": 2.9304,
      "step": 1490
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6906287670135498,
      "learning_rate": 9.997003923807551e-05,
      "loss": 2.9874,
      "step": 1495
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6665874123573303,
      "learning_rate": 9.996982410233615e-05,
      "loss": 2.9974,
      "step": 1500
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6874006390571594,
      "learning_rate": 9.996960819719068e-05,
      "loss": 2.9395,
      "step": 1505
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6323267221450806,
      "learning_rate": 9.996939152264242e-05,
      "loss": 2.9654,
      "step": 1510
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6188246011734009,
      "learning_rate": 9.99691740786947e-05,
      "loss": 2.9102,
      "step": 1515
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6180820465087891,
      "learning_rate": 9.996895586535086e-05,
      "loss": 2.9988,
      "step": 1520
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7066469192504883,
      "learning_rate": 9.996873688261428e-05,
      "loss": 3.0628,
      "step": 1525
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6971163749694824,
      "learning_rate": 9.996851713048831e-05,
      "loss": 3.0725,
      "step": 1530
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6099901795387268,
      "learning_rate": 9.996829660897637e-05,
      "loss": 2.9595,
      "step": 1535
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7307775616645813,
      "learning_rate": 9.996807531808181e-05,
      "loss": 3.0753,
      "step": 1540
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6743276715278625,
      "learning_rate": 9.996785325780806e-05,
      "loss": 3.0507,
      "step": 1545
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6724675893783569,
      "learning_rate": 9.996763042815853e-05,
      "loss": 2.9571,
      "step": 1550
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6878014802932739,
      "learning_rate": 9.996740682913668e-05,
      "loss": 3.0095,
      "step": 1555
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6683099269866943,
      "learning_rate": 9.996718246074591e-05,
      "loss": 3.0054,
      "step": 1560
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6748294830322266,
      "learning_rate": 9.99669573229897e-05,
      "loss": 2.9908,
      "step": 1565
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6642842292785645,
      "learning_rate": 9.996673141587151e-05,
      "loss": 3.0206,
      "step": 1570
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6406435966491699,
      "learning_rate": 9.996650473939482e-05,
      "loss": 3.007,
      "step": 1575
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6542099118232727,
      "learning_rate": 9.996627729356314e-05,
      "loss": 3.0519,
      "step": 1580
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6434016227722168,
      "learning_rate": 9.996604907837993e-05,
      "loss": 2.9962,
      "step": 1585
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6845762133598328,
      "learning_rate": 9.996582009384874e-05,
      "loss": 2.9858,
      "step": 1590
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6658374071121216,
      "learning_rate": 9.996559033997307e-05,
      "loss": 2.9897,
      "step": 1595
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6587632298469543,
      "learning_rate": 9.996535981675648e-05,
      "loss": 2.9072,
      "step": 1600
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6414953470230103,
      "learning_rate": 9.996512852420251e-05,
      "loss": 2.9837,
      "step": 1605
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6526081562042236,
      "learning_rate": 9.996489646231471e-05,
      "loss": 2.999,
      "step": 1610
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7001782059669495,
      "learning_rate": 9.996466363109667e-05,
      "loss": 2.8893,
      "step": 1615
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6682369112968445,
      "learning_rate": 9.996443003055196e-05,
      "loss": 3.0231,
      "step": 1620
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7254176139831543,
      "learning_rate": 9.99641956606842e-05,
      "loss": 2.9728,
      "step": 1625
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6807917952537537,
      "learning_rate": 9.996396052149698e-05,
      "loss": 2.9534,
      "step": 1630
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6344894766807556,
      "learning_rate": 9.996372461299392e-05,
      "loss": 2.9067,
      "step": 1635
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6123491525650024,
      "learning_rate": 9.996348793517866e-05,
      "loss": 2.9103,
      "step": 1640
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6646186709403992,
      "learning_rate": 9.996325048805484e-05,
      "loss": 2.9041,
      "step": 1645
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6371365189552307,
      "learning_rate": 9.996301227162612e-05,
      "loss": 2.9974,
      "step": 1650
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6955332159996033,
      "learning_rate": 9.996277328589618e-05,
      "loss": 3.0168,
      "step": 1655
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6499887108802795,
      "learning_rate": 9.996253353086866e-05,
      "loss": 2.9981,
      "step": 1660
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6034931540489197,
      "learning_rate": 9.996229300654729e-05,
      "loss": 2.9212,
      "step": 1665
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6491664052009583,
      "learning_rate": 9.996205171293576e-05,
      "loss": 2.9033,
      "step": 1670
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6661312580108643,
      "learning_rate": 9.996180965003778e-05,
      "loss": 3.0415,
      "step": 1675
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6164848804473877,
      "learning_rate": 9.996156681785708e-05,
      "loss": 2.9594,
      "step": 1680
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5869746804237366,
      "learning_rate": 9.996132321639741e-05,
      "loss": 2.9452,
      "step": 1685
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6178178191184998,
      "learning_rate": 9.996107884566252e-05,
      "loss": 2.9365,
      "step": 1690
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6178844571113586,
      "learning_rate": 9.996083370565616e-05,
      "loss": 2.9443,
      "step": 1695
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6911107301712036,
      "learning_rate": 9.99605877963821e-05,
      "loss": 2.9838,
      "step": 1700
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6590896844863892,
      "learning_rate": 9.996034111784413e-05,
      "loss": 2.9883,
      "step": 1705
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6115434765815735,
      "learning_rate": 9.996009367004607e-05,
      "loss": 2.9015,
      "step": 1710
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6378669142723083,
      "learning_rate": 9.995984545299171e-05,
      "loss": 2.9652,
      "step": 1715
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6293317675590515,
      "learning_rate": 9.995959646668488e-05,
      "loss": 2.9267,
      "step": 1720
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6136114001274109,
      "learning_rate": 9.995934671112941e-05,
      "loss": 2.917,
      "step": 1725
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6761816143989563,
      "learning_rate": 9.995909618632915e-05,
      "loss": 2.9885,
      "step": 1730
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6426324844360352,
      "learning_rate": 9.995884489228794e-05,
      "loss": 2.9601,
      "step": 1735
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6569941639900208,
      "learning_rate": 9.995859282900967e-05,
      "loss": 2.9025,
      "step": 1740
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5927703976631165,
      "learning_rate": 9.995833999649821e-05,
      "loss": 2.901,
      "step": 1745
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6033011674880981,
      "learning_rate": 9.995808639475746e-05,
      "loss": 2.9061,
      "step": 1750
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6786762475967407,
      "learning_rate": 9.995783202379131e-05,
      "loss": 2.9531,
      "step": 1755
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6934323310852051,
      "learning_rate": 9.99575768836037e-05,
      "loss": 2.9172,
      "step": 1760
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6619399785995483,
      "learning_rate": 9.995732097419855e-05,
      "loss": 2.9246,
      "step": 1765
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6082199215888977,
      "learning_rate": 9.995706429557978e-05,
      "loss": 2.9967,
      "step": 1770
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6528149843215942,
      "learning_rate": 9.995680684775137e-05,
      "loss": 3.0095,
      "step": 1775
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7223584055900574,
      "learning_rate": 9.995654863071729e-05,
      "loss": 2.7954,
      "step": 1780
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7765888571739197,
      "learning_rate": 9.995628964448147e-05,
      "loss": 2.9265,
      "step": 1785
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6559895277023315,
      "learning_rate": 9.995602988904793e-05,
      "loss": 3.003,
      "step": 1790
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5960218906402588,
      "learning_rate": 9.995576936442068e-05,
      "loss": 2.8787,
      "step": 1795
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5916800498962402,
      "learning_rate": 9.995550807060371e-05,
      "loss": 2.9118,
      "step": 1800
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6400606632232666,
      "learning_rate": 9.995524600760105e-05,
      "loss": 2.8409,
      "step": 1805
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6375911831855774,
      "learning_rate": 9.995498317541674e-05,
      "loss": 2.9476,
      "step": 1810
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6217756271362305,
      "learning_rate": 9.995471957405483e-05,
      "loss": 2.7785,
      "step": 1815
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7918213605880737,
      "learning_rate": 9.995445520351935e-05,
      "loss": 2.9981,
      "step": 1820
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5906363725662231,
      "learning_rate": 9.995419006381442e-05,
      "loss": 2.9044,
      "step": 1825
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5905340313911438,
      "learning_rate": 9.995392415494407e-05,
      "loss": 2.9472,
      "step": 1830
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6386330723762512,
      "learning_rate": 9.995365747691243e-05,
      "loss": 2.9446,
      "step": 1835
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6660987734794617,
      "learning_rate": 9.995339002972359e-05,
      "loss": 2.9167,
      "step": 1840
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6718746423721313,
      "learning_rate": 9.995312181338168e-05,
      "loss": 2.957,
      "step": 1845
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5916494727134705,
      "learning_rate": 9.995285282789082e-05,
      "loss": 2.9731,
      "step": 1850
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6081727743148804,
      "learning_rate": 9.995258307325515e-05,
      "loss": 2.8996,
      "step": 1855
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6379761099815369,
      "learning_rate": 9.995231254947883e-05,
      "loss": 2.9089,
      "step": 1860
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6183826923370361,
      "learning_rate": 9.995204125656603e-05,
      "loss": 2.9124,
      "step": 1865
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7123430371284485,
      "learning_rate": 9.995176919452092e-05,
      "loss": 2.8321,
      "step": 1870
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.653268039226532,
      "learning_rate": 9.995149636334769e-05,
      "loss": 2.9323,
      "step": 1875
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5845485329627991,
      "learning_rate": 9.995122276305054e-05,
      "loss": 2.8274,
      "step": 1880
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5870990753173828,
      "learning_rate": 9.995094839363368e-05,
      "loss": 2.879,
      "step": 1885
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6692662239074707,
      "learning_rate": 9.995067325510132e-05,
      "loss": 3.046,
      "step": 1890
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6163268089294434,
      "learning_rate": 9.995039734745774e-05,
      "loss": 2.8931,
      "step": 1895
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6048374176025391,
      "learning_rate": 9.995012067070715e-05,
      "loss": 2.7977,
      "step": 1900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5814905762672424,
      "learning_rate": 9.994984322485383e-05,
      "loss": 2.8357,
      "step": 1905
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.587144672870636,
      "learning_rate": 9.994956500990205e-05,
      "loss": 2.8379,
      "step": 1910
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6135372519493103,
      "learning_rate": 9.994928602585608e-05,
      "loss": 2.8005,
      "step": 1915
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6610756516456604,
      "learning_rate": 9.994900627272023e-05,
      "loss": 2.8722,
      "step": 1920
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6361928582191467,
      "learning_rate": 9.99487257504988e-05,
      "loss": 2.8837,
      "step": 1925
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5890988111495972,
      "learning_rate": 9.994844445919612e-05,
      "loss": 2.9208,
      "step": 1930
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6039892435073853,
      "learning_rate": 9.994816239881648e-05,
      "loss": 2.8886,
      "step": 1935
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6212635040283203,
      "learning_rate": 9.99478795693643e-05,
      "loss": 2.9129,
      "step": 1940
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5875033736228943,
      "learning_rate": 9.994759597084384e-05,
      "loss": 2.8012,
      "step": 1945
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.606772780418396,
      "learning_rate": 9.994731160325955e-05,
      "loss": 2.9052,
      "step": 1950
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5782520771026611,
      "learning_rate": 9.994702646661577e-05,
      "loss": 2.907,
      "step": 1955
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6093077659606934,
      "learning_rate": 9.994674056091689e-05,
      "loss": 2.8991,
      "step": 1960
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5934308171272278,
      "learning_rate": 9.994645388616732e-05,
      "loss": 2.8572,
      "step": 1965
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5909088253974915,
      "learning_rate": 9.994616644237148e-05,
      "loss": 2.8354,
      "step": 1970
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6410349607467651,
      "learning_rate": 9.994587822953378e-05,
      "loss": 2.9032,
      "step": 1975
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6037827134132385,
      "learning_rate": 9.994558924765866e-05,
      "loss": 2.805,
      "step": 1980
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5677337646484375,
      "learning_rate": 9.99452994967506e-05,
      "loss": 2.8627,
      "step": 1985
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6104411482810974,
      "learning_rate": 9.994500897681401e-05,
      "loss": 2.831,
      "step": 1990
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5663310289382935,
      "learning_rate": 9.994471768785339e-05,
      "loss": 2.889,
      "step": 1995
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5860994458198547,
      "learning_rate": 9.994442562987324e-05,
      "loss": 2.8099,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7345673441886902,
      "learning_rate": 9.994413280287803e-05,
      "loss": 2.8285,
      "step": 2005
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6739687919616699,
      "learning_rate": 9.994383920687229e-05,
      "loss": 2.8197,
      "step": 2010
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.584872841835022,
      "learning_rate": 9.994354484186053e-05,
      "loss": 2.9376,
      "step": 2015
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6866011619567871,
      "learning_rate": 9.994324970784727e-05,
      "loss": 2.9073,
      "step": 2020
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5657792091369629,
      "learning_rate": 9.994295380483709e-05,
      "loss": 2.8273,
      "step": 2025
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6064589023590088,
      "learning_rate": 9.99426571328345e-05,
      "loss": 2.9005,
      "step": 2030
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5676794052124023,
      "learning_rate": 9.994235969184411e-05,
      "loss": 2.7046,
      "step": 2035
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6180962920188904,
      "learning_rate": 9.994206148187046e-05,
      "loss": 2.7834,
      "step": 2040
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6457920670509338,
      "learning_rate": 9.994176250291818e-05,
      "loss": 2.7516,
      "step": 2045
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6263876557350159,
      "learning_rate": 9.994146275499186e-05,
      "loss": 2.8006,
      "step": 2050
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6252774596214294,
      "learning_rate": 9.994116223809611e-05,
      "loss": 2.7711,
      "step": 2055
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7848661541938782,
      "learning_rate": 9.994086095223557e-05,
      "loss": 2.6706,
      "step": 2060
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.659628689289093,
      "learning_rate": 9.994055889741485e-05,
      "loss": 2.8816,
      "step": 2065
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6629912257194519,
      "learning_rate": 9.994025607363862e-05,
      "loss": 2.8892,
      "step": 2070
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5890025496482849,
      "learning_rate": 9.993995248091155e-05,
      "loss": 2.8246,
      "step": 2075
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6077884435653687,
      "learning_rate": 9.993964811923831e-05,
      "loss": 2.8843,
      "step": 2080
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5962379574775696,
      "learning_rate": 9.993934298862358e-05,
      "loss": 2.8614,
      "step": 2085
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6014490127563477,
      "learning_rate": 9.993903708907205e-05,
      "loss": 2.8212,
      "step": 2090
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6425051689147949,
      "learning_rate": 9.993873042058845e-05,
      "loss": 2.8085,
      "step": 2095
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5633202791213989,
      "learning_rate": 9.993842298317749e-05,
      "loss": 2.8895,
      "step": 2100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.638590931892395,
      "learning_rate": 9.993811477684392e-05,
      "loss": 2.8179,
      "step": 2105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6229696869850159,
      "learning_rate": 9.993780580159246e-05,
      "loss": 2.7942,
      "step": 2110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6182060837745667,
      "learning_rate": 9.993749605742788e-05,
      "loss": 2.8442,
      "step": 2115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6393762230873108,
      "learning_rate": 9.993718554435495e-05,
      "loss": 2.9508,
      "step": 2120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.594402551651001,
      "learning_rate": 9.993687426237846e-05,
      "loss": 2.8063,
      "step": 2125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5687596201896667,
      "learning_rate": 9.993656221150318e-05,
      "loss": 2.8609,
      "step": 2130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6220025420188904,
      "learning_rate": 9.993624939173394e-05,
      "loss": 2.7422,
      "step": 2135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5810810923576355,
      "learning_rate": 9.993593580307553e-05,
      "loss": 2.8206,
      "step": 2140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5967430472373962,
      "learning_rate": 9.99356214455328e-05,
      "loss": 2.7649,
      "step": 2145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.588529109954834,
      "learning_rate": 9.993530631911059e-05,
      "loss": 2.8731,
      "step": 2150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6125682592391968,
      "learning_rate": 9.993499042381373e-05,
      "loss": 2.8575,
      "step": 2155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5827690362930298,
      "learning_rate": 9.993467375964711e-05,
      "loss": 2.7779,
      "step": 2160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6052179932594299,
      "learning_rate": 9.99343563266156e-05,
      "loss": 2.9429,
      "step": 2165
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6475282907485962,
      "learning_rate": 9.993403812472408e-05,
      "loss": 2.7794,
      "step": 2170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6134741306304932,
      "learning_rate": 9.993371915397745e-05,
      "loss": 2.8014,
      "step": 2175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6013492941856384,
      "learning_rate": 9.993339941438062e-05,
      "loss": 2.8495,
      "step": 2180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5772548317909241,
      "learning_rate": 9.993307890593853e-05,
      "loss": 2.7981,
      "step": 2185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6609306335449219,
      "learning_rate": 9.993275762865609e-05,
      "loss": 2.832,
      "step": 2190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6265547275543213,
      "learning_rate": 9.993243558253826e-05,
      "loss": 2.8032,
      "step": 2195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6226460337638855,
      "learning_rate": 9.993211276759e-05,
      "loss": 2.8619,
      "step": 2200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.600709855556488,
      "learning_rate": 9.993178918381628e-05,
      "loss": 2.8337,
      "step": 2205
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5936564803123474,
      "learning_rate": 9.993146483122207e-05,
      "loss": 2.7985,
      "step": 2210
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5660296082496643,
      "learning_rate": 9.993113970981238e-05,
      "loss": 2.8352,
      "step": 2215
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5602771043777466,
      "learning_rate": 9.993081381959221e-05,
      "loss": 2.8656,
      "step": 2220
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5585789084434509,
      "learning_rate": 9.99304871605666e-05,
      "loss": 2.769,
      "step": 2225
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5992918610572815,
      "learning_rate": 9.993015973274053e-05,
      "loss": 2.8855,
      "step": 2230
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5875166654586792,
      "learning_rate": 9.992983153611908e-05,
      "loss": 2.8846,
      "step": 2235
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5779223442077637,
      "learning_rate": 9.99295025707073e-05,
      "loss": 2.8132,
      "step": 2240
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6243892312049866,
      "learning_rate": 9.992917283651024e-05,
      "loss": 2.797,
      "step": 2245
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6116815209388733,
      "learning_rate": 9.992884233353298e-05,
      "loss": 2.7691,
      "step": 2250
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.591465413570404,
      "learning_rate": 9.992851106178063e-05,
      "loss": 2.8762,
      "step": 2255
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6097704768180847,
      "learning_rate": 9.992817902125826e-05,
      "loss": 2.8234,
      "step": 2260
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5803406834602356,
      "learning_rate": 9.9927846211971e-05,
      "loss": 2.8682,
      "step": 2265
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6173955798149109,
      "learning_rate": 9.992751263392397e-05,
      "loss": 2.789,
      "step": 2270
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5837902426719666,
      "learning_rate": 9.992717828712232e-05,
      "loss": 2.8457,
      "step": 2275
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5767691135406494,
      "learning_rate": 9.992684317157117e-05,
      "loss": 2.7776,
      "step": 2280
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5765328407287598,
      "learning_rate": 9.992650728727571e-05,
      "loss": 2.8023,
      "step": 2285
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5791749358177185,
      "learning_rate": 9.99261706342411e-05,
      "loss": 2.8468,
      "step": 2290
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9553482532501221,
      "learning_rate": 9.99258332124725e-05,
      "loss": 2.8294,
      "step": 2295
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5476585626602173,
      "learning_rate": 9.992549502197514e-05,
      "loss": 2.766,
      "step": 2300
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5768662095069885,
      "learning_rate": 9.992515606275423e-05,
      "loss": 2.7767,
      "step": 2305
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5564925670623779,
      "learning_rate": 9.992481633481495e-05,
      "loss": 2.8214,
      "step": 2310
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5379332900047302,
      "learning_rate": 9.992447583816257e-05,
      "loss": 2.7077,
      "step": 2315
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5444866418838501,
      "learning_rate": 9.992413457280231e-05,
      "loss": 2.7534,
      "step": 2320
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5604690909385681,
      "learning_rate": 9.992379253873943e-05,
      "loss": 2.7819,
      "step": 2325
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.666120171546936,
      "learning_rate": 9.99234497359792e-05,
      "loss": 2.7529,
      "step": 2330
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5457541346549988,
      "learning_rate": 9.992310616452689e-05,
      "loss": 2.8092,
      "step": 2335
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6704611778259277,
      "learning_rate": 9.99227618243878e-05,
      "loss": 2.8705,
      "step": 2340
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5706437230110168,
      "learning_rate": 9.992241671556724e-05,
      "loss": 2.8518,
      "step": 2345
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5592533349990845,
      "learning_rate": 9.99220708380705e-05,
      "loss": 2.743,
      "step": 2350
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5520803332328796,
      "learning_rate": 9.99217241919029e-05,
      "loss": 2.8164,
      "step": 2355
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6195580363273621,
      "learning_rate": 9.992137677706982e-05,
      "loss": 2.8266,
      "step": 2360
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6236368417739868,
      "learning_rate": 9.99210285935766e-05,
      "loss": 2.8168,
      "step": 2365
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6074082851409912,
      "learning_rate": 9.992067964142856e-05,
      "loss": 2.7844,
      "step": 2370
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5505146980285645,
      "learning_rate": 9.992032992063109e-05,
      "loss": 2.7745,
      "step": 2375
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5659940838813782,
      "learning_rate": 9.99199794311896e-05,
      "loss": 2.7655,
      "step": 2380
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5549494028091431,
      "learning_rate": 9.991962817310947e-05,
      "loss": 2.758,
      "step": 2385
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5577967762947083,
      "learning_rate": 9.991927614639612e-05,
      "loss": 2.7967,
      "step": 2390
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5398279428482056,
      "learning_rate": 9.991892335105493e-05,
      "loss": 2.7662,
      "step": 2395
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5548604726791382,
      "learning_rate": 9.991856978709137e-05,
      "loss": 2.7732,
      "step": 2400
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5554727911949158,
      "learning_rate": 9.99182154545109e-05,
      "loss": 2.7963,
      "step": 2405
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.641569197177887,
      "learning_rate": 9.991786035331893e-05,
      "loss": 2.8123,
      "step": 2410
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5932303071022034,
      "learning_rate": 9.991750448352096e-05,
      "loss": 2.6855,
      "step": 2415
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6493214964866638,
      "learning_rate": 9.991714784512243e-05,
      "loss": 2.7081,
      "step": 2420
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5881122350692749,
      "learning_rate": 9.991679043812887e-05,
      "loss": 2.83,
      "step": 2425
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5996439456939697,
      "learning_rate": 9.99164322625458e-05,
      "loss": 2.7966,
      "step": 2430
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5461181402206421,
      "learning_rate": 9.991607331837867e-05,
      "loss": 2.747,
      "step": 2435
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5765201449394226,
      "learning_rate": 9.991571360563306e-05,
      "loss": 2.8224,
      "step": 2440
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5844093561172485,
      "learning_rate": 9.99153531243145e-05,
      "loss": 2.8055,
      "step": 2445
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5607630014419556,
      "learning_rate": 9.991499187442852e-05,
      "loss": 2.7217,
      "step": 2450
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5314213633537292,
      "learning_rate": 9.99146298559807e-05,
      "loss": 2.7377,
      "step": 2455
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5622405409812927,
      "learning_rate": 9.991426706897661e-05,
      "loss": 2.758,
      "step": 2460
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5749192833900452,
      "learning_rate": 9.991390351342185e-05,
      "loss": 2.6864,
      "step": 2465
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5853544473648071,
      "learning_rate": 9.991353918932198e-05,
      "loss": 2.7541,
      "step": 2470
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5883823037147522,
      "learning_rate": 9.991317409668264e-05,
      "loss": 2.8416,
      "step": 2475
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5632338523864746,
      "learning_rate": 9.991280823550945e-05,
      "loss": 2.7668,
      "step": 2480
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5229887366294861,
      "learning_rate": 9.991244160580803e-05,
      "loss": 2.7354,
      "step": 2485
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5903317332267761,
      "learning_rate": 9.991207420758404e-05,
      "loss": 2.7345,
      "step": 2490
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5514800548553467,
      "learning_rate": 9.991170604084314e-05,
      "loss": 2.7589,
      "step": 2495
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6564476490020752,
      "learning_rate": 9.991133710559097e-05,
      "loss": 2.7603,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6076124906539917,
      "learning_rate": 9.991096740183324e-05,
      "loss": 2.8327,
      "step": 2505
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6043878197669983,
      "learning_rate": 9.991059692957563e-05,
      "loss": 2.8157,
      "step": 2510
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5622871518135071,
      "learning_rate": 9.991022568882384e-05,
      "loss": 2.7797,
      "step": 2515
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5332220792770386,
      "learning_rate": 9.99098536795836e-05,
      "loss": 2.7628,
      "step": 2520
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5546334981918335,
      "learning_rate": 9.990948090186063e-05,
      "loss": 2.8003,
      "step": 2525
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5108422636985779,
      "learning_rate": 9.990910735566065e-05,
      "loss": 2.7761,
      "step": 2530
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5122519135475159,
      "learning_rate": 9.990873304098946e-05,
      "loss": 2.7268,
      "step": 2535
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.676216185092926,
      "learning_rate": 9.990835795785278e-05,
      "loss": 2.7806,
      "step": 2540
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.586461067199707,
      "learning_rate": 9.990798210625642e-05,
      "loss": 2.7773,
      "step": 2545
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5285787582397461,
      "learning_rate": 9.990760548620613e-05,
      "loss": 2.7266,
      "step": 2550
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5570098757743835,
      "learning_rate": 9.990722809770773e-05,
      "loss": 2.8205,
      "step": 2555
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5485754013061523,
      "learning_rate": 9.990684994076702e-05,
      "loss": 2.7751,
      "step": 2560
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5630438327789307,
      "learning_rate": 9.990647101538983e-05,
      "loss": 2.7252,
      "step": 2565
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5532368421554565,
      "learning_rate": 9.990609132158201e-05,
      "loss": 2.7761,
      "step": 2570
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5329113602638245,
      "learning_rate": 9.990571085934938e-05,
      "loss": 2.7448,
      "step": 2575
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5626487135887146,
      "learning_rate": 9.990532962869783e-05,
      "loss": 2.7335,
      "step": 2580
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.517505407333374,
      "learning_rate": 9.990494762963318e-05,
      "loss": 2.7641,
      "step": 2585
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5552113056182861,
      "learning_rate": 9.990456486216136e-05,
      "loss": 2.7006,
      "step": 2590
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5434961915016174,
      "learning_rate": 9.990418132628824e-05,
      "loss": 2.722,
      "step": 2595
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5944434404373169,
      "learning_rate": 9.990379702201974e-05,
      "loss": 2.7018,
      "step": 2600
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5404393672943115,
      "learning_rate": 9.990341194936173e-05,
      "loss": 2.7499,
      "step": 2605
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5616127252578735,
      "learning_rate": 9.990302610832021e-05,
      "loss": 2.7176,
      "step": 2610
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5197855830192566,
      "learning_rate": 9.990263949890108e-05,
      "loss": 2.8037,
      "step": 2615
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5537593364715576,
      "learning_rate": 9.990225212111029e-05,
      "loss": 2.6442,
      "step": 2620
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5792954564094543,
      "learning_rate": 9.990186397495383e-05,
      "loss": 2.7893,
      "step": 2625
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5314340591430664,
      "learning_rate": 9.990147506043766e-05,
      "loss": 2.7168,
      "step": 2630
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5506948828697205,
      "learning_rate": 9.990108537756775e-05,
      "loss": 2.6876,
      "step": 2635
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5075097680091858,
      "learning_rate": 9.990069492635015e-05,
      "loss": 2.7664,
      "step": 2640
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.563379168510437,
      "learning_rate": 9.990030370679082e-05,
      "loss": 2.6777,
      "step": 2645
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5424590110778809,
      "learning_rate": 9.989991171889581e-05,
      "loss": 2.7762,
      "step": 2650
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5738586783409119,
      "learning_rate": 9.989951896267115e-05,
      "loss": 2.8039,
      "step": 2655
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5715847015380859,
      "learning_rate": 9.989912543812289e-05,
      "loss": 2.7106,
      "step": 2660
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5335726141929626,
      "learning_rate": 9.989873114525709e-05,
      "loss": 2.7817,
      "step": 2665
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5716210007667542,
      "learning_rate": 9.989833608407981e-05,
      "loss": 2.691,
      "step": 2670
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.540751039981842,
      "learning_rate": 9.989794025459715e-05,
      "loss": 2.71,
      "step": 2675
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5484097599983215,
      "learning_rate": 9.989754365681519e-05,
      "loss": 2.6558,
      "step": 2680
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5558391809463501,
      "learning_rate": 9.989714629074006e-05,
      "loss": 2.7702,
      "step": 2685
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.525719940662384,
      "learning_rate": 9.989674815637784e-05,
      "loss": 2.7225,
      "step": 2690
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.513150691986084,
      "learning_rate": 9.989634925373468e-05,
      "loss": 2.7614,
      "step": 2695
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5460965633392334,
      "learning_rate": 9.989594958281673e-05,
      "loss": 2.6378,
      "step": 2700
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.546363890171051,
      "learning_rate": 9.989554914363015e-05,
      "loss": 2.7108,
      "step": 2705
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5618521571159363,
      "learning_rate": 9.989514793618109e-05,
      "loss": 2.6784,
      "step": 2710
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5239582657814026,
      "learning_rate": 9.989474596047572e-05,
      "loss": 2.8489,
      "step": 2715
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5998889803886414,
      "learning_rate": 9.989434321652024e-05,
      "loss": 2.7929,
      "step": 2720
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6108545660972595,
      "learning_rate": 9.989393970432085e-05,
      "loss": 2.7515,
      "step": 2725
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5883182883262634,
      "learning_rate": 9.989353542388378e-05,
      "loss": 2.6468,
      "step": 2730
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6191084980964661,
      "learning_rate": 9.989313037521524e-05,
      "loss": 2.6375,
      "step": 2735
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5718762874603271,
      "learning_rate": 9.989272455832145e-05,
      "loss": 2.707,
      "step": 2740
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5779682993888855,
      "learning_rate": 9.989231797320869e-05,
      "loss": 2.7221,
      "step": 2745
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5891039371490479,
      "learning_rate": 9.989191061988319e-05,
      "loss": 2.7767,
      "step": 2750
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5651604533195496,
      "learning_rate": 9.989150249835125e-05,
      "loss": 2.6954,
      "step": 2755
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5214573740959167,
      "learning_rate": 9.989109360861913e-05,
      "loss": 2.7909,
      "step": 2760
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5371715426445007,
      "learning_rate": 9.989068395069313e-05,
      "loss": 2.7602,
      "step": 2765
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.532345175743103,
      "learning_rate": 9.989027352457958e-05,
      "loss": 2.8279,
      "step": 2770
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5614933967590332,
      "learning_rate": 9.988986233028478e-05,
      "loss": 2.7481,
      "step": 2775
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5751219987869263,
      "learning_rate": 9.988945036781507e-05,
      "loss": 2.704,
      "step": 2780
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5688990950584412,
      "learning_rate": 9.988903763717679e-05,
      "loss": 2.6255,
      "step": 2785
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.564172625541687,
      "learning_rate": 9.988862413837628e-05,
      "loss": 2.683,
      "step": 2790
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5096065402030945,
      "learning_rate": 9.988820987141993e-05,
      "loss": 2.7548,
      "step": 2795
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5488696694374084,
      "learning_rate": 9.988779483631409e-05,
      "loss": 2.7881,
      "step": 2800
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5133165121078491,
      "learning_rate": 9.98873790330652e-05,
      "loss": 2.668,
      "step": 2805
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5278681516647339,
      "learning_rate": 9.988696246167961e-05,
      "loss": 2.7369,
      "step": 2810
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5096678733825684,
      "learning_rate": 9.988654512216378e-05,
      "loss": 2.7268,
      "step": 2815
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5189996957778931,
      "learning_rate": 9.988612701452409e-05,
      "loss": 2.6634,
      "step": 2820
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5260164141654968,
      "learning_rate": 9.988570813876701e-05,
      "loss": 2.697,
      "step": 2825
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.566125214099884,
      "learning_rate": 9.988528849489897e-05,
      "loss": 2.7568,
      "step": 2830
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5293660163879395,
      "learning_rate": 9.988486808292646e-05,
      "loss": 2.6333,
      "step": 2835
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5165902972221375,
      "learning_rate": 9.988444690285592e-05,
      "loss": 2.7245,
      "step": 2840
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5085412859916687,
      "learning_rate": 9.988402495469385e-05,
      "loss": 2.6695,
      "step": 2845
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5047938227653503,
      "learning_rate": 9.988360223844673e-05,
      "loss": 2.7279,
      "step": 2850
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5740139484405518,
      "learning_rate": 9.98831787541211e-05,
      "loss": 2.6985,
      "step": 2855
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5511103272438049,
      "learning_rate": 9.988275450172346e-05,
      "loss": 2.8694,
      "step": 2860
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6081477999687195,
      "learning_rate": 9.988232948126037e-05,
      "loss": 2.7226,
      "step": 2865
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5188899636268616,
      "learning_rate": 9.988190369273833e-05,
      "loss": 2.7731,
      "step": 2870
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5195143222808838,
      "learning_rate": 9.988147713616391e-05,
      "loss": 2.7258,
      "step": 2875
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5492423176765442,
      "learning_rate": 9.988104981154371e-05,
      "loss": 2.688,
      "step": 2880
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5299638509750366,
      "learning_rate": 9.988062171888429e-05,
      "loss": 2.7256,
      "step": 2885
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5158653855323792,
      "learning_rate": 9.988019285819221e-05,
      "loss": 2.6708,
      "step": 2890
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5343019962310791,
      "learning_rate": 9.987976322947411e-05,
      "loss": 2.7536,
      "step": 2895
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5206443071365356,
      "learning_rate": 9.98793328327366e-05,
      "loss": 2.666,
      "step": 2900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5335565209388733,
      "learning_rate": 9.98789016679863e-05,
      "loss": 2.7717,
      "step": 2905
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5125073194503784,
      "learning_rate": 9.987846973522985e-05,
      "loss": 2.629,
      "step": 2910
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5397815704345703,
      "learning_rate": 9.98780370344739e-05,
      "loss": 2.6899,
      "step": 2915
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5438279509544373,
      "learning_rate": 9.987760356572512e-05,
      "loss": 2.6915,
      "step": 2920
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5073192119598389,
      "learning_rate": 9.987716932899018e-05,
      "loss": 2.7419,
      "step": 2925
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5216573476791382,
      "learning_rate": 9.987673432427576e-05,
      "loss": 2.6845,
      "step": 2930
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.534911572933197,
      "learning_rate": 9.987629855158855e-05,
      "loss": 2.7858,
      "step": 2935
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5791166424751282,
      "learning_rate": 9.98758620109353e-05,
      "loss": 2.6968,
      "step": 2940
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.545536994934082,
      "learning_rate": 9.98754247023227e-05,
      "loss": 2.6637,
      "step": 2945
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5529240965843201,
      "learning_rate": 9.987498662575747e-05,
      "loss": 2.6915,
      "step": 2950
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5895755887031555,
      "learning_rate": 9.987454778124639e-05,
      "loss": 2.8227,
      "step": 2955
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.528200626373291,
      "learning_rate": 9.987410816879619e-05,
      "loss": 2.6331,
      "step": 2960
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5059030652046204,
      "learning_rate": 9.987366778841365e-05,
      "loss": 2.7899,
      "step": 2965
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5321470499038696,
      "learning_rate": 9.987322664010554e-05,
      "loss": 2.6965,
      "step": 2970
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.49717816710472107,
      "learning_rate": 9.987278472387867e-05,
      "loss": 2.6409,
      "step": 2975
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5020269751548767,
      "learning_rate": 9.987234203973986e-05,
      "loss": 2.6829,
      "step": 2980
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5025157928466797,
      "learning_rate": 9.987189858769586e-05,
      "loss": 2.6508,
      "step": 2985
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5445405840873718,
      "learning_rate": 9.987145436775357e-05,
      "loss": 2.6039,
      "step": 2990
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5124613046646118,
      "learning_rate": 9.98710093799198e-05,
      "loss": 2.6557,
      "step": 2995
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5134322643280029,
      "learning_rate": 9.987056362420138e-05,
      "loss": 2.7006,
      "step": 3000
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5479865670204163,
      "learning_rate": 9.98701171006052e-05,
      "loss": 2.7969,
      "step": 3005
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5121752023696899,
      "learning_rate": 9.986966980913817e-05,
      "loss": 2.656,
      "step": 3010
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5319463014602661,
      "learning_rate": 9.986922174980709e-05,
      "loss": 2.6208,
      "step": 3015
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5362205505371094,
      "learning_rate": 9.986877292261893e-05,
      "loss": 2.6864,
      "step": 3020
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5482008457183838,
      "learning_rate": 9.986832332758058e-05,
      "loss": 2.7126,
      "step": 3025
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5209033489227295,
      "learning_rate": 9.986787296469895e-05,
      "loss": 2.7388,
      "step": 3030
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5194149017333984,
      "learning_rate": 9.986742183398099e-05,
      "loss": 2.6547,
      "step": 3035
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5551761388778687,
      "learning_rate": 9.986696993543366e-05,
      "loss": 2.7355,
      "step": 3040
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5037550926208496,
      "learning_rate": 9.986651726906387e-05,
      "loss": 2.6912,
      "step": 3045
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5208343267440796,
      "learning_rate": 9.986606383487862e-05,
      "loss": 2.5746,
      "step": 3050
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5090872645378113,
      "learning_rate": 9.98656096328849e-05,
      "loss": 2.5891,
      "step": 3055
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5009158253669739,
      "learning_rate": 9.98651546630897e-05,
      "loss": 2.6721,
      "step": 3060
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5027348399162292,
      "learning_rate": 9.986469892550001e-05,
      "loss": 2.6029,
      "step": 3065
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5191223621368408,
      "learning_rate": 9.986424242012287e-05,
      "loss": 2.6359,
      "step": 3070
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5179228186607361,
      "learning_rate": 9.986378514696528e-05,
      "loss": 2.7535,
      "step": 3075
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5449337363243103,
      "learning_rate": 9.98633271060343e-05,
      "loss": 2.7222,
      "step": 3080
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5349432826042175,
      "learning_rate": 9.9862868297337e-05,
      "loss": 2.7279,
      "step": 3085
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5382620096206665,
      "learning_rate": 9.98624087208804e-05,
      "loss": 2.6671,
      "step": 3090
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.517651379108429,
      "learning_rate": 9.98619483766716e-05,
      "loss": 2.6219,
      "step": 3095
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.565244734287262,
      "learning_rate": 9.98614872647177e-05,
      "loss": 2.6853,
      "step": 3100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48961013555526733,
      "learning_rate": 9.986102538502579e-05,
      "loss": 2.676,
      "step": 3105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5306321382522583,
      "learning_rate": 9.986056273760297e-05,
      "loss": 2.6327,
      "step": 3110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5058478116989136,
      "learning_rate": 9.986009932245637e-05,
      "loss": 2.7486,
      "step": 3115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5205589532852173,
      "learning_rate": 9.985963513959313e-05,
      "loss": 2.7049,
      "step": 3120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5357959270477295,
      "learning_rate": 9.985917018902041e-05,
      "loss": 2.6962,
      "step": 3125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5069162845611572,
      "learning_rate": 9.985870447074535e-05,
      "loss": 2.5962,
      "step": 3130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5101613998413086,
      "learning_rate": 9.985823798477512e-05,
      "loss": 2.6788,
      "step": 3135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.516661524772644,
      "learning_rate": 9.985777073111691e-05,
      "loss": 2.632,
      "step": 3140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5397171974182129,
      "learning_rate": 9.985730270977791e-05,
      "loss": 2.6219,
      "step": 3145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5125150084495544,
      "learning_rate": 9.985683392076534e-05,
      "loss": 2.7078,
      "step": 3150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5294501781463623,
      "learning_rate": 9.98563643640864e-05,
      "loss": 2.6996,
      "step": 3155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.534930408000946,
      "learning_rate": 9.985589403974833e-05,
      "loss": 2.5993,
      "step": 3160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5082909464836121,
      "learning_rate": 9.985542294775838e-05,
      "loss": 2.6252,
      "step": 3165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5503858923912048,
      "learning_rate": 9.985495108812377e-05,
      "loss": 2.6587,
      "step": 3170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5094603896141052,
      "learning_rate": 9.985447846085181e-05,
      "loss": 2.6526,
      "step": 3175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5308571457862854,
      "learning_rate": 9.985400506594974e-05,
      "loss": 2.6714,
      "step": 3180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49667343497276306,
      "learning_rate": 9.985353090342487e-05,
      "loss": 2.6966,
      "step": 3185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4810599982738495,
      "learning_rate": 9.985305597328449e-05,
      "loss": 2.726,
      "step": 3190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5117978453636169,
      "learning_rate": 9.985258027553593e-05,
      "loss": 2.6943,
      "step": 3195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.505811870098114,
      "learning_rate": 9.98521038101865e-05,
      "loss": 2.7316,
      "step": 3200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5139271020889282,
      "learning_rate": 9.985162657724354e-05,
      "loss": 2.5637,
      "step": 3205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5018622875213623,
      "learning_rate": 9.98511485767144e-05,
      "loss": 2.7099,
      "step": 3210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5111027359962463,
      "learning_rate": 9.985066980860642e-05,
      "loss": 2.6401,
      "step": 3215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4807913303375244,
      "learning_rate": 9.9850190272927e-05,
      "loss": 2.6788,
      "step": 3220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5409941673278809,
      "learning_rate": 9.984970996968352e-05,
      "loss": 2.6626,
      "step": 3225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4858187735080719,
      "learning_rate": 9.984922889888335e-05,
      "loss": 2.6412,
      "step": 3230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5396366715431213,
      "learning_rate": 9.984874706053393e-05,
      "loss": 2.7071,
      "step": 3235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.514211118221283,
      "learning_rate": 9.984826445464266e-05,
      "loss": 2.6606,
      "step": 3240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.527515709400177,
      "learning_rate": 9.984778108121698e-05,
      "loss": 2.6524,
      "step": 3245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5172006487846375,
      "learning_rate": 9.984729694026431e-05,
      "loss": 2.6682,
      "step": 3250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4940512478351593,
      "learning_rate": 9.984681203179214e-05,
      "loss": 2.6602,
      "step": 3255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5335553288459778,
      "learning_rate": 9.984632635580792e-05,
      "loss": 2.5849,
      "step": 3260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.51326584815979,
      "learning_rate": 9.984583991231913e-05,
      "loss": 2.7283,
      "step": 3265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5399080514907837,
      "learning_rate": 9.984535270133324e-05,
      "loss": 2.7139,
      "step": 3270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.501043975353241,
      "learning_rate": 9.984486472285777e-05,
      "loss": 2.7503,
      "step": 3275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4906523525714874,
      "learning_rate": 9.984437597690024e-05,
      "loss": 2.7105,
      "step": 3280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5123621225357056,
      "learning_rate": 9.984388646346817e-05,
      "loss": 2.6445,
      "step": 3285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.528542160987854,
      "learning_rate": 9.984339618256909e-05,
      "loss": 2.703,
      "step": 3290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4952125549316406,
      "learning_rate": 9.984290513421057e-05,
      "loss": 2.7118,
      "step": 3295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5416468381881714,
      "learning_rate": 9.984241331840015e-05,
      "loss": 2.6502,
      "step": 3300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5029064416885376,
      "learning_rate": 9.984192073514538e-05,
      "loss": 2.7902,
      "step": 3305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4966199994087219,
      "learning_rate": 9.98414273844539e-05,
      "loss": 2.5977,
      "step": 3310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5337540507316589,
      "learning_rate": 9.984093326633325e-05,
      "loss": 2.6941,
      "step": 3315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5469340682029724,
      "learning_rate": 9.98404383807911e-05,
      "loss": 2.6653,
      "step": 3320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5816378593444824,
      "learning_rate": 9.983994272783502e-05,
      "loss": 2.7109,
      "step": 3325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5296834707260132,
      "learning_rate": 9.983944630747266e-05,
      "loss": 2.6872,
      "step": 3330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4935688078403473,
      "learning_rate": 9.983894911971167e-05,
      "loss": 2.7075,
      "step": 3335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5713367462158203,
      "learning_rate": 9.983845116455969e-05,
      "loss": 2.6023,
      "step": 3340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5040152072906494,
      "learning_rate": 9.983795244202439e-05,
      "loss": 2.6675,
      "step": 3345
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4885009825229645,
      "learning_rate": 9.983745295211347e-05,
      "loss": 2.6463,
      "step": 3350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4754915237426758,
      "learning_rate": 9.983695269483459e-05,
      "loss": 2.6291,
      "step": 3355
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5501618981361389,
      "learning_rate": 9.983645167019546e-05,
      "loss": 2.6428,
      "step": 3360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5042656064033508,
      "learning_rate": 9.983594987820381e-05,
      "loss": 2.6324,
      "step": 3365
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49827688932418823,
      "learning_rate": 9.983544731886735e-05,
      "loss": 2.6634,
      "step": 3370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.509573221206665,
      "learning_rate": 9.983494399219386e-05,
      "loss": 2.5538,
      "step": 3375
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5233283638954163,
      "learning_rate": 9.983443989819103e-05,
      "loss": 2.6592,
      "step": 3380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47309717535972595,
      "learning_rate": 9.983393503686665e-05,
      "loss": 2.6319,
      "step": 3385
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49493810534477234,
      "learning_rate": 9.983342940822849e-05,
      "loss": 2.4953,
      "step": 3390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6002669334411621,
      "learning_rate": 9.983292301228432e-05,
      "loss": 2.6549,
      "step": 3395
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5734832882881165,
      "learning_rate": 9.983241584904197e-05,
      "loss": 2.684,
      "step": 3400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4951913356781006,
      "learning_rate": 9.983190791850925e-05,
      "loss": 2.5379,
      "step": 3405
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4619002342224121,
      "learning_rate": 9.983139922069394e-05,
      "loss": 2.6891,
      "step": 3410
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5143390893936157,
      "learning_rate": 9.983088975560391e-05,
      "loss": 2.7138,
      "step": 3415
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5225774049758911,
      "learning_rate": 9.983037952324699e-05,
      "loss": 2.5836,
      "step": 3420
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49718183279037476,
      "learning_rate": 9.982986852363102e-05,
      "loss": 2.6585,
      "step": 3425
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5280039310455322,
      "learning_rate": 9.98293567567639e-05,
      "loss": 2.6383,
      "step": 3430
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4971632659435272,
      "learning_rate": 9.982884422265349e-05,
      "loss": 2.6651,
      "step": 3435
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.531157374382019,
      "learning_rate": 9.982833092130768e-05,
      "loss": 2.5817,
      "step": 3440
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5042031407356262,
      "learning_rate": 9.982781685273438e-05,
      "loss": 2.5735,
      "step": 3445
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5079023241996765,
      "learning_rate": 9.982730201694153e-05,
      "loss": 2.6192,
      "step": 3450
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5610727071762085,
      "learning_rate": 9.982678641393701e-05,
      "loss": 2.7344,
      "step": 3455
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4902513027191162,
      "learning_rate": 9.98262700437288e-05,
      "loss": 2.6081,
      "step": 3460
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.496755450963974,
      "learning_rate": 9.982575290632481e-05,
      "loss": 2.6675,
      "step": 3465
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4732847809791565,
      "learning_rate": 9.982523500173304e-05,
      "loss": 2.6409,
      "step": 3470
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48057082295417786,
      "learning_rate": 9.982471632996144e-05,
      "loss": 2.6923,
      "step": 3475
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5159416794776917,
      "learning_rate": 9.9824196891018e-05,
      "loss": 2.6338,
      "step": 3480
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48506852984428406,
      "learning_rate": 9.982367668491075e-05,
      "loss": 2.6765,
      "step": 3485
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5037724375724792,
      "learning_rate": 9.982315571164765e-05,
      "loss": 2.5883,
      "step": 3490
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5094142556190491,
      "learning_rate": 9.982263397123675e-05,
      "loss": 2.6223,
      "step": 3495
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.499828040599823,
      "learning_rate": 9.982211146368608e-05,
      "loss": 2.6433,
      "step": 3500
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5333167314529419,
      "learning_rate": 9.982158818900369e-05,
      "loss": 2.6439,
      "step": 3505
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48766425251960754,
      "learning_rate": 9.982106414719762e-05,
      "loss": 2.5852,
      "step": 3510
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47875019907951355,
      "learning_rate": 9.982053933827595e-05,
      "loss": 2.7333,
      "step": 3515
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49952203035354614,
      "learning_rate": 9.982001376224677e-05,
      "loss": 2.7494,
      "step": 3520
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4896669387817383,
      "learning_rate": 9.981948741911816e-05,
      "loss": 2.711,
      "step": 3525
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4818333685398102,
      "learning_rate": 9.981896030889821e-05,
      "loss": 2.6949,
      "step": 3530
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5047662854194641,
      "learning_rate": 9.981843243159507e-05,
      "loss": 2.5664,
      "step": 3535
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48330235481262207,
      "learning_rate": 9.981790378721684e-05,
      "loss": 2.6685,
      "step": 3540
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4814155697822571,
      "learning_rate": 9.98173743757717e-05,
      "loss": 2.5899,
      "step": 3545
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47834208607673645,
      "learning_rate": 9.981684419726775e-05,
      "loss": 2.7353,
      "step": 3550
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5021136403083801,
      "learning_rate": 9.981631325171318e-05,
      "loss": 2.6149,
      "step": 3555
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5237367749214172,
      "learning_rate": 9.981578153911616e-05,
      "loss": 2.5783,
      "step": 3560
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5064594149589539,
      "learning_rate": 9.981524905948487e-05,
      "loss": 2.6868,
      "step": 3565
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48829689621925354,
      "learning_rate": 9.981471581282753e-05,
      "loss": 2.6666,
      "step": 3570
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5007501244544983,
      "learning_rate": 9.981418179915234e-05,
      "loss": 2.5823,
      "step": 3575
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4974224269390106,
      "learning_rate": 9.981364701846752e-05,
      "loss": 2.6331,
      "step": 3580
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49039095640182495,
      "learning_rate": 9.981311147078128e-05,
      "loss": 2.5939,
      "step": 3585
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5201197862625122,
      "learning_rate": 9.981257515610192e-05,
      "loss": 2.5924,
      "step": 3590
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.451112300157547,
      "learning_rate": 9.981203807443764e-05,
      "loss": 2.6238,
      "step": 3595
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49229520559310913,
      "learning_rate": 9.981150022579676e-05,
      "loss": 2.6513,
      "step": 3600
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49071210622787476,
      "learning_rate": 9.981096161018754e-05,
      "loss": 2.6573,
      "step": 3605
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49310067296028137,
      "learning_rate": 9.981042222761826e-05,
      "loss": 2.5975,
      "step": 3610
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4742829203605652,
      "learning_rate": 9.980988207809726e-05,
      "loss": 2.6434,
      "step": 3615
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5171051025390625,
      "learning_rate": 9.98093411616328e-05,
      "loss": 2.5883,
      "step": 3620
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5007846355438232,
      "learning_rate": 9.980879947823327e-05,
      "loss": 2.6359,
      "step": 3625
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49619126319885254,
      "learning_rate": 9.980825702790696e-05,
      "loss": 2.5719,
      "step": 3630
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4895338714122772,
      "learning_rate": 9.980771381066227e-05,
      "loss": 2.6439,
      "step": 3635
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4817723035812378,
      "learning_rate": 9.980716982650753e-05,
      "loss": 2.6251,
      "step": 3640
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.474238783121109,
      "learning_rate": 9.980662507545113e-05,
      "loss": 2.613,
      "step": 3645
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5235281586647034,
      "learning_rate": 9.980607955750146e-05,
      "loss": 2.6449,
      "step": 3650
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4708310067653656,
      "learning_rate": 9.98055332726669e-05,
      "loss": 2.5951,
      "step": 3655
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4771590232849121,
      "learning_rate": 9.980498622095588e-05,
      "loss": 2.7138,
      "step": 3660
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48100796341896057,
      "learning_rate": 9.980443840237681e-05,
      "loss": 2.593,
      "step": 3665
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5168555974960327,
      "learning_rate": 9.980388981693816e-05,
      "loss": 2.6732,
      "step": 3670
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47479844093322754,
      "learning_rate": 9.980334046464832e-05,
      "loss": 2.5961,
      "step": 3675
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4561765491962433,
      "learning_rate": 9.980279034551579e-05,
      "loss": 2.6482,
      "step": 3680
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5105358362197876,
      "learning_rate": 9.980223945954902e-05,
      "loss": 2.6596,
      "step": 3685
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4837929904460907,
      "learning_rate": 9.980168780675651e-05,
      "loss": 2.6389,
      "step": 3690
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4949181079864502,
      "learning_rate": 9.980113538714675e-05,
      "loss": 2.5648,
      "step": 3695
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4795752465724945,
      "learning_rate": 9.980058220072825e-05,
      "loss": 2.6128,
      "step": 3700
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4581420123577118,
      "learning_rate": 9.98000282475095e-05,
      "loss": 2.5719,
      "step": 3705
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4843868017196655,
      "learning_rate": 9.979947352749906e-05,
      "loss": 2.6884,
      "step": 3710
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49813541769981384,
      "learning_rate": 9.979891804070545e-05,
      "loss": 2.5402,
      "step": 3715
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4655483365058899,
      "learning_rate": 9.979836178713724e-05,
      "loss": 2.7318,
      "step": 3720
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5929327011108398,
      "learning_rate": 9.979780476680298e-05,
      "loss": 2.663,
      "step": 3725
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49324893951416016,
      "learning_rate": 9.979724697971125e-05,
      "loss": 2.7141,
      "step": 3730
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5129790306091309,
      "learning_rate": 9.979668842587065e-05,
      "loss": 2.6824,
      "step": 3735
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4779525101184845,
      "learning_rate": 9.979612910528976e-05,
      "loss": 2.6675,
      "step": 3740
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5094065070152283,
      "learning_rate": 9.979556901797723e-05,
      "loss": 2.6056,
      "step": 3745
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4714033305644989,
      "learning_rate": 9.979500816394165e-05,
      "loss": 2.6331,
      "step": 3750
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47872960567474365,
      "learning_rate": 9.979444654319165e-05,
      "loss": 2.6289,
      "step": 3755
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4854588210582733,
      "learning_rate": 9.979388415573591e-05,
      "loss": 2.5073,
      "step": 3760
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5043526887893677,
      "learning_rate": 9.979332100158307e-05,
      "loss": 2.697,
      "step": 3765
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47291815280914307,
      "learning_rate": 9.979275708074179e-05,
      "loss": 2.6535,
      "step": 3770
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47586941719055176,
      "learning_rate": 9.979219239322078e-05,
      "loss": 2.6713,
      "step": 3775
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49423152208328247,
      "learning_rate": 9.979162693902871e-05,
      "loss": 2.6115,
      "step": 3780
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4946158230304718,
      "learning_rate": 9.97910607181743e-05,
      "loss": 2.6793,
      "step": 3785
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5269312262535095,
      "learning_rate": 9.979049373066626e-05,
      "loss": 2.6325,
      "step": 3790
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5132317543029785,
      "learning_rate": 9.978992597651334e-05,
      "loss": 2.6793,
      "step": 3795
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5273664593696594,
      "learning_rate": 9.978935745572424e-05,
      "loss": 2.5032,
      "step": 3800
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4658631384372711,
      "learning_rate": 9.978878816830776e-05,
      "loss": 2.578,
      "step": 3805
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4793817102909088,
      "learning_rate": 9.978821811427263e-05,
      "loss": 2.6431,
      "step": 3810
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48498842120170593,
      "learning_rate": 9.978764729362766e-05,
      "loss": 2.6647,
      "step": 3815
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4684382975101471,
      "learning_rate": 9.978707570638162e-05,
      "loss": 2.6551,
      "step": 3820
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47810858488082886,
      "learning_rate": 9.978650335254332e-05,
      "loss": 2.5889,
      "step": 3825
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45369502902030945,
      "learning_rate": 9.978593023212155e-05,
      "loss": 2.6527,
      "step": 3830
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47577059268951416,
      "learning_rate": 9.978535634512515e-05,
      "loss": 2.6373,
      "step": 3835
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4781218469142914,
      "learning_rate": 9.978478169156298e-05,
      "loss": 2.6247,
      "step": 3840
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4626234769821167,
      "learning_rate": 9.978420627144385e-05,
      "loss": 2.5582,
      "step": 3845
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45888105034828186,
      "learning_rate": 9.978363008477665e-05,
      "loss": 2.6591,
      "step": 3850
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5434486269950867,
      "learning_rate": 9.978305313157022e-05,
      "loss": 2.6681,
      "step": 3855
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5075468420982361,
      "learning_rate": 9.978247541183348e-05,
      "loss": 2.6347,
      "step": 3860
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.469722181558609,
      "learning_rate": 9.978189692557529e-05,
      "loss": 2.5285,
      "step": 3865
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4554632604122162,
      "learning_rate": 9.978131767280456e-05,
      "loss": 2.568,
      "step": 3870
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46973398327827454,
      "learning_rate": 9.978073765353026e-05,
      "loss": 2.6907,
      "step": 3875
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4850519597530365,
      "learning_rate": 9.978015686776125e-05,
      "loss": 2.6312,
      "step": 3880
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47379451990127563,
      "learning_rate": 9.977957531550652e-05,
      "loss": 2.5547,
      "step": 3885
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5005493760108948,
      "learning_rate": 9.977899299677502e-05,
      "loss": 2.6441,
      "step": 3890
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48443296551704407,
      "learning_rate": 9.97784099115757e-05,
      "loss": 2.5607,
      "step": 3895
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5234101414680481,
      "learning_rate": 9.977782605991754e-05,
      "loss": 2.6019,
      "step": 3900
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49355748295783997,
      "learning_rate": 9.977724144180953e-05,
      "loss": 2.6662,
      "step": 3905
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.475903183221817,
      "learning_rate": 9.977665605726067e-05,
      "loss": 2.6125,
      "step": 3910
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49247586727142334,
      "learning_rate": 9.977606990628001e-05,
      "loss": 2.6056,
      "step": 3915
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49697259068489075,
      "learning_rate": 9.977548298887651e-05,
      "loss": 2.6317,
      "step": 3920
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4894625246524811,
      "learning_rate": 9.977489530505927e-05,
      "loss": 2.6068,
      "step": 3925
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4799943268299103,
      "learning_rate": 9.97743068548373e-05,
      "loss": 2.5216,
      "step": 3930
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4934372305870056,
      "learning_rate": 9.977371763821966e-05,
      "loss": 2.6251,
      "step": 3935
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.456089049577713,
      "learning_rate": 9.977312765521545e-05,
      "loss": 2.6642,
      "step": 3940
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.43914005160331726,
      "learning_rate": 9.977253690583372e-05,
      "loss": 2.6076,
      "step": 3945
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4740755259990692,
      "learning_rate": 9.97719453900836e-05,
      "loss": 2.6524,
      "step": 3950
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4478985369205475,
      "learning_rate": 9.977135310797418e-05,
      "loss": 2.5523,
      "step": 3955
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4642377197742462,
      "learning_rate": 9.977076005951458e-05,
      "loss": 2.5893,
      "step": 3960
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.511378824710846,
      "learning_rate": 9.977016624471392e-05,
      "loss": 2.6728,
      "step": 3965
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5151898264884949,
      "learning_rate": 9.976957166358137e-05,
      "loss": 2.6865,
      "step": 3970
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4587767422199249,
      "learning_rate": 9.976897631612607e-05,
      "loss": 2.5667,
      "step": 3975
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4739585518836975,
      "learning_rate": 9.97683802023572e-05,
      "loss": 2.5529,
      "step": 3980
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4672941267490387,
      "learning_rate": 9.976778332228389e-05,
      "loss": 2.6175,
      "step": 3985
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4913594424724579,
      "learning_rate": 9.97671856759154e-05,
      "loss": 2.5984,
      "step": 3990
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48126938939094543,
      "learning_rate": 9.976658726326088e-05,
      "loss": 2.5715,
      "step": 3995
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49017399549484253,
      "learning_rate": 9.976598808432957e-05,
      "loss": 2.6564,
      "step": 4000
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5115654468536377,
      "learning_rate": 9.976538813913068e-05,
      "loss": 2.5752,
      "step": 4005
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4924398362636566,
      "learning_rate": 9.976478742767347e-05,
      "loss": 2.575,
      "step": 4010
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4573129117488861,
      "learning_rate": 9.976418594996716e-05,
      "loss": 2.5553,
      "step": 4015
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46202531456947327,
      "learning_rate": 9.976358370602103e-05,
      "loss": 2.5485,
      "step": 4020
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4698021411895752,
      "learning_rate": 9.976298069584438e-05,
      "loss": 2.6654,
      "step": 4025
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4943801164627075,
      "learning_rate": 9.976237691944643e-05,
      "loss": 2.5356,
      "step": 4030
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46160221099853516,
      "learning_rate": 9.976177237683653e-05,
      "loss": 2.4932,
      "step": 4035
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46950432658195496,
      "learning_rate": 9.976116706802395e-05,
      "loss": 2.5692,
      "step": 4040
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47877827286720276,
      "learning_rate": 9.976056099301805e-05,
      "loss": 2.477,
      "step": 4045
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4829055964946747,
      "learning_rate": 9.975995415182812e-05,
      "loss": 2.6717,
      "step": 4050
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48518285155296326,
      "learning_rate": 9.975934654446355e-05,
      "loss": 2.5859,
      "step": 4055
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47944504022598267,
      "learning_rate": 9.975873817093367e-05,
      "loss": 2.757,
      "step": 4060
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4779786169528961,
      "learning_rate": 9.975812903124782e-05,
      "loss": 2.543,
      "step": 4065
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48569634556770325,
      "learning_rate": 9.975751912541544e-05,
      "loss": 2.5283,
      "step": 4070
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4903237819671631,
      "learning_rate": 9.975690845344588e-05,
      "loss": 2.5293,
      "step": 4075
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4669572412967682,
      "learning_rate": 9.975629701534854e-05,
      "loss": 2.5721,
      "step": 4080
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.43018680810928345,
      "learning_rate": 9.975568481113285e-05,
      "loss": 2.6062,
      "step": 4085
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4799908399581909,
      "learning_rate": 9.975507184080823e-05,
      "loss": 2.5437,
      "step": 4090
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44065147638320923,
      "learning_rate": 9.975445810438413e-05,
      "loss": 2.5789,
      "step": 4095
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48151564598083496,
      "learning_rate": 9.975384360186999e-05,
      "loss": 2.5724,
      "step": 4100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5080844759941101,
      "learning_rate": 9.975322833327527e-05,
      "loss": 2.6129,
      "step": 4105
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48089006543159485,
      "learning_rate": 9.975261229860944e-05,
      "loss": 2.6058,
      "step": 4110
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4649292230606079,
      "learning_rate": 9.975199549788199e-05,
      "loss": 2.6061,
      "step": 4115
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5042335391044617,
      "learning_rate": 9.975137793110243e-05,
      "loss": 2.6169,
      "step": 4120
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48404693603515625,
      "learning_rate": 9.975075959828026e-05,
      "loss": 2.5844,
      "step": 4125
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.469649076461792,
      "learning_rate": 9.975014049942499e-05,
      "loss": 2.5999,
      "step": 4130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.43247804045677185,
      "learning_rate": 9.974952063454615e-05,
      "loss": 2.5827,
      "step": 4135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46738842129707336,
      "learning_rate": 9.974890000365332e-05,
      "loss": 2.5354,
      "step": 4140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4725681245326996,
      "learning_rate": 9.9748278606756e-05,
      "loss": 2.6349,
      "step": 4145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47017788887023926,
      "learning_rate": 9.974765644386382e-05,
      "loss": 2.5986,
      "step": 4150
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48136472702026367,
      "learning_rate": 9.974703351498631e-05,
      "loss": 2.6282,
      "step": 4155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46830618381500244,
      "learning_rate": 9.974640982013307e-05,
      "loss": 2.6155,
      "step": 4160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45107367634773254,
      "learning_rate": 9.974578535931374e-05,
      "loss": 2.6954,
      "step": 4165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.43282821774482727,
      "learning_rate": 9.974516013253789e-05,
      "loss": 2.6959,
      "step": 4170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46773773431777954,
      "learning_rate": 9.974453413981517e-05,
      "loss": 2.5461,
      "step": 4175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4431322515010834,
      "learning_rate": 9.974390738115521e-05,
      "loss": 2.6256,
      "step": 4180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46605169773101807,
      "learning_rate": 9.974327985656767e-05,
      "loss": 2.5847,
      "step": 4185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.447367787361145,
      "learning_rate": 9.974265156606221e-05,
      "loss": 2.5217,
      "step": 4190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4886632561683655,
      "learning_rate": 9.974202250964849e-05,
      "loss": 2.6944,
      "step": 4195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46943023800849915,
      "learning_rate": 9.974139268733623e-05,
      "loss": 2.5384,
      "step": 4200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4830293655395508,
      "learning_rate": 9.974076209913508e-05,
      "loss": 2.602,
      "step": 4205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4505039155483246,
      "learning_rate": 9.974013074505477e-05,
      "loss": 2.6145,
      "step": 4210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4503353536128998,
      "learning_rate": 9.973949862510503e-05,
      "loss": 2.5785,
      "step": 4215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4471443295478821,
      "learning_rate": 9.97388657392956e-05,
      "loss": 2.6213,
      "step": 4220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48380884528160095,
      "learning_rate": 9.97382320876362e-05,
      "loss": 2.5912,
      "step": 4225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45759817957878113,
      "learning_rate": 9.973759767013662e-05,
      "loss": 2.5274,
      "step": 4230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48003727197647095,
      "learning_rate": 9.973696248680659e-05,
      "loss": 2.6848,
      "step": 4235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4756498336791992,
      "learning_rate": 9.97363265376559e-05,
      "loss": 2.6387,
      "step": 4240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4764418601989746,
      "learning_rate": 9.973568982269436e-05,
      "loss": 2.618,
      "step": 4245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46902626752853394,
      "learning_rate": 9.973505234193177e-05,
      "loss": 2.6138,
      "step": 4250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4530481994152069,
      "learning_rate": 9.973441409537793e-05,
      "loss": 2.5846,
      "step": 4255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48749569058418274,
      "learning_rate": 9.973377508304269e-05,
      "loss": 2.6662,
      "step": 4260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45778825879096985,
      "learning_rate": 9.973313530493586e-05,
      "loss": 2.6972,
      "step": 4265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47211283445358276,
      "learning_rate": 9.973249476106732e-05,
      "loss": 2.5342,
      "step": 4270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4639548361301422,
      "learning_rate": 9.97318534514469e-05,
      "loss": 2.639,
      "step": 4275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45356884598731995,
      "learning_rate": 9.97312113760845e-05,
      "loss": 2.5823,
      "step": 4280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4618416130542755,
      "learning_rate": 9.973056853499001e-05,
      "loss": 2.6278,
      "step": 4285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44997209310531616,
      "learning_rate": 9.972992492817331e-05,
      "loss": 2.5398,
      "step": 4290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5342674255371094,
      "learning_rate": 9.972928055564432e-05,
      "loss": 2.5547,
      "step": 4295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46561747789382935,
      "learning_rate": 9.972863541741297e-05,
      "loss": 2.6474,
      "step": 4300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47535789012908936,
      "learning_rate": 9.972798951348918e-05,
      "loss": 2.5772,
      "step": 4305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45316630601882935,
      "learning_rate": 9.972734284388288e-05,
      "loss": 2.5786,
      "step": 4310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4569714069366455,
      "learning_rate": 9.972669540860406e-05,
      "loss": 2.6677,
      "step": 4315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4416045844554901,
      "learning_rate": 9.972604720766268e-05,
      "loss": 2.6203,
      "step": 4320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4491463601589203,
      "learning_rate": 9.97253982410687e-05,
      "loss": 2.5425,
      "step": 4325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47782695293426514,
      "learning_rate": 9.972474850883213e-05,
      "loss": 2.6296,
      "step": 4330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4748191237449646,
      "learning_rate": 9.972409801096299e-05,
      "loss": 2.5663,
      "step": 4335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4803968369960785,
      "learning_rate": 9.972344674747125e-05,
      "loss": 2.579,
      "step": 4340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4710930585861206,
      "learning_rate": 9.972279471836698e-05,
      "loss": 2.541,
      "step": 4345
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4304599165916443,
      "learning_rate": 9.972214192366021e-05,
      "loss": 2.5922,
      "step": 4350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4736495614051819,
      "learning_rate": 9.972148836336098e-05,
      "loss": 2.5805,
      "step": 4355
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.43438127636909485,
      "learning_rate": 9.972083403747937e-05,
      "loss": 2.6771,
      "step": 4360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5093146562576294,
      "learning_rate": 9.972017894602542e-05,
      "loss": 2.5476,
      "step": 4365
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47126978635787964,
      "learning_rate": 9.971952308900925e-05,
      "loss": 2.6206,
      "step": 4370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44014111161231995,
      "learning_rate": 9.971886646644096e-05,
      "loss": 2.5823,
      "step": 4375
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4550796449184418,
      "learning_rate": 9.971820907833065e-05,
      "loss": 2.5393,
      "step": 4380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4763948321342468,
      "learning_rate": 9.971755092468843e-05,
      "loss": 2.6761,
      "step": 4385
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5066975355148315,
      "learning_rate": 9.971689200552445e-05,
      "loss": 2.6534,
      "step": 4390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44651156663894653,
      "learning_rate": 9.971623232084886e-05,
      "loss": 2.6679,
      "step": 4395
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45041322708129883,
      "learning_rate": 9.971557187067179e-05,
      "loss": 2.5766,
      "step": 4400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44005414843559265,
      "learning_rate": 9.971491065500345e-05,
      "loss": 2.629,
      "step": 4405
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47684261202812195,
      "learning_rate": 9.9714248673854e-05,
      "loss": 2.5209,
      "step": 4410
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.478615939617157,
      "learning_rate": 9.971358592723361e-05,
      "loss": 2.6056,
      "step": 4415
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44478839635849,
      "learning_rate": 9.97129224151525e-05,
      "loss": 2.5939,
      "step": 4420
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.451492577791214,
      "learning_rate": 9.971225813762094e-05,
      "loss": 2.6849,
      "step": 4425
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5021209120750427,
      "learning_rate": 9.971159309464906e-05,
      "loss": 2.4992,
      "step": 4430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4687957763671875,
      "learning_rate": 9.971092728624718e-05,
      "loss": 2.575,
      "step": 4435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4451132118701935,
      "learning_rate": 9.97102607124255e-05,
      "loss": 2.498,
      "step": 4440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4498390555381775,
      "learning_rate": 9.970959337319433e-05,
      "loss": 2.5256,
      "step": 4445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47291824221611023,
      "learning_rate": 9.970892526856392e-05,
      "loss": 2.6109,
      "step": 4450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44330450892448425,
      "learning_rate": 9.970825639854454e-05,
      "loss": 2.497,
      "step": 4455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4706554412841797,
      "learning_rate": 9.970758676314652e-05,
      "loss": 2.528,
      "step": 4460
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46172359585762024,
      "learning_rate": 9.970691636238017e-05,
      "loss": 2.5172,
      "step": 4465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4847109317779541,
      "learning_rate": 9.970624519625578e-05,
      "loss": 2.6284,
      "step": 4470
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45727211236953735,
      "learning_rate": 9.970557326478371e-05,
      "loss": 2.5652,
      "step": 4475
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46407824754714966,
      "learning_rate": 9.97049005679743e-05,
      "loss": 2.5416,
      "step": 4480
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46202969551086426,
      "learning_rate": 9.970422710583791e-05,
      "loss": 2.5849,
      "step": 4485
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4335814118385315,
      "learning_rate": 9.970355287838491e-05,
      "loss": 2.5125,
      "step": 4490
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44231608510017395,
      "learning_rate": 9.970287788562567e-05,
      "loss": 2.5517,
      "step": 4495
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4590959846973419,
      "learning_rate": 9.97022021275706e-05,
      "loss": 2.5945,
      "step": 4500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4547059237957001,
      "learning_rate": 9.97015256042301e-05,
      "loss": 2.5673,
      "step": 4505
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43108028173446655,
      "learning_rate": 9.97008483156146e-05,
      "loss": 2.5556,
      "step": 4510
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43963751196861267,
      "learning_rate": 9.970017026173448e-05,
      "loss": 2.5059,
      "step": 4515
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4631103277206421,
      "learning_rate": 9.969949144260022e-05,
      "loss": 2.5253,
      "step": 4520
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42648255825042725,
      "learning_rate": 9.969881185822227e-05,
      "loss": 2.6674,
      "step": 4525
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46012362837791443,
      "learning_rate": 9.969813150861111e-05,
      "loss": 2.534,
      "step": 4530
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45968350768089294,
      "learning_rate": 9.969745039377717e-05,
      "loss": 2.6159,
      "step": 4535
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46120530366897583,
      "learning_rate": 9.969676851373098e-05,
      "loss": 2.5867,
      "step": 4540
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46591633558273315,
      "learning_rate": 9.969608586848301e-05,
      "loss": 2.477,
      "step": 4545
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48572370409965515,
      "learning_rate": 9.969540245804377e-05,
      "loss": 2.6431,
      "step": 4550
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4353214204311371,
      "learning_rate": 9.96947182824238e-05,
      "loss": 2.6583,
      "step": 4555
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44570955634117126,
      "learning_rate": 9.969403334163364e-05,
      "loss": 2.4987,
      "step": 4560
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43207380175590515,
      "learning_rate": 9.969334763568383e-05,
      "loss": 2.6108,
      "step": 4565
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47050824761390686,
      "learning_rate": 9.969266116458491e-05,
      "loss": 2.6217,
      "step": 4570
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43929001688957214,
      "learning_rate": 9.969197392834747e-05,
      "loss": 2.5402,
      "step": 4575
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4635504186153412,
      "learning_rate": 9.969128592698207e-05,
      "loss": 2.6524,
      "step": 4580
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4375777542591095,
      "learning_rate": 9.969059716049934e-05,
      "loss": 2.6242,
      "step": 4585
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.437782347202301,
      "learning_rate": 9.968990762890986e-05,
      "loss": 2.5533,
      "step": 4590
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4642016291618347,
      "learning_rate": 9.968921733222425e-05,
      "loss": 2.6043,
      "step": 4595
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4626062512397766,
      "learning_rate": 9.968852627045313e-05,
      "loss": 2.5393,
      "step": 4600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44650930166244507,
      "learning_rate": 9.968783444360716e-05,
      "loss": 2.5475,
      "step": 4605
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43003106117248535,
      "learning_rate": 9.968714185169698e-05,
      "loss": 2.6237,
      "step": 4610
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4509531855583191,
      "learning_rate": 9.968644849473326e-05,
      "loss": 2.5957,
      "step": 4615
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44457072019577026,
      "learning_rate": 9.968575437272667e-05,
      "loss": 2.4696,
      "step": 4620
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44891107082366943,
      "learning_rate": 9.968505948568792e-05,
      "loss": 2.4885,
      "step": 4625
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4464776813983917,
      "learning_rate": 9.968436383362766e-05,
      "loss": 2.6104,
      "step": 4630
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4448380768299103,
      "learning_rate": 9.968366741655663e-05,
      "loss": 2.5229,
      "step": 4635
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45508697628974915,
      "learning_rate": 9.968297023448558e-05,
      "loss": 2.5234,
      "step": 4640
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4814097285270691,
      "learning_rate": 9.96822722874252e-05,
      "loss": 2.4993,
      "step": 4645
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4739972651004791,
      "learning_rate": 9.968157357538625e-05,
      "loss": 2.6228,
      "step": 4650
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44550636410713196,
      "learning_rate": 9.968087409837951e-05,
      "loss": 2.613,
      "step": 4655
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44849979877471924,
      "learning_rate": 9.968017385641574e-05,
      "loss": 2.5296,
      "step": 4660
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5624348521232605,
      "learning_rate": 9.967947284950569e-05,
      "loss": 2.5324,
      "step": 4665
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4593139886856079,
      "learning_rate": 9.967877107766021e-05,
      "loss": 2.5204,
      "step": 4670
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4467976689338684,
      "learning_rate": 9.967806854089006e-05,
      "loss": 2.5745,
      "step": 4675
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4501049220561981,
      "learning_rate": 9.967736523920607e-05,
      "loss": 2.503,
      "step": 4680
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4531939625740051,
      "learning_rate": 9.967666117261907e-05,
      "loss": 2.5296,
      "step": 4685
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4667191505432129,
      "learning_rate": 9.96759563411399e-05,
      "loss": 2.4684,
      "step": 4690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4281299412250519,
      "learning_rate": 9.967525074477943e-05,
      "loss": 2.4717,
      "step": 4695
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46667036414146423,
      "learning_rate": 9.96745443835485e-05,
      "loss": 2.529,
      "step": 4700
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46279165148735046,
      "learning_rate": 9.967383725745799e-05,
      "loss": 2.5195,
      "step": 4705
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4385567009449005,
      "learning_rate": 9.96731293665188e-05,
      "loss": 2.4803,
      "step": 4710
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.455239862203598,
      "learning_rate": 9.967242071074181e-05,
      "loss": 2.5206,
      "step": 4715
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43868812918663025,
      "learning_rate": 9.967171129013797e-05,
      "loss": 2.5314,
      "step": 4720
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4491572678089142,
      "learning_rate": 9.967100110471817e-05,
      "loss": 2.5726,
      "step": 4725
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44071635603904724,
      "learning_rate": 9.967029015449334e-05,
      "loss": 2.5378,
      "step": 4730
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.443033903837204,
      "learning_rate": 9.966957843947444e-05,
      "loss": 2.5152,
      "step": 4735
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41849708557128906,
      "learning_rate": 9.966886595967243e-05,
      "loss": 2.5727,
      "step": 4740
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4407452642917633,
      "learning_rate": 9.966815271509828e-05,
      "loss": 2.5078,
      "step": 4745
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4368855357170105,
      "learning_rate": 9.966743870576297e-05,
      "loss": 2.495,
      "step": 4750
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43668389320373535,
      "learning_rate": 9.96667239316775e-05,
      "loss": 2.5355,
      "step": 4755
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4357081949710846,
      "learning_rate": 9.966600839285286e-05,
      "loss": 2.5682,
      "step": 4760
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4498099982738495,
      "learning_rate": 9.966529208930008e-05,
      "loss": 2.4986,
      "step": 4765
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4471588730812073,
      "learning_rate": 9.96645750210302e-05,
      "loss": 2.5376,
      "step": 4770
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47415876388549805,
      "learning_rate": 9.966385718805423e-05,
      "loss": 2.5498,
      "step": 4775
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43391546607017517,
      "learning_rate": 9.966313859038325e-05,
      "loss": 2.5408,
      "step": 4780
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.510086715221405,
      "learning_rate": 9.966241922802832e-05,
      "loss": 2.5398,
      "step": 4785
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4366661012172699,
      "learning_rate": 9.96616991010005e-05,
      "loss": 2.4871,
      "step": 4790
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45814430713653564,
      "learning_rate": 9.96609782093109e-05,
      "loss": 2.5386,
      "step": 4795
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.451948344707489,
      "learning_rate": 9.96602565529706e-05,
      "loss": 2.5776,
      "step": 4800
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43737974762916565,
      "learning_rate": 9.965953413199071e-05,
      "loss": 2.569,
      "step": 4805
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4532112181186676,
      "learning_rate": 9.965881094638238e-05,
      "loss": 2.551,
      "step": 4810
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43967023491859436,
      "learning_rate": 9.965808699615674e-05,
      "loss": 2.6556,
      "step": 4815
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4336446225643158,
      "learning_rate": 9.965736228132492e-05,
      "loss": 2.583,
      "step": 4820
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42914679646492004,
      "learning_rate": 9.965663680189807e-05,
      "loss": 2.5372,
      "step": 4825
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4583548605442047,
      "learning_rate": 9.965591055788738e-05,
      "loss": 2.5626,
      "step": 4830
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46772143244743347,
      "learning_rate": 9.965518354930405e-05,
      "loss": 2.6046,
      "step": 4835
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45584800839424133,
      "learning_rate": 9.965445577615922e-05,
      "loss": 2.5033,
      "step": 4840
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4444721043109894,
      "learning_rate": 9.965372723846415e-05,
      "loss": 2.4902,
      "step": 4845
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44349899888038635,
      "learning_rate": 9.965299793623003e-05,
      "loss": 2.5471,
      "step": 4850
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4365882873535156,
      "learning_rate": 9.96522678694681e-05,
      "loss": 2.5466,
      "step": 4855
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4728853702545166,
      "learning_rate": 9.965153703818959e-05,
      "loss": 2.6155,
      "step": 4860
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4355456233024597,
      "learning_rate": 9.965080544240575e-05,
      "loss": 2.5652,
      "step": 4865
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44645026326179504,
      "learning_rate": 9.965007308212786e-05,
      "loss": 2.5727,
      "step": 4870
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4526258409023285,
      "learning_rate": 9.964933995736721e-05,
      "loss": 2.4737,
      "step": 4875
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4396340548992157,
      "learning_rate": 9.964860606813504e-05,
      "loss": 2.6569,
      "step": 4880
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4376806616783142,
      "learning_rate": 9.96478714144427e-05,
      "loss": 2.489,
      "step": 4885
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4288381338119507,
      "learning_rate": 9.964713599630146e-05,
      "loss": 2.4526,
      "step": 4890
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.454735666513443,
      "learning_rate": 9.964639981372267e-05,
      "loss": 2.4526,
      "step": 4895
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43699678778648376,
      "learning_rate": 9.964566286671766e-05,
      "loss": 2.557,
      "step": 4900
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44667163491249084,
      "learning_rate": 9.964492515529777e-05,
      "loss": 2.5501,
      "step": 4905
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4327116310596466,
      "learning_rate": 9.964418667947438e-05,
      "loss": 2.463,
      "step": 4910
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47008413076400757,
      "learning_rate": 9.964344743925884e-05,
      "loss": 2.5935,
      "step": 4915
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4614291489124298,
      "learning_rate": 9.964270743466253e-05,
      "loss": 2.5838,
      "step": 4920
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48439037799835205,
      "learning_rate": 9.964196666569686e-05,
      "loss": 2.5857,
      "step": 4925
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44462206959724426,
      "learning_rate": 9.964122513237322e-05,
      "loss": 2.6719,
      "step": 4930
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43672817945480347,
      "learning_rate": 9.964048283470305e-05,
      "loss": 2.4652,
      "step": 4935
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4677668809890747,
      "learning_rate": 9.963973977269776e-05,
      "loss": 2.5397,
      "step": 4940
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4408162236213684,
      "learning_rate": 9.963899594636879e-05,
      "loss": 2.5196,
      "step": 4945
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4364597797393799,
      "learning_rate": 9.96382513557276e-05,
      "loss": 2.4694,
      "step": 4950
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4129132032394409,
      "learning_rate": 9.963750600078565e-05,
      "loss": 2.4727,
      "step": 4955
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43847817182540894,
      "learning_rate": 9.963675988155443e-05,
      "loss": 2.5986,
      "step": 4960
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.449704110622406,
      "learning_rate": 9.963601299804541e-05,
      "loss": 2.5175,
      "step": 4965
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4559851586818695,
      "learning_rate": 9.963526535027012e-05,
      "loss": 2.5108,
      "step": 4970
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4575880765914917,
      "learning_rate": 9.963451693824003e-05,
      "loss": 2.6154,
      "step": 4975
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4101316034793854,
      "learning_rate": 9.963376776196669e-05,
      "loss": 2.4985,
      "step": 4980
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41284433007240295,
      "learning_rate": 9.963301782146163e-05,
      "loss": 2.5158,
      "step": 4985
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48281750082969666,
      "learning_rate": 9.96322671167364e-05,
      "loss": 2.5124,
      "step": 4990
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45164093375205994,
      "learning_rate": 9.963151564780256e-05,
      "loss": 2.5492,
      "step": 4995
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.455078661441803,
      "learning_rate": 9.963076341467165e-05,
      "loss": 2.4855,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4493098855018616,
      "learning_rate": 9.963001041735531e-05,
      "loss": 2.5465,
      "step": 5005
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4205504059791565,
      "learning_rate": 9.962925665586509e-05,
      "loss": 2.6066,
      "step": 5010
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42429250478744507,
      "learning_rate": 9.962850213021262e-05,
      "loss": 2.6035,
      "step": 5015
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4484236240386963,
      "learning_rate": 9.96277468404095e-05,
      "loss": 2.449,
      "step": 5020
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43770748376846313,
      "learning_rate": 9.962699078646735e-05,
      "loss": 2.5405,
      "step": 5025
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42430028319358826,
      "learning_rate": 9.962623396839785e-05,
      "loss": 2.3869,
      "step": 5030
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4589504897594452,
      "learning_rate": 9.962547638621262e-05,
      "loss": 2.5442,
      "step": 5035
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4459128677845001,
      "learning_rate": 9.962471803992334e-05,
      "loss": 2.5284,
      "step": 5040
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4252271056175232,
      "learning_rate": 9.962395892954168e-05,
      "loss": 2.5589,
      "step": 5045
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4504375457763672,
      "learning_rate": 9.962319905507934e-05,
      "loss": 2.5514,
      "step": 5050
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42302483320236206,
      "learning_rate": 9.9622438416548e-05,
      "loss": 2.557,
      "step": 5055
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4391854405403137,
      "learning_rate": 9.962167701395937e-05,
      "loss": 2.5549,
      "step": 5060
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47206124663352966,
      "learning_rate": 9.962091484732521e-05,
      "loss": 2.6016,
      "step": 5065
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4422343075275421,
      "learning_rate": 9.962015191665722e-05,
      "loss": 2.6199,
      "step": 5070
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44093284010887146,
      "learning_rate": 9.961938822196716e-05,
      "loss": 2.4781,
      "step": 5075
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4397658109664917,
      "learning_rate": 9.961862376326678e-05,
      "loss": 2.6339,
      "step": 5080
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5068472027778625,
      "learning_rate": 9.961785854056788e-05,
      "loss": 2.6363,
      "step": 5085
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43679526448249817,
      "learning_rate": 9.961709255388221e-05,
      "loss": 2.5617,
      "step": 5090
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4748583436012268,
      "learning_rate": 9.961632580322156e-05,
      "loss": 2.618,
      "step": 5095
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47490161657333374,
      "learning_rate": 9.961555828859777e-05,
      "loss": 2.6415,
      "step": 5100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43888208270072937,
      "learning_rate": 9.961479001002264e-05,
      "loss": 2.583,
      "step": 5105
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.414302259683609,
      "learning_rate": 9.961402096750799e-05,
      "loss": 2.4053,
      "step": 5110
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43851080536842346,
      "learning_rate": 9.961325116106566e-05,
      "loss": 2.5356,
      "step": 5115
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44866058230400085,
      "learning_rate": 9.961248059070751e-05,
      "loss": 2.5612,
      "step": 5120
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.432462215423584,
      "learning_rate": 9.961170925644542e-05,
      "loss": 2.4849,
      "step": 5125
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46267688274383545,
      "learning_rate": 9.961093715829127e-05,
      "loss": 2.6004,
      "step": 5130
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4562717378139496,
      "learning_rate": 9.961016429625691e-05,
      "loss": 2.5889,
      "step": 5135
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4471887946128845,
      "learning_rate": 9.960939067035426e-05,
      "loss": 2.4824,
      "step": 5140
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46789994835853577,
      "learning_rate": 9.960861628059524e-05,
      "loss": 2.4364,
      "step": 5145
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4580118656158447,
      "learning_rate": 9.960784112699177e-05,
      "loss": 2.5037,
      "step": 5150
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.455715537071228,
      "learning_rate": 9.960706520955578e-05,
      "loss": 2.5148,
      "step": 5155
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4822584390640259,
      "learning_rate": 9.960628852829923e-05,
      "loss": 2.5339,
      "step": 5160
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42242109775543213,
      "learning_rate": 9.960551108323406e-05,
      "loss": 2.5022,
      "step": 5165
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4775131344795227,
      "learning_rate": 9.960473287437224e-05,
      "loss": 2.4575,
      "step": 5170
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4330013692378998,
      "learning_rate": 9.960395390172579e-05,
      "loss": 2.4646,
      "step": 5175
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4541088342666626,
      "learning_rate": 9.960317416530665e-05,
      "loss": 2.6381,
      "step": 5180
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4417051374912262,
      "learning_rate": 9.960239366512686e-05,
      "loss": 2.4165,
      "step": 5185
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4233478903770447,
      "learning_rate": 9.960161240119843e-05,
      "loss": 2.4925,
      "step": 5190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42705217003822327,
      "learning_rate": 9.96008303735334e-05,
      "loss": 2.5304,
      "step": 5195
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43817585706710815,
      "learning_rate": 9.960004758214379e-05,
      "loss": 2.5145,
      "step": 5200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4757373034954071,
      "learning_rate": 9.959926402704165e-05,
      "loss": 2.6099,
      "step": 5205
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4676450490951538,
      "learning_rate": 9.959847970823908e-05,
      "loss": 2.6378,
      "step": 5210
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4203851819038391,
      "learning_rate": 9.959769462574813e-05,
      "loss": 2.5508,
      "step": 5215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4587145447731018,
      "learning_rate": 9.959690877958088e-05,
      "loss": 2.4252,
      "step": 5220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4631746709346771,
      "learning_rate": 9.959612216974946e-05,
      "loss": 2.6348,
      "step": 5225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.451335072517395,
      "learning_rate": 9.959533479626596e-05,
      "loss": 2.5148,
      "step": 5230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41803300380706787,
      "learning_rate": 9.95945466591425e-05,
      "loss": 2.4702,
      "step": 5235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4509798586368561,
      "learning_rate": 9.959375775839124e-05,
      "loss": 2.4572,
      "step": 5240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.430500328540802,
      "learning_rate": 9.95929680940243e-05,
      "loss": 2.5784,
      "step": 5245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43727821111679077,
      "learning_rate": 9.959217766605384e-05,
      "loss": 2.4859,
      "step": 5250
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43553853034973145,
      "learning_rate": 9.959138647449205e-05,
      "loss": 2.4885,
      "step": 5255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4576265215873718,
      "learning_rate": 9.95905945193511e-05,
      "loss": 2.5775,
      "step": 5260
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4429559111595154,
      "learning_rate": 9.958980180064318e-05,
      "loss": 2.4148,
      "step": 5265
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.39115047454833984,
      "learning_rate": 9.958900831838051e-05,
      "loss": 2.3913,
      "step": 5270
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46966734528541565,
      "learning_rate": 9.958821407257531e-05,
      "loss": 2.5523,
      "step": 5275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5082553625106812,
      "learning_rate": 9.958741906323978e-05,
      "loss": 2.5491,
      "step": 5280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43193185329437256,
      "learning_rate": 9.95866232903862e-05,
      "loss": 2.4209,
      "step": 5285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4479002356529236,
      "learning_rate": 9.958582675402677e-05,
      "loss": 2.5524,
      "step": 5290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4230886995792389,
      "learning_rate": 9.958502945417382e-05,
      "loss": 2.5588,
      "step": 5295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4087367653846741,
      "learning_rate": 9.958423139083958e-05,
      "loss": 2.5101,
      "step": 5300
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41850173473358154,
      "learning_rate": 9.958343256403635e-05,
      "loss": 2.5179,
      "step": 5305
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4485442042350769,
      "learning_rate": 9.958263297377643e-05,
      "loss": 2.4583,
      "step": 5310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43585070967674255,
      "learning_rate": 9.958183262007215e-05,
      "loss": 2.4991,
      "step": 5315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41171038150787354,
      "learning_rate": 9.95810315029358e-05,
      "loss": 2.5638,
      "step": 5320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4388819932937622,
      "learning_rate": 9.958022962237974e-05,
      "loss": 2.5274,
      "step": 5325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42040616273880005,
      "learning_rate": 9.957942697841632e-05,
      "loss": 2.584,
      "step": 5330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43130549788475037,
      "learning_rate": 9.957862357105785e-05,
      "loss": 2.4085,
      "step": 5335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42507386207580566,
      "learning_rate": 9.957781940031676e-05,
      "loss": 2.49,
      "step": 5340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42620331048965454,
      "learning_rate": 9.957701446620543e-05,
      "loss": 2.6375,
      "step": 5345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4413948953151703,
      "learning_rate": 9.957620876873621e-05,
      "loss": 2.5472,
      "step": 5350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43932124972343445,
      "learning_rate": 9.957540230792152e-05,
      "loss": 2.4953,
      "step": 5355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4090133011341095,
      "learning_rate": 9.957459508377381e-05,
      "loss": 2.4882,
      "step": 5360
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41920241713523865,
      "learning_rate": 9.957378709630548e-05,
      "loss": 2.5208,
      "step": 5365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4339069724082947,
      "learning_rate": 9.957297834552898e-05,
      "loss": 2.5839,
      "step": 5370
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4434177279472351,
      "learning_rate": 9.957216883145675e-05,
      "loss": 2.5564,
      "step": 5375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4229823052883148,
      "learning_rate": 9.957135855410127e-05,
      "loss": 2.4761,
      "step": 5380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43733927607536316,
      "learning_rate": 9.957054751347501e-05,
      "loss": 2.4847,
      "step": 5385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.415409654378891,
      "learning_rate": 9.956973570959045e-05,
      "loss": 2.6518,
      "step": 5390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4181080460548401,
      "learning_rate": 9.956892314246012e-05,
      "loss": 2.4936,
      "step": 5395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44642311334609985,
      "learning_rate": 9.956810981209647e-05,
      "loss": 2.549,
      "step": 5400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4356335699558258,
      "learning_rate": 9.956729571851211e-05,
      "loss": 2.5454,
      "step": 5405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.438887357711792,
      "learning_rate": 9.956648086171949e-05,
      "loss": 2.5344,
      "step": 5410
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4093167781829834,
      "learning_rate": 9.956566524173121e-05,
      "loss": 2.528,
      "step": 5415
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42105576395988464,
      "learning_rate": 9.956484885855982e-05,
      "loss": 2.4986,
      "step": 5420
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42270317673683167,
      "learning_rate": 9.956403171221786e-05,
      "loss": 2.4976,
      "step": 5425
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4256037771701813,
      "learning_rate": 9.956321380271796e-05,
      "loss": 2.5724,
      "step": 5430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4129668176174164,
      "learning_rate": 9.956239513007268e-05,
      "loss": 2.5627,
      "step": 5435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42029595375061035,
      "learning_rate": 9.956157569429464e-05,
      "loss": 2.5848,
      "step": 5440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42088839411735535,
      "learning_rate": 9.956075549539643e-05,
      "loss": 2.4287,
      "step": 5445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42138537764549255,
      "learning_rate": 9.955993453339072e-05,
      "loss": 2.542,
      "step": 5450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4296362102031708,
      "learning_rate": 9.955911280829011e-05,
      "loss": 2.4648,
      "step": 5455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47869762778282166,
      "learning_rate": 9.95582903201073e-05,
      "loss": 2.497,
      "step": 5460
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4281044900417328,
      "learning_rate": 9.955746706885492e-05,
      "loss": 2.4977,
      "step": 5465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45564088225364685,
      "learning_rate": 9.955664305454564e-05,
      "loss": 2.5451,
      "step": 5470
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4287124574184418,
      "learning_rate": 9.955581827719216e-05,
      "loss": 2.4588,
      "step": 5475
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41181036829948425,
      "learning_rate": 9.955499273680719e-05,
      "loss": 2.4857,
      "step": 5480
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4273780584335327,
      "learning_rate": 9.955416643340343e-05,
      "loss": 2.5489,
      "step": 5485
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44192609190940857,
      "learning_rate": 9.95533393669936e-05,
      "loss": 2.5777,
      "step": 5490
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4140126705169678,
      "learning_rate": 9.955251153759044e-05,
      "loss": 2.448,
      "step": 5495
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4196762144565582,
      "learning_rate": 9.955168294520672e-05,
      "loss": 2.5404,
      "step": 5500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4200664460659027,
      "learning_rate": 9.955085358985514e-05,
      "loss": 2.4624,
      "step": 5505
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42223796248435974,
      "learning_rate": 9.955002347154851e-05,
      "loss": 2.4939,
      "step": 5510
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4249913990497589,
      "learning_rate": 9.954919259029962e-05,
      "loss": 2.521,
      "step": 5515
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41143882274627686,
      "learning_rate": 9.954836094612123e-05,
      "loss": 2.5243,
      "step": 5520
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41552168130874634,
      "learning_rate": 9.954752853902619e-05,
      "loss": 2.4583,
      "step": 5525
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4282611310482025,
      "learning_rate": 9.954669536902728e-05,
      "loss": 2.5578,
      "step": 5530
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4456869661808014,
      "learning_rate": 9.954586143613733e-05,
      "loss": 2.5506,
      "step": 5535
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41114768385887146,
      "learning_rate": 9.954502674036919e-05,
      "loss": 2.4921,
      "step": 5540
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45204275846481323,
      "learning_rate": 9.954419128173572e-05,
      "loss": 2.5328,
      "step": 5545
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44963258504867554,
      "learning_rate": 9.954335506024978e-05,
      "loss": 2.4121,
      "step": 5550
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4164026081562042,
      "learning_rate": 9.954251807592423e-05,
      "loss": 2.5536,
      "step": 5555
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41360342502593994,
      "learning_rate": 9.954168032877198e-05,
      "loss": 2.5751,
      "step": 5560
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4418202340602875,
      "learning_rate": 9.95408418188059e-05,
      "loss": 2.5004,
      "step": 5565
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44671615958213806,
      "learning_rate": 9.954000254603894e-05,
      "loss": 2.6019,
      "step": 5570
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42313751578330994,
      "learning_rate": 9.9539162510484e-05,
      "loss": 2.5832,
      "step": 5575
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4158114492893219,
      "learning_rate": 9.953832171215401e-05,
      "loss": 2.4017,
      "step": 5580
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42053520679473877,
      "learning_rate": 9.953748015106193e-05,
      "loss": 2.4835,
      "step": 5585
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4337523281574249,
      "learning_rate": 9.953663782722069e-05,
      "loss": 2.5245,
      "step": 5590
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44196808338165283,
      "learning_rate": 9.953579474064329e-05,
      "loss": 2.5238,
      "step": 5595
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4276723563671112,
      "learning_rate": 9.95349508913427e-05,
      "loss": 2.5784,
      "step": 5600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.417923241853714,
      "learning_rate": 9.953410627933193e-05,
      "loss": 2.555,
      "step": 5605
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43664753437042236,
      "learning_rate": 9.953326090462397e-05,
      "loss": 2.525,
      "step": 5610
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41672369837760925,
      "learning_rate": 9.953241476723181e-05,
      "loss": 2.5353,
      "step": 5615
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4442696273326874,
      "learning_rate": 9.953156786716851e-05,
      "loss": 2.5486,
      "step": 5620
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.40515774488449097,
      "learning_rate": 9.953072020444711e-05,
      "loss": 2.4949,
      "step": 5625
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4318718910217285,
      "learning_rate": 9.952987177908066e-05,
      "loss": 2.4858,
      "step": 5630
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41956329345703125,
      "learning_rate": 9.95290225910822e-05,
      "loss": 2.593,
      "step": 5635
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4449191689491272,
      "learning_rate": 9.952817264046486e-05,
      "loss": 2.5665,
      "step": 5640
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46397972106933594,
      "learning_rate": 9.952732192724167e-05,
      "loss": 2.5151,
      "step": 5645
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4269982576370239,
      "learning_rate": 9.952647045142576e-05,
      "loss": 2.5527,
      "step": 5650
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41480764746665955,
      "learning_rate": 9.952561821303021e-05,
      "loss": 2.4777,
      "step": 5655
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43861159682273865,
      "learning_rate": 9.952476521206818e-05,
      "loss": 2.4412,
      "step": 5660
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4196626842021942,
      "learning_rate": 9.952391144855278e-05,
      "loss": 2.4684,
      "step": 5665
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5045226216316223,
      "learning_rate": 9.952305692249717e-05,
      "loss": 2.5188,
      "step": 5670
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4083325266838074,
      "learning_rate": 9.952220163391451e-05,
      "loss": 2.4927,
      "step": 5675
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4224720001220703,
      "learning_rate": 9.952134558281793e-05,
      "loss": 2.5608,
      "step": 5680
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.41653189063072205,
      "learning_rate": 9.952048876922066e-05,
      "loss": 2.5345,
      "step": 5685
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4333929419517517,
      "learning_rate": 9.951963119313588e-05,
      "loss": 2.607,
      "step": 5690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43696242570877075,
      "learning_rate": 9.951877285457678e-05,
      "loss": 2.4009,
      "step": 5695
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4371735453605652,
      "learning_rate": 9.951791375355659e-05,
      "loss": 2.4876,
      "step": 5700
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4209734797477722,
      "learning_rate": 9.951705389008852e-05,
      "loss": 2.5155,
      "step": 5705
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42070284485816956,
      "learning_rate": 9.951619326418582e-05,
      "loss": 2.5161,
      "step": 5710
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46585139632225037,
      "learning_rate": 9.951533187586178e-05,
      "loss": 2.5702,
      "step": 5715
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41560661792755127,
      "learning_rate": 9.951446972512958e-05,
      "loss": 2.5576,
      "step": 5720
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44494864344596863,
      "learning_rate": 9.951360681200256e-05,
      "loss": 2.508,
      "step": 5725
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4059827923774719,
      "learning_rate": 9.9512743136494e-05,
      "loss": 2.5144,
      "step": 5730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43018990755081177,
      "learning_rate": 9.951187869861715e-05,
      "loss": 2.5701,
      "step": 5735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4327108561992645,
      "learning_rate": 9.951101349838538e-05,
      "loss": 2.5283,
      "step": 5740
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3157150745391846,
      "learning_rate": 9.9510147535812e-05,
      "loss": 2.4851,
      "step": 5745
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44111132621765137,
      "learning_rate": 9.950928081091032e-05,
      "loss": 2.5226,
      "step": 5750
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4190981984138489,
      "learning_rate": 9.95084133236937e-05,
      "loss": 2.4622,
      "step": 5755
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44815823435783386,
      "learning_rate": 9.950754507417547e-05,
      "loss": 2.5215,
      "step": 5760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4043428599834442,
      "learning_rate": 9.950667606236906e-05,
      "loss": 2.4835,
      "step": 5765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4303000569343567,
      "learning_rate": 9.95058062882878e-05,
      "loss": 2.4663,
      "step": 5770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43617507815361023,
      "learning_rate": 9.950493575194509e-05,
      "loss": 2.4667,
      "step": 5775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44052234292030334,
      "learning_rate": 9.950406445335434e-05,
      "loss": 2.5739,
      "step": 5780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4307103455066681,
      "learning_rate": 9.950319239252896e-05,
      "loss": 2.3969,
      "step": 5785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45380836725234985,
      "learning_rate": 9.95023195694824e-05,
      "loss": 2.4671,
      "step": 5790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4191444218158722,
      "learning_rate": 9.950144598422809e-05,
      "loss": 2.5225,
      "step": 5795
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4318788945674896,
      "learning_rate": 9.950057163677944e-05,
      "loss": 2.4582,
      "step": 5800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42679736018180847,
      "learning_rate": 9.949969652714997e-05,
      "loss": 2.3649,
      "step": 5805
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4104940891265869,
      "learning_rate": 9.949882065535312e-05,
      "loss": 2.5358,
      "step": 5810
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3968827426433563,
      "learning_rate": 9.949794402140239e-05,
      "loss": 2.4712,
      "step": 5815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4292668402194977,
      "learning_rate": 9.949706662531127e-05,
      "loss": 2.5055,
      "step": 5820
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4163782000541687,
      "learning_rate": 9.949618846709327e-05,
      "loss": 2.4331,
      "step": 5825
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4434222877025604,
      "learning_rate": 9.949530954676194e-05,
      "loss": 2.5192,
      "step": 5830
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4047832489013672,
      "learning_rate": 9.949442986433076e-05,
      "loss": 2.4865,
      "step": 5835
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4600178599357605,
      "learning_rate": 9.949354941981332e-05,
      "loss": 2.429,
      "step": 5840
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.431003600358963,
      "learning_rate": 9.949266821322315e-05,
      "loss": 2.5856,
      "step": 5845
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42228439450263977,
      "learning_rate": 9.949178624457384e-05,
      "loss": 2.4921,
      "step": 5850
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4291303753852844,
      "learning_rate": 9.949090351387895e-05,
      "loss": 2.5224,
      "step": 5855
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43365296721458435,
      "learning_rate": 9.949002002115208e-05,
      "loss": 2.411,
      "step": 5860
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4267596900463104,
      "learning_rate": 9.948913576640683e-05,
      "loss": 2.5243,
      "step": 5865
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4833824038505554,
      "learning_rate": 9.948825074965682e-05,
      "loss": 2.5369,
      "step": 5870
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4124198853969574,
      "learning_rate": 9.948736497091569e-05,
      "loss": 2.5029,
      "step": 5875
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4524277448654175,
      "learning_rate": 9.948647843019705e-05,
      "loss": 2.4858,
      "step": 5880
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4206031560897827,
      "learning_rate": 9.948559112751456e-05,
      "loss": 2.4594,
      "step": 5885
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42888301610946655,
      "learning_rate": 9.948470306288189e-05,
      "loss": 2.545,
      "step": 5890
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4724574387073517,
      "learning_rate": 9.948381423631273e-05,
      "loss": 2.5237,
      "step": 5895
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4244762361049652,
      "learning_rate": 9.948292464782072e-05,
      "loss": 2.4414,
      "step": 5900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43399712443351746,
      "learning_rate": 9.948203429741961e-05,
      "loss": 2.3952,
      "step": 5905
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3920097351074219,
      "learning_rate": 9.948114318512307e-05,
      "loss": 2.5336,
      "step": 5910
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4308924078941345,
      "learning_rate": 9.948025131094482e-05,
      "loss": 2.508,
      "step": 5915
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.684937596321106,
      "learning_rate": 9.947935867489862e-05,
      "loss": 2.5592,
      "step": 5920
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.420612096786499,
      "learning_rate": 9.94784652769982e-05,
      "loss": 2.5802,
      "step": 5925
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41123536229133606,
      "learning_rate": 9.947757111725732e-05,
      "loss": 2.5351,
      "step": 5930
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40832725167274475,
      "learning_rate": 9.947667619568974e-05,
      "loss": 2.5141,
      "step": 5935
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43194207549095154,
      "learning_rate": 9.947578051230925e-05,
      "loss": 2.4846,
      "step": 5940
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4277208745479584,
      "learning_rate": 9.947488406712962e-05,
      "loss": 2.5235,
      "step": 5945
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4192441701889038,
      "learning_rate": 9.947398686016469e-05,
      "loss": 2.5105,
      "step": 5950
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38758358359336853,
      "learning_rate": 9.947308889142824e-05,
      "loss": 2.4372,
      "step": 5955
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4232860207557678,
      "learning_rate": 9.94721901609341e-05,
      "loss": 2.4407,
      "step": 5960
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4191145896911621,
      "learning_rate": 9.947129066869614e-05,
      "loss": 2.481,
      "step": 5965
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.408853143453598,
      "learning_rate": 9.947039041472817e-05,
      "loss": 2.5193,
      "step": 5970
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41409456729888916,
      "learning_rate": 9.946948939904407e-05,
      "loss": 2.5069,
      "step": 5975
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40104952454566956,
      "learning_rate": 9.946858762165772e-05,
      "loss": 2.5272,
      "step": 5980
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40809035301208496,
      "learning_rate": 9.946768508258299e-05,
      "loss": 2.4117,
      "step": 5985
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41133978962898254,
      "learning_rate": 9.946678178183377e-05,
      "loss": 2.4383,
      "step": 5990
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40951696038246155,
      "learning_rate": 9.946587771942401e-05,
      "loss": 2.4786,
      "step": 5995
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45598047971725464,
      "learning_rate": 9.946497289536758e-05,
      "loss": 2.4928,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4362678825855255,
      "learning_rate": 9.946406730967846e-05,
      "loss": 2.4591,
      "step": 6005
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42664480209350586,
      "learning_rate": 9.946316096237054e-05,
      "loss": 2.4982,
      "step": 6010
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41220349073410034,
      "learning_rate": 9.946225385345782e-05,
      "loss": 2.4709,
      "step": 6015
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43856361508369446,
      "learning_rate": 9.946134598295424e-05,
      "loss": 2.5667,
      "step": 6020
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4276503920555115,
      "learning_rate": 9.94604373508738e-05,
      "loss": 2.4786,
      "step": 6025
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41963067650794983,
      "learning_rate": 9.945952795723048e-05,
      "loss": 2.4428,
      "step": 6030
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48083385825157166,
      "learning_rate": 9.945861780203828e-05,
      "loss": 2.526,
      "step": 6035
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44299736618995667,
      "learning_rate": 9.945770688531122e-05,
      "loss": 2.4939,
      "step": 6040
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4383508265018463,
      "learning_rate": 9.945679520706332e-05,
      "loss": 2.4496,
      "step": 6045
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43519842624664307,
      "learning_rate": 9.945588276730862e-05,
      "loss": 2.5721,
      "step": 6050
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42347609996795654,
      "learning_rate": 9.945496956606115e-05,
      "loss": 2.5208,
      "step": 6055
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43728137016296387,
      "learning_rate": 9.945405560333502e-05,
      "loss": 2.4588,
      "step": 6060
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.424650102853775,
      "learning_rate": 9.945314087914426e-05,
      "loss": 2.5403,
      "step": 6065
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40075138211250305,
      "learning_rate": 9.945222539350296e-05,
      "loss": 2.4019,
      "step": 6070
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3839583992958069,
      "learning_rate": 9.945130914642524e-05,
      "loss": 2.3934,
      "step": 6075
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4093738794326782,
      "learning_rate": 9.945039213792518e-05,
      "loss": 2.4921,
      "step": 6080
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4153226912021637,
      "learning_rate": 9.94494743680169e-05,
      "loss": 2.5639,
      "step": 6085
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4155246615409851,
      "learning_rate": 9.944855583671455e-05,
      "loss": 2.4787,
      "step": 6090
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4315229654312134,
      "learning_rate": 9.944763654403226e-05,
      "loss": 2.5068,
      "step": 6095
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4219866991043091,
      "learning_rate": 9.94467164899842e-05,
      "loss": 2.4869,
      "step": 6100
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4205532670021057,
      "learning_rate": 9.944579567458452e-05,
      "loss": 2.5307,
      "step": 6105
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4196615219116211,
      "learning_rate": 9.944487409784741e-05,
      "loss": 2.503,
      "step": 6110
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4189133048057556,
      "learning_rate": 9.944395175978705e-05,
      "loss": 2.4497,
      "step": 6115
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4121706485748291,
      "learning_rate": 9.944302866041764e-05,
      "loss": 2.4762,
      "step": 6120
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4065507650375366,
      "learning_rate": 9.944210479975342e-05,
      "loss": 2.4848,
      "step": 6125
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41278985142707825,
      "learning_rate": 9.944118017780857e-05,
      "loss": 2.5647,
      "step": 6130
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4232842028141022,
      "learning_rate": 9.944025479459735e-05,
      "loss": 2.4884,
      "step": 6135
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3934057354927063,
      "learning_rate": 9.943932865013404e-05,
      "loss": 2.4549,
      "step": 6140
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40570321679115295,
      "learning_rate": 9.943840174443284e-05,
      "loss": 2.445,
      "step": 6145
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40578344464302063,
      "learning_rate": 9.943747407750805e-05,
      "loss": 2.4283,
      "step": 6150
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3879859745502472,
      "learning_rate": 9.943654564937397e-05,
      "loss": 2.5032,
      "step": 6155
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4264591932296753,
      "learning_rate": 9.943561646004488e-05,
      "loss": 2.5933,
      "step": 6160
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4077552258968353,
      "learning_rate": 9.943468650953508e-05,
      "loss": 2.4904,
      "step": 6165
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.486481636762619,
      "learning_rate": 9.943375579785889e-05,
      "loss": 2.4287,
      "step": 6170
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43164318799972534,
      "learning_rate": 9.943282432503067e-05,
      "loss": 2.3814,
      "step": 6175
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4128531515598297,
      "learning_rate": 9.943189209106472e-05,
      "loss": 2.4367,
      "step": 6180
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3966301381587982,
      "learning_rate": 9.943095909597541e-05,
      "loss": 2.4793,
      "step": 6185
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3899100124835968,
      "learning_rate": 9.943002533977713e-05,
      "loss": 2.5425,
      "step": 6190
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4200935661792755,
      "learning_rate": 9.942909082248423e-05,
      "loss": 2.4768,
      "step": 6195
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4275478720664978,
      "learning_rate": 9.942815554411109e-05,
      "loss": 2.372,
      "step": 6200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4672972559928894,
      "learning_rate": 9.942721950467214e-05,
      "loss": 2.4602,
      "step": 6205
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4355321228504181,
      "learning_rate": 9.942628270418177e-05,
      "loss": 2.4668,
      "step": 6210
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40181246399879456,
      "learning_rate": 9.942534514265442e-05,
      "loss": 2.4313,
      "step": 6215
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4078441560268402,
      "learning_rate": 9.942440682010452e-05,
      "loss": 2.4663,
      "step": 6220
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4134707450866699,
      "learning_rate": 9.942346773654652e-05,
      "loss": 2.4639,
      "step": 6225
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39584508538246155,
      "learning_rate": 9.942252789199487e-05,
      "loss": 2.6145,
      "step": 6230
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4283214807510376,
      "learning_rate": 9.942158728646406e-05,
      "loss": 2.5054,
      "step": 6235
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41903436183929443,
      "learning_rate": 9.942064591996855e-05,
      "loss": 2.4998,
      "step": 6240
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4014042615890503,
      "learning_rate": 9.941970379252284e-05,
      "loss": 2.5493,
      "step": 6245
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44338393211364746,
      "learning_rate": 9.941876090414146e-05,
      "loss": 2.4667,
      "step": 6250
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4306994080543518,
      "learning_rate": 9.941781725483889e-05,
      "loss": 2.5246,
      "step": 6255
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39664509892463684,
      "learning_rate": 9.941687284462969e-05,
      "loss": 2.4982,
      "step": 6260
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40283602476119995,
      "learning_rate": 9.941592767352839e-05,
      "loss": 2.4605,
      "step": 6265
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4373075067996979,
      "learning_rate": 9.941498174154954e-05,
      "loss": 2.4391,
      "step": 6270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4079284965991974,
      "learning_rate": 9.94140350487077e-05,
      "loss": 2.4637,
      "step": 6275
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.400416761636734,
      "learning_rate": 9.941308759501747e-05,
      "loss": 2.4492,
      "step": 6280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39027532935142517,
      "learning_rate": 9.941213938049342e-05,
      "loss": 2.4864,
      "step": 6285
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3920605182647705,
      "learning_rate": 9.941119040515017e-05,
      "loss": 2.4649,
      "step": 6290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44276800751686096,
      "learning_rate": 9.941024066900229e-05,
      "loss": 2.5184,
      "step": 6295
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39533501863479614,
      "learning_rate": 9.940929017206445e-05,
      "loss": 2.5142,
      "step": 6300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41311994194984436,
      "learning_rate": 9.940833891435124e-05,
      "loss": 2.3874,
      "step": 6305
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4152023494243622,
      "learning_rate": 9.940738689587735e-05,
      "loss": 2.4046,
      "step": 6310
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4222840964794159,
      "learning_rate": 9.940643411665742e-05,
      "loss": 2.526,
      "step": 6315
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40121543407440186,
      "learning_rate": 9.940548057670612e-05,
      "loss": 2.4279,
      "step": 6320
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4088546931743622,
      "learning_rate": 9.940452627603813e-05,
      "loss": 2.4637,
      "step": 6325
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40210849046707153,
      "learning_rate": 9.940357121466813e-05,
      "loss": 2.4456,
      "step": 6330
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4052463471889496,
      "learning_rate": 9.940261539261086e-05,
      "loss": 2.4327,
      "step": 6335
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41920721530914307,
      "learning_rate": 9.940165880988102e-05,
      "loss": 2.5574,
      "step": 6340
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4081125855445862,
      "learning_rate": 9.940070146649333e-05,
      "loss": 2.5172,
      "step": 6345
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40494611859321594,
      "learning_rate": 9.939974336246254e-05,
      "loss": 2.3408,
      "step": 6350
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.424580454826355,
      "learning_rate": 9.93987844978034e-05,
      "loss": 2.4383,
      "step": 6355
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39831721782684326,
      "learning_rate": 9.939782487253067e-05,
      "loss": 2.464,
      "step": 6360
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43550965189933777,
      "learning_rate": 9.939686448665915e-05,
      "loss": 2.5528,
      "step": 6365
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41427865624427795,
      "learning_rate": 9.939590334020359e-05,
      "loss": 2.5097,
      "step": 6370
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45534589886665344,
      "learning_rate": 9.939494143317881e-05,
      "loss": 2.4459,
      "step": 6375
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41332027316093445,
      "learning_rate": 9.939397876559962e-05,
      "loss": 2.5712,
      "step": 6380
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4081014394760132,
      "learning_rate": 9.939301533748084e-05,
      "loss": 2.3074,
      "step": 6385
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4047027826309204,
      "learning_rate": 9.939205114883731e-05,
      "loss": 2.53,
      "step": 6390
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42956018447875977,
      "learning_rate": 9.939108619968386e-05,
      "loss": 2.4842,
      "step": 6395
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42118537425994873,
      "learning_rate": 9.939012049003537e-05,
      "loss": 2.5033,
      "step": 6400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39760950207710266,
      "learning_rate": 9.93891540199067e-05,
      "loss": 2.4271,
      "step": 6405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43220454454421997,
      "learning_rate": 9.938818678931271e-05,
      "loss": 2.5299,
      "step": 6410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38811275362968445,
      "learning_rate": 9.938721879826832e-05,
      "loss": 2.4631,
      "step": 6415
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41012826561927795,
      "learning_rate": 9.938625004678844e-05,
      "loss": 2.5525,
      "step": 6420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41467025876045227,
      "learning_rate": 9.938528053488796e-05,
      "loss": 2.422,
      "step": 6425
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6718437075614929,
      "learning_rate": 9.938431026258182e-05,
      "loss": 2.4627,
      "step": 6430
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38922634720802307,
      "learning_rate": 9.938333922988497e-05,
      "loss": 2.4099,
      "step": 6435
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3912346363067627,
      "learning_rate": 9.938236743681235e-05,
      "loss": 2.4088,
      "step": 6440
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39553752541542053,
      "learning_rate": 9.938139488337891e-05,
      "loss": 2.533,
      "step": 6445
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39396369457244873,
      "learning_rate": 9.938042156959966e-05,
      "loss": 2.4396,
      "step": 6450
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3992331027984619,
      "learning_rate": 9.937944749548956e-05,
      "loss": 2.3884,
      "step": 6455
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4010012149810791,
      "learning_rate": 9.937847266106359e-05,
      "loss": 2.4891,
      "step": 6460
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.416885644197464,
      "learning_rate": 9.937749706633682e-05,
      "loss": 2.4163,
      "step": 6465
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41220128536224365,
      "learning_rate": 9.937652071132422e-05,
      "loss": 2.5174,
      "step": 6470
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43417608737945557,
      "learning_rate": 9.937554359604084e-05,
      "loss": 2.4028,
      "step": 6475
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41461363434791565,
      "learning_rate": 9.937456572050172e-05,
      "loss": 2.391,
      "step": 6480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4043256640434265,
      "learning_rate": 9.937358708472193e-05,
      "loss": 2.3454,
      "step": 6485
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41102898120880127,
      "learning_rate": 9.937260768871654e-05,
      "loss": 2.4133,
      "step": 6490
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40208908915519714,
      "learning_rate": 9.93716275325006e-05,
      "loss": 2.4914,
      "step": 6495
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4110158681869507,
      "learning_rate": 9.937064661608924e-05,
      "loss": 2.4322,
      "step": 6500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3935650587081909,
      "learning_rate": 9.936966493949752e-05,
      "loss": 2.4234,
      "step": 6505
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39359191060066223,
      "learning_rate": 9.936868250274059e-05,
      "loss": 2.3596,
      "step": 6510
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3873981833457947,
      "learning_rate": 9.936769930583359e-05,
      "loss": 2.3661,
      "step": 6515
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42440375685691833,
      "learning_rate": 9.93667153487916e-05,
      "loss": 2.4492,
      "step": 6520
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41155171394348145,
      "learning_rate": 9.936573063162984e-05,
      "loss": 2.4563,
      "step": 6525
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40671977400779724,
      "learning_rate": 9.936474515436342e-05,
      "loss": 2.5597,
      "step": 6530
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4052463471889496,
      "learning_rate": 9.936375891700753e-05,
      "loss": 2.4422,
      "step": 6535
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3891323208808899,
      "learning_rate": 9.936277191957737e-05,
      "loss": 2.4227,
      "step": 6540
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41244983673095703,
      "learning_rate": 9.936178416208813e-05,
      "loss": 2.479,
      "step": 6545
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40125420689582825,
      "learning_rate": 9.9360795644555e-05,
      "loss": 2.4402,
      "step": 6550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4157456159591675,
      "learning_rate": 9.935980636699322e-05,
      "loss": 2.4452,
      "step": 6555
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4074021875858307,
      "learning_rate": 9.935881632941801e-05,
      "loss": 2.4117,
      "step": 6560
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4032139182090759,
      "learning_rate": 9.935782553184463e-05,
      "loss": 2.4789,
      "step": 6565
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39190277457237244,
      "learning_rate": 9.935683397428833e-05,
      "loss": 2.3725,
      "step": 6570
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40465593338012695,
      "learning_rate": 9.935584165676436e-05,
      "loss": 2.4394,
      "step": 6575
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41066768765449524,
      "learning_rate": 9.935484857928801e-05,
      "loss": 2.4753,
      "step": 6580
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.408014714717865,
      "learning_rate": 9.935385474187459e-05,
      "loss": 2.4508,
      "step": 6585
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41270387172698975,
      "learning_rate": 9.93528601445394e-05,
      "loss": 2.5185,
      "step": 6590
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6306276917457581,
      "learning_rate": 9.935186478729771e-05,
      "loss": 2.4834,
      "step": 6595
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41581469774246216,
      "learning_rate": 9.93508686701649e-05,
      "loss": 2.2969,
      "step": 6600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4046902358531952,
      "learning_rate": 9.934987179315627e-05,
      "loss": 2.4463,
      "step": 6605
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.407663494348526,
      "learning_rate": 9.934887415628719e-05,
      "loss": 2.4972,
      "step": 6610
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41052407026290894,
      "learning_rate": 9.934787575957301e-05,
      "loss": 2.5008,
      "step": 6615
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3814753592014313,
      "learning_rate": 9.934687660302912e-05,
      "loss": 2.4693,
      "step": 6620
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4147699475288391,
      "learning_rate": 9.934587668667088e-05,
      "loss": 2.4807,
      "step": 6625
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39891317486763,
      "learning_rate": 9.93448760105137e-05,
      "loss": 2.4919,
      "step": 6630
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40257900953292847,
      "learning_rate": 9.934387457457299e-05,
      "loss": 2.3545,
      "step": 6635
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40402868390083313,
      "learning_rate": 9.934287237886417e-05,
      "loss": 2.3374,
      "step": 6640
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.407966673374176,
      "learning_rate": 9.934186942340265e-05,
      "loss": 2.4887,
      "step": 6645
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4142005145549774,
      "learning_rate": 9.934086570820392e-05,
      "loss": 2.46,
      "step": 6650
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40551379323005676,
      "learning_rate": 9.933986123328337e-05,
      "loss": 2.4072,
      "step": 6655
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3947029411792755,
      "learning_rate": 9.933885599865651e-05,
      "loss": 2.527,
      "step": 6660
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41236764192581177,
      "learning_rate": 9.933785000433882e-05,
      "loss": 2.43,
      "step": 6665
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4008924663066864,
      "learning_rate": 9.933684325034577e-05,
      "loss": 2.4215,
      "step": 6670
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40346482396125793,
      "learning_rate": 9.933583573669287e-05,
      "loss": 2.4366,
      "step": 6675
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41285115480422974,
      "learning_rate": 9.933482746339564e-05,
      "loss": 2.4023,
      "step": 6680
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4007401764392853,
      "learning_rate": 9.933381843046959e-05,
      "loss": 2.37,
      "step": 6685
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4013449251651764,
      "learning_rate": 9.933280863793026e-05,
      "loss": 2.4735,
      "step": 6690
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39859235286712646,
      "learning_rate": 9.933179808579321e-05,
      "loss": 2.4448,
      "step": 6695
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4158281683921814,
      "learning_rate": 9.933078677407399e-05,
      "loss": 2.458,
      "step": 6700
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4227816164493561,
      "learning_rate": 9.932977470278818e-05,
      "loss": 2.4755,
      "step": 6705
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39157384634017944,
      "learning_rate": 9.932876187195135e-05,
      "loss": 2.476,
      "step": 6710
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.446099191904068,
      "learning_rate": 9.932774828157911e-05,
      "loss": 2.3965,
      "step": 6715
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39338812232017517,
      "learning_rate": 9.932673393168705e-05,
      "loss": 2.4295,
      "step": 6720
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4035241901874542,
      "learning_rate": 9.932571882229081e-05,
      "loss": 2.5213,
      "step": 6725
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41839835047721863,
      "learning_rate": 9.932470295340599e-05,
      "loss": 2.4257,
      "step": 6730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4051453173160553,
      "learning_rate": 9.932368632504826e-05,
      "loss": 2.5016,
      "step": 6735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.384147584438324,
      "learning_rate": 9.932266893723325e-05,
      "loss": 2.4694,
      "step": 6740
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40205591917037964,
      "learning_rate": 9.932165078997666e-05,
      "loss": 2.4619,
      "step": 6745
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3942875862121582,
      "learning_rate": 9.932063188329412e-05,
      "loss": 2.4509,
      "step": 6750
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3972095847129822,
      "learning_rate": 9.931961221720136e-05,
      "loss": 2.3298,
      "step": 6755
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40668410062789917,
      "learning_rate": 9.931859179171405e-05,
      "loss": 2.3663,
      "step": 6760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42049479484558105,
      "learning_rate": 9.931757060684792e-05,
      "loss": 2.3934,
      "step": 6765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41439497470855713,
      "learning_rate": 9.93165486626187e-05,
      "loss": 2.4709,
      "step": 6770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.406587690114975,
      "learning_rate": 9.93155259590421e-05,
      "loss": 2.4662,
      "step": 6775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39916422963142395,
      "learning_rate": 9.931450249613389e-05,
      "loss": 2.4228,
      "step": 6780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3894968628883362,
      "learning_rate": 9.931347827390982e-05,
      "loss": 2.4019,
      "step": 6785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4106947183609009,
      "learning_rate": 9.931245329238566e-05,
      "loss": 2.3937,
      "step": 6790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39689967036247253,
      "learning_rate": 9.931142755157719e-05,
      "loss": 2.4428,
      "step": 6795
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41577035188674927,
      "learning_rate": 9.931040105150019e-05,
      "loss": 2.4256,
      "step": 6800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38176754117012024,
      "learning_rate": 9.93093737921705e-05,
      "loss": 2.48,
      "step": 6805
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3934345245361328,
      "learning_rate": 9.93083457736039e-05,
      "loss": 2.3768,
      "step": 6810
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3936603367328644,
      "learning_rate": 9.930731699581626e-05,
      "loss": 2.4233,
      "step": 6815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40052396059036255,
      "learning_rate": 9.930628745882339e-05,
      "loss": 2.5167,
      "step": 6820
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4123167097568512,
      "learning_rate": 9.930525716264114e-05,
      "loss": 2.4972,
      "step": 6825
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.402942419052124,
      "learning_rate": 9.93042261072854e-05,
      "loss": 2.4452,
      "step": 6830
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3888356685638428,
      "learning_rate": 9.930319429277202e-05,
      "loss": 2.468,
      "step": 6835
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.37647855281829834,
      "learning_rate": 9.93021617191169e-05,
      "loss": 2.3422,
      "step": 6840
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.37467947602272034,
      "learning_rate": 9.930112838633592e-05,
      "loss": 2.4159,
      "step": 6845
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40984708070755005,
      "learning_rate": 9.930009429444503e-05,
      "loss": 2.3843,
      "step": 6850
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.441827654838562,
      "learning_rate": 9.929905944346012e-05,
      "loss": 2.3928,
      "step": 6855
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.412013441324234,
      "learning_rate": 9.929802383339713e-05,
      "loss": 2.319,
      "step": 6860
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41906899213790894,
      "learning_rate": 9.9296987464272e-05,
      "loss": 2.5268,
      "step": 6865
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39814114570617676,
      "learning_rate": 9.929595033610072e-05,
      "loss": 2.4993,
      "step": 6870
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39326849579811096,
      "learning_rate": 9.929491244889921e-05,
      "loss": 2.5017,
      "step": 6875
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4035642147064209,
      "learning_rate": 9.929387380268349e-05,
      "loss": 2.5794,
      "step": 6880
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3980746567249298,
      "learning_rate": 9.929283439746955e-05,
      "loss": 2.4599,
      "step": 6885
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39549529552459717,
      "learning_rate": 9.929179423327336e-05,
      "loss": 2.4716,
      "step": 6890
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3970012068748474,
      "learning_rate": 9.929075331011097e-05,
      "loss": 2.4208,
      "step": 6895
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42309442162513733,
      "learning_rate": 9.928971162799838e-05,
      "loss": 2.4253,
      "step": 6900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3859192132949829,
      "learning_rate": 9.928866918695165e-05,
      "loss": 2.5436,
      "step": 6905
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39164263010025024,
      "learning_rate": 9.928762598698684e-05,
      "loss": 2.4985,
      "step": 6910
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4092990458011627,
      "learning_rate": 9.928658202811999e-05,
      "loss": 2.4547,
      "step": 6915
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3924550414085388,
      "learning_rate": 9.928553731036719e-05,
      "loss": 2.4905,
      "step": 6920
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3956751227378845,
      "learning_rate": 9.928449183374451e-05,
      "loss": 2.4556,
      "step": 6925
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4207506775856018,
      "learning_rate": 9.928344559826806e-05,
      "loss": 2.5001,
      "step": 6930
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3894541263580322,
      "learning_rate": 9.928239860395394e-05,
      "loss": 2.4301,
      "step": 6935
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4091970920562744,
      "learning_rate": 9.928135085081827e-05,
      "loss": 2.5448,
      "step": 6940
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41059577465057373,
      "learning_rate": 9.928030233887721e-05,
      "loss": 2.3999,
      "step": 6945
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3968183398246765,
      "learning_rate": 9.927925306814686e-05,
      "loss": 2.4984,
      "step": 6950
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39499032497406006,
      "learning_rate": 9.927820303864341e-05,
      "loss": 2.5482,
      "step": 6955
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3749392032623291,
      "learning_rate": 9.927715225038303e-05,
      "loss": 2.4329,
      "step": 6960
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39309263229370117,
      "learning_rate": 9.927610070338187e-05,
      "loss": 2.421,
      "step": 6965
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40027090907096863,
      "learning_rate": 9.927504839765615e-05,
      "loss": 2.4163,
      "step": 6970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4124675691127777,
      "learning_rate": 9.927399533322206e-05,
      "loss": 2.4822,
      "step": 6975
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.394878089427948,
      "learning_rate": 9.92729415100958e-05,
      "loss": 2.3698,
      "step": 6980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3886864185333252,
      "learning_rate": 9.927188692829363e-05,
      "loss": 2.3865,
      "step": 6985
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.36872586607933044,
      "learning_rate": 9.927083158783177e-05,
      "loss": 2.3402,
      "step": 6990
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39966264367103577,
      "learning_rate": 9.926977548872644e-05,
      "loss": 2.4148,
      "step": 6995
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40319275856018066,
      "learning_rate": 9.926871863099396e-05,
      "loss": 2.5272,
      "step": 7000
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40935972332954407,
      "learning_rate": 9.926766101465056e-05,
      "loss": 2.4173,
      "step": 7005
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3994516134262085,
      "learning_rate": 9.926660263971254e-05,
      "loss": 2.4131,
      "step": 7010
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44502273201942444,
      "learning_rate": 9.92655435061962e-05,
      "loss": 2.4601,
      "step": 7015
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39007487893104553,
      "learning_rate": 9.926448361411784e-05,
      "loss": 2.4931,
      "step": 7020
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43532663583755493,
      "learning_rate": 9.926342296349378e-05,
      "loss": 2.4953,
      "step": 7025
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39114102721214294,
      "learning_rate": 9.926236155434034e-05,
      "loss": 2.4639,
      "step": 7030
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4052385687828064,
      "learning_rate": 9.926129938667388e-05,
      "loss": 2.3582,
      "step": 7035
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3917314410209656,
      "learning_rate": 9.926023646051075e-05,
      "loss": 2.4139,
      "step": 7040
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4162963926792145,
      "learning_rate": 9.925917277586734e-05,
      "loss": 2.5009,
      "step": 7045
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41129741072654724,
      "learning_rate": 9.925810833275998e-05,
      "loss": 2.3752,
      "step": 7050
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4191969037055969,
      "learning_rate": 9.925704313120508e-05,
      "loss": 2.4152,
      "step": 7055
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4316924512386322,
      "learning_rate": 9.925597717121906e-05,
      "loss": 2.5632,
      "step": 7060
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4402734637260437,
      "learning_rate": 9.925491045281832e-05,
      "loss": 2.4542,
      "step": 7065
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41813838481903076,
      "learning_rate": 9.925384297601928e-05,
      "loss": 2.4068,
      "step": 7070
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4104415774345398,
      "learning_rate": 9.925277474083835e-05,
      "loss": 2.4301,
      "step": 7075
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4105343222618103,
      "learning_rate": 9.925170574729204e-05,
      "loss": 2.469,
      "step": 7080
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3795762062072754,
      "learning_rate": 9.925063599539677e-05,
      "loss": 2.4478,
      "step": 7085
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38717755675315857,
      "learning_rate": 9.924956548516901e-05,
      "loss": 2.4716,
      "step": 7090
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3971962332725525,
      "learning_rate": 9.924849421662528e-05,
      "loss": 2.4466,
      "step": 7095
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40029284358024597,
      "learning_rate": 9.924742218978203e-05,
      "loss": 2.5116,
      "step": 7100
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41997361183166504,
      "learning_rate": 9.924634940465577e-05,
      "loss": 2.4209,
      "step": 7105
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39156413078308105,
      "learning_rate": 9.924527586126305e-05,
      "loss": 2.4113,
      "step": 7110
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40273407101631165,
      "learning_rate": 9.924420155962037e-05,
      "loss": 2.4281,
      "step": 7115
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38489866256713867,
      "learning_rate": 9.924312649974427e-05,
      "loss": 2.3841,
      "step": 7120
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41676950454711914,
      "learning_rate": 9.924205068165134e-05,
      "loss": 2.3848,
      "step": 7125
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37713316082954407,
      "learning_rate": 9.924097410535811e-05,
      "loss": 2.3624,
      "step": 7130
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4255376160144806,
      "learning_rate": 9.923989677088116e-05,
      "loss": 2.4039,
      "step": 7135
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41981804370880127,
      "learning_rate": 9.92388186782371e-05,
      "loss": 2.4194,
      "step": 7140
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41376930475234985,
      "learning_rate": 9.92377398274425e-05,
      "loss": 2.4817,
      "step": 7145
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40257972478866577,
      "learning_rate": 9.923666021851398e-05,
      "loss": 2.4569,
      "step": 7150
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40416043996810913,
      "learning_rate": 9.92355798514682e-05,
      "loss": 2.4525,
      "step": 7155
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4198589026927948,
      "learning_rate": 9.923449872632173e-05,
      "loss": 2.4718,
      "step": 7160
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.426847904920578,
      "learning_rate": 9.923341684309128e-05,
      "loss": 2.374,
      "step": 7165
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40475407242774963,
      "learning_rate": 9.923233420179347e-05,
      "loss": 2.4623,
      "step": 7170
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4202999174594879,
      "learning_rate": 9.923125080244497e-05,
      "loss": 2.4811,
      "step": 7175
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4112209975719452,
      "learning_rate": 9.923016664506248e-05,
      "loss": 2.425,
      "step": 7180
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3937538266181946,
      "learning_rate": 9.922908172966268e-05,
      "loss": 2.5316,
      "step": 7185
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4081346094608307,
      "learning_rate": 9.922799605626227e-05,
      "loss": 2.4028,
      "step": 7190
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4191742241382599,
      "learning_rate": 9.9226909624878e-05,
      "loss": 2.473,
      "step": 7195
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40102726221084595,
      "learning_rate": 9.922582243552654e-05,
      "loss": 2.4487,
      "step": 7200
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40936776995658875,
      "learning_rate": 9.922473448822469e-05,
      "loss": 2.4319,
      "step": 7205
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3840581476688385,
      "learning_rate": 9.922364578298916e-05,
      "loss": 2.4561,
      "step": 7210
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41058847308158875,
      "learning_rate": 9.922255631983673e-05,
      "loss": 2.3306,
      "step": 7215
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3945527672767639,
      "learning_rate": 9.922146609878418e-05,
      "loss": 2.4007,
      "step": 7220
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3918132781982422,
      "learning_rate": 9.922037511984829e-05,
      "loss": 2.4763,
      "step": 7225
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39406219124794006,
      "learning_rate": 9.921928338304584e-05,
      "loss": 2.4057,
      "step": 7230
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39223161339759827,
      "learning_rate": 9.921819088839369e-05,
      "loss": 2.3902,
      "step": 7235
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42004266381263733,
      "learning_rate": 9.92170976359086e-05,
      "loss": 2.3575,
      "step": 7240
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4019894003868103,
      "learning_rate": 9.921600362560744e-05,
      "loss": 2.5206,
      "step": 7245
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3854498863220215,
      "learning_rate": 9.921490885750706e-05,
      "loss": 2.5375,
      "step": 7250
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4116158187389374,
      "learning_rate": 9.92138133316243e-05,
      "loss": 2.491,
      "step": 7255
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43389198184013367,
      "learning_rate": 9.921271704797602e-05,
      "loss": 2.3506,
      "step": 7260
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4043695330619812,
      "learning_rate": 9.921162000657913e-05,
      "loss": 2.4247,
      "step": 7265
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40360504388809204,
      "learning_rate": 9.921052220745048e-05,
      "loss": 2.4271,
      "step": 7270
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3835066258907318,
      "learning_rate": 9.920942365060702e-05,
      "loss": 2.4355,
      "step": 7275
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3952276408672333,
      "learning_rate": 9.920832433606564e-05,
      "loss": 2.4251,
      "step": 7280
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3937224745750427,
      "learning_rate": 9.920722426384326e-05,
      "loss": 2.3264,
      "step": 7285
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41895556449890137,
      "learning_rate": 9.920612343395683e-05,
      "loss": 2.5043,
      "step": 7290
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3932245075702667,
      "learning_rate": 9.92050218464233e-05,
      "loss": 2.4279,
      "step": 7295
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39219963550567627,
      "learning_rate": 9.920391950125963e-05,
      "loss": 2.367,
      "step": 7300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3895873725414276,
      "learning_rate": 9.920281639848278e-05,
      "loss": 2.4918,
      "step": 7305
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4108986556529999,
      "learning_rate": 9.920171253810975e-05,
      "loss": 2.4788,
      "step": 7310
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3783574402332306,
      "learning_rate": 9.920060792015754e-05,
      "loss": 2.4888,
      "step": 7315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3836260139942169,
      "learning_rate": 9.919950254464316e-05,
      "loss": 2.4407,
      "step": 7320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4092741906642914,
      "learning_rate": 9.919839641158361e-05,
      "loss": 2.4177,
      "step": 7325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37531915307044983,
      "learning_rate": 9.919728952099593e-05,
      "loss": 2.3812,
      "step": 7330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40291279554367065,
      "learning_rate": 9.919618187289717e-05,
      "loss": 2.4877,
      "step": 7335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4033949673175812,
      "learning_rate": 9.919507346730437e-05,
      "loss": 2.3532,
      "step": 7340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39707833528518677,
      "learning_rate": 9.919396430423463e-05,
      "loss": 2.4352,
      "step": 7345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.373808354139328,
      "learning_rate": 9.9192854383705e-05,
      "loss": 2.3805,
      "step": 7350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4724747836589813,
      "learning_rate": 9.919174370573257e-05,
      "loss": 2.4291,
      "step": 7355
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3892482817173004,
      "learning_rate": 9.919063227033445e-05,
      "loss": 2.4302,
      "step": 7360
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38751205801963806,
      "learning_rate": 9.918952007752776e-05,
      "loss": 2.4297,
      "step": 7365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38450437784194946,
      "learning_rate": 9.91884071273296e-05,
      "loss": 2.4574,
      "step": 7370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.395622193813324,
      "learning_rate": 9.918729341975712e-05,
      "loss": 2.37,
      "step": 7375
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5536355376243591,
      "learning_rate": 9.918617895482749e-05,
      "loss": 2.5123,
      "step": 7380
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39615893363952637,
      "learning_rate": 9.918506373255784e-05,
      "loss": 2.4712,
      "step": 7385
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39179477095603943,
      "learning_rate": 9.918394775296536e-05,
      "loss": 2.5454,
      "step": 7390
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3762317895889282,
      "learning_rate": 9.91828310160672e-05,
      "loss": 2.4483,
      "step": 7395
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4103586971759796,
      "learning_rate": 9.918171352188062e-05,
      "loss": 2.5511,
      "step": 7400
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41817358136177063,
      "learning_rate": 9.918059527042275e-05,
      "loss": 2.4477,
      "step": 7405
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3743829131126404,
      "learning_rate": 9.917947626171086e-05,
      "loss": 2.4858,
      "step": 7410
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39710715413093567,
      "learning_rate": 9.917835649576217e-05,
      "loss": 2.4682,
      "step": 7415
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38651755452156067,
      "learning_rate": 9.91772359725939e-05,
      "loss": 2.4578,
      "step": 7420
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39019668102264404,
      "learning_rate": 9.917611469222333e-05,
      "loss": 2.4343,
      "step": 7425
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39105936884880066,
      "learning_rate": 9.917499265466772e-05,
      "loss": 2.4834,
      "step": 7430
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39419859647750854,
      "learning_rate": 9.917386985994432e-05,
      "loss": 2.4779,
      "step": 7435
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4166622459888458,
      "learning_rate": 9.917274630807047e-05,
      "loss": 2.3691,
      "step": 7440
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38331329822540283,
      "learning_rate": 9.917162199906341e-05,
      "loss": 2.3368,
      "step": 7445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39266878366470337,
      "learning_rate": 9.917049693294049e-05,
      "loss": 2.4844,
      "step": 7450
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40055006742477417,
      "learning_rate": 9.916937110971902e-05,
      "loss": 2.3604,
      "step": 7455
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40987130999565125,
      "learning_rate": 9.916824452941635e-05,
      "loss": 2.5089,
      "step": 7460
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3698435127735138,
      "learning_rate": 9.916711719204981e-05,
      "loss": 2.4089,
      "step": 7465
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4144148826599121,
      "learning_rate": 9.916598909763675e-05,
      "loss": 2.4649,
      "step": 7470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37636470794677734,
      "learning_rate": 9.916486024619455e-05,
      "loss": 2.4684,
      "step": 7475
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3827991485595703,
      "learning_rate": 9.916373063774062e-05,
      "loss": 2.4011,
      "step": 7480
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40807825326919556,
      "learning_rate": 9.91626002722923e-05,
      "loss": 2.3435,
      "step": 7485
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.408422589302063,
      "learning_rate": 9.916146914986703e-05,
      "loss": 2.3013,
      "step": 7490
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43231138586997986,
      "learning_rate": 9.916033727048223e-05,
      "loss": 2.4364,
      "step": 7495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.399453729391098,
      "learning_rate": 9.915920463415531e-05,
      "loss": 2.501,
      "step": 7500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9023413062095642,
      "learning_rate": 9.915807124090371e-05,
      "loss": 2.3835,
      "step": 7505
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.36874958872795105,
      "learning_rate": 9.915693709074487e-05,
      "loss": 2.4157,
      "step": 7510
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3889729082584381,
      "learning_rate": 9.91558021836963e-05,
      "loss": 2.3536,
      "step": 7515
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38608017563819885,
      "learning_rate": 9.915466651977543e-05,
      "loss": 2.5221,
      "step": 7520
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37839657068252563,
      "learning_rate": 9.915353009899976e-05,
      "loss": 2.4154,
      "step": 7525
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38729095458984375,
      "learning_rate": 9.915239292138678e-05,
      "loss": 2.4068,
      "step": 7530
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3784966766834259,
      "learning_rate": 9.915125498695402e-05,
      "loss": 2.3974,
      "step": 7535
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3970995247364044,
      "learning_rate": 9.9150116295719e-05,
      "loss": 2.4421,
      "step": 7540
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3773486316204071,
      "learning_rate": 9.914897684769921e-05,
      "loss": 2.4785,
      "step": 7545
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37721654772758484,
      "learning_rate": 9.914783664291225e-05,
      "loss": 2.5114,
      "step": 7550
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3855530917644501,
      "learning_rate": 9.914669568137565e-05,
      "loss": 2.4007,
      "step": 7555
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41357412934303284,
      "learning_rate": 9.914555396310698e-05,
      "loss": 2.4582,
      "step": 7560
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37612685561180115,
      "learning_rate": 9.914441148812382e-05,
      "loss": 2.4253,
      "step": 7565
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41444817185401917,
      "learning_rate": 9.914326825644376e-05,
      "loss": 2.3858,
      "step": 7570
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3748912215232849,
      "learning_rate": 9.91421242680844e-05,
      "loss": 2.3962,
      "step": 7575
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37750405073165894,
      "learning_rate": 9.914097952306338e-05,
      "loss": 2.38,
      "step": 7580
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40947234630584717,
      "learning_rate": 9.913983402139828e-05,
      "loss": 2.4995,
      "step": 7585
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4007430374622345,
      "learning_rate": 9.913868776310677e-05,
      "loss": 2.4834,
      "step": 7590
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39499083161354065,
      "learning_rate": 9.913754074820649e-05,
      "loss": 2.3868,
      "step": 7595
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37979787588119507,
      "learning_rate": 9.913639297671511e-05,
      "loss": 2.3856,
      "step": 7600
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41183364391326904,
      "learning_rate": 9.91352444486503e-05,
      "loss": 2.5077,
      "step": 7605
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37465187907218933,
      "learning_rate": 9.913409516402974e-05,
      "loss": 2.4466,
      "step": 7610
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3971143364906311,
      "learning_rate": 9.913294512287112e-05,
      "loss": 2.3629,
      "step": 7615
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3925729990005493,
      "learning_rate": 9.913179432519215e-05,
      "loss": 2.4556,
      "step": 7620
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38120999932289124,
      "learning_rate": 9.913064277101055e-05,
      "loss": 2.4416,
      "step": 7625
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40581458806991577,
      "learning_rate": 9.912949046034406e-05,
      "loss": 2.3256,
      "step": 7630
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3752162754535675,
      "learning_rate": 9.912833739321042e-05,
      "loss": 2.4947,
      "step": 7635
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3919074833393097,
      "learning_rate": 9.912718356962738e-05,
      "loss": 2.4256,
      "step": 7640
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3681202530860901,
      "learning_rate": 9.91260289896127e-05,
      "loss": 2.3353,
      "step": 7645
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41241446137428284,
      "learning_rate": 9.912487365318417e-05,
      "loss": 2.4419,
      "step": 7650
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39689916372299194,
      "learning_rate": 9.912371756035957e-05,
      "loss": 2.411,
      "step": 7655
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4310096502304077,
      "learning_rate": 9.912256071115671e-05,
      "loss": 2.4337,
      "step": 7660
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40198689699172974,
      "learning_rate": 9.91214031055934e-05,
      "loss": 2.4894,
      "step": 7665
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4186384081840515,
      "learning_rate": 9.912024474368745e-05,
      "loss": 2.4854,
      "step": 7670
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38108059763908386,
      "learning_rate": 9.911908562545671e-05,
      "loss": 2.3469,
      "step": 7675
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41650480031967163,
      "learning_rate": 9.911792575091903e-05,
      "loss": 2.3926,
      "step": 7680
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38261598348617554,
      "learning_rate": 9.911676512009226e-05,
      "loss": 2.4683,
      "step": 7685
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3748529553413391,
      "learning_rate": 9.911560373299426e-05,
      "loss": 2.4104,
      "step": 7690
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3981565833091736,
      "learning_rate": 9.911444158964296e-05,
      "loss": 2.4742,
      "step": 7695
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38296014070510864,
      "learning_rate": 9.91132786900562e-05,
      "loss": 2.3562,
      "step": 7700
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39533358812332153,
      "learning_rate": 9.911211503425191e-05,
      "loss": 2.4471,
      "step": 7705
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3848271667957306,
      "learning_rate": 9.911095062224798e-05,
      "loss": 2.3965,
      "step": 7710
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3788564205169678,
      "learning_rate": 9.91097854540624e-05,
      "loss": 2.444,
      "step": 7715
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4057249426841736,
      "learning_rate": 9.910861952971304e-05,
      "loss": 2.3739,
      "step": 7720
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39320170879364014,
      "learning_rate": 9.910745284921791e-05,
      "loss": 2.4353,
      "step": 7725
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38510292768478394,
      "learning_rate": 9.910628541259495e-05,
      "loss": 2.4294,
      "step": 7730
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3958958089351654,
      "learning_rate": 9.91051172198621e-05,
      "loss": 2.4759,
      "step": 7735
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39086902141571045,
      "learning_rate": 9.910394827103742e-05,
      "loss": 2.457,
      "step": 7740
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38251712918281555,
      "learning_rate": 9.910277856613886e-05,
      "loss": 2.4473,
      "step": 7745
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38484612107276917,
      "learning_rate": 9.910160810518442e-05,
      "loss": 2.4713,
      "step": 7750
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39433446526527405,
      "learning_rate": 9.910043688819218e-05,
      "loss": 2.3366,
      "step": 7755
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40028220415115356,
      "learning_rate": 9.909926491518012e-05,
      "loss": 2.4683,
      "step": 7760
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38515549898147583,
      "learning_rate": 9.909809218616629e-05,
      "loss": 2.4093,
      "step": 7765
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40134966373443604,
      "learning_rate": 9.909691870116877e-05,
      "loss": 2.4577,
      "step": 7770
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3859964609146118,
      "learning_rate": 9.909574446020561e-05,
      "loss": 2.4089,
      "step": 7775
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3677075207233429,
      "learning_rate": 9.90945694632949e-05,
      "loss": 2.3087,
      "step": 7780
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39916500449180603,
      "learning_rate": 9.909339371045474e-05,
      "loss": 2.4633,
      "step": 7785
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39440280199050903,
      "learning_rate": 9.90922172017032e-05,
      "loss": 2.489,
      "step": 7790
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38181072473526,
      "learning_rate": 9.909103993705845e-05,
      "loss": 2.4666,
      "step": 7795
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4084306061267853,
      "learning_rate": 9.908986191653857e-05,
      "loss": 2.403,
      "step": 7800
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3949647545814514,
      "learning_rate": 9.90886831401617e-05,
      "loss": 2.4522,
      "step": 7805
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39706915616989136,
      "learning_rate": 9.908750360794603e-05,
      "loss": 2.4559,
      "step": 7810
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3866342008113861,
      "learning_rate": 9.908632331990968e-05,
      "loss": 2.4574,
      "step": 7815
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38194456696510315,
      "learning_rate": 9.908514227607084e-05,
      "loss": 2.4329,
      "step": 7820
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4228445291519165,
      "learning_rate": 9.90839604764477e-05,
      "loss": 2.4378,
      "step": 7825
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39185288548469543,
      "learning_rate": 9.908277792105845e-05,
      "loss": 2.413,
      "step": 7830
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39325347542762756,
      "learning_rate": 9.90815946099213e-05,
      "loss": 2.4297,
      "step": 7835
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38388702273368835,
      "learning_rate": 9.908041054305446e-05,
      "loss": 2.3253,
      "step": 7840
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38563692569732666,
      "learning_rate": 9.907922572047618e-05,
      "loss": 2.4153,
      "step": 7845
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3844965398311615,
      "learning_rate": 9.90780401422047e-05,
      "loss": 2.41,
      "step": 7850
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38265910744667053,
      "learning_rate": 9.907685380825825e-05,
      "loss": 2.3994,
      "step": 7855
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3885512351989746,
      "learning_rate": 9.907566671865512e-05,
      "loss": 2.429,
      "step": 7860
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4092221260070801,
      "learning_rate": 9.907447887341359e-05,
      "loss": 2.393,
      "step": 7865
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39177924394607544,
      "learning_rate": 9.907329027255194e-05,
      "loss": 2.4219,
      "step": 7870
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3880070745944977,
      "learning_rate": 9.907210091608848e-05,
      "loss": 2.4683,
      "step": 7875
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38256946206092834,
      "learning_rate": 9.907091080404152e-05,
      "loss": 2.413,
      "step": 7880
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3984758257865906,
      "learning_rate": 9.906971993642937e-05,
      "loss": 2.348,
      "step": 7885
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3896706998348236,
      "learning_rate": 9.906852831327038e-05,
      "loss": 2.3218,
      "step": 7890
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3767906725406647,
      "learning_rate": 9.906733593458289e-05,
      "loss": 2.3778,
      "step": 7895
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38337501883506775,
      "learning_rate": 9.906614280038529e-05,
      "loss": 2.4005,
      "step": 7900
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3817296326160431,
      "learning_rate": 9.90649489106959e-05,
      "loss": 2.448,
      "step": 7905
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.397694855928421,
      "learning_rate": 9.906375426553314e-05,
      "loss": 2.415,
      "step": 7910
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38217830657958984,
      "learning_rate": 9.90625588649154e-05,
      "loss": 2.3976,
      "step": 7915
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3991295397281647,
      "learning_rate": 9.906136270886107e-05,
      "loss": 2.3574,
      "step": 7920
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3935030400753021,
      "learning_rate": 9.90601657973886e-05,
      "loss": 2.454,
      "step": 7925
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3950061798095703,
      "learning_rate": 9.905896813051636e-05,
      "loss": 2.4486,
      "step": 7930
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37913620471954346,
      "learning_rate": 9.905776970826286e-05,
      "loss": 2.3663,
      "step": 7935
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3849557042121887,
      "learning_rate": 9.905657053064651e-05,
      "loss": 2.4091,
      "step": 7940
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.388558954000473,
      "learning_rate": 9.905537059768578e-05,
      "loss": 2.4393,
      "step": 7945
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37208279967308044,
      "learning_rate": 9.905416990939915e-05,
      "loss": 2.4227,
      "step": 7950
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3677532970905304,
      "learning_rate": 9.905296846580512e-05,
      "loss": 2.3472,
      "step": 7955
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38295331597328186,
      "learning_rate": 9.905176626692217e-05,
      "loss": 2.3864,
      "step": 7960
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40648430585861206,
      "learning_rate": 9.905056331276884e-05,
      "loss": 2.4342,
      "step": 7965
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38737648725509644,
      "learning_rate": 9.904935960336359e-05,
      "loss": 2.4047,
      "step": 7970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5467673540115356,
      "learning_rate": 9.904815513872503e-05,
      "loss": 2.4811,
      "step": 7975
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39807960391044617,
      "learning_rate": 9.904694991887164e-05,
      "loss": 2.3539,
      "step": 7980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4612131416797638,
      "learning_rate": 9.904574394382203e-05,
      "loss": 2.4822,
      "step": 7985
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4088277518749237,
      "learning_rate": 9.904453721359474e-05,
      "loss": 2.2927,
      "step": 7990
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38123083114624023,
      "learning_rate": 9.904332972820837e-05,
      "loss": 2.4544,
      "step": 7995
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4041150212287903,
      "learning_rate": 9.904212148768148e-05,
      "loss": 2.3577,
      "step": 8000
    }
  ],
  "logging_steps": 5,
  "max_steps": 126689,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 2.173076069626675e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}